Masaryk University
The development of the IEclusters obstruent + t/s(dh
)
[habilitation thesis]
Brno 2020 Ondřej Šefčík
2
3
Hereby I state myself the sole author of the presented thesis, I used only the quoted literature as listed below.
4
5
1 On IE obstruents, examined clusters and used methods
1.0 Introduction remarks
The purpose of the present study is to describe the development of a particular set of
consonantal clusters in the Indo-European languages and, in hindsight, to shed some light both
on its earlier phase (shared in various degrees by the various sub-branches of the IE language
family) and on the main tendencies present in the development of given Indo-European
branches.
The set of consonantal clusters we will study are two-consonantal clusters (with a few
exceptions, mentioned as such in the following text, used from necessity given by the lack of
more suitable data) formed either by an Indo-European obstruent (in the context of the present
study, an obstruent will be any IE plosive and a sibilant) in the left position and by either by *tor
*s- or *dh
- in the right position (the terms ‘rightʼ and ‘leftʼ are arbitrarily used to mark the
mutual relative position within the speech act).
The method used is the principally traditional structural analysis of clusters of our
interest, primarily those synchronically productive, secondarily those etymological (i.e.,
synchronically unanalysable and with structure revealed only through the etymological
analysis). The analysis is based on the assumption that the phonemic alternation (indifferent if
synchronic or diachronic) could reveal, in its nature, functions and relations otherwise invisible
to the pure phonemic analysis in the way of simple registration of elements and their
description; this is the reason why we use the term ‘structural analysisʼ since we are focused on
mutual relations between segments as much as on the segments themselves.
On the following lines of this chapter in we will bring forth the reconstructed set of IE
obstruents, the classification of given consonantal classes and a few methodological remarks.
1.1 Indo-European obstruents
The set of the Indo-European obstruents consists of a single voiceless sibilant *s (split in
languages affected by the Pedersen’s Law/the ruki-rule into two sibilants – in such languages,
we will deal with both sibilants independently) and a numerous set of plosives, their number
differing according to the used models.
The main differences between given models of the plosive sets used could be classified
according to the number of modal classes used and on local series; the main points of divergence
6
we will sum up on following lines to express reasons for the list of plosive phonemes we will
be using below.
1.1.1 Indo-European plosives I: the modal classes
The classification of IE obstruents according to their modality is a matter of debate both of their
number and phonemic nature. The models used could be classified thus:
i. the quaternary model, containing the voiceless-nonaspirated, voiced-nonaspirated, voiceless
aspirated and voiced aspirated classes, is, in fact, a projection of the Vedic system backwards
in time, was first proposed by Curtius (1853), and became the standard model for another
century, being used by influential grammarians, especially by Brugmann. Hirt (1927: 218–
219, 224, 240-241; Hirt 1939: 161) followed the quaternary model, though considering the
secondary origin of the voiceless aspirates (as proposed by ternary models). From later
supporters we have to mention Hiersche (1964) and especially Szemerényi (1967: 84, 88–
89; Szemerényi 1996: 54); Rasmussen (1987: 81–109 = 1998: 216–243), Elbourne (1998:
1–30; 2000: 2‒28).
ii. the ternary model has numerous variants, all having in common a denial of the IndoEuropean
origin of the voiceless-nonaspirates (since de Saussure 1892: 118), for its classical
form see especially Pedersen (1926: 48); Kuryłowicz (1927: 202‒204; Kuryłowicz 1973:
68–69); Lehmann (1952: 99). From its variants, we could mention the one assuming that the
voiced-aspirates were originally the voiced spirants (cf. Brücke 1856: 59–60; Walde 1897:
466; Prokosch 1918–1919; Prokosch 1939: 39–41; Hammerich 1967: 839–849). More
popular are various glottal models, denying the traditional values of three modal classes in
various degrees, cf. Pedersen 1951: 10–16; Andreev 1957: 7–8; Griffen 1989. Often variant
are purely glottalic models such as those by Martinet (1953: 67–68); Gamkrelidze/Ivanov
(1972: 15–18); Hopper (1973: 141–166); Kortlandt (1978b: 107–108; Korlandt 1985); Huld
(1984: 140); Collinge (1985: 259–269); Salmons (1993); Fallon (2002: 284–288, 317–318).
An interesting variant replaces the voiced non-aspirated with an implosive, cf. Haider
(1985); Kümmel (2012: 303–306); Brett Miller (2012: 95, 236‒266).
Note: The binary model, using the simple opposition of the voice was introduced by Schleicher (18611
: 136–137;
18662
: 162–163) as the first stage in the common development of the IE languages; even Schleicher assumes
the later existence of the ternary model. Erhart (1956; Erhart 1982: 39) later brings a similar model of the twostages
development with a later split of the voiced class on two modal classes.
Within the frame of this study, we will use the traditional variant of the ternary model, since it
satisfyingly fits our purposes, especially since we will deal with plosives in the neutralization
positions, where the distinction between the modal classes is subjected to various alternations.
As we will see below, the contrast between the voiceless and voiced non-aspirates is always
neutralized (as often are the voiced aspirates), and the distinction between both non-aspirated
plosives on one side and the voiced aspirate on the other side is relevant only in some contexts
of our interest for the Indo-Iranian languages (the contexts of Bartholomae’s law).
7
The phonetic properties of the reconstructed IE phonemes are always only better or
worse approximatively due to their necessarily abstract nature, this being the result of the
reconstruction, not of direct observation. As we saw above, the conventional modal classes are
variously interpreted. The terms voiceless non-aspirates, voiced non-aspirates and voiced
aspirates will be used on following lines, with approximate values given by the names (except
with the third modal class where the phonetic nature as voiced aspirate could be successfully
doubted, cf. Kümmel 2015: 293; we would prefer the values of voiced spirants, as mentioned
above, but we will use both the traditional values and frame)1
. However, Jakobson (1958: 22–
23) and Hopper (1973: 141) are wrong when arguing that two voiced plosives in a single triadic
system are impossible (cf. Kümmel 2012: 294–295), as demonstrate the typological parallel of
Madurese or Kelabit (Blust 2009: 174–175, 182; added could be probably even Bintulu).
Note: The existence of the IE voiceless-aspirates we cannot be merely rejected, but we can surely assume that if
such modal class did exist, it was not proportional to other classes (cf. Šefčík 2012; Šefčík 2016), as it is in
Indic (the Indic state is a secondary leveling of the system, as we will demonstrate below). It should be noted
that OIA, where the voiceless aspirates are a singular class (unlike in other languages – Greek voiceless
aspirates usually reflect the IE voiced-aspirates), they never enter the context we will examine below, hence
clearly demonstrating their unusual position in the whole phonemic system.
1.1.2 Indo-European plosives II: the local series
The reconstruction of the dental and labial series is not in doubt (though the status of the IE
phoneme *b is often questioned, the reconstructed existence rest of labials is accepted).
However, the number and phonetic realization of a reconstructed velar series is a matter of
debate.
To sum up, there are the following approaches to the ‘guttural questionʼ:
i. the monic model reconstructs only three series: labial, dental and (plain) velar series. This
type of model was first used by Schleicher (18611
: 136–1379; Schleicher 18662
: 162–165),
who presupposed that other velar series (the terminology used today was not used in his
days) developed due to the set of processes he even tried to list. Schleicherʼs model in the
original form was abandoned in favour of either dyadic or triadic models; however, even
later the monic model was considered as a working model for an earlier stage of
development of guttural series, i.e. used as a predecessor of later models with more series,
cf. Pedersen (1897: 192; Pedersen 1900: 292–300; Pedersen 1908: 354; 1 Pedersen 951: 3);
Ribezzo (1903; Ribezzo 1922–1923; Ribezzo 1929); Hirt (1927: 234–236; Hirt 1939: 162);
Sturtevant (1930); Specht (1944: 316–317); Safarewicz (1945: 37); Vaillant (1950: 25);
Otrębski (1963: 11–15). To the possibilities of the palatalization and labialization of velars
see Solta (1965), who adds typological parallels. The monic model is, as an earlier stage of
the development of IE guttural system, presupposed later by Markey (1980) and Szemerényi
1
Peeters (1971) prefers to reconstruct non-phonemic voice (development later in context with the later creation of
voiceless aspirates). Peeters assumes the non-occlusive character of IE *Dh
, though he does not directly state
the spirant value.
8
(1996: 149), first via the split of plain velars and labiovelars, followed (in satəm-languages)
by the phonemization of palatovelars and the subsequent delabialization of labiovelars2
.
ii. the dyadic centum-model presupposes that system of guttural series we meet in centumlanguages,
is already traceable back to Indo-European and consequently, the satəm-language
model is considered a later innovation. According to this model, the three guttural series
model is a pure phantom given by a generalization and merging of two models in a single
one, without any real existence. The first proponent of the dyadic centum model was Meillet
(1893), later Hirt (1899; also Hirt 1927: 226f; 1939); Vaillant (1950: 25), Lehmann (1952:
8; Lehmann 1993: 100–102); Sihler (1995: 151–165). Generally, the satəmization is
considered a process similar in its nature to later palatalizations of velars (in Indo-Iranian,
Slavic, Romance etc.), though the specification of its conditions is quite vague. This
satəmization lead to the split of original plain velar series into two, the palatovelars later
(af)fricativized. Numerous authors pointed out that satəm-languages form a single
innovation area (i.e. area of satəmization)3
, cf. József Schmidt (1912: 45); Sköld (1931, 56-
79); Pisani (1961); Porzig (1954: 76); W. P. Schmid (1966: 11); Shields (1981: 210–211);
Sihler (1995: 153); Schmitt-Brand (1998, 88–90). Burrows (1955: 72–73) considers
satəmization in the satəm-languages as the first stage of a general process of palatalization
(‘first palatalizationʼ), followed by a ‘second palatalizationʼ of velars (< original IE *K and
*Kṷ
), occurring independently in the satəm Balto-Slavic and Armenian (Burrows 1955: 76-
77), similarly Sihler (1995: 154–155); Sims-Williams N. (2017: 268–270). To generalize the
thesis: the original state of IE was with plain velars and labiovelar, and some of the plain
velars became secondarily palatalized (except in some positions like before r, a or after u, ū
etc.) in the future satəm-languages. With a new marker in work, the original labiovelars lost
their labiality and merged with original IE plain velar.
iii. the dyadic satəm-model mirrors the dyadic centum-model in presupposing that the threevalued
model is a purely reconstructional phantom. The difference in the idea that the satəmmodel
(i.e. the model with distinguished palatovelars and plain velars) was the original
model of IE since in this model, the labiovelar series is a secondary series, arising due to
some process of labialization of plain velars (‘centumizationʼ), resulting in the secondary
split of plain velar series into two. It is worthy of mention that this model was originally
accepted by Brugmann in the first edition of his Grundriss (Brugmann 1886), though it was
later abandoned and replaced by the triadic model. However, similar ideas were later stated
by authors like Joh. Schmidt (KZ 25, 1881: 134); Ribezzo (1903; 1922–23; 1929); Jószef
Schmidt (1912: 54) and Reichelt (1922: 40–81). Later the main proponent of this model was
Kuryłowicz, who voiced his opinion repeatedly (Kuryłowicz 1956: 356–375; Kuryłowicz
1964: 12). The mechanism of centumization was described by Szemerényi as the result of
simplification of *Ku̯ clusters (where K is any plain velar) (Szemerényi 1964: 401;
Szemerényi 1996: 145–146); in this he follows Vaillant (1950: 171–173).
iv. the dyadic equipollent model assumes that there were two marked series (i.e. the labiovelar
and the palatovelar series) and that the plain velar series, which are reconstructed for IE in
an only very limited number (note that absolutely higher number of plain series in attested
IE languages is given by the merging of one of marked guttural series with plain velar series,
not inherited from IE), were not present in reconstructed Indo-European. As a proponent of
this model, we can list Meillet (1893; Meillet 1934: 91–95), since in his model, plain velar
is just an allophone of palatovelar and hence the phoneme is palatovelar in opposition to
2
That the palatovelars must be arisen before the merging of plain velars and labiovelars is clear from the fact that
there are no palatovelars arisen from original labiovelars since both sets of gutturals are clearly and strictly
distinguished.
3
Less probably is the idea supported by Georgiev (1937: 124; 1966: 46) and Abaev (1965: 140f.), who suppose
the independent process of satəmization in given IE languages.
9
labiovelar. Steensland (1973, cf. especially 96–127) presupposes two series, marked in a
purely algebraic way as KA and KB, the first of them developing in palatovelar series in the
satəm-languages and plain velars in the centum-languages, the second of them developing
into labiovelar series in the centum-languages and plain velars in the satəm-languages; plain
velars are then a result of neutralization of one of his series. A very similar model was
developed by Kortlandt (1978a; Kortlandt 1994a: 2–3) and Woodhouse (1998; 2000), who
advances the opposition between prevelars and backvelars. Even Beekes (2011: 124–126),
who otherwise reconstructed three guttural series, seems to tend to accept the opinion that
plain velars were just positional allophones either of palatovelars or labiovelars (cf. Cavoto
2001: 51).
v. the triadic model assumes the original existence of all three velar series and was first
introduced by Bezzenberger (Bezzenberger 1890), named by von Bradke (1890) and
accepted by Brugmann (1897) (both by the second edition of his great comparative grammar
and also by his concise grammar). Since that time, this model has been used by many authors;
for example by Szemerényi (1990: 71). This model became a widely accepted standard,
having the advantage of covering all possible guttural series. However, it could still contain
a reconstructional error since the possibility of reconstructing three guttural series does not
necessarily mean that the system with three series ever existed at the same time (cf. Sihler
1995: 154). This traditional model is accepted by Allen (1978); Tischler (1990: 93–94);
Kapović (2017: 14–15, 21–28). The fact that attested Indo-European languages have only
two guttural series4
, led Burrows (1955: 75–76), Kuryłowicz (1956: 356; Kuryłowicz 1973:
64); Meillet (1893: 278); Lehmann (1952: 100), Bernabé Pajares (1971: 84) and others to
state that the triadic model is therefore impossible, but since there are external parallels in
the Northern Caucasian and other languages, it is hardly possible to say that the system of
three guttural series is impossible from the typological point of view, and we will
demonstrate that some version of it even exists in Indo-European language. A slightly
modified statement is that by Cavoto (2001: 51), who namely considers three phonetically
distinct guttural series, but only two phonemic series for IE (cf. Beekes 2011: 126, too). A
very similar statement was already earlier made by Safarewicz (1945: 37). Speaking about
the triadic system, we have to mention the variants given by Huld (1986; 1997) and by
Kümmel (2007: 310-327), working with the triad with traditional labiovelars, but where
traditional palatalovelars are plain velars (hence there is no depalatalization in the centumlanguages,
but there is a palatalization in the satəm-languages), and traditional plain velars
are uvulars (de-uvularized in both branches) (cf. also Huld 1986: 144–147; Huld 1997;
Woodhouse 1998). Lipp (2009a, especially: 5–19) postulates two stages system, similar to
that by Szemerényi (1996: 60–61), with an original distinction between plain- and
labiovelars, but with a later split of palatovelars from the plain velars in languages which
later became the satəm-languages; the satəmization as a first step in the general development
of palatalization (cf. two stages of palatalization in Burrows 1955: 72–73).
The aforementioned lack of the whole triadic system in attested IE languages led numerous
authors to raise grave objections against the triadic system (cf. Meillet 1893: 278; Lehmann
1952: 100; Burrows 1955: 75; Kuryłowicz 1956: 356; Kuryłowicz 1973: 64 as examples).
However the traces of a triadic system were stated to be found in Albanian, since Pre-Albanian
labiovelars *kṷ
and *gṷ
(< IE *gṷ
, *gṷh
) were later palatalized to Albanian s, z before the palatal
4
The question of preserving at least of traces of three guttural series in IE we will deal with below.
10
vowel but merged with plain velars in other positions and hence giving, as plain velars, *k and
*g (< IE *g, *gh
); cf. Alb. sjell ‘to bringʼ < IE *√kṷ
elH1- but Alb. pjek ‘to bakeʼ < IE *√pekṷ
-;
Alb. zorrë ‘gutʼ < IE *√ghṷ
ērn- but Alb. djeg ‘to burnʼ < IE *√dh
egṷh
-. Moreover, since Albanian
is a satəm-language (IE palatovelar *ḱ developed into Alb. th, the IE palatovelars *ǵ, ǵh
merged
into Albanian dh), Albanian preserved traces of the original triadic system. The first proponent
of this theory was already Pedersen (1900a), similarly Jokl (1937), Huld (1984: 144; Huld
1997); Rusakov (2017: 569–571); for the opposing points of view cf. Ölberg (1976), Kortlandt
(1980a), Orël (2000: 66–74).
A similar development was stated for Armenian, in which, again, the IE voiceless
labiovelar5
was in some cases palatalized before the merging of labiovelars and plain velars
(cf. Stempel 1994; Kortlandt 1980; Kümmel 2007: 311; Olsen 2017: 426–428; for data see at
least Schmitt 2007: 62–65): IE *√ḱḗrdi- > Arm. sirt ‘heartʼ; IE *√kér-ō > Arm. k̒erem ‘I scratchʼ;
IE *√kṷ
et(ṷ)ores > Arm. č ̒ork ‘fourʼ. A contrary view that the distinction between original
labiovelars and plain velars is secondary and accidental is held by Kortlandt (1975), Beekes
(2003: 176–179) and Martirosyan (2010: 711), who consider the palatalization as regular for
both original labiovelars and plain velars, the regularity of this process later being disrupted by
the excessive analogy.
Note: A model of three velars series was proposed for Luvo-Lycian by Melchert (1987: 182–204; Melchert 1994:
251–256)6
but was later distinctly decreased by Melchert himself (Melchert 2012: 206–218), not being
unconditioned, but having conditioned palatalization of original palatovelars (but not plain velars) before front
vowels, i̯ and ṷ (the phenomenon seems to be proved for voiceless palatovelars, but not fully affirmed for
voiced palatovelars); cf. Cuneiform Luv. ziyari, Lyc. sijẽni ‘liesʼ < IE *ḱē̆sāi̯e- vs Cuneiform Luv. kišā(i)- ‘to
combʼ < IE *kē̆s- vs Cuneiform Luv. kui-, Lyc. ti ‘whoʼ < IE *kṷ
i-. In contrast, the traces of three guttural series
in Phrygian are assumed by Woodhouse (2005) as the result of the further development of the original IE
dyadic two-valued model, based originally on front velar and back velar (in this respect he is following
Steensland 1973: 96–107; Kortlandt 1978a: 237; see above).
Other traces of the original existence of the labiovelars in the future satəm-languages are
considered to be found in original reduced grades such as OIA guru- ‘heavyʼ and gūrta‘welcomeʼ,
similarly in OCS gъnati (cf. ženǫ ‘propel, drive, chaseʼ) from IE *√gṷh
en- (cf.
Burrows 1957; Pisani 1961; Mayrhofer 1986: 104–105).
The striking phenomenon of the satəm-languages, especially present in Balto-Slavic,
but traceable in Indo-Iranian, too, is the existence of parallel roots with plain velar or original
5
Surprisingly, this change did not affect IE *gṷ
.
6
Melchert (1987: 204) mentions an independent statement made by Warren Cowgill in his unpublished manuscript
from the early seventies, though based on less evidence (which was probably the reason why the paper was
never published). For another independent statement on the preservation of the IE triadic system in Luwian,
see Morpurgo Davies/Hawkins (1988: 169‒182).
11
palatovelar, not caused by any noticeable first-hand alternation-trigger, cf. OIA √klam- ‘be tiredʼ
vs √ślam- ‘be tiredʼ; OIA √ruc- ‘shineʼ vs ruśant- ‘be brightʼ; OIA √śru- ‘hearʼ, OCS slyšati
‘hearʼ vs Lith. klausyti ‘hearʼ; Lith. pekus ‘domestic animalʼ vs OIA paśu- ‘cattleʼ; OIA śvaśura‘father-in-lawʼ
vs OCS svekrъ ‘father-in-lawʼ; OIA aśman- ‘stoneʼ, Lith. ašmuõ, Latv. asmens
‘sharpnessʼ vs Lith. akmuõ, Latv. akmens, OCS kamy ‘stoneʼ etc. (cf. Hirt 1927: 238–241;
Vaillant 1950: 171–173; Burrows 1955: 75–76; Steensland 1972: 102–104; Čekman 1974;
Allen 1978: 103; Shields 1981: 210–211; Mayrhofer 1986: 105–106; Szemerényi 1996: 146;
Lipp 2009a: 5–98; Kapović 2017: 26–27; Young 2017: 497).
To the phonemic status of given velars: the labiovelars were determined as such7
at least
since Zupitza (1896: 1–2). Sköld doubted the monophonemic value of labiovelars (Sköld 1924:
128; similarly Whatmough 1937: 52–56) and preferred the double articulated labial-velar
plosives (i.e. as Ewe ͡kp, ͡gb), Salmons and Smith (2005) returned to the monophonemic status
of (voiced) labiovelars, independently confirming it. Palatovelars are not attested as such in a
single satəm-language (here we usually meet sibilants or affricates); their phonetic value is then
a reconstruction8
.
We will use the two-stages triadic model based on Lipp (2009a: 5–19), assuming the older
distinction between the plain velar (unmarked) and the labiovelar (marked series). This model
was replaced in the area of the later satəm-languages by the classical triadic model (this explains
why there are traces of the old labiovelar vs plain velar distinction in the satəm-languages, but
not the traces of the palatovelars in the centum-languages). However, it is highly probable that
the ‘palatovelarsʼ were affricates at the same time, not true palatovelar plosives.9
With such
massive remodelling, the primordial distinction between old labiovelars and plain velars was
neutralized in all positions, resulting in a merging of both guttural series (similar processes are
known from the centum-languages, e.g. k-Celtic, Tocharian, later Germanic languages).
7
Similar sounds are present in North-Caucasian, the Suto-Chuana subgroup of Bantu languages and Salish
languages.
8
Again, palatovelars as such are attested in Ubykh and other North-Caucasian languages.
9
However, the classical triadic system has typological parallels in some non-Indo-European languages, especially
in North-Caucasian; we should mention especially recently extinct Ubykh or in Abaza and Abkhaz. Edeľman
(1973: 540–546) demonstrated that the guttural system based on three series (plain, palatalized, labialized) is
attested in Yazghulami, an East Iranian dialect in the North Pamir, and it should be noted that this system is
not a directly inheritance from IE, but secondarily created. It is worthy of mention that Yazghulani has not only
three guttural series, but a plain uvular and labialized uvular series, too. Ossetian has a pair of guttural and
uvular series in opposition plain vs labialized, too, again independently developed, not inherited from IE. That
the triadic system of related series could exist, is demonstrated in OIA in Šefčík 2012.
12
However, for simplicity, we will use the traditional marking for the palatovelars, as we do for
other reconstructed Indo-European phonemes.
1.1.3 Indo-European obstruents: the list
Considering the issues as mentioned above, the set of Indo-European obstruents we work with
is then the seemingly conventional set (including the ruki-sibilant *š), we can put as follows:
plosives sibilant
labiovelars: ku̯
gu̯
gu̯ h
plain velars: k g gh
palatovelars: ḱ ǵ ǵh
(š)
dentals: t d dh
s
labials: p b bh
This model does not probably reflect any actual state in the development of the Common IndoEuropean
but serves as a panchronic working model (as we have already mentioned above, the
guttural triad was limited to future satəm-languages and palatovelars were probably not
phonetic palatovelar plosives, and we have to repeat that the modal values are probably
simplified, etc.). For panchronic reasons, the ruki-palatal š is included in the list, though never
reconstructable outside the satəm-area and even so it’s existence is questionable for PreAlbanian
and Pre-Armenian).
1.2 Segments and contexts
The focus of our study is clusters of IE obstruents in three different contexts.
The obstruent segments (i.e., the left part) of clusters could be split into the following blocks:
i. the central block10
containing IE dentals (phonetically could also be realized as alveovelars
in some of the languages) and palatovelars (in the satəm-languages only);
ii. the peripheral block11
containing IE labials, plain velars and labiovelar series (in the
centum-languages only);
iii. the sibilant block12
containing the old IE sibilant *s, the most satəm-languages have a
secondary sibilant *š, arising due to Pedersen’s Law/the ruki-law.
Similarly, the context (i.e., the right part) of clusters could be classified as:
i. the context of *t-, causing the devoicing and deaspiration of the left obstruent in all IndoEuropean
languages; the exception is languages with Bartholomae’s Law;
10
In terms of Jakobson / Halle (1956: 31), we can term it an acute block.
11
Using the terminology of Jakobson/Halle again, the block could be termed a grave block.
12
This block is defined, as we see, not by its localization, but its sonority.
13
ii. the context of *dh
-; causing the voicing and deaspiration of the left obstruent in languages
where the voicing is preserved (here especially in Indo-European);
iii. the context of *s-;13
also causing the devoicing and deaspiration of the left obstruent; again;
these clusters could be affected by Bartholomae’s Law, the feature well preserved in Iranian
only.
1.3 On trajectories of the development in general
The development from the Indo-European stage to a given language stage is the transformation
between two (or more) states-of-arts; each state is a set of elements.
Within this analysis, we generally distinguish at least three stages of the development
from the input Indo-European into a given output language:
i. the (Late) Indo-European stage, the initial state, arrived at by reconstruction;
ii. the intermediate stage, the transitional stage (or, more appropriately, a set of sub-stages)
between Indo-European and an output language;
iii. the given language stage, the actually attested language14
.
The Indo-European-stage and the given-language-stage are hence both termini of the whole
complex of the development processes, and both serve either as an input or respectively output
of the complex transformation of the reconstructed Indo-European phonemic system into a
given phonemic system. This transformation, concerning the processes involved, could be
described more appropriately as a set of minor transformations, not as a single giant
transformation of its own.
Note that the input (i.e., the Indo-European-stage) is a result of reconstruction, but the
output is actual language matter, recorded as a hard fact. Both are fixed sets objects, though
results of a different approach (output is simply observed, the input is reconstructed).
The intermediate-stage (or stages) is, on the contrary, a set of trajectories between the
input and the output and hence a matter of more or less analytic nature.
It is a paradox that often the reconstructed stage (the input) can be quite securely
reconstructed without a more profound analysis, e.g. the merging of both voiced modal classes
of plosives in Iranian, Balto-Slavic, etc., since it can be based on a simple observation.
The intermediate stage is based both on a comparative reconstruction and an internal
reconstruction and in this very aspect the intermediate-stage is a kind of black box since we do
13
In some context realized as *š-; the distribution, in this case, is affected by the left phoneme in the cluster, i.e.,
by the Pedersen’s Law/ruki-rule; securely attested in Indo-Iranian and Balto-Slavic);
14
At this moment, we leave aside the often insecure knowledge of the phonetic realization (cf. the phonetic
nature of the ‘tau gallicumʼ, the value of Hittite plosives, etc.).
14
not see immediately what is inside, but we project the possible trajectories in it and choose that
which fits most the known data both of the input and the output.
INPUT
→
TRANSFORMATIONS
→
OUTPUT
Indo-European intermediate
stage(s)
a given language
(given by
reconstruction)
(‘the black boxʼ; the
modelled trajectories)
(attested language
facts)
When speaking about the trajectory from the input to the output, there are often multiple ways
of plotting the trajectory, i.e., there could be competing scenarios, though leading from the same
income towards the same outcome, as we will see on the examples of the development of
various clusters below (a typical case is the development of the peripheral clusters + t/s- in Pre-
Slavic).
Note: On the following lines, the terms ‘majorʼ and ‘minorʼ developments will be sometimes used. A minor
development is a development which in its results differs from the expected lautgesetzlich development
according to a given known sound law but falls with a possible range of outputs, though the precise reasons
for this special development are not clear (otherwise they would fit within a defined sound law).
1.4 On language material and its analysis
The focus of the present analysis is on clusters of Indo-European obstruents (in a sense
described above) + *t/s- clusters. The reason we have picked these clusters is obvious and
apparent: formations of these two types are very numerous and well-attested, and the
observation of their function could be applied, in the basic features, to all other clusters formed
by voiceless obstruents.
The clusters of the same obstruents with *dh
- were used only with languages with
extensive and productive use of this kind of formation, namely in Indo-Iranian (here especially
since the comparison of the dh
-context clusters are important due to Bartholomae’s Law and
subsequently for the t-context clusters), Baltic (more to illustrate the situation outside of IndoIranian)
and Greek (besides the purely illustrative use, it was important to demonstrate the *dh
context
clusters after the phonemic revaluating of voice and subsequent remodelling of whole
*dh
-context clusters in Greek). Again, the function of the *dh
-context clusters can be applied
and generalized to other IE voiced aspirated contexts, namely to IE *bh
-context.
The reader will notice that a natural complement to the present study could be the
‘thornʼ-question, which was deliberately and willingly omitted, since the sheer number of
examples would increase the number of pages required – the author of the present lines could
only recommend comprehensive study by Lipp (2009b – covering 350 print pages!).
15
The work on this analysis was made possible due to preceding important works, quoted in the
lines below. Especially worthy of mention are authors of LIV (the second edition of it was
used), the more so since our focus was mainly on the productive verbal formations, NIL (used
when verbal data were not sufficient) and of various quoted etymological dictionaries of given
languages (often of the Leiden origin) but also with the use of the classical works like Pokorny
(IEW). Of valuable help were lists of OIA verbal forms by Whitney (1885) and MacDonell
(1916), the list of the Avestan verbal forms by Kellens (1984) and the list of Tocharian verbal
forms by Malzahn (2010). From monographs at least few of many should be noted in alphabetic
order: Görtzen (1998), Hill (2003), Kümmel (2007) and two-part monumental volume by Lipp
(2009a; 2009b). To the quoted literature, we should attribute all the positive sides of the present
works; all mistakes are fully the author’s alone.
The analysis is segmented in accordance with given branches of the IE languages, represented
either by a single language (Old Church Slavonic for the whole Slavic family, Gothic for the
whole Germanic) or more languages (OIA, Avestan, Persian and Nūrīstānī represent the IndoIranian
branch; Italic languages are represented by both by Latin and Sabellian languages) if
the data they bring are substantially and significantly different (the aforementioned Italic
languages differ in their development of the peripheral series, Indo-Iranian languages
substantially differ in the development of all local series). The ordering of the given branches
is not accidental – first, we will follow the satəm-languages, followed by the centum-languages
and finally by two peripheral (technically centum) languages with a remarkable difference in
the development of the clusters of dental plosive + *t/s (Hittite for Anatolian languages, and
Tocharian). Each chapter on a given branch/language is then written as an independent study,
both for convenience of the reader and for the simplicity of the present treatise as a whole.
17
2 The development of the two-obstruent clusters from Indo-European into
Indo-Iranian
2.0 Indo-Iranian languages
The Indo-Iranian (Aryan) branch of Indo-European languages is one of the most influential sources of the whole
discipline, due to its huge documentation.
Old Indo-Aryan is known from extensive Vedic texts, the most important source being R̥ g-veda, assumed
to be compiled by the middle of the second millennium BC, this oldest stratum is followed by younger documents,
notably brāhmaṇas and post-Vedic works. The Old Indo-Aryan language was replaced by its Middle and New
Indo-Aryan heirs, forming a natural laboratory of a language continuum (cf. Bloch 11–31; Masica 1991: 32–60;
Cadorna 2017). The influence of Vedic (or precisely, of Classical Sanskrit) on the development (and results) of
Indo-European comparative linguistics was enormous, especially in the 19th
century.
There are two relatively well Old Iranian languages: Avestan, orally composed in the 2nd
and 1st
millennia
BC (written down in the Sassanian Period ca 600 AD), the Old and Young Avestan are not different stages of a
single language but are distinguished even spacially. The second language is Old Persian, attested in contemporary
documents in cuneiform script, dated since the rule of Darius I until Alexander (cf. Skjærvo 2017: 472–474).
Nūristānī language are, on the other hand, a relatively small group of languages with an unclear relation
to both main branches, attested from the 20th
century AD and relatively worse both documented and examined.
2.1 On the reconstruction of the trajectory of the Indo-Iranian development
The development of two-obstruent clusters from Indo-European into given Indo-Iranian
languages should be understood as a complex of sequences of at least three stages, which can
be demonstrated by the development of Old Indo-Aryan: there is the IE stage, which is an input
to the process, the Old Indo-Aryan stage, which is the output of the process and the intermediate
transitory stage (or stages), which we can term the Indo-Iranian stage. We can similarly model
the development of Avestan and Old Persian, differing in their respective outputs and later
developments, specific for the Iranian branch of the Indo-European languages. Old Indo-Aryan,
Avestan and Old Persian stages are mutually equivalent, being attested roughly at the same
time. Nūristānī, on the other hand, which probably forms the third branch of Indo-Iranian
languages, according to the prevailing opinion (but see note below), is attested since the early
20th
century, which creates remarkable problems because of the relationship of its data to the
old Indo-Iranian languages.
Note: The position of Nūristānī within the Indo-Iranian family was considered the third independent branch
(Mayrhoffer 1951: 15; Morgenstierne 1961; Fussman 1972: 391; Strand 1973; Nelson 1986: 104–116; Kausen
2012: 661–667), closer to Iranian (Mayrhofer 1984; Mayrhoffer 1997: 107–108; Lipp 2009a: 156–157) or to
Indo-Aryan languages (Bloch 1965: 54; Buddruss 1977: 33; Degener 2002; Blažek/Hegedüs 2012; Weba
2016), for the most actual discussions on the theme also see Cardona/Jain (2003: 22–25).
In the following lines, our analysis will be primarily focused on the development of Old Indic
(represented by Vedic) and of Old Iranian languages (i.e., Avestan and Old Persian), since they
are fully equivalent in the relative chronology, though Old Persian data are far worse attested
than those of Avestan and Avestan worse than those of Vedic. Nūristānī data will serve more
18
as a commentary and appendices to the main course of analysis, quoted when possible, visually
marked from the bulk of Old Indo-Iranian languages by the size of the font used.
Note that the output stages contain the ‘hard dataʼ and these data are almost empirically
approached, but both IE and II stages are reconstructed models15
, reconstructed via the
comparison either of IE or Indo-Iranian languages, the approach being more ‘algebraicʼ than
‘empiricalʼ, ‘modelsʼ than ‘factsʼ in its very nature. The changes between given stages are
expressed through trajectories; in some cases, parallel trajectories could be modelled, as we will
demonstrate below.
Our primary focus is on two-obstruent clusters, formed either by any left-standing
plosive or a sibilant in contexts of the right standing *t-, *dh
- or *s/š. The main point of our
interest will be clusters formed on the morphemic boundary between two morphs, i.e., on the
clusters of obstruents synchronically produced; the ‘etymologicalʼ clusters (not resulting from
synchronic alternations) will be used only if the ‘alternatingʼ clusters will not be at hand. The
alternating clusters have precedence, since being a very ‘living fleshʼ of a given language, such
alternations show active structures and can serve as a cornerstone for an internal reconstruction.
The verbal systems offer a huge thesaurus of such productive alternations; nominal morphology
will be used only occasionally.
Note: The two-obstruent clusters could either be preceded or followed by a non-obstruent consonant but since
Indo-Iranian non-obstruents (liquids, sonants, nasals16
) are always peripheries of the consonantal clusters, if
the obstruents are present, such clusters will be accepted as well as obstruent-only clusters, especially since
non-obstruents do not generally trigger any alternations of obstruents in observed languages (the exceptions
will be pointed out in case of need).
2.2 The development of two-obstruent clusters in Vedic
In the forming of two-obstruent clusters in Vedic, there are the following general tendencies
and rules (usually shared with other old Indo-Iranian languages as well):
i. voiceless and voiced non-aspirated plosives have the same developments before voiceless
obstruents;
ii. voiced aspirated plosives, when concatenated by a voiceless non-aspirate plosive, form
clusters of DDh
; this process is known as Bartholomae’s law;
iii. the same result is given by the concatenation of two voiced aspirated plosives or by a
concatenation of a voiceless or voiced non-aspirate with a voiced aspirate plosive17
;
iv. in contrast with above, all plosives are neutralized on a voiceless plosive before s(Bartholomae’s
Law is not applied).
15
However, even ‘hard” Vedic, Avesta, Old Persian or Nūristānī data form some kind of a mental structure; they
form a model sui generis, though based on empirical evidence.
16
For good reasons, we deal with nasals as with phonemic non-obstruents, though phonetically, they are obstruents
as well.
17
But we will see below, there are some interesting exceptions of this generally assumed rule.
19
Beside those, we have to remark that:
v. the OIA voiceless aspirates never form a cluster with the following obstruent in the same
way as their counterparts. The OIA voiceless aspirated plosives are always separated by i
from the following obstruent; hence, all examples of voiceless aspirates will be absent from
our overview;
vi. for both for diachronic and synchronic reasons we have to distinguish two series: the original
Indo-European palatovelars (and though this series is in no way palatalovelar in its phonetic
realization, we will stick for simplicity to this name): ś, j1, h1 and true palatals: c, j2, h2 arising
from the later palatalization from merged original IE plain velars and labiovelars. The
synchronic reason to distinguish both series is as follows: original palatovelars are realized
before plosives in OIA as cerebrals (or as zero), as we will see below, the palatals are
neutralized to plain velars (in contrast with that both series are neutralized on plain velar
before the sibilant).
vii. Since the number of examples on the alternations of left standing cerebrals is insignificant,
the cerebrals (diachronically moreover of the late origin) are in generally omitted, except -ṣ,
which is generally the result of Pedersen’s Law and in its diachronic aspect definitely at least
of Indo-Iranian origin. Some notes on cerebrals are included as a special note to section 2.3
dealing with dentals.
2.2.1 Clusters of labial + t/dh
/s
The clusters with the labial obstruent on the left form following patterns:
P + t = OIA pt:
ppp. āptá-, inf. ā́ptum (√āp- ‘obtainʼ; cf. pr. āpnóti; < IE *√H2ep-; cf. Hitt. ēpzi, appanzi
‘grabʼ, OL. apiō ‘fastenʼ; cf. Whitney 1885: 6; MacDonell 1916: 371; Pokorny IEW:
50–51; EWAi I: 167; LIV2
: 237; NIL: 311–317);
ao. táptam, ppp. taptá- (√tap- ‘heatʼ; cf. pr. tápati; < IE *√tep-; cf. L. tepeō ‘be warmʼ,
OCS teplostъ ‘heatʼ; cf. Whitney 1885: 61; MacDonell 1916: 386; Pokorny IEW:
1169–1170; EWAi I: 623–624; LIV2
: 629–630; NIL: 698–670);
ppp. tṛptá- (√tṛp- ‘be pleasedʼ; cf. pr. tṛpṇóti; < IE *√terp-; cf. Lith. tarpstù ‘satisfyʼ; cf.
Whitney 1885: 65–66; MacDonell 1916: 386; Pokorny IEW: 1077–1078; EWAi I:
634–635; LIV2
: 636);
pr. sváptu, ppp. suptá- (√svap- ‘sleepʼ; cf. pr. svápati; < IE *√su̯ep-; cf. L. sopiō, OCS
sъpljǫ ‘sleepʼ; cf. Whitney 1885: 201; MacDonell 1916: 432; Pokorny IEW: 1048–
1049; EWAi II: 791; LIV2
: 612; NIL: 675–680);
num. saptá- ‘sevenʼ (< IE *√septm̥-; cf. Gr. ἑπτά, L. septem ‘sevenʼ; cf. Pokorny IEW:
1048–1049; Emmerick 1992a: 169–170, 181–182; Blažek 1999: 246; EWai II: 700);
bh
+ t = OIA bdh
:
ppp. ubdhá- (√ubh- ‘stickʼ; cf. pr. áumbhat; < IE *√u̯ebh
-; cf. Hitt. wepta, OHG weban
‘weaveʼ; cf. Whitney 1885: 13; MacDonell 1916: 373; Pokorny IEW: 1114; EWAi II:
506; LIV2
: 658);
ppp. rabdhá- (√rambh- ‘graspʼ; cf. pr. rábhate; < IE *√lembh
-; cf. Gr. λάφῡρα ‘spoils of
warʼ; cf. Whitney 1885:136; MacDonell 1916: 411; Pokorny IEW: 652; EWAi II: 434–
435; LIV2
: 411–412); etymologically the same root as the following one:
20
ppp. labdhá-, gd. labdhvā́ (√labh- ‘takeʼ; cf. pr. lábhate; < IE *√lembh
-; cf. Gr. λάφῡρα
‘spoils of warʼ; cf. Whitney 1885: 145–146; MacDonell 1916: 414; Pokorny IEW: 652;
EWAi II: 434–435; LIV2
: 411–412);
ppp. dabdhá- (√dabh- ‘harmʼ; cf. pr. dábhati; < IE *√dh
ebh
-; cf. Hitt. tepnuzzi ‘downsizeʼ,
Lith. dóbiu ‘invalidateʼ; cf. Whitney 1885: 70; MacDonell 1916: 388; Pokorny IEW:
240; EWAi II: 694–696; LIV2
: 132–133; NIL: 85–86);
P + dh
= OIA *bdh
:
not attested
bh
+ dh
= OIA *bdh
:
not attested
P + s = OIA ps:
ft. āpsyáti, ds. ī́psati (√āp- ‘obtainʼ; cf. pr. āpnóti; < IE *√H2ep-; cf. Hitt. ēpzi, appanzi
‘grabʼ, OL apiō ‘fastenʼ; cf. Whitney 1885: 6; MacDonell 1916: 371; Pokorny IEW:
50–51; EWAi I: 167; LIV2
: 237; NIL 311–317);
ao. átapsīt (√tap- ‘heatʼ; cf. pr. tápati; < IE *√tep-; cf. L. tepeō ‘be warmʼ, OCS teplostъ
‘heatʼ; cf. Whitney 1885: 61; MacDonell 1916: 386; Pokorny IEW: 1169–1170; EWAi
I: 623–624; LIV2
: 629–630; NIL: 698–670);
co. átarpsyat B, ds. títṛpsati (√tṛp- ‘be pleasedʼ; cf. pr. tṛpṇóti; < IE *√terp-; cf. Lith.
tarpstù ‘satisfyʼ; cf. Whitney 1885: 65–66; MacDonell 1916: 386; Pokorny IEW:
1077–1078; EWAi I: 634–635; LIV2
: 636);
ft. svapsyáti B (√svap- ‘sleepʼ; cf. pr. svápati; < IE *√su̯ep-; cf. L. sopiō, OCS sъpljǫ
‘sleepʼ; cf. Whitney 1885: 201; MacDonell 1916: 432; Pokorny IEW: 1048–1049;
EWAi II: 791; LIV2
: 612; NIL:675–680);
bh
+ s = OIA ps:
ps. rípsate B (√rambh- ‘graspʼ; cf. pr. rábhate; < IE *√lembh
-; cf. Gr. λάφῡρα ‘spoils of
warʼ; cf. Whitney 1885:136; MacDonell 1916: 411; Pokorny IEW: 652; EWAi II: 434–
435; LIV2
: 411–412); etymologically the same root as the following one
ds. álapsata, ft. lapsyáti, ds. lípsate AV, ps. lipsyáte B (√labh- ‘takeʼ; < IE *√lembh
-; cf.
Gr. λάφῡρα ‘spoils of warʼ; cf. Whitney 1885: 145–146; MacDonell 1916: 414;
Pokorny IEW: 652; EWAi II: 434–435; LIV2
: 411–412);
ds. dípsati (√dabh- ‘harmʼ; cf. pr. dábhati; < IE *√dh
ebh
-; cf. Hitt. tepnuzzi ‘downsizeʼ,
Lith. dóbiu ‘invalidateʼ; cf. Whitney 1885: 70; MacDonell 1916: 388; Pokorny IEW:
240; EWAi II: 694–696; LIV2
: 132–133; NIL: 85–86);
The bh
t-clusters are subjected to Bartholomae’s law (but not bh
s-clusters), as are all clusters
resulting from original voiced aspirated plosive + t. There are no secure examples of clusters
from p/b + dh
and bh
+ dh
, though we can assume, due to analogy with other series (see below),
for both as *dbh
, i.e., the same outcome as for clusters of bh
t.
21
2.2.2 Clusters of velar palatal + t/dh
/s
The true velars are generally scarce at the end of roots, and the (secondary) palatals are products
of the palatalization of original IE plain velars and labiovelars. Both series are in privative
opposition to the related series in the sense of Trubetzkoy, and palatals neutralize to plain velars
before obstruents, hence, both series are treated as a single one.
The clusters with the velar/palatal obstruent on the left form the following patterns:
K(u̯ )
+ t = OIA kt:
ao. śaktám, inf. śaktave (√śak- ‘be ableʼ: cf. ao. śákat; < IE *√ḱek(u̯ )
-; cf. OIr cecht ‘mightʼ;
cf. Whitney 1885: 169; MacDonell 1916: 422; Pokorny IEW: 522; EWAi II: 600–601;
LIV2
: 322);
pf. mumóktu, ppp. muktá- (√muc- ‘releaseʼ; cf. ao. ámoci; < IE *√meu̯ k-; cf. L. ē-mungō
‘blow outʼ, Lith. munkù ‘get looseʼ; cf. Whitney 1885: 122; MacDonell 1916: 406;
Pokorny IEW: 744; EWAi II: 382–383; LIV2
: 443–444);
pr. pṛṅkté, impf. pipṛktá (√pṛc- ‘mixʼ; cf. pr. pṛňcáte; < IE *√perk-; cf. L. parcō ‘spareʼ;
cf. Whitney 1885: 101; MacDonell 1916: 398–399; Pokorny IEW: 820; EWAi II: 96;
LIV2
: 476);
inf. páktave, gd. paktvā́ , nom. paktar- (√pac- ‘cookʼ; cf. pr. pácati; < IE *√peku̯
-; cf. L.
coquō, OCS pekǫ ‘cookʼ; cf. Whitney 1885: 92–93; MacDonell 1916: 396; Pokorny
IEW: 798; EWAi II: 64; LIV2
: 468; NIL: 548–552);
pr. vívakti, ppp. uktá- (√vac- ‘speakʼ; cf. ao. ávāci; < IE *√u̯ eku̯
-; cf. Gr. εἶπον ‘sayʼ; cf.
Whitney 1885: 151; MacDonell 1916: 415; Pokorny IEW: 1135–1136; EWAi II: 489–
491; LIV2
: 673–674);
pr. niniktá, ppp. niktá-, int. nenikté (√nij- ‘washʼ; cf. ao. ánijam; < IE *√ne gu̯
-; cf. Gr.
νίζω ‘washʼ, OIr. -nenaig ‘washʼ; cf. Whitney 1885: 90; MacDonell 1916: 395;
Pokorny IEW: 761; EWAi II: 54; LIV2
: 450; NIL: 519–520);
ppp. tyaktá- B (√tyaj- ‘forsakeʼ; cf. pf. tityā́ja; < IE *√t egu̯
-; cf. Gr. σέβομαι ‘feel
awe/shameʼ; cf. Whitney 1885: 66; MacDonell 1916: 387; Pokorny IEW: 1086; EWAi
I: 673–674; LIV2
: 643; NIL: 660–661);
pr. yunákti, yunákta, ao. yukta (√yuj- ‘joinʼ; cf. pr. yuňjánti; < IE *√ eu̯ g-; cf. L. iungō
‘harnesssʼ, OCS igo ‘yokeʼ; cf. Whitney 1885: 132–133; MacDonell 1916: 410;
Pokorny IEW: 508–510; EWAi II: 417–418; LIV2
: 316; NIL: 397–404);
ao. viktá, ps. viktá- (√vij- ‘trembleʼ; cf. pr. vijánte; < IE *√u̯ e g-; cf. Gr. εἴκω, OE wīcan
‘give wayʼ; cf. Whitney 1885: 159; MacDonell 1916: 418; Pokorny IEW: 1130–1131;
EWAi II: 577–578; LIV2
: 667–668);
num. pakthá- ‘fifthʼ, pa(ṅ)ktí- ‘set of fiveʼ (< IE *√penku̯
-t-; cf. L. quīnctus, OCS pȩtъ
‘fifthʼ; cf. Pokorny IEW: 1048–1049; Emmerick 1992a: 168–169, 180–181; Blažek
1999: 246; EWai II: 65–66);
g(u̯ )h
+ t = OIA gdh
:
prc. dhaktám (√dagh- ‘reach toʼ; cf. pr. daghnuyā́t B; < IE *√dh
eu̯ gu̯ h
H2-; cf. Gr. Hom.
φθάνω ‘come firstʼ; cf. Whitney 1885: 69; MacDonell 1916: 388; Pokorny IEW: 250;
EWAi I: 691; LIV2
: 134–135);
ppp. dagdhá- (√dah- ‘burnʼ; cf. pr. dáhati; < IE *√dh
egu̯ h
-; cf. Lith. degù ‘burnʼ; cf.
Whitney 1885: 71; MacDonell 1916: 388–389; Pokorny IEW: 240–241; EWAi I: 712–
713; LIV2
: 133–134);
22
pr. dógdhi, dugdhé, ppp. dugdhá- (√duh- ‘milkʼ; cf. pr. duhánti; < IE *√dh
eu̯ gh
-; cf. Gr.
τεύχω ‘make readyʼ, OE ge-dȳgan ‘overcomeʼ; cf. Whitney 1885: 76; MacDonell
1916: 390; Pokorny IEW: 271; EWAi I: 747–748; LIV2
: 148–149);
K(u̯ )
+ dh
= OIA gdh
:
ao. śagdhí (√śak- ‘be ableʼ: cf. ao. śákat; < IE *√ḱek(u̯ )
-; cf. OIr cecht ‘mightʼ; cf. Whitney
1885: 169; MacDonell 1916: 422; Pokorny IEW: 522; EWAi II: 600–601; LIV2
: 322);
pr. pípṛgdhí, pṛṅdhí (= pṛṅgdhí) (√pṛc- ‘mixʼ; cf. pr. pṛňcáte; < IE *√perk-; cf. L. parcō
‘spareʼ; cf. Whitney 1885: 101; MacDonell 1916: 398–399; Pokorny IEW: 820; EWAi
II: 96; LIV2
: 476);
pf. mumugdhí, ao. ámugdhvam (√muc- ‘releaseʼ; cf. ao. ámoci; < IE *√meu̯ k-; cf. L. ēmungō
‘blow outʼ, Lith. munkù ‘get looseʼ; cf. Whitney 1885: 122; MacDonell 1916:
406; Pokorny IEW: 744; EWAi II: 382–383; LIV2
: 443–444);
pf. tityagdhí (√tyaj- ‘forsakeʼ; cf. pf. tityā́ja; < IE *√t egu̯
-; cf. Gr. σέβομαι ‘feel
awe/shameʼ; cf. Whitney 1885: 66; MacDonell 1916: 387; Pokorny IEW: 1086; EWAi
I: 673–674; LIV2
: 643);
int. nenigdhí (√nij- ‘washʼ; cf. ao. ánijam; < IE *√ne gu̯
-; cf. Gr. νίζω ‘washʼ, OIr. -nenaig
‘washʼ; cf. Whitney 1885: 90; MacDonell 1916: 395; Pokorny IEW: 761; EWAi II:
54; LIV2
: 450; NIL: 519–520; NIL: 660–661);
ao. áyugdham (√yuj- ‘joinʼ; cf. pr. yuňjánti; < IE *√ eu̯ g-; cf. L. iungō ‘harnessʼ, OCS igo
‘yokeʼ; cf. Whitney 1885: 132–133; MacDonell 1916: 410; Pokorny IEW: 508–510;
EWAi II: 417–418; LIV2
: 316; NIL: 397–404);
g(u̯ )h
+ dh
= *gdh
:
not attested
K(u̯ )
+ s = OIA kṣ:
ft. śakṣyáti (√śak- ‘be ableʼ: cf. ao. śákat; < IE *√ḱek(u̯ )
-; cf. OIr cecht ‘mightʼ; cf. Whitney
1885: 169; MacDonell 1916: 422; Pokorny IEW: 522; EWAi II: 600–601; LIV2
: 322);
ao. ápṛkṣi (√pṛc- ‘mixʼ; cf. pr. pṛňcáte; < IE *√perk-; cf. L. parcō ‘spareʼ; cf. Whitney
1885: 101; MacDonell 1916: 398–399; Pokorny IEW: 820; EWAi II: 96; LIV2
: 476);
ao. mukṣata, ds. múmukṣati (√muc- ‘releaseʼ; cf. ao. ámoci; < IE *√meu̯ k-; cf. L. ē-mungō
‘blow outʼ, Lith. munkù ‘get looseʼ; cf. Whitney 1885: 122; MacDonell 1916: 406;
Pokorny IEW: 744; EWAi II: 382–383; LIV2
: 443–444);
ao. pákṣat, ft. pakṣyáti (√pac- ‘cookʼ; cf. pr. pácati; < IE *√peku̯
-; cf. L. coquō, OCS pekǫ
‘cookʼ; cf. Whitney 1885: 92–93; MacDonell 1916: 396; Pokorny IEW: 798; EWAi
II: 64; LIV2
: 468; NIL: 548–552);
ft. vakṣyáti, ds. vívakṣati (√vac- ‘speakʼ; cf. ao. ávāci; < IE *√u̯ eku̯
-; cf. Gr. εἶπον ‘sayʼ;
cf. Whitney 1885: 151; MacDonell 1916: 415; Pokorny IEW: 1135–1136; EWAi II:
489–491; LIV2
: 673–674);
ao. nikṣi, ánaikṣīt (√nij- ‘washʼ; cf. ao. ánijam; < IE *√ne gu̯
-; cf. Gr. νίζω ‘washʼ, OIr. nenaig
‘washʼ; cf. Whitney 1885: 90; MacDonell 1916: 395; Pokorny IEW: 761;
EWAi II: 54; LIV2
: 450; NIL: 519–520);
g(u̯ )h
+ s = OIA kṣ:
ao. ádhākṣīt, ft. dhakṣyáti (√dah- ‘burnʼ; cf. pr. dáhati; < IE *√dh
egu̯ h
-; cf. Lith. degù
‘burnʼ; cf. Whitney 1885: 71; MacDonell 1916: 388–389; Pokorny IEW: 240–241;
EWAi I: 712–713; LIV2
: 133–134);
23
ao. ádhukṣata, dhukṣata (√duh- ‘milkʼ; cf. pr. duhánti; < IE *√dh
eu̯ gh
-; cf. Gr. τεύχω
‘make readyʼ, OE ge-dȳgan ‘overcomeʼ; cf. Whitney 1885: 76; MacDonell 1916: 390;
Pokorny IEW: 271; EWAi I: 747–748; LIV2
: 148–149);
As above, a solid example of clusters resulting from a concatenation of gh/
h2+dh
is missing, the
reconstruction is given purely by analogy either with ‘Bartholomae’s clustersʼ of gh
t/h2t or with
the dental series. As in other cases, the gh
s/h2s clusters are not subjected to Bartholomae’s Law,
in contrast with gh
t/h2t.
2.2.3 Clusters of palatovelar + t/dh
/s
The original IE palatovelars (realized in OIA as ś, j1, h1 respectively) form clusters of the
following patterns:
Ḱ + t = OIA ṣṭ:
pf. dídeṣṭu, ao. ádiṣṭa, ppp. diṣṭá-, int. dédiṣṭi (√diś- ‘pointʼ; cf. pr. diśátu; < IE *√de ḱ-;
cf. L. dīcō ‘sayʼ; cf. Whitney 1885: 73; MacDonell 1916: 389; Pokorny IEW: 188–
189; EWAi I: 744–746; LIV2
: 108–109);
ppp. dṛṣṭá-, gd. dṛṣṭvā́ (√dṛś- ‘see’; cf. pf. dadárśa; < IE *√derḱ-; cf. Gr. δέρκομαι ‘seeʼ;
cf. Whitney 1885: 391; MacDonell 1916: 78; Pokorny IEW: 213; EWAi I: x; LIV2
:
122);
pr. váṣṭi, váṣṭu, vivaṣṭi, vivaṣṭu (√vaś- ‘desireʼ; cf. pr. váśmi; < IE *√u̯ eḱ-; cf. Hitt. wēkzi
‘wishʼ; cf. Whitney 1885: 155; MacDonell 1916: 416; Pokorny IEW: 1135; EWAi II:
527–528; LIV2
: 672–673);
ao. áyaṣṭa, ppp. iṣṭá- (√yaj- ‘sacrificeʼ; cf. pr. yájati; < IE *√H aǵ-; cf. Gr. ἅζομαι ‘stand
in aweʼ (?); cf. Whitney 1885: 129; MacDonell 1916: 408; Pokorny IEW: 501; EWAi
II: 392–394; LIV2
: 254–255);
ppp. mṛṣtá- (√mṛj- ‘wipeʼ; cf. pr. mṛjánti; < IE *√H2merǵ-; cf. Gr. ἀμέργω ‘pluck, pullʼ;
cf. Whitney 1885: 125; MacDonell 1916: 407; Pokorny IEW: 738; EWAi II: 324–326;
LIV2
: 280–281);
ao. ásṛṣṭa, ppp. sṛṣtá- (√sṛj- ‘emitʼ; cf. pr. sṛjáti; < IE *√selǵ-; cf. Gr. λαγαίω, ‘releaseʼ
(?); cf. Whitney 1885: 189–190; MacDonell 1916: 428–429; Pokorny IEW: 900–901;
EWAi II: 709; LIV2
: 528–529);
ǵh
+ t = OIA 0ḍh
:
ppp. gūḍhá-, gd. gūḍhvī́ (√guh- ‘hideʼ; cf. ao. guháḥ; < IE *√gu̯ h
eu̯ ǵh
-; cf. Whitney 1885:
38; MacDonell 1916: 379–380; Pokorny IEW: 450; EWAi I: 502–503; LIV2
: 199);
pr. léḍhi B (√lih- ‘lickʼ; cf. caus. leháyati; < IE *√le ǵh
-; cf. Gr. λείχω, OCS ližǫ ‘lickʼ; cf.
Whitney 1885: 148; MacDonell 1916: 414; Pokorny IEW: 668; EWAi II: 463; LIV2
:
404);
ppp. rūḍhá-, gd. rūḍhvā́ (√ruh- ‘ascend’; cf. pr. róhati; < IE *√H1leu̯ dh
-; cf. Gr. Hom.
ἤλῠθον ‘come, startʼ; cf. Whitney 1885: 143–144; MacDonell 1916: 414; Pokorny
IEW: 306–307, 684–685; EWAi II: 467–469; LIV2
: 248–249; NIL: 245–246);
ppp. ūḍhá-, inf. vóḍhum (√vah- ‘carry’; cf. pr. váhati; < IE *√u̯ eǵh
-; cf. L. uehō, OCS
vezǫ ‘rideʼ; cf. Whitney 1885: 157; MacDonell 1916: 417; Pokorny IEW: 1118–1120;
EWAi II: 535–537; LIV2
: 661–662);
24
ppp. sāḍhá- (√sah- ‘prevailʼ; cf. pr. sáhate; < IE *√seǵh
-; cf. Gr. ἔχω ‘have, holdʼ; cf.
Whitney 1885: 184–185; MacDonell 1916: 427; Pokorny IEW: 888–889; EWAi II:
717–718; LIV2
: 515–516; NIL: 600–604);
Ḱ + dh
= OIA ḍḍh
:
pf. didiḍḍhí (√diś- ‘pointʼ; cf. pr. diśátu; < IE *√de ḱ-; cf. L. dīcō ‘sayʼ; cf. Whitney 1885:
73; MacDonell 1916: 389; Pokorny IEW: 188–189; EWAi II: 744–746; LIV2
: 108–
109);
pr. mṛḍḍhvám (√mṛj- ‘wipeʼ; cf. pr. mṛjánti; < IE *√H2merǵ-; cf. Gr. ἀμέργω ‘pluck,
pullʼ; cf. Whitney 1885: 125; MacDonell 1916: 407; Pokorny IEW: 738; EWAi II:
324–326; LIV2
: 280–281);
ǵh
+ dh
= OIA 0ḍh
:
ao. voḷhám, voḍhvám (√vah- ‘carry’; cf. pr. váhati; < IE *√u̯ eǵh
-; cf. L. uehō, OCS vezǫ
‘rideʼ;cf. Whitney 1885: 157; MacDonell 1916: 417; Pokorny IEW: 1118–1120;
EWAi II: 535–537; LIV2
: 661–662);
Ḱ + s = OIA kṣ:
ao. ádikṣi, ádikṣat B (√diś- ‘pointʼ; cf. pr. diśátu; < IE *√de ḱ-; cf. L. dīcō ‘sayʼ; cf.
Whitney 1885: 73; MacDonell 1916: 389; Pokorny IEW: 188–189; EWAi II: 744–
746; LIV2
: 108–109);
pr. dadṛkṣé, ao. ádrākṣīt B, ds. dídṛkṣate (√dṛś- ‘see’; cf. pf. dadárśa; < IE *√derḱ-; cf.
Gr. δέρκομαι ‘seeʼ; cf. Whitney 1885: 391; MacDonell 1916: 78; Pokorny IEW: 213;
EWAi I: x; LIV2
: 122);
pr. vákṣi, vavákṣi (√vaś- ‘desireʼ; cf. pr. váśmi; < IE *√u̯ eḱ-; cf. Hitt. wēkzi ‘wishʼ; cf.
Whitney 1885: 155; MacDonell 1916: 416; Pokorny IEW: 1135; EWAi II: 527–528;
LIV2
: 672–673);
ao. yákṣva, yákṣat, ft. yákṣyáte, ds. íyakṣati18
(√yaj- ‘sacrificeʼ; cf. pr. yájati; < IE *√H aǵ;
cf. Gr. ἅζομαι ‘stand in aweʼ (?); cf. Whitney 1885: 129; MacDonell 1916: 408;
Pokorny IEW: 501; EWAi II: 392–394; LIV2
: 254–255);
pr. mṛkṣvá, ao. ámārkṣīt B, ft. mrakṣyáte B (√mṛj- ‘wipeʼ; cf. pr. mṛjánti; < IE *√H2merǵ;
cf. Gr. ἀμέργω ‘pluck, pullʼ; cf. Whitney 1885: 125; MacDonell 1916: 407; Pokorny
IEW: 738; EWAi II: 324–326; LIV2
: 280–281);
ao. ásṛkṣí, ásṛkṣmahí, ft. srakṣyáti B, ds. sísṛkṣati B (√sṛj- ‘emitʼ; cf. pr. sṛjáti; < IE
*√selǵ-; cf. Gr. λαγαίω, ‘releaseʼ (?); cf. Whitney 1885: 189–190; MacDonell 1916:
428–429; Pokorny IEW: 900–901; EWAi II: 709; LIV2
: 528–529);
ǵh
+ s = OIA kṣ:
ao. ághukṣat, ds. júgukṣati (√guh- ‘hideʼ; cf. ao. guháḥ; < IE *√gu̯ h
eu̯ ǵh
-; cf. Whitney
1885: 38; MacDonell 1916: 379–380; Pokorny IEW: 450; EWAi I: 502–503; LIV2
:
199);
ao. rukṣās, árukṣat, ft. rokṣyáti B, ds. rúrukṣati (√ruh- ‘ascend’; cf. pr. róhati; < IE
*√H1leu̯ dh
-; cf. Gr. Hom. ἤλῠθον ‘come, startʼ; cf. Whitney 1885: 143–144;
MacDonell 1916: 414; Pokorny IEW: 306–307, 684–685; EWAi II: 467–469; LIV2
:
248–249; NIL: 245–246);
18
An atypical reduplication, the expecting outcome would be *yíyakṣati (cf. caus. ao. áyīyajat B. from the same
root).
25
ao. ávākṣur, ft. vakṣyáti (√vah- ‘carry’; cf. pr. váhati; < IE *√u̯ eǵh
-; cf. L. uehō, OCS vezǫ
‘rideʼ; cf. Whitney 1885: 157; MacDonell 1916: 417; Pokorny IEW: 1118–1120;
EWAi II: 535–537; LIV2
: 661–662);
ao. ásākṣi, sákṣat, ft. sakṣyáte B (√sah- ‘prevailʼ; cf. pr. sáhate; < IE *√seǵh
-; cf. Gr. ἔχω
‘have, holdʼ; cf. Whitney 1885: 184–185; MacDonell 1916: 427; Pokorny IEW: 888–
889; EWAi II: 717–718; LIV2
: 515–516; NIL: 600–604);
An interesting feature in the development is the secondary cerebralization of clusters, which is
triggered by the ruki-rule (Pedersen’s Law), affecting clusters with a right dental plosive,
clearly distinguishing them from clusters of velars/palatals + plosive (see above). What is
remarkable is that h1 + t (‘Bartholomae’s clustersʼ) and h1 + dh
clusters both tend to the outcome
0ḍh
, where 0 stands for the lost element (probably an approximant), causing either
diphthongization of the preceding a to o or e, dependent on the vowel of the next syllable (cf.
voḍhvám from √vah- ‘carryʼ, tṛṇédhi from √tṛh- ‘crushʼ for diphthongic outcome; ūḍhá- from
√vah- ‘carryʼ, sāḍhá- from √sah- ‘prevailʼ for a simple lengthening of a preceding vowel). In
contrast, clusters resulting from ś/j1+dh
have as the outcome a geminate ḍḍh
(cf. pf. didiḍḍhí
from √diś- ‘point’ or pr. mṛḍḍhvám from √mṛj- ‘wipeʼ). As in other cases, the h1s cluster is not
subjected to Bartholomae’s Law, in contrast with h1t.
2.2.4 Clusters of dental + t/dh
/s
The dentals form clusters of the following patterns:
T + t = OIA tt:
ppp. kṛttá- (√kṛt- ‘cutʼ; cf. pr. kṛntáti; < IE *√(s)kert-; cf. Lith. kertù ‘cut offʼ; cf. Whitney
1885: 22–23; MacDonell 1916: 376; Pokorny IEW: 941–942; EWAi I: 315–316; LIV2
:
559–560);
ppp. vṛtṭá- (√vṛt- ‘turnʼ; cf. pr. vártate; < IE *√u̯ ert-; cf. L uertor, Goth. waírþan ‘turnʼ;
cf. Whitney 1885: 164; MacDonell 1916: 420–421; Pokorny IEW: 1156–158; EWAi
II: 518–519; LIV2
: 691–692);
pr. unátti, ppp. uttá- B (√ud- ‘wetʼ; cf. pr. undánti; < IE *√u̯ ed-; cf. Whitney 1885: 13;
MacDonell 1916: 373; Pokorny IEW: 346; EWAi II: 215–216, 279; LIV2
: 658–659;
NIL: 706–715);
pr. átti inf. áttum (√ad- ‘eatʼ; cf. pr. ádmi; < IE *√H1ed-; cf. Hitt. ēdmi, L. edō ‘eatʼ; cf.
Whitney 1885: 3; MacDonell 1916: 370; Pokorny IEW: 287–289; EWAi I: 61–62;
LIV2
: 230–231; NIL: 208–220);
pr. bhinátti, gd. bhittvā́ (√bhid- ‘splitʼ; cf. pr. bhinádmi; < IE *√bh
e d-; cf. L. findō ‘splitʼ,
Goth. beitan ‘bitʼ; cf. Whitney 1885: 111; MacDonell 1916: 402; Pokorny IEW: 116–
117; EWAi II: 273–274; LIV2
: 70–71; NIL: 11–12);
ppp. vittá-, gd. vittvā́ (√vid- ‘findʼ; cf. pr. vindáti; < IE *√u̯ e d-; cf. L. uīdī ‘seeʼ, Arm.
egit ‘findʼ; cf. Whitney 1885: 159–160; MacDonell 1916: 418; Pokorny IEW: 1125–
1127; EWAi II: 579–581; LIV2
: 665–667; NIL: 717–722);
dh
+ t = OIA ddh
:
26
pr. inddhé, ppp. iddhá- (√idh- ‘kindleʼ; cf. pr. indháte; < IE *√H2e dh
-; cf. Gr. αἴθω
‘kindleʼ; cf. Whitney 1885: 8; MacDonell 1916: 371–372; Pokorny IEW: 11-12;
EWAi II: 267; LIV2
: 259);
ppp. baddhá-, gd. baddhvā́ (√bandh- ‘bindʼ; cf. pr. badhnáte; < IE *√bh
endh
-; cf. Goth.
band ‘bindʼ; cf. Whitney 1885: 105; MacDonell 1916: 400; Pokorny IEW: 127; EWAi
II: 208; LIV2
: 75);
ppp. buddhá- (√budh- ‘wakeʼ; cf. pr. bódhati; < IE *√bh
eu̯ dh
-; cf. Gr. πεύθομαι ‘give
noticeʼ, OCS bljudǫ ‘bewareʼ; cf. Whitney 1885: 106–107; MacDonell 1916: 400–
401; Pokorny IEW: 150–152; EWAi II: 233–235; LIV2
: 82–83; NIL: 36–37);
ppp. yuddhá-, gd. yuddhvī́ (√yudh- ‘fightʼ; cf. pr. yúdhyate; < IE *√H eu̯ dh
-; cf. OL. ioubē
‘commandʼ; cf. Whitney 1885: 133; MacDonell 1916: 410; Pokorny IEW: 511–512;
EWAi II: 418–419; LIV2
: 225–226);
T + dh
= OIA ddh
/0dh
:
pr. addhí (√ad- ‘eatʼ; cf. pr. ádmi; < IE *√H1ed-; cf. Hitt. ēdmi, L. edō ‘eatʼ; cf. Whitney
1885: 3; MacDonell 1916: 370; Pokorny IEW: 287–289; EWAi I: 61–62; LIV2
: 230–
231; NIL: 208–220);
pr. viddhí (√vid- ‘findʼ; cf. pr. vindáti; < IE *√u̯ e d-; cf. L. uīdī ‘seeʼ, Arm. egit ‘findʼ; cf.
Whitney 1885: 159–160; MacDonell 1916: 418; Pokorny IEW: 1125–1127; EWAi II:
579–581; LIV2
: 665–667; NIL: 717–722);
pr. dehí,19
daddhí (√dā- ‘giveʼ; cf. pr. dádāti; < IE *√deH3-; cf. Gr. δίδωμι, L. dō, OLith.
duosti ‘giveʼ; cf. Whitney 1885: 71–72; MacDonell 1916: 388–389; Pokorny IEW:
223–225; EWAi I: 713–715; LIV2
: 105–106; NIL: 60–69);
pr. undhí (-ddh-) (√ud- ‘wetʼ; cf. pr. undánti; < IE *√u̯ ed-; cf. Whitney 1885: 13;
MacDonell 1916: 373; Pokorny IEW: 346; EWAi II: 215–216, 279; LIV2
: 658–659;
NIL: 706–715);
pr. bhindhí (-ddh-) (√bhid- ‘splitʼ; cf. pr. bhinádmi; < IE *√bh
e d-; cf. L. findō ‘splitʼ,
Goth. beitan ‘bitʼ; cf. Whitney 1885: 111; MacDonell 1916: 402; Pokorny IEW: 116–
117; EWAi II: 273–274; LIV2
: 70–71; NIL: 11–12);
dh
+ dh
= OIA 0dh
:
pr. bodhí (-ddh-) (√budh- ‘wakeʼ; cf. pr. bódhati; < IE *√bh
eu̯ dh
-; cf. Gr. πεύθομαι ‘give
noticeʼ, OCS bljudǫ ‘bewareʼ; cf. Whitney 1885: 106–107; MacDonell 1916: 400–
401; Pokorny IEW: 150–152; EWAi II: 233–235; LIV2
: 82–83; NIL: 36–37);
ao. yodhí (-ddh-) (√yudh- ‘fightʼ; cf. pr. yúdhyate; < IE *√H eu̯ dh
-; cf. OL. ioubē
‘commandʼ; cf. Whitney 1885: 133; MacDonell 1916: 410; Pokorny IEW: 511–512;
EWAi II: 418–419; LIV2
: 225–226);
pr. dhehí20
(√dhā- ‘putʼ; cf. pr. dádhāmi; < IE *√dh
eH1-; cf. Gr. τίθημι ‘putʼ, Lith. desti;
cf. Whitney 1885: 82; MacDonell 1916: 392–393; Pokorny IEW: 235–239; EWAi I:
783–786; LIV2
: 136–138; NIL: 99–117);
pr. indhvám (-ddh-) (√idh- ‘kindleʼ; cf. pr. indháte; < IE *√H2e dh
-; cf. Gr. αἴθω ‘kindleʼ;
cf. Whitney 1885: 8; MacDonell 1916: 371–372; Pokorny IEW: 11-12; EWAi II: 267;
LIV2
: 259);
19
From -d+dh
, output is simplified instead of the expected daddh
i or dedh
i (= -0dh
-).
2020
From -dh
+dh
, output is simplified instead of the expected dh
edh
i (= -0dh
-).
27
T + s = OIA ts:
ft. kartsyā́ mi (√kṛt- ‘cutʼ; cf. pr. kṛntáti; < IE *√(s)kert-; cf. Lith. kertù ‘cut offʼ; cf.
Whitney 1885: 22–23; MacDonell 1916: 376; Pokorny IEW: 941–942; EWAi I: 315–
316; LIV2
: 559–560);
ao. ávṛtsaṭa, ft. vartsyáti (√vṛt- ‘turnʼ; cf. pr. vártate; < IE *√u̯ ert-; cf. L uertor, Goth.
waírþan ‘turnʼ; cf. Whitney 1885: 164; MacDonell 1916: 420–421; Pokorny IEW:
1156–158; EWAi II: 518–519; LIV2
: 691–692);
pr. átsi (√ad- ‘eatʼ; cf. pr. ádmi; < IE *√H1ed-; cf. Hitt. ēdmi, L. edō ‘eatʼ; cf. Whitney
1885: 3; MacDonell 1916: 370; Pokorny IEW: 287–289; EWAi I: 61–62; LIV2
: 230–
231; NIL: 208–220);
pr. bhinátsi, ft. bhetsyáte B, ds. bíbhitsati (√bhid- ‘splitʼ; cf. pr. bhinádmi; < IE *√bh
e d-;
cf. L. findō ‘splitʼ, Goth. beitan ‘bitʼ; cf. Whitney 1885: 111; MacDonell 1916: 402;
Pokorny IEW: 116–117; EWAi II: 273–274; LIV2
: 70–71; NIL: 11–12);
pr. vitsé, ao. ávitsi, ds. vívitsati B (√vid- ‘findʼ; cf. pr. vindáti; < IE *√u̯ e d-; cf. L. uīdī
‘seeʼ, Arm. egit ‘findʼ; cf. Whitney 1885: 159–160; MacDonell 1916: 418; Pokorny
IEW: 1125–1127; EWAi II: 579–581; LIV2
: 665–667; NIL: 717–722);
ds. part. dítsant- (√dā- ‘giveʼ; cf. pr. dádāti; < IE *√deH3-; cf. Gr. δίδωμι, L. dō, OLith.
duosti ‘giveʼ; cf. Whitney 1885: 71–72; MacDonell 1916: 388–389; Pokorny IEW:
223–225; EWAi I: 713–715; LIV2
: 105–106; NIL: 60–69);
dh
+ s = OIA ts:
ft. bhantsyáti (√bandh- ‘bindʼ; cf. pr. badhnáte; < IE *√bh
endh
-; cf. Goth. band ‘bindʼ; cf.
Whitney 1885: 105; MacDonell 1916: 400; Pokorny IEW: 127; EWAi II: 208; LIV2
:
75);
ao. ábhutsi, ft. bhotsáti B (√budh- ‘wakeʼ; cf. pr. bódhati; < IE *√bh
eu̯ dh
-; cf. Gr. πεύθομαι
‘give noticeʼ, OCS bljudǫ ‘bewareʼ; cf. Whitney 1885: 106–107; MacDonell 1916:
400–401; Pokorny IEW: 150–152; EWAi II: 233–235; LIV2
: 82–83; NIL: 36–37);
pr. yótsi, ds. yúyutsati (√yudh- ‘fightʼ; cf. pr. yúdhyate; < IE *√H eu̯ dh
-; cf. OL. ioubē
‘commandʼ; cf. Whitney 1885: 133; MacDonell 1916: 410; Pokorny IEW: 511–512;
EWAi II: 418–419; LIV2
: 225–226);
ds. dhítsati (√dhā- ‘putʼ; cf. pr. dádhāmi; < IE *√dh
eH1-; cf. Gr. τίθημι ‘putʼ, Lith. desti;
cf. Whitney 1885: 82; MacDonell 1916: 392–393; Pokorny IEW: 235–239; EWAi I:
783–786; LIV2
: 136–138; NIL: 99–117);
Especially worthy of note are t/d + dh
clusters,
since they result either in ddh
(as do
‘Bartholomae’s dh
t clustersʼ), but sometimes in 0dh
; cf. the root √dā- ‘giveʼ which has both
forms: daddhí, dehí. The cluster of dh
+ dh
results in 0dh
regularly. As in other cases, the dh
s
cluster is not subjected to Bartholomae’s Law, in contrast with dh
t.
Note: Few examples demonstrate the development of clusters with a cerebral plosive in the left. From the root
√īḍ- ‘praiseʼ there is ppp. ī́ṭṭe. from √taḍ- ‘beatʼ there is pr. tāḍhi (from roots √pīḍ- ‘pressʼ, √vīḍ- ‘make strongʼ,
√hīḍ- ‘be hostileʼ etc., have no attested forms with direct contact of a left cerebral plosive + t/dh
/s, since this
clusters are prevented often by the anaptyctic i; cf. ppp. vīḷitá-, hīḍitá-). The clusters ending in -ṣ are dealt with
below.
2.2.5 Clusters of sibilant + t/dh
/s
Both OIA sibilants are inherited from Indo-Iranian, ṣ arose from Indo-Iranian *š (and from IE
*s due to Pedersen’s law/ruki-rule).
28
s + t = OIA st:
pr. ásti, stás, ástu (√as- ‘be’; cf. pr. ásat; < IE *√H1es-; cf. Gr. ἐστί, L. est ‘beʼ; cf.
Whitney 1885: 5; MacDonell 1916: 370–371; Pokorny IEW: 340–341; EWAi I: 144;
LIV2
: 241–242; NIL: 235–238);
ao. ághasta (√ghas- ‘eat’; cf. pf. jaghása; < IE *√g(u̯ )h
es-; cf. Whitney 1885: 42;
MacDonell 1916: 381; Pokorny IEW: 452; EWAi I: 514; LIV2
: 198–199);
ppp. ustá-, inf. vástave (√vas- ‘shine’; cf. ao. ávasran; < IE *√u̯ es-; cf. Lith. au͂šti ‘break
dawnʼ; cf. Whitney 1885: 155-156; MacDonell 1916: 417; Pokorny IEW: 86–87;
EWAi II: 530–532; LIV2
: 292–293; NIL: 357–367);
pr. śāste (√śās- ‘order’; cf. pr. śā́smi; < IE *√ḱeHs-; cf. Alb. thom ‘sayʼ; cf. Whitney 1885:
172; MacDonell 1916: 423; Pokorny IEW: 533; EWAi II: 632–633; LIV2
: 318–319);
ṣ + t = OIA ṣṭ:
pr. dvéṣṭi, ppp. dviṣṭá- (√dviṣ- ‘hate’; cf. pr. dvéṣat; < IE *√du̯ e s-; cf. OAv. dai
bišəṇtī; cf.
Whitney 1885: 81; MacDonell 1916: 392; Pokorny IEW: 228; EWAi I: 770–771;
LIV2
: 131);
pr. pináṣṭi, ppp. piṣṭá- (√piṣ- ‘crush’; cf. pf. pipéṣa; < IE *√pe s-; cf. Lith. pisù ‘copulateʼ,
OCS pьxomъ ‘pushʼ; cf. Whitney 1885: 97–98; MacDonell 1916: 398; Pokorny IEW:
796; EWAi II: 169; LIV2
: 466–467);
pr. víveṣṭi, viviṣṭás, ppp. viṣṭá- (√viṣ- ‘be active’; cf. pr. víveṣaḥ; < IE *√u̯ e s- 3; cf.
Whitney 1885: 161; MacDonell 1916: 419; EWAi II: 585–586; LIV2
: 672);
s + dh
= OIA 0dh
:21
pr. edhí (√as- ‘be’; cf. pr. ásat; < IE *√H1es-; cf. Gr. ἐστί, L. est ‘beʼ; cf. Whitney 1885:
5; MacDonell 1916: 370–371; Pokorny IEW: 340–341; EWAi I: 144; LIV2
: 241–242;
NIL: 235–238);
pr. ā́ dhvam (√ās- ‘sit’; cf. pr. ā́sāthe; < IE *√x-; cf. Hith. ēsa, Gr. ἡσται ‘sitʼ; cf. Whitney
1885: 6-7; MacDonell 1916: 371; Pokorny IEW: 342–343; EWAi II: x; LIV2
: 232);
pr. śādhí, pf. śaśādhí (√śās- ‘order’; cf. pr. śā́smi; < IE *√ḱeHs-; cf. Alb. thom ‘sayʼ; cf.
Whitney 1885: 172; MacDonell 1916: 423; Pokorny IEW: 533; EWAi II: 632–633;
LIV2
: 318–319);
ṣ + dh
= OIA ḍḍh
:
pf. viviḍḍhí (√viṣ- ‘be active’; cf. pr. víveṣaḥ; < IE *√u̯ e s- 3; cf. Whitney 1885: 161;
MacDonell 1916: 419; EWAi II: 585–586; LIV2
: 672);
s + s = OIA ts/ss/0s:
ds. jíghatsati AV (√ghas- ‘eat’; cf. pf. jaghása; < IE *√g(u̯ )h
es-; cf. X x ‘xʼ; cf. Whitney
1885: 42; MacDonell 1916: 381; Pokorny IEW: 452; EWAi I: 514; LIV2
: 198–199);
co. ávatsyat B (√vas- ‘shine’; cf. ao. ávasran; < IE *√u̯ es-; cf. Lith. au͂šti ‘break dawnʼ;
cf. Whitney 1885: 155-156; MacDonell 1916: 417; Pokorny IEW: 86–87; EWAi II:
530–532; LIV2
: 292–293; NIL: 357–367);
but
pr. śāssi (√śās- ‘order’; cf. pr. śā́smi; < IE *√ḱeHs-; cf. Alb. thom ‘sayʼ; cf. Whitney
1885: 172; MacDonell 1916: 423; Pokorny IEW: 533; EWAi II: 632–633; LIV2
: 318–
319);
21
Cf. later √vas- 2. ‘wear’: pr. vaddhvam S. for the form with two plosives. This form is clearly construed as
analogical to šdh
> ḍḍh
.
29
but
pr. ási22
(√as- ‘be’; cf. pr. ásat; < IE *√H1es-; cf. Gr. ἐστί, L. est ‘beʼ; cf. Whitney 1885:
5; MacDonell 1916: 370–371; Pokorny IEW: 340–341; EWAi I: 144; LIV2
: 241–242;
NIL: 235–238);
ṣ + s = OIA kṣ:
ao. dvikṣát, dvikṣata AV (√dviṣ- ‘hate’; cf. pr. dvéṣat; < IE *√du̯ e s-; cf. OAv. dai
bišəṇtī;
cf. Whitney 1885: 81; MacDonell 1916: 392; Pokorny IEW: 228; EWAi I: 770–771;
LIV2
: 131);
ao. apikṣan ŚB (√piṣ- ‘crush’; cf. pf. pipéṣa; < IE *√pe s-; cf. Lith. pisù ‘copulateʼ, OCS
pьxomъ ‘pushʼ; cf. Whitney 1885: 97–98; MacDonell 1916: 398; Pokorny IEW: 796;
EWAi II: 169; LIV2
: 466–467);
pr. vívekṣi, ft. vekṣyáti (√viṣ- ‘be active’; cf. pr. víveṣaḥ; < IE *√u̯ e s- 3; cf. Whitney 1885:
161; MacDonell 1916: 419; EWAi II: 585–586; LIV2
: 672);
Clusters with t- and dh
- are cerebralized after ṣ. Clusters from ṣdh
are realized in two variants,
either as ḍḍh
or as 0ḍh
(cf. development of clusters of palatovelars + dh
and dentals + dh
above).
Clusters of two sibilants are usually dissimilated as a plosive + sibilant.
2.2.6 Overview of attested Vedic alternations
For a summary of the attested alternations of Vedic clusters formed either by any plosive +
t/dh
/s or a sibilant + t/dh
/s see the following table. Not attested forms are omitted, and the
assumed analogous forms are in brackets:
IE OIA t- dh
- s-
-k(ṷ)
/g(ṷ)
-k/c/g/j2 kt gdh
kṣ
-g(ṷ)h
-gh
/h2 gdh
(kṣ)
-ḱ/ǵ -ś/j1 ṣṭ ḍḍh
kṣ
-ǵh
-h1 0ḍh
0ḍh
(kṣ)
-t/d -t/d tt ddh
0dh
ts
-dh
-dh
ddh
0dh
(ts)
-p/b -p/b pt ps
-bh
-bh
bdh
(ps)
-s -s st 0dh
ts
-š -ṣ ṣṭ ḍḍh
kṣ
Note: Since Indo-Iranian does not distinguish between original IE plain velars and labiovelars, we omit this
distinction for its irrelevance for the analysed data.
Note: For simplicity, the contrast between s and š is ahistorically translated to Indo-European, though the
applicance of Pedersen´s Law (ruki-law) was limited just to the part of the IE dialect continuum, similar
‘ahistoricalʼ introduction of š will also be used below for other languages.
22
It seems that in this case the simplification of IE *ss to *s is already Indo-European, hence older than here listed
examples of the later development, since we meet not only Av. ahi, but also Gr. εἶ, OLith. esi, OCS jesi, see
more below.
30
2.3 The development of two-obstruent clusters in Avestan
Iranian development differs in some of its aspects from that of Indic, though on the other hand,
there are striking similarities.
The Old-Iranian phase is represented by two languages, Avestan and Old Persian. Our
primary source language is Avestan, which is better attested and hence, offers more data. Old
Persian will serve as complementary material to Avestan (an overview will be given below
independently).
From the typical features of Avestan development, we have to emphasize:
i. the original voiced non-aspirates and voiced aspirated are neutralized to voiced non-aspirated
in all positions, the originally voiced aspirates could be internally reconstructed only
using Bartholomae’s Law (see iv);
ii. the left plosives are generally spirantized in the cluster, dentals are sibilantized (and see v);
iii. though Bartholomae’s Law is operating in Iranian, it is often eroded due to analogy;
iv. in contrast with Indo-Aryan, clusters of original voiced aspirate + s are generally subjected
to Bartholomae’s law (again, this process is often eroded by analogy);
v. the clusters of original palatovelar plosive + t/dh
are sibilantized in Iranian to št/žd(h)
, the
analogical clusters of *Ḱ+s and *ǵh
+s are realized as 0š and 0ž;
vi. the clusters of two sibilants are simplified as a single sibilant.
2.3.1 Clusters of labial + t/dh
/s
A remarkable feature of the Avestan development is that the IE cluster *Pt is realized, contrary
to the expected results, as Av. pt (see below about this peculiarity), similarly the *bh
t and *Pdh
clusters, both realized as bd without spirantization. the IE cluster *Ps is regularly realized as
fs/fš (irregularly subjected to Pedersen’s Law!):
P + t = Av. pt:
ppp. YAv. hąm.tapta- (√tap- ‘heatʼ; cf. caus. YAv. tāpaiieiti; < IE *√tep-; cf. L. tepeō ‘be
warmʼ, OCS teplostъ ‘heatʼ; cf. Pokorny IEW: 1069; Kellens 1995: 24; LIV2
: 629–
630; NIL: 698–700; Cheung 2007: 378–380);
pr. OAv. haptī (√hap- ‘keepʼ; < IE *√sep-; cf. Gr. ἕπω ‘to be aboutʼ, L. sepeliō ‘buryʼ; cf.
Pokorny IEW: 909; Kellens 1995: 71; LIV2
: 534; Cheung 2007: 129);
num. hapta- ‘sevenʼ (< IE *√septm̥-; cf. Gr. ἑπτά, L. septem ‘sevenʼ; cf. Pokorny IEW:
909; Emmerick 1992b: 299; Blažek 1999: 246);
ppp. YAv. vipta- (vaēp- ‘engage in homosexual activitiesʼ; cf. pr. YAv. vaēpəṇti; < IE
*√u̯e p-; cf. OIA vépate ‘tremble, shakeʼ, L. uibrāre ‘vibrateʼ; cf. Pokorny IEW: 1131–
1132; Kellens 1995: 55; LIV2
: 671; Cheung 2007: 415);
bh
+ t = OAv. bd/YAv. βδ (pt):
nom. YAv. dərəβδa ‘bundle of musclesʼ (< Ir. *√darb- ‘joinʼ; < IE *√dh
erbh
-; cf. Lith.
dirbù ‘workʼ; cf. Pokorny IEW: 211-212, 257; LIV2
: 121; Cheung 2007: 60);
but without the Bartholomae’s Law:
31
ppp. YAv. dapta- ‘cheatedʼ (√dab- ‘deceiveʼ; cf. pr. OAv. dəbənaotā < IE *√dh
ebh
-; cf.
Lith. dóbiu ‘subdueʼ; cf. Pokorny IEW: 240; Kellens 1995: 27; LIV2
: 132–133; NIL:
85–86; Cheung 2007: 42–43);
P + dh
= Av. bd:
pr. YAv. auuaŋhabdaēta, caus. YAv. nixv
abdaiieiti (√xv
ap- ‘sleepʼ, pf. YAv. hušxv
afa <
IE *√su̯ep-; cf. L. sopiō, OCS sъpljǫ ‘sleepʼ; cf. Pokorny IEW: 1048–1149; Kellens
1995: 17–18; LIV2
: 612–613; NIL: 675–680; Cheung 2007: 145–146);
bh
+ dh
= Av. *bd:
not attested
P + s = Av. fš/fs:
pr. hafšī OAv. (√hap- ‘keepʼ; < IE *√sep-; cf. Gr. ἕπω ‘to be aboutʼ, L. sepeliō ‘buryʼ; cf.
Pokorny IEW: 909; Kellens 1995: 71; LIV2
: 534; Cheung 2007: 129);
inch. YAv. xv
afsa, xv
afsata (√xv
ap- ‘sleepʼ, pf. YAv. hušxv
afa < IE *√su̯ep-; cf. L. sopiō,
OCS sъpljǫ ‘sleepʼ; cf. Pokorny IEW: 1048–1149; Kellens 1995: 17–18; LIV2
: 612–
613; NIL: 675–680; Cheung 2007: 145–146);
pr. inch. YAv. tafsat̰ (√tap- ‘heatʼ; cf. caus. YAv. tāpaiieiti; < IE *√tep-; cf. L. tepeō ‘be
warmʼ, OCS teplostъ ‘heatʼ; cf. Pokorny IEW: 1069; Kellens 1995: 24; LIV2
: 629–
630; NIL: 698–700; Cheung 2007: 378–380);
bh
+ s = Av. βž:
ds. OAv diβžaidiiāi (√dab- ‘deceiveʼ; cf. pr. OAv. dəbənaotā; < IE *√dh
ebh
-; cf. Lith.
dóbiu ‘subdueʼ; cf. Pokorny IEW: 240; Kellens 1995: 27; LIV2
: 132–133; NIL: 85–86;
Cheung 2007: 42–43);
YAv. vaβžaka- ‘scorpionʼ (< IE *u̯obh
sā; cf. L. uespa, OHG wafsa; cf. Pokorny IEW:
1179;);
but without the Bartholomae’s law:
ao. subj. YAv. haṇgərfšānē (√grab- ‘seizeʼ, ao. inj. grabəm; < IE *√gh
rebH2-/gh
rebh
- (?);
cf. Lith. grė́biu, OCS grablǫ ‘robʼ; cf. Pokorny IEW: 455; Kellens 1995: 20–21; LIV2
:
201; Cheung 2007: 119–121);
pr. inch. YAV. xšufsąn (√xšub- ‘rustle, trembleʼ; < IE *√k(u̯)
seu̯bh
-; cf. OIA cukṣubhe
‘quakeʼ; cf. Pokorny IEW: 625; Kellens 1995: 17; LIV2
: 372; Cheung 2007: 454–455);
A remarkable feature is the preservation of the pt in Avestan clusters as two-plosive clusters,
which is in strong contrast to the development of analogous clusters (see below). The Avestan
development of the cluster Pt is irregular, not only in comparison to the development of the
analogous clusters but also in other Iranian languages, since Av. hapta ‛seven’ has the Pahlavi
and New Persian counterpart haft (< OIr. *hafta). The question is whether the Avestan pt is an
archaism, or whether it represents an innovation (despirantization/occlusivization), since, in
Avestan, we encounter the following forms of pitar- ‛father’: nom. sg. OAv. ptā, tā, YAv. ptā,
pita, dat. sg. OAv. fə
δrōi, piϑrē, YAv. piϑre. The form tā is easy to explain as the result of
development from *ftā (i.e., pt > ft > ht > 0t) but Hoffmann and Forssman (1996: 94) assume
the direct simplification of the word-initial pt- > t-). Beekes (1988: 73) and Hoffmann and
32
Forssman (1996: 94) reckon with the preservation of inherited pt, contrary to Reichelt (1909:
40), whereas for eastern dialects of Iranian Kümmel (2007: 65) assumes partial restitution of
the spirants *f, *θ, *χ by the aspirates ph
, th
, kh
, which partially, however, can be considered as
original according to Morgenstierne (cf. Lipp 2009a: 158–160 with further ref.). The outcome
of the cluster *bh
dh
is not directly attested; it is modelled according to the assumed analogy with
‘Bartholomae’s clusterʼ *bh
t.
Note: The confusing outcomes of labial + s as pš or ps could be due to morphology, since inchoatives and sapresents
(both parallel to OIA -ccha- and from IE *sḱo-) regularly have ps but aorists and desideratives fš. The
extension to the ‘rukiʼ-rule in Avestan is hence limited to original single sibilants, not clusters.
2.3.2 Clusters of velar/palatal + t/dh
/s
The Indo-European plain and labiovelars are merged, secondarily often palatalized, but this
palatalization is neutralized before obstruents. The plosives are spirantized before *t-/dh/s-,
the
clusters ending in *g(u̯)h
are subjected to Bartholomae’s Law (often neglected by the analogy
with voiceless clusters):
K(u̯)
+ t: = Av. xt:
ppp. YAv. yuxta- (√yuj- ‘yoke’; cf. ao. OAv. yaojā; < IE *√ eu̯ g-; cf. L. iungō ‘harnessʼ,
OCS igo ‘yokeʼ; cf. Pokorny IEW: 508–509; Kellens 1995: 47; LIV2
: 316; NIL: 397–
404; Cheung 2007: 217–218);
YAv. pr. irinaxti (√raēc- ‘leaveʼ; cf. caus. YAv. raēcaiieiṇti; < IE *√le ku̯
-; cf. Gr. λείπω,
L. linquō ‘leaveʼ; cf. Pokorny IEW: 669–670; Kellens 1995: 58; LIV2
: 406–408; NIL:
451–452; Cheung 2007: 307–308);
g(u̯)h
+ t: = OAv. gd/YAv. γδ:
ppp. YAv. bərəγδa- (√barj- ‘honourʼ; cf. ps. bərəjaēm; < IE *√bh
ergh
-; cf. Goth. bairgan
‘protect, shelterʼ, OCS ne-brěšti ‘neglectʼ; cf. Pokorny IEW: 145; Kellens 1995: 38;
LIV2
: 79–80; Cheung 2007: 10–12);
pr. OAv. aogə
dā but. pr. YAV aoxte (√aoj- ‘sayʼ; cf. pr. OAV. aojōi; < IE *√H1eu̯g(u̯)h
(?);
cf. Gr. εὔχομαι ‘prayʼ, L. uoueō ‘vowʼ; cf. Pokorny IEW: 348; Kellens 1995: 9,14;
LIV2
: x; Cheung 2007: 169–170);
but without Bartholomae’s law:
ppp. YAv. handraxta-, int. YAv. dądrąxti (√draṇj- ‘fixʼ; cf. caus. drəṇjaiieiti; < IE
*√dregh
-; cf. Gr. δράσσομαι ‘graspʼ, OCS drьžǫ ‘holdʼ; cf. Pokorny IEW: 254; Kellens
1995: 32; LIV2
: 126; Cheung 2007: 76);
YAv. ppp. anādruxta- (√druj- ‘lie, deceiveʼ; cf. pr. YAv. družaite; < IE *√dh
reu̯gh
-; cf.
OHG triugan ‘deceitʼ; cf. Pokorny IEW: 276; Kellens 1995: 32; LIV2
: 157; Cheung
2007: 80–81);
K(u̯)
+ dh
= Av. gd:
OAv. mərəṇgəduiiē (√marc- ‘destroyʼ; cf. pr. mərəṇcaite; < IE *√melku̯
-; cf. Gr. βλάπτω
‘damageʼ; cf. Pokorny IEW: 737; Kellens 1995: 43; LIV2
: 434–435; Cheung 2007:
265–266);
33
g(u̯)h
+ dh
= Av. *gd:
not attested;
K(u̯)
+ s = Av. xš:
ao. OAv. uruuāxšat̰ (√uruuaj- ‘go forthʼ; < IE *√u̯reg-; cf. L. urgēre ‘urgeʼ, Goth. wrikan
‘pursueʼ; cf. Pokorny IEW: 1181; Kellens 1995: 60; LIV2
: 697; Cheung 2007: 438);
YAv. ao. raēkšīša (√raēc- ‘leaveʼ; cf. caus. YAv. raēcaiieiṇti; < IE *√le ku̯
-; cf. Gr. λείπω,
L. linquō ‘leaveʼ; cf. Pokorny IEW: 669–670; Kellens 1995: 58; LIV2
: 406–408; NIL:
451–452; Cheung 2007: 307–308);
pr. YAv. takše, ds. YAv. x
tíxšəṇti (√tac- ‘runʼ, pr. YAv. fratacaiti; < IE *√teku̯
-; cf. OCS
tekǫ, Lith. tekù ‘run, flowʼ; cf. Pokorny IEW: 1059–1060; Kellens 1995: 24; LIV2
:
620–621; Cheung 2007: 372–374);
nom. vāxš ‘voiceʼ (~ vācəm), fut. OAv. vaxšiiā (√vac- ‘speakʼ; cf. OAv. ao vaocat̰; < IE
*√u̯eku̯
-; cf. Gr. εἶπον ‘sayʼ; cf. Pokorny IEW: 1135–1136; Kellens 1995: 48–49; LIV2
:
673–674; Cheung 2007: 402–404);
g(u̯)h
+ s = Av. γž (xš):
ds. OAv. dīdraγžō.duiiē (√draṇj- ‘fixʼ; cf. caus. drəṇjaiieiti; < IE *√dregh
-; cf. Gr.
δράσσομαι ‘graspʼ, OCS drьžǫ ‘holdʼ; cf. Pokorny IEW: 254; Kellens 1995: 32; LIV2
:
126; Cheung 2007: 76);
ds. OAv. mimaγžō (√mag- ‘offerʼ; cf. OIA maghá- ‘gift, rewardʼ; < IE *√meH2gh
- (?); cf.
Goth. mag, OCS mogǫ ‘canʼ, Lith. magù ‘want, likeʼ; cf. Pokorny IEW: 695; Kellens
1995: 43; LIV2
: 422; Cheung 2007: 254);
pr. OAv. pai
rii-aoγžā (√aoj- ‘sayʼ; cf. pr. OAV. aojōi; < IE *√H1eu̯g(u̯)h
- (?); cf. Gr.
εὔχομαι ‘prayʼ, L. uoueō ‘vowʼ; cf. Pokorny IEW: 348; Kellens 1995: 9,14; LIV2
: x;
Cheung 2007: 169–170);
but without the Bartholomae´s Law:
nom. miϑrō.druxš ‘oathbreakerʼ (√druj- ‘lie, deceiveʼ; cf. pr. YAv. družaite; < IE
*√dh
reu̯gh
-; cf. OHG triugan ‘deceitʼ; cf. Pokorny IEW: 276; Kellens 1995: 32; LIV2
:
157; Cheung 2007: 80–81);
Again, the outcome of gh
dh
/jh
dh
clusters is not directly attested, being modelled according to
the assumed analogy with ‘Bartholomae’s clusterʼ gh
t/jh
t.
2.3.3 Clusters of palatovelar + t/dh
/s
The old IE palatovelars are realized as a palatal sibilant before obstruents and lost before a
sibilant (which is, according to Pedersen’s Law/the ruki-rule, usually palatalized). The final
original aspirate, if followed by a voiceless obstruent, is subjected to Bartholomae’s Law, if it
is not neglected by the analogy:
Ḱ + t = Av. št:
int. OAv. daēdōišt (√daēs- ‘showʼ, pr. YAv. daēsaiiən; < IE *√de ḱ-; cf. cf. Gr. δείκνυμι,
‘showʼ; L. dīcō ‘sayʼ; cf. Pokorny IEW: 188–189; Kellens 1995: 30; LIV2
: 108–109;
Cheung 2007: 51–52);
pr. OAv vaštī (√vas- ‘desireʼ, pr. OAv. vasəmi; < IE *√u̯ eḱ-; cf. Hitt. wēkzi ‘wishʼ; cf.
Pokorny IEW: 1135; Kellens 1995: 52; LIV2
: 672–673; Cheung 2007: 427);
34
ppp. YAv. pairi.aŋharšta- (√harz- ‘releaseʼ, pr. YAv. auuaŋhərəzāmi; < IE *√selǵ-; cf.
Hith. salk- ‘knead, mingleʼ, OHG selken ‘fall downʼ; cf. Pokorny IEW: 900–901;
Kellens 1995: 72; LIV2
: 528–529; Cheung 2007: 132–133);
pr. YAv. ište (√is- ‘be able, ruleʼ, pr. OAv. isāmaidē; < IE *√H2e ḱ-; cf. OCS iskati ‘seekʼ;
cf. Pokorny IEW: 16; Kellens 1995: 13; LIV2
: 260; Cheung 2007: 158);
ppp. YAv. uzvaršta- (√varz- ’workʼ, OAv. vərəziiāmahī; < IE *√u̯ erǵ-; cf. Gr. ἔργω
‘workʼ; cf. Pokorny IEW: 1168–1169; Kellens 1995: 51–52; LIV2
: 689–690; Cheung
2007: 425–427);
num. ašta ‘eightʼ (< IE *oḱt -; cf. OIA aṣṭáu, L. octō ‘eightʼ; cf. Pokorny IEW: 775;
Emmerick 1992b: 299–300; Blažek 1999: 263);
ǵh
+ t = Av. žd (št):
OAv. pr. gərəždā (√garz- ‘complain, pr. OAV. gərəzōi; < IE *√g(u̯ )
eRǵh
-; cf. OIA gṛhate
‘complainʼ; cf. Pokorny IEW: 350–351; Kellens 1995: 19–20; LIV2
: 187; Cheung
2007: 111–112);
OAv. nom. važdra- but without Bartholomae´s law: ppp. YAv. vašta-, nom. vaštar- ‘drag
animalʼ (?) (√vaz- ‘driveʼ, pr. YAv. vazaiti; < IE *√u̯ eǵh
-; < IE *√u̯ eǵh
-; cf. L. uehō,
OCS vezǫ ‘rideʼ; cf. Pokorny IEW: 1118–1119 ; Kellens 1995: 52; LIV2
: 661–662;
Cheung 2007: 429–432);
Ḱ + dh
= Av žd:
OAv. inf. āždiiāi (√(n)as/š- ‘reachʼ; cf. ao. OAv. nąsat̰; < IE *√H2neḱ-; cf. Gr. διηνεκής
‘continuousʼ; cf. Pokorny IEW: 316–317; Kellens 1995: 40–41; LIV2
: 282–283;
Cheung 2007: 183-184);
OAv. inf. mərąždiiāi (√marz- ‘rubʼ; cf. pr. OAv. marəzaiti; < IE *√H2merǵ-; cf. Gr.
ἀμέργω ‘pluck, pullʼ; cf. Pokorny IEW: 722–723; Kellens 1995: 44; LIV2
: 280–281;
Cheung 2007: 180–182);
ǵh
+ dh
= Av. *žd:
not attested;
Ḱ + s = Av. 0š/0s:
ao. OAv. dāiš, dōišā (√daēs- ‘showʼ, pr. YAv. daēsaiiən; < IE *√de ḱ-; cf. cf. Gr.
δείκνυμι, ‘showʼ; L. dīcō ‘sayʼ; cf. Pokorny IEW: 188–189; Kellens 1995: 30; LIV2
:
108–109; Cheung 2007: 51–52);
pr. OAv. vašī (√vas- ‘desireʼ, pr. OAv. vasəmi; < IE *√u̯ eḱ-; cf. Hitt. wēkzi ‘wishʼ; cf.
Pokorny IEW: 1135; Kellens 1995: 52; LIV2
: 672–673; Cheung 2007: 427);
inch. OAv. pərəsā, YAv. pərəse, ao. OAv. frašī (√fras- ‘askʼ, pr. pərəsaniieiti; < IE
*√preḱ-; cf. L. poscō ‘demandʼ, Toch. A prak-, B prek-, OCS prositi ‘askʼ; cf. Pokorny
IEW: 821–822; Kellens 1995: 35; LIV2
: 490–491; Cheung 2007: 88–90);
ft. YAv. harəšiiamna- (√harz- ‘releaseʼ, pr. YAv. auuaŋhərəzāmi; < IE *√selǵ-; cf. Hith.
salk- ‘knead, mingleʼ, OHG selken ‘fall downʼ; cf. Pokorny IEW: 900–901; Kellens
1995: 72; LIV2
: 528–529; Cheung 2007: 132–133);
pr. YAv. išē (√is- ‘be able, ruleʼ, pr. OAv. isāmaidē; < IE *√H2e ḱ-; cf. OCS iskati ‘seekʼ;
cf. Pokorny IEW: 16; Kellens 1995: 13; LIV2
: 260; Cheung 2007: 158);
OAv. ao. varəšā (√varz- ’workʼ, OAv. vərəziiāmahī; < IE *√u̯ erǵ-; cf. Gr. ἔργω ‘workʼ;
cf. Pokorny IEW: 1168–1169; Kellens 1995: 51–52; LIV2
: 689–690; Cheung 2007:
425–427);
35
ǵh
+ s = Av. 0ž (0š):
YAv. ao. uzuuažat (√vaz- ‘driveʼ, pr. YAv. vazaiti; < IE *√u̯ eǵh
-; < IE *√u̯ eǵh
-; cf. L.
uehō, OCS vezǫ ‘rideʼ; cf. Pokorny IEW: 1118–1119 ; Kellens 1995: 52; LIV2
: 661–
662; Cheung 2007: 429–432); but analogical without the Bartholomae´s Law ao. YAv. vašata
(? – cf. Narten 1964: 368 fn1; Kellens 1995: 52; Cheung 430) ;
Note: It should be mentioned that Kellens (1976) found examples for a minor development *Ḱt in Iranian xšt; Av.
paiti.fraxštar- ‘interrogatorʼ, yaoxšti ‘branchʼ, spaxšti- ‘visionʼ, Kellens considers this a proof that clusters from
Ḱ+t and š+t did not merge fully even in Proto-Iranian, which contrasts with the known merging of both clusters
not only in Indic but also in Baltic (and Slavic). Lubotsky (2018: 1884) follows such data as a proof of a
dialectal development in Eastern Iranian (similar reflexes are found in Sogdian and Bactrian). Bartholomae
(1895–1901: 36) considers the anlauting clusters of xšt- (not only those from *Ḱt-, since even YAv. xštāt̰
‘standsʼ has such a for ) as a prothetic consonant, which is acceptable only in a word-initial, but hardly for the
internal consonantal clusters (Kellens 1976: 68 rejects the idea of the prothesis as whole). To this question we
will return below.
As previously, the outcome of the ǵh
dh
cluster is not directly attested, being modelled according
to the assumed analogy with ‘Bartholomae’s clusterʼ of *ǵh
t.
2.3.4 Clusters of dental + t/dh
/s
The old IE dentals are realized as a non-palatal sibilant before obstruents and lost before a
sibilant. The final original aspirate, if followed by a voiceless obstruent, is subjected to the
Bartholomae’s Law, if it is not neglected by analogy:
T + t = Av. st:
ao. OAv. cistā (<√cōit- ‘observeʼ; cf. pf. cikaēθa; < IE *√(s)ku̯
e t-; cf. OCS čьtǫ ‘count,
readʼ; cf. Pokorny IEW: 636–637; Kellens 1995: 22; LIV2
: 382; Cheung 2007: 31);
pr. OAv. vīnastī, pr. YAv. vinasti (√vid- ‘findʼ; cf. YAv. viṇdəṇti; < IE *√u̯ e d-; cf. L. uīdī
‘seeʼ, Arm. egit ‘findʼ; cf. Pokorny IEW: 1125–1127; Kellens 1995: 54–55; LIV2
:
665–667; Cheung 2007: 409–410);
pf. OAv. vōistā (√vid- ‘knowʼ; cf. OAv. pf. vaēdā; < IE *√u̯ e d-; cf. L. uīdī ‘seeʼ, Arm.
egit ‘findʼ; cf. Pokorny IEW: 1125–1127; Kellens 1995: 54; LIV2
: 665–667; NIL:
717–722; Cheung 2007: 408–409);
ppp. YAv. xv
āsta (√hv
ad- ‘to become savouryʼ, OAv. nom. hudəma- ‘sweetness; < IE
*√su̯ed-; cf. OIA svádant ‘make tastyʼ, OE swēte ‘sweetʼ; cf. Pokorny IEW: 1039–
1040; LIV2
: 606–607; NIL: 670–672; Cheung 2007: 141);
ppp. YAv. xšusta (√xšud- ‘to become liquidʼ, YAv. nom. xšudra- ‘liquidʼ ;< IE *√kseu̯d-;
cf. OIA kṣódante ‘dissolveʼ; cf. Pokorny IEW: 625; LIV2
: 372; Cheung 2007: 455–
456);
dh
+ t > Av. zd (st):
pr. OAv. dazdā but YAv dasta (√dā- ‘putʼ; cf. impf. YAv. ādadat̰; < IE *√dh
eH1-; cf. Gr.
τίθημι ‘putʼ, Lith. desti; cf. Pokorny IEW: 235–236; Kellens 1995: 29; LIV2
: 136–137;
Cheung 2007: 45–46);
ppp. OAv. vərəzda- (√vard- ‘growʼ, OAv. pr. varədaitī; < IE *√Hu̯eRdh
-; cf. OIA
várdhate ‘growʼ; cf. Pokorny IEW: 1167; Kellens 1995: 51; LIV2
: 228; Cheung 2007:
208);
36
ppf. OAv. urūraost (√raud- ‘hinderʼ; cf. ps. YAv. ºraoδəṇti; < IE *√leu̯dh
-; cf. OIA
arodham ‘restrainʼ; cf. Kellens 1995: 59–60; LIV2
: 415; Cheung 2007: 317);
ppp. YAv. niuruzda- but without the Bartholomae’s law: YAv. urusta- (√raud- ‘grow biggerʼ;
cf. pr. raoδahe; < IE *√H1leu̯dh
-; cf. Gr. Hom. ἤλῠθον ‘come, startʼ; cf. Pokorny IEW:
306–307, 684–685; Kellens 1995: 59; LIV2
: 248–249; NIL: 245–246; Cheung 2007:
193–194);
but without the Bartholomae’s law:
ppp. YAv. busta- (√baod- ‘feel, senseʼ, part. ps. OAv. baodaṇt-; < IE *√bh
eu̯ dh
-; cf. Gr.
πεύθομαι ‘give noticeʼ, OCS bljudǫ ‘bewareʼ; cf. Pokorny IEW: 150–151; Kellens
1995: 39; LIV2
: 83–84; NIL : 36–37; Cheung 2007: 14–15);
T + dh
= Av. zd:
pr. imp. YAv. dazdi, OAv. mąz-dazdūm (√dā- ‘giveʼ; cf. OAv. pr. dadē; < IE *√deH3-;
cf. Gr. δίδωμι, L. dō, OLith. duosti ‘giveʼ; cf. Pokorny IEW: 223–225; Kellens 1995:
29; LIV2
: 105–106; NIL: 60–69; Cheung 2007: 43–45);
inf. OAv. vōizdiiāi, ao. OAv. frauuōizdūm (√vid- ‘knowʼ; cf. OAv. pf. vaēdā; < IE *√u̯ e d;
cf. L. uīdī ‘seeʼ, Arm. egit ‘findʼ; cf. Pokorny IEW: 1125–1127; Kellens 1995: 54;
LIV2
: 665–667; NIL: 717–722; Cheung 2007: 408–409);
dh
+ dh
= Av. zd:
inf. YAv. dazdiiāi (√dā- ‘putʼ; cf. impf. YAv. ādadat̰; < IE *√dh
eH1-; cf. Gr. τίθημι ‘putʼ,
Lith. desti; cf. Pokorny IEW: 235–236; Kellens 1995: 29; LIV2
: 136–137; NIL: 99–
11; Cheung 2007: 45–46);
T + s = Av. 0s:
YAv. nom. masiia- ‘fishʼ (< IE *√mad-s o-; cf. OIA mátsya-; cf. Pokorny IEW: 694 ;
NIL: 455–457);
OAv. nom. grəguuasū (< -t-su) ‘liableʼ (√x- ‘xʼ; cf. pr. x; < IE *√x-; cf. X x ‘xʼ; cf. Pokorny
IEW: x; Kellens 1995: x; LIV2
: x; Cheung 2007: x);
ao. OAv. sąs (√səṇd- ‘appearʼ, YAv. pr. saδaiiemi; < IE *√(s)ḱend-; cf. OIA chádayati
‘appearʼ; cf. Kellens 1995: 61; LIV2
: 546; Cheung 2007: 332–334
dh
+ s = Av. 0z:
(?) pr. YAv. uruuāza- (√uruuād- ‘be proudʼ; cf. Kellens 1995: 60; Cheung 2007: 438);
The sibilantization of all dental plosives before *t/*dh
is systematic (and shared with other IndoEuropean
languages but not with OIA). The development before *s could be considered as a
simplification of the original *ss-cluster.
2.3.5 Clusters of sibilant +t/dh
/s
Sibilants are, in contrast to Vedic, preserved in all clusters. The voiceless sibilants became
voiced before *dh
, while clusters of two sibilants are simplified on a single one (degemination):
*st = Av. st:
pr. OAv. astī, YAv. asti (√ah- ‘beʼ; cf. pr. OAv. ahmī; < IE *√H1es-; cf. Gr. ἐστί, L. est
‘beʼ; cf. IEW: 340–341; Kellens 1995: 10–11; LIV2
: 241; Cheung 2007: 151–152);
37
pr. YAv. vastē (√vah- ‘be dressedʼ; cf. pr. YAv. vaŋhata; < IE *√u̯es-; cf. Hitt. wēsta
‘wearʼ, Gr. ἕννῡμι, ἑννύω ‘put clothes onʼ, L. uestis ‘clothʼ; cf. Pokorny IEW: 1172–
1173; Kellens 1995: 53; LIV2
: 692–693; Cheung 2007: 405);
ppp. YAv. aiβi.sasta-, inf. ao. OAv. sasté (√səṇh- ‘declareʼ, pr. YAv. saŋhaite; < IE
*√ḱens-; cf. L. cēnseō ‘judgeʼ; cf. Pokorny IEW: 566; Kellens 1995: 62; LIV2
: 326;
Cheung 2007: 334–335);
*št = Av. št:
ao. OAv. cōišt, YAv. cōišta (√ciš- ‘assignʼ; cf. pr. OAv. cīšmahī < IE *√ku̯
e s-; cf. OIr.
ad-cí ‘seeʼ; cf. Pokorny IEW: 637; Kellens 1995: 22–23; LIV2
: 381–382; Cheung
2007: 30);
nom. OAv. būštīš ‘endeavoursʼ (√būš- ‘endeavourʼ; < IE *√bh
eu̯H2-s-; cf. Lith. bùs ‘will
beʼ; cf. Pokorny IEW: 146–147; Kellens 1995: 39–40; LIV2
: 98–101; Cheung 2007:
25–26);
ppp. YAv. paiti.θβaršta- (√θβars- ‘cutʼ; cf. pr. YAv. θβərəsaiti; < IE *√tu̯ers- ( ?); cf.
Pokorny IEW: 1102; Kellens 1995: 26; LIV2
: 656; Cheung 2007: 399–400);
*sdh
= Av. zd:
pr. OAv. zdī (√ah- ‘beʼ; cf. pr. OAv. ahmī; < IE *√H1es-; cf. Gr. ἐστί, L. est ‘beʼ; cf. IEW:
340–341; Kellens 1995: 10–11; LIV2
: 241; Cheung 2007: 151–152);
ao. OAv. θrāzdūm (√θrā- ‘protectʼ; cf. YAv. pr. θrāiieṇte; < IE *√treH-; cf. OIA trā́yate
‘protect, saveʼ; cf. Pokorny IEW: 1075; Kellens 1995: 27; LIV2
: 646; Cheung 2007:
394);
inf. OAv. sazdiiiāi23
(√səṇh- ‘declareʼ, pr. YAv. saŋhaite; < IE *√ḱens-; cf. L. cēnseō
‘judgeʼ; cf. Pokorny IEW: 566; Kellens 1995: 62; LIV2
: 326; Cheung 2007: 334–335);
*šdh
= Av. žd:
ao. OAv. θβarōždūm (√θβars- ‘cutʼ; cf. pr. YAv. θβərəsaiti; < IE *√tu̯ers- ( ?); cf.
Pokorny IEW: 1102; Kellens 1995: 26; LIV2
: 656; Cheung 2007: 399–400);
pr. (ao.?) OAv. cīždī (√ciš- ‘assignʼ; cf. pr. OAv. cīšmahī < IE *√ku̯
e s-; cf. OIr. ad-cí
‘seeʼ; cf. Pokorny IEW: 637; Kellens 1995: 22–23; LIV2
: 381–382; Cheung 2007: 30);
Av. inf. būždiiāi (√būš- ‘endeavourʼ; < IE *√bh
eu̯H2-s-; cf. Lith. bùs ‘will beʼ; cf. Pokorny
IEW: 146–147; Kellens 1995: 39–40; LIV2
: 98–101; Cheung 2007: 25–26);
*ss = Av. 0s:24
YAv. part. inch. (vī)usaitī- f. (√vah- ‘shineʼ, < IE *√u̯ es-; cf. Lith. au͂šti ‘break dawnʼ; cf.
Pokorny IEW: 86–87; Kellens 1995: 53; LIV2
: 292–293; Cheung 2007: 202);
ppp. (?) YAv. ustriiamna- (= us-√stər- ‘throw downʼ; < IE *√ster-; cf. L. prosternō ‘cause
to fallʼ; cf. Pokorny IEW: 1029–1030; Kellens 1995: 64; LIV2
: 597–598; Cheung 2007:
363–364);
*šs = Av. 0š/0s:
pr. inch. YAv. tusən (√tuš- ‘be emptyʼ; cf. pr. caus. YAv. -taošaiieiti; < IE *√teu̯s-; cf.
OCS tъštь ‘empty, vainʼ, Lith. tùščias ‘empty, poorʼ; cf. Pokorny IEW: 1085; Kellens
1995: 26; LIV2
: 642; Cheung 2007: 388–389);
23
This from could be from √saṇd- ‘appear”, cf. Cheung 334, but it seems to be doubtful. Kellens lists it as derived
from √səṇh- ‘declare” without any doubts (Kellens 62).
24
But from √ah- ‘be” is 2nd
sg. OAv. ahī, YAv. ahi, probably because the cluster of *ss was merged into *s already
in IE, cf. OIA asi from √as- ‘be”, cf. Hoffmann/Forssmann: 109.
38
Again, the cluster of two sibilants is simplified, as it was in case of (secondary) sibilants from
dentals and palatovelars + s (cf. above).
2.3.6 The overview of the Avestan development
See the following table for a summary of the attested alternations of the Avestan clusters formed
either by any plosive + t/dh
/s or a sibilant + t/dh
/s. Not attested forms are omitted, and the forms
clearly formed by an analogy (especially those parallel to forms formed according to the
Bartholomae’s Law) are marked by round brackets:
IE Avestan t- d(h)
- s-ḱ/ǵ
-s/z št žd 0š
(0s)
-ǵh
-z žd
(št)
0ž
(0š)
-k(ṷ)
/g(ṷ)
-k/c/g/j xt gd xš
-g(ṷ)h
-g/j gd
(xt)
γž
(xš)
-t/d -t/d st zd 0s
-dh
-d zd
(st)
zd 0z
(0s)
-p/b -p/b pt
(*ft)
bd fš
(fs)
-bh
-b bd
(pt)
βž
(fš, fs)
-s -s st zd 0s
-š -š št žd 0š
(0s)
2.4 The development of two-obstruent clusters in Old Persian
Old Persian data are far more scarcely attested than those of Avestan (or Vedic) but offer some
useful parallels to those of Avestan. The remarkable advantage of Old Persian is its antiquity;
the remarkable disadvantage is the attested corpus, which is far more limited than that of
Avestan or Vedic. For these reasons, the Old Persian data are overshadowed by those of
Avestan.
The whole development of Old Persian is in its nature very similar and close to that of
Avestan, as just the original IE palatovelars are realized as follows: *ḱ > OP ϑ, *ǵ and *ǵh
> OP
d.
In its general features Old Persian follows same trajectories as Avestan do, we should
mention specially that:
39
i. the originally voiced non-aspirates and voiced aspirates are neutralized as voiced nonaspirates
as in Avestan;
ii. the dentals are sibilantized before dentals as happens in Avestan;
iii. the left plosives of the clusters are always neutralized according to the right plosive or
sibilant in voice. There are no traces of Bartholomae’s Law, which was probably replaced
by an analogy, similar to processes of leveling known from Avestan, being possibly
regular in Old Persian (this does not affect, of course, regular clusters with the dominance
of the right voiced element *dh
-, where the voiced nature of the outcome cluster is regular
and in accord with the routine dominance of a right plosive over a left one).
P + t = OP *ft:
*haftā (reconstructed on np. haft, Yaghnobi aft, avt etc.; < IE *septm̥ -; cf. Av. hapta-, Gr.
ἑπτά, L. septem ‘sevenʼ; cf. Pokorny IEW: 909; Brandenstein/Mayrhofer 1964: 37;
Emmerich 1992b: 299; Blažek 1999: 246);
K + s = OP xš:
nom. xšaça- ‘ruleʼ (cf. Av. xšaϑra- ‘rule, controlʼ, OIA. kṣatrā ‘powerʼ; cf. Kent 1950:
181; Pokorny IEW: 626; Brandenstein/Mayrhofer 1964: 126; LIV2
: 618–616; Cheung
2007: 451–452; Brust 2018: 165–166);
gh
+ t = OP xt:
nom. duruxta- (√du
ruj- ‘lieʼ; cf. impf. adu
rujiya; < IE *√dh
reu̯gh
-; cf. Av. √druj-, OE
driogan ‘deceitʼ; cf. Kent 1950: 192; Pokorny IEW: 276; Brandenstein/Mayrhofer
1964: 117; LIV2
: 157; Cheung 2007: 80–81; Brust 2018: 227–228);
Ḱ + t = OP št/st:
ppp. ufrasta-/ufrašta- (√fraϑ- ‘punish, ask, inquireʼ, ps. fraϑiyaiš; < IE *√preḱ-; cf. Av.
√fras- ‘inquireʼ; cf. Kent 1950: 198; Pokorny IEW: 321–322; Brandenstein /Mayrhofer
1964: 119; LIV2
: 490–491; Cheung 2007: 88–90; Brust 2018: 260);
ppp. ni-pišta-, inf. ni-pištanayi (√paiϑ- ‘cut, engrave, adornʼ; cf. impf. ps. x
niyapiϑiya; <
IE *√pei̯ ḱ-; cf. Av. √paēs- ‘adornʼ, L. pingī ‘paintʼ, OCS pьšǫ ‘writeʼ; cf. Kent 1950:
194; Pokorny IEW: 794–795; Brandenstein/Mayrhofer 1964: 136; LIV2
: 465-466;
Cheung 2007: 291–292; NIL: 546–548; Brust 2018: 239–240);
adj. arštā f. ‘justiceʼ, adj. rāsta- ‘rightʼ (cf. Av. arštāt f. ‘Iustitia’, Av. rāšta- ‘ordered’;
< IE *√H3reǵ -; cf. L. regō ‘ruleʼ; cf. Kent 1950: 206; Pokorny IEW: 854–855;
Brandenstein/Mayrhofer 1964: 106, 141; LIV2
: 304–305; Cheung 2007: 196–197;
Brust 2018: 123–124, 298);
Ḱ + s = OP 0s/0š:
pr. inch. p(a)rsāmiy, ap(a)rsam (√fraϑ- ‘punish, ask, inquireʼ, ps. fraϑiyaiš; < IE
*√preḱ-; cf. Av. √fras- ‘inquireʼ; cf. Kent 1950: 198; Pokorny IEW: 321–322;
Brandenstein /Mayrhofer 1964: 119; LIV2
: 490–491; Cheung 2007: 88–90; Brust
2018: 260);
ao. niya-paišam (√paiϑ- ‘cut, engrave, adornʼ; cf. impf. ps. x
niyapiϑiya; < IE *√pei̯ ḱ-; cf.
Av. √paēs- ‘adornʼ, L. pingī ‘paintʼ, OCS pьšǫ ‘writeʼ; cf. Kent 1950: 194; Pokorny
IEW: 794–795; Brandenstein/Mayrhofer 1964: 136; LIV2
: 465-466; Cheung 2007:
291–292; Brust 2018: 239–240);
40
T + t = OP st:
nom. pastiš ‘foot soldierʼ (= *ped-ti-; < IE *√ped-; cf. Av. pad-, L. pēs ‘footʼ; cf. Kent
1950: 197; Pokorny IEW: 790-792; Brandenstein/Mayrhofer 1964: 138; Brust 2018:
254–255);
nom. ustašanām f. ‘staircaseʼ (= ud-√taϑ- ‘workʼ, < IE *√tetḱ-; cf. Gr. τέκτων ‘carpenterʼ;
cf. Kent 1950: 178, 185; Pokorny IEW: 1058–1059; Brandenstein/Mayrhofer 1964:
148; LIV2
: 638–639; Cheung 2007: 384–385; Brust 2018: 151);
dh
+ t = OP st:
without the Bartholomae’s Law (due to analogy?)
ppp. basta- (√band- ‘bindʼ, pr. YAv. baṇdāmi; < IE *√bh
endh
-; cf. Goth. band ‘bindʼ; cf.
Kent 1950: 199; Pokorny IEW: 127; Brandenstein/Mayrhofer 1964: 110; Kellens
1995: 37; LIV2
: 75–76; Cheung 2007: 4–6; Brust 2018: 265);
ppp. gasta- ‘evil, repugnantʼ (√gant- ‘stinkʼ; < IE *√gu̯
edh
-; cf. OIA gandháḥ ‘smellʼ; cf.
Kent 1950: 183; Pokorny IEW: 466–467; Brandenstein/Mayrhofer 1964: 121; Cheung
2007: 103–104; Brust 2018: 177–178);
dh
+ dh
= OP zd:
adv. azdā (?) ‘knownʼ (cf. Av. azdā, OIA addhā; but it could be either from adh
-tā or adh
tā
or ad-dā, < IE *√H2edh
- (?); cf. OIA āha ‘sayʼ; cf. Kent 1950: 173–174; Pokorny
IEW: 291; EWAi 1: 64, Brandenstein/ Mayrhofer 1964: 109; LIV2
222; Lipp 2009b:
87; Cheung 2007: 153; Brust 2018: 132–133);
s + t = OP st:
pr. astiy (√ah- ‘beʼ; cf. impf. āham; < IE *√H1es-; cf. OIA asti, Gr. εἰμί, L. est ‘beʼ; cf.
Kent 1950: 174; Pokorny IEW: 340–341; Brandenstein/Mayrhofer 1964: 101; LIV2
:
241; Cheung 2007: 151–152; Brust 2018: 133);
caus. impf. avāstāyam (√stā- ‘standʼ; < IE *√steH2-; cf. OIA tiṣṭhati, Gr. ἵστημι ‘standʼ;
cf. Kent 1950: 210; Pokorny IEW: 1004–1005; Brandenstein/Mayrhofer 1964: 146;
LIV2
: 590–591; Cheung 2007: 358–361; Brust 2018: 311–313);
š + t = OP št:
nom. dauštā ‘friendʼ (√x- ‘xʼ; cf. pr. x; < IE *√x-; cf. X x ‘xʼ; cf. Kent 1950: 189; Pokorny
IEW: 399–400; Brandenstein/Mayrhofer 1964: 116; Cheung 2007: 473; Brust 2018:
209);
impf. aištatā, caus. niyaštāyam, frāstāyam (√stā- ‘standʼ; < IE *√steH2-; cf. OIA tiṣṭhati,
Gr. ἵστημι ‘standʼ; cf. Kent 1950: 210; Pokorny IEW: 1004–1005;
Brandenstein/Mayrhofer 1964: 146; LIV2
: 590–591; Cheung 2007: 358–361; Brust
2018: 311–313);
s + dh
= OP zd:
nom. Auramazdā m. (= aura-+mazdā; cf. Av. Ahurō mazdā̊ ; mazdā = *mas-+ dh
ā < IE
*mn̥ s- + *√dh
eH1; cf. OIA medhā ; cf. Kent 1950: 165, 188; Pokorny IEW: 235–236;
EWAi 2: 378, Brandenstein/Mayrhofer 1964: 108 ; LIV2
: 136–137; Cheung 2007:
45–46; Brust 2018: 93–95);
To sum up, for the attested Old Persian development, see the following table. As in the case of
Old Indo-Aryan or Avestan above, round brackets mark the attested forms given by an analogy
(not attested forms are not reconstructed here):
41
IE OP t- d(h)
- s-ḱ/ǵ
-ϑ/d št
(st)
0š
(0s)
-ǵh
-z
-k(ṷ)
/g(ṷ)
-k/c/g/j xš
-g(ṷ)h
-g/j (xt)
-t/d -t/d st
-dh
-d (st) zd
-p/b -p/b *ft
-bh
-b
-s -s st zd
-š -š št
2.5 The development of Nūristānī clusters
As we have said above, the Nūristānī data are not principally on a similar level as the above mentioned Vedic,
Avestan or Old Persian data, especially since we are dealing with living dialects, not with a long-dead language.
It is hence not of the age comparable with Old Indo-Iranian languages examined above, but a result of more than
a thousand years of development necessarily affecting the state of the data. However, and surprisingly, due to the
political situation in the area, the data are scarce, and furthermore, from a very ‘thinʼ source, since the linguistic
description of Nūristānī languages is sketchy and incomplete25
.
For these reasons, we will treat Nūristānī as a background to data from Old Indo-Iranian languages, with
its credibility well beyond them. Moreover, there is another important distinction: the OIA, Avestan, even Old
Persian clustering-data in general are taken from synchronic situations, i.e., from the living synchronic alternations
on a contact of morphs, but Nūristānī data are etymological in general, reflecting synchronically closed situations.
Comparing Nūristānī to Old Persian, we can state that Old Persian is both ancient and relict, Nūristānī then modern
and evasive, for our use both languages are necessarily fragmentary, but this fragmental nature is different for Old
Persian (being by its archaicity and structural composition closer to other ancient Indo-Iranian languages) than for
Nūristānī (which is sketchy and more than two thousand years younger than any of the ancient Indo-Iranian
languages we dealt with until now).
It should be noted that the independent status of Nūristānī as the third branch of Indo-Iranian is often
disputed and though Morgenstierne (1965) and others support the third-branch status (cf. Mayrhofer 1951: 15;
Fussman 1972: 391; Nelson 1986: 104–116; Kausen 2012: 662–663), others stand either for Indian (cf. Bloch
1965: 54; Budruss 1977: 33; Degener 2002; Blažek/Hegedüs 2012; Werba 2016) or Iranian origin of Nūristānī
(Mayrhofer 1984; Mayrhofer 1997: 107–108; Lipp 2009a: 156–157).
From attested Nūristānī data (quotations of sources are given directly with data) we can assume the following
development of old clusters plosive + t, plosive + dh
, plosive + s (the reconstructed forms are approximative, for
the reasons stated above, each symbol marks the whole local series):
25
The first survey by Morgenstierne (1926) was done from Kabul for safety reasons, which did not improve at all
for another century.
42
P + t = N. 0t:
Ashkun nōt, Prasun natī, nätix, nətik, Kati, Waigali nūt ‘little girlʼ (< IE *nept-; cf. OIA naptī, OAv. nafšu;
cf. Fussman 1972: 271–272; Pokorny IEW: 746; EWai II: 11–12);
Kati lot, Waigali lāt ‘peaceʼ, Kalaṣa/Waigali lātoy ‘he has foundʼ (< Indo-Iranian *√labh
-ta-;26
< IE
*√lembh
-; cf. OIA labdhá-; cf. Nelson 1986: 99; Degener 2002: 109; Pokorny IEW: 652; EWai II: 434;
LIV2
: 411–412);
Kati sut, Waigali sōt, Ashkun sūt (< IE *septm̥-; cf. OIA sapta-, Av. hapta ‘sevenʼ; cf. Gr. ἑπτά, L. septem
‘sevenʼ; cf. Nelson 1986: 99; Pokorny IEW: 909; Blažek 1999: 246; EWai II: 700; Lipp 2009a: 158;
Werba 2016: 347);
K + t = N. 0t:
Kati yit ‘a pairʼ (< IE *√ eu̯ g-; cf. OIA yuktá-, yukti, YAv. yuxta; Nelson 1986: 99; Pokorny IEW: 508–
509; EWai II: 417; LIV2
: 316; Cheung 2007: 217–218);
gh
+ t = N. 0g:
Prasun ḍogū, ḍūgu, ḍuge ‘milkʼ (< IE *dh
eu̯ g- (?); cf. OIA dugdhá-; Fussman 1972: 200–201; Pokorny
IEW: 271; EWai I: 747–748; LIV2
: 153; Cheung 2007: 66–67);
Note: The outcome of the cluster gh
t seems to be: gh
t > gdh
> gg(h)
with atypical assilation of location (?).
Ḱ + t = N. ṣṭ:
Kati (w)uṣṭ, Waigali oṣṭ, Ashkun ōṣt, Prasun āstë, Tregani wūṣṭ ‘eightʼ (< IE *oḱto-; cf. OIA aṣṭáu, Av.
ašta, L. octō; cf. Fussman 1972: 196–197; Nelson 1986: 58; Pokorny IEW: 775; Blažek 1999: 263;
EWai I: 142);
T + t = N. 0t:
Kati ptå, Kalaṣa/Waigali pratoy ‘he has givenʼ (< *pra-tta- < IE *√deH3-to-; cf. Morgenstierne 1926: 60;
Degener 2002: 105);
Kati, Waigali, Prasun čit ‘aimʼ (< IE *√(s)ku̯
e t-; cf. OIA cittá-, Av. cisti-; cf. Lipp 2009a: 169; Werba 2016:
349; Pokorny IEW: 636–637; LIV2
: 382; EWai I: 547; Cheung 2007: 31);
dh
+ t = N. ṛ/0d:
Ashkun būṛə ‘mind, spiritʼ, Waigali buṛā́, buṛṓk ‘meaning, intentʼ (< *būḍhi < *būṣḍhi < *bh
udz
dhi < IE
*bh
udh
-ti; cf. Turner 1964; Turner 1966: 525; Hill 2003: 44–45), but cf. Kati bədi ‘mindʼ, Prasun büdü,
büt ‘sense, mindʼ (cf. Degener 2002: 105; Lipp 2009a: 169; Morgenstierne 1926: 60 assumes it a
loanword); Ashkun batu ‘understoodʼ (< *budh
-ta; cf. OIA buddhá-, YAv. busta-; < IE *√bh
eu̯ dh
-; cf.
Degener 2002: 109; Lipp 2009a: 169; Pokorny IEW: 150–151; EWai II: 233; LIV2
: 83–84; Cheung
2007: 14–15);
The above quoted Nūristānī form of *dh
t has cerebralization (from older palatalization ?) more probably not due
to Pedersen’s Law, but due to analogy or some later process (the parallel froms with 0d are allegedly from similar
formations!), since it would be a singlular example of working of the ruki-rule on a sibilant from an original dental.
The extraordinariness of the process is clear especially from the fact that from cluster Tt has Nūristānī an an
outcome not ṭ, but t; cf. examples above.
K + s = N. č:
Ashkun čuʾai ‘this nightʼ, c̣ū, čū ‘last nightʼ (< IE *√k(u̯)
sep-; cf. OIA kṣáp- ‘nightʼ; YAV. xšap ‘darknessʼ;
cf. Pokorny IEW: 649; EWai I: 424);
Ashkun aʾci, acī, Kati achē͂, ācē ‘eyeʼ, Prasun ižʾi, ižī, Waigali ačʾē͂, ačʾē, Tregami ac̣ē͂ ‘eyeʼ (< IE *H3eku̯
-;
cf. OIA akṣi-, Av. aši ‘eyeʼ (cf. Fussman 1972: 248; Hegedüs 2012: 151–153; Pokorny IEW: 775–777;
EWai I: 42–43);
Kati čuŕi, c̣ūŕi, ču(ŕ)ī, Waigali čō̆i ‘sickleʼ (< IE *√kseu̯-ro; cf. OIA kṣura- ‘sickleʼ, Fussman 1972: 155–
156; Hegedüs 2012: 151; Pokorny IEW: 586; EWai I: 435–436);
26
Not attested in Iranian.
43
Ḱ + s = N. c:
Ashkun daʾcun, Kati daciē͂ ‘rightʼ (< IE *deḱs-in-o-; cf. OIA dakṣiṇa-, Av. dašinō ‘rightʼ; cf. Nelson 1986:
82; Degener 2002: 105; Lipp 2009a:150–151, 155; Hegedüs 2012: 148–149; Werba 2016: 351; Pokorny
IEW: 190–191; EWai I: 690–691);
Ashkun kūc ‘middle; bellyʼ, Kati kǖc, kuc, kyǖc ‘stomach, bellyʼ, Prasun abʾuc (?) ‘side (direction)ʼ (< IE
*(s)keu̯ḱs-; cf. OIA kukṣí- ‘belly, wombʼ, Lith. kūšỹs ‘female pubic hairʼ; Hegedüs 2012: 148–149;
Pokorny IEW: 953; EWai I: 360–361);
T + s = N. ć:
Ashkun mōċ, Kati. maċi, Tregani mə̄ċ, Waigali maċë, maċa ‘fishʼ (< IE *√mad-s o-; cf. OIA mátsya-; cf.
Pokorny IEW: 694; Fussman 1972: 279–280; EWai II: 297–298; Lipp 2009a: 150, 169);
Waigali ūċ ‘sourceʼ (< Indo-Iranian *ud-sa-; cf. OIA útsa- ‘sourceʼ; Fussman 1972: 345–346; Pokorny
IEW: 78–79; EWai I: 213);
s + t = N. st/št:
Kati mrestə ‘corpseʼ, Waigali mo͂sta, Ashkun mərəsta ‘deadʼ (< Indo-Iranian *mṛta-sta-< IE *√mer-; cf.
Nelson 1986: 58; Pokorny IEW: 735; EWai II: 318–319; LIV2
: 439–440; Cheung 2007: 264–265);
Kati dušt, Waigali došt, Ashkun dōšt ‘handʼ (< IE *ǵh
est-to-; cf. OIA hastá-, Av. zasta, OP. dasta-; cf.
Pokorny IEW: 447; Nelson 1986: 58, 74; EWai II: 812);
š + t = N. ṣṭ/0ṭ:
Kat. uṣṭ, Waigali ūṣṭ, Ashkun ōṣt ‘mouthʼ (< IE *Heu̯ s-tH-; cf. OIA óṣṭha-, YAv. aošta, ‘mouth, lipʼ; cf.
Pokorny: 785; Nelson 1986: 58; Pokorny IEW: 785; EWai I: 282–283);
but with the outcome ṭ:
Kati pṭī Waigali yā-paṭī (but Ashkun piṣṭī) ‘backʼ (cf. OIA pṛṣṭhá-, YAv. paršta- ‘backʼ; cf. Morgenstierne
1926: 63; Pokorny IEW: 735; EWai II: 165–166);
Kati kāṭ ‘branchʼ (< IE *k(u̯ )
olsth
o- (?); OIA kā́ṣṭha-; cf. Morgenstierne 1926: 63; EWai I: 354–346);
š + dh
= N. (ž)d:
Kati piždå, pidå ‘avalancheʼ (< IE *pi-sd-; cf. OIA pīḍā- ‘damageʼ; cf. Morgenstierne 1926: 61; Pokorny
IEW: 887; EWai II: 136–137);
Bartholomae’s Law was either not functional with the precursor of Nūristānī, or its results were leveled, as
happened in Old Persian or in some cases in Avestan or Vedic (cf. Aṣhkun batu ‘understoodʼ but Kati bədi ‘mindʼ
above; both are traced to the root √*budh
- + t-).
An interesting feature of Nūristānī is the different outcomes of clusters K + s and of Ḱ + s: though both
are realized as affricates, the results probably affected by later developments (cf. Morgenstierne 1926: 58–59;
Nelson 1986: 84; Degener 2002: 105; Hegedüs 2012: 148–153). Indic has a single outcome for both clusters,
Avestan has two outcomes; cf. above).
The outcomes of s + t, š + t and Ḱ + t seems to be reworked in the subsequent development, it seems that
the ruki-rule was either never-functional or its result were leveled later (cf. Nelson 1986: 96–98; Degener 2002:
104), the second solution seems to be most probable, since similar process also affected Middle Indo-Aryan. The
confusion of št, st could be probably a result of a secondary palatalizations (Nelson 1986: 94).
Hegedüs states that IE *ḱs does not trigger the shift of *s to *š (Ashkun daʾcun, Kati datziē͂ ‘rightʼ; cf.
OIA dakṣina-, Av. dašinō ‘rightʼ; cf. Hegedüs 2012: 148–149), in contrast with Indic and Iranian. The outcome of
IE *ḱs in Nūristānī is an affricate *c, but the distribution of them is not clear even in the same dialect. However,
this outcome is usually different from the outcome of IE *ks, which results in Nūristānī *č (see above). It seems
very improbable the IE *ḱs would not undergo the ruki-development, so the outcome has to be result of a some
later secondary depalatalization.
44
Nūristānī seems to follow a similar trajectory in the development of clusters of two obstruents to that
which Indo-Aryan does, since all two-plosive clusters are reduced to a single plosive, the single exception being
Ḱt > N. ṣṭ (with št as a probably intermediate state). Indo-Aryan followed the trajectory of gemination of
heterogenous clusters in Middle Indo-Aryan (T1T2 > T2T2; cf. OIA bhuktá - > Pāli bhutta), followed by later
simplification of geminates. In this feature Nūristānī differs from the Iranian development, which followed (at
least at its earliest stage) the trajectory of spirantization of the peripheral clusters Pt, Kt etc. (cf. Degener 2002:
104, 109)27
. On the other hand, this ‘Indicʼ development did not affect clusters with a sibilant (either an old one or
resulting from an original palatovelar), which are preserved, contrary to the development in MIA, where OIA ST
regularly gives MIA TTh
(cf. OIA asti ‘isʼ, aṣṭau ‘eightʼ vs. Pā. atthi, aṭṭha).
The attested Nūristānī outcomes could be summed in the following table:
IE Nūr. t- d(h)
- s-ḱ/ǵ
-ϑ/d ṣṭ c
-ǵh
-z
-k(ṷ)
/g(ṷ)
-k/c/g/j 0t
-g(ṷ)h
-g/j 0g(?)
-t/d -t/d 0t ć
-dh
-d ṛ
-p/b -p/b 0t
-bh
-b
-s -s st
št
-š -š ṣṭ
0ṭ
žd
0ḍ
2.6 The trajectories of the Indo-Iranian developments
To describe the trajectories of the development of two-obstruent clusters, we need to explain
all changes and exceptions as fully as possible within the known Indo-Iranian and IndoEuropean
contexts.28
We used the term ‘trajectoriesʼ not only to express the proper multitude of the
trajectories in their number (i.e the trajectory of velars, the trajectory of dentals, etc.), but to
emphasize the possibilities of developments from the input, i.e., from the Indo-Iranian state of
arts, given by the reconstruction, to the output, i.e., to the given states as attested in the Old
Indo-Iranian languages.
27
The same trajectory is valid in Nūristānī as in Indic, for the dental cluster of TT.
28
It should be remarked that traditional reconstruction generally followed the thesis of the innate archaism of
Sanskrit, hence any difference from Sanskrit in any Indo-European language was primarily considered an
aberration. With two centuries gone since the ‘invention of Sanskrit” by the newly founded Indogermanistik,
the psychological dominance of OIA is lesser and its archaic status could be rightly doubted, since the ‘archaic”
state of OIA could often be just a phantom, a result of a methodological prejudice based on the mental inertia,
which could be demonstrated especially by the development of the Tt clusters, as we dare to demonstrate below.
45
2.6.1 Modelling the Indo-Iranian obstruent phonemes
We reconstruct the following inventory of Indo-Iranian obstruents:
plosives sibilants
palatovelars29
: *ḱ *ǵ *ǵh
(plain) velars30
: k g gh
palatals31
: c j jh
š (ž)
dentals: t d dh
s (z)
labials: p b bh
To our reconstruction model, we have to add that:
i. the question of the reconstruction of the palatovelars will be dealt with below in the section
devoted to this: ‘palatovelarsʼ. We use sibilants to mark given Indo-Iranian phonemes for
simplicity, though today is more often to reconstruct them as affricates, at least for the
first phase (the second being usually sibilants). For simplicity (and to avoid the debatable
Indo-Iranian reconstruction) and the historical reasons the reconstructed palatovelars are
present here as plosives;
ii. we have willingly omitted the question of voiceless aspirates in our reconstruction, though
we assume their existence for the Common Indo-Iranian stage; this omission is purely
due to lack of their development in the present paper. We suppose (in contrast with
Lubotsky 2018: 1876, 1879, who directly rejects to accept the existence of the voiceless
aspirates on the Indo-Iranian level) that at least three voiceless aspirates (kh
, th
, ph
) could
be securely reconstructed on the common Indo-Iranian level (though it is possible such
phonemes were in fact spirants), but there is no secure reconstruction leading either
towards śh
nor ch
(OIA ch
is a result of an independent, specially Indic development of
the IE cluster *sḱ, paralleled by Iranian s; cf. OIA gácchati vs. YAv. jasaiti, both from
Indo-Iranian *√gam- ‘goʼ or OIA pṛccháti vs. OAv. pərəsaētē, YAv. pərəsaitē̆ from IndoIranian
√praś- ‘askʼ).
However, we suppose that though the states of realization of our clusters differ in Old Indic and
Old Iranian languages (and Nūristānī as well), they are the results of the regular developments
in each branch, being the result of a development of the common Indo-Iranian state, i.e., that
each difference (or preservation of an older state) between Indic and Iranian is a result of a
further development of a given branch since Common Indo-Iranian, but never the result of a
different development before the Common period.
29
For simplicity, the term ‘palatovelar” will be used, though since we do not consider given set of phonemes to be
‘velar” in its phonetic form on the Common Indo-Iranian level at all, the term will be used purely as a
convenient (and unmistakable) term, which helps to avoid a possible confusion resulting from the use of terms
‘older” and ‘younger” palatals.
30
The result of merging of IE labiovelars and plain velars.
31
Until the distinction between IE *e and *o (and *a) was preserved, velars and true palatals were just allophones
of a singular series of velars, since complementary distributed according to a (non-)palatal context.
46
The points of difference between Old Indic and Old Iranian in the development of clusters of
the type obstruent +t/s/dh
- could be summarized thus:
i. There is a spirantization of the left voiceless plosive before a voiceless obstruent for velar/
palatal and labial series in Iranian, but this feature is not known from Indic;
ii. There is a sibilantization of the left dental plosive before any plosive (either voiceless or
voiced) in Iranian, but again this feature is not known from Indic, where dentals are
either realized as a plosive, or as a null before voiced aspirates;
iii. Though the original IE palatovelars are realized as depalatalized sibilants in antevocalic
positions in Iranian, the same palatovelars are realized in the same position either as
palatalized sibilants (for *ḱ) or affricates (for *ǵ, *ǵh
) in Indic; palatovelars are realized
as voiceless plosives before t and s or either as ḍ or 0 before the voiced plosive in Indic
but as voiceless or voiced sibilants (according to the context) in Iranian;
iv. The original labial and velar (and palatal) plosives are realized as spirants before IE *s
(including Indo-Iranian *š) in Iranian, though in Indic they are realized as plosives in
the same environments;
v. The original palatalovelar and dental plosives are lost before s in Iranian, but are realized as
plosives in Indic;
vi. Bartholomae’s Law does not affect the clusters formed by a voiced aspirate and *s/š in Indic,
though it is fully functional for same clusters in Iranian (except in cases corrupted by
the analogy: ppp. YAv. anādruxta- instead of the expected regular -γža-);
vii. The sibilants *s/š are realized either as a voiced plosive or null before voiced plosives in
Indic, but are uniformly realized as voiced sibilants in Iranian;
viii. Original sibilants *s/š are realized before a sibilant as a plosive in Indic32
, but Iranian
sibilants are lost before sibilants.
(ix. Indic has cerebral ṣ where Iranian has š; dental plosives following OIA ṣ are subsequently
cerebralized.33
)
2.6.2 Towards the analysis of the Indo-Iranian clusters
Within the analysis of the Indo-Iranian languaes, we have to distinguish three contexts:
i. the context of t- (could be affected by Bartholomae’s law, if the left phoneme is a voiced
aspirate);
ii. the context of dh
- (the resulting cluster is usually the same as that of Dh
+ t affected by
Bartholomae’s Law);
iii. the context of s- (could be also affected by Bartholomae’s law under the same circumstances,
the sibilant could be *š if in the context of Pedersen’s Law);
Into this context enter the following blocks:
i. the central/acute block, including both dentals (including OIA ‘cerebralsʼ) and original IE
palatovelars, which will be here dealt with as with plosives, according to their origin and
function;
32
Speaking about verbs, there are different outcomes for sibilants in the noun-declension of s-stems, see below.
33
For the sake of continuity, we will stick to the traditional terms cerebral and cerebralization instead of retroflex
and reflexivization or cacuminal and cacuminalization, and for the same reason we will use the traditional
marking with the subscribed dot (ṣ, ṭ, ḍ(h)
, ṇ).
47
ii. The peripheral/grave block containing labials and velars (including old IE labiovelars and
Indo-Iranian palatalized velars, since the latter are neutralized on plain velars in the
examined contexts.);
iii. the sibilant block formed by the preserved IE sibilant *s and *š, arising from it due to
Pedersen’s law (the ruki-law).
The development of the central block differs from the development of the peripheral block. We
can safely suppose that the development of both its series precedes the development of series
of the peripheral block, since the developments of the central block are shared either with all
IE branches (the development of dental series being the best example) or with all satəm
languages (for the development of original palatovelar series). However, before we examine
the trajectories of development of both blocks and the sibilant block, we have to mention the
specifics related to the forming of clusters of two obstruents in Indo-Iranian languages34
.
Note: In the following lines, the inputs and outputs of diachronic processes will be marked bold; the intermediate
states are without marking. The symbol > is used for regular (i.e., according to given sound laws) developments,
the symbol → for analogical developments.
2.6.2 The cluster-forming and Bartholomae’s Law
A typical feature of the development of the two-obstruent clusters is their equalization of the
modal features, which is usually in its final value given by the value of the right obstruent, with
the left one is assimilated to it.
The exception to this rule is well attested in the old Indo-Iranian languages but very
doubtful outside of this language family. The scope of this exception affects clusters formed by
the left voiced aspirate (Dh
) and the right standing voiceless obstruent (T, S), in this case the
output is DDh
for the input cluster Dh
T and D(h)
Z for the input Dh
S. The model presented here
is based on the generalization of the processes operational in Indic (where Dh
T > DDh
but Dh
S
> TS without the operational Bartholomae’s Law) and Iranian (where Dh
T > DD/ÐÐ and Dh
S
> DZ), so the generalization is hence nothing more than a model, not a directly attested as such
in any language.
The outcome of the cluster Dh
T is shared with the outcomes of the clusters Dh
Dh
and
TDh
both in Indic and Iranian, giving in Indic DDh
and DD in Iranian.
Note: From the above described full operational status of Bartholomae’s Law on the two-plosive cluster, we can
see pr. dhattá, derived from √dhā- ‘putʼ, as a sole exemption, having arisen probably due to analogy with dattá
etc., from √dā- ‘giveʼ.
Note: A remarkable feature is that the Indic outcome has aspiration only on the right plosive, not on both (†Dh
Dh
),
the nature of the Indic voiced aspirates clearly blocking the presence of the aspiration before the right standing
34
There are no secure examples of Bartholomaeʼs Law in Nūristānī.
48
plosive. This is not the case of the Greek voiceless aspirates, resulting from the IE voiced aspirates, since in
Greek the clusters formed by two voiceless aspirates Th
Th
are possible and even a natural outcome of the
clustering of T/D + Th
(see details in the chapter on Greek; cf. Gr. ἐτρέφθην vs. ἐτράπην).
Iranian, beside the assumed loss of aspiration is more close to the reconstructed Indo-Iranian
state, since the law is fully operational on the s-clusters, unlike in Indic, where for this cluster
we have to presume the secondary leveling (as it often happen in Iranian, too, as we saw above
in examples). What is worth of reconsidering is if the original result in Indo-Iranian was not
really DZh
, analogical to DDh
; the final aspirated voiced sibilant is reconstructed by Burrow
(1955: 94) or Lipp (2009a: 172). The outcome, i.e., the aspirated voiced sibilant, is not directly
attested in any Indo-Iranian language (neither is a voiceless aspirated sibilant attested, nor any
aspirated sibilant in other old IE languages) but such phonemes are not so uncommon: voiceless
aspirated fricatives exist (often just as allophones of plain fricatives) in Korean and in Chodi (a
member of the Tibeto-Burman languages, Gansu province, Central China), in Amerindian Ofo
(a member of the wider Siouan family), in some Oto-Manguean languages (located in the
Central America, especially southern Mexico) and in Hmu (located in Guizhou province,
Southwest China), but voiced aspirated sibilants seems to be extremely rare. The aspirated
voiced sibilant *žh
is often reconstructed as the outcome of the IE palatovelar *ǵh
. It has to be
noted that a development of Dh
S towards DZ (without any translatio aspirationis on a sibilant)
is an appealing possibility.
Note: Bartholomae’s Law is perpendicular for its orthogonality on location series, its applications crosses the
series, i.e., it affects all of them without any regard for the location value of a given series.
Bartholomae was to first to describe the law (Bartholomae 1882; Bartholomae 1883: 48;
Bartholomae 1885: 206; Bartholomae 1895–1901: 21–23; for further references see especially
Collinge 1985: 7–11 and Mayrhofer 1986: 115–118; Szemerényi 1990: 106–109; Mayrhofer
2004: 46). Bartholomae simply stated that the result of the concatenation of any voiced
aspirated obstruent and voiceless unaspirated obstruent would be a cluster of voiced and voiced
aspirated plosives (schematically: -Dh
+ T- = DDh
, -Dh
+ s- = -Dz-)35
.
The development of two-plosive clusters was modelled by Anderson (1970: 388) as a
two-stage development: in the first phase, a cluster of two voiced aspirated plosives is created
(-Dh
+ T- = Dh
Dh
), the second phase is a deaspiration of the first plosive (Dh
Dh
> DDh
). Sag
(1974: 593) states a paradox: Bartholomae’s law (and subsequent deaspiration) should precede
35
Note that Barthlomaeʼs careful wording covers even clusters arising from -T + Dh
- and clusters with sibilants,
since he does not state its limitation on plosives!
49
Grassmann’s Law in case of bhotsya- but Grassman’s Law should precede Bartholomae’s Law
in case of buddha-! Deaspiration as a second step of the whole process is accepted by Schindler
(1976), though other authors assume that Indian aspirates are of biphonematic nature (Ejerhed
1981). Mey (1972) forms a complex of processes, where a deaspiration with a subsequent
devoicing before an obstruent goes through a series of shifts of inter-exclusive operations.
Schindler (1976) simply states that an obstruent becomes aspirated (and inherently voiced) after
a voiced aspirate. D. G. Miller (1977) follows the influence of the root structure on the process,
considering the voicing process as a primary trajectory, followed by the aspiration as a later
process. Lombardi (1991: 140) tries to explain the the unexpected voicing of the right obstruent
and the transfer of aspiration to it as ‘spreading of the entire Laryngeal nodeʼ, i.e., by aspiration
of the whole cluster. Kobayashi (2004: 117–125) speaks about the ‘aspiration throwbackʼ, using
the instrumentality of Optimality theory and following the morphemic structure of clusters.
The origin of Bartholomae’s Law is either assumed to be Indo-European36
(as stated in
e.g., Kuryłowicz 1935: 50–51; Lubotsky 2018: 1879) or specifically Indo-Iranian (e.g.,
Szemerényi 1990: 107; Hoffmann/Forsmann 1996: 95–96) but is outside the scope of this paper,
since only its validity for Indo-Iranian is undoubted.
Note: There are no secure examples of Bartholomae’s Law in Nūristānī; hence, the following model is applied
solely to Old Indo-Iranian languages, though it could be applied even for Nūristānī.
Our proposed model for the development of Dh
T clusters follows the trajectory of spirantization
(Walde 1897: 466 assumes that the original value of the Dh
was originally the voiced spirant Δ)
and subsequent fortition with the following steps:
i. the left plosive (an originally voiced aspirate) becomes a voiced spirant and the right voiceless
plosive also becomes a voiced spirant (Dh
+ T → ΔΔ)37
;
ii. in the second phase both spirants became a subject of fortition to plosives, the left spirant
became a voiced plosive, the right spirant changed into a voiced aspirate (ΔΔ → DDh
). The
result is preserved in Indic, but the second plosive lost its aspiration in Iranian (either directly
due the same process or later; Walde 1897 assumes aspiration a later feature both of OIA
and Gr. aspirates).
i. Dh
+ T > ΔΔ > DDh
(Indic)
ii. Dh
+ T > ΔΔ > DD/ÐÐ (Iranian)
36
The possible validity of the Bartholomae’s Law for Germanic was examined in last years especially by Görtzen
(1998: 444–448) and Hill (2003: 218–220, for older references see Collinge (1985: 5–11) and Szemerényi
(1990: 115–117).
37
We assume, similarly to D. G. Miller (1977) that voice was a primary quality, not aspiration.
50
As we have already mentioned above, the outcome of Bartholomae’s Law for Dh
T clusters is
essentially the same as the outcome of the regular regressive development of the cluster of TDh
.
Here we have to point out that this time the process is under the dominance of the right standing
plosive, so the process could be sketched:
i. both plosives become voiced spirants;
ii. the cluster develops as described above, with a fortition of both plosives.
i. T/D + Dh
> ΔΔ > DDh
(Indic)
ii. T/D + Dh
> ΔΔ > DD/ÐÐ (Iranian)
Similarly, we also assume the spirantization and the subsequent fortition for Dh
Dh
clusters,
which develop as follows:
i. Dh
+ Dh
> ΔΔ > DDh
(Indic)
ii. Dh
+ Dh
> ΔΔ > DD/ÐÐ (Iranian)
The development of Dh
s clusters is in its trajectory and properties similar to the development
of the Dh
T clusters:
i. a voiced aspirate becomes a voiced spirant;
ii. a sibilant becomes voiced
iii. a ΔZ cluster is despirantized in the left part of it in Iranian; Ts replaces the expected †DZ
due to analogy in Indic, the Iranian state is assumed to be archaic:
i. Dh
+ S > ΔZ (→ TS) (Indic)
ii. Dh
+ S > ΔZ > DZ (Iranian)
The spirantization model of Bartholomae’s law has one prominent advantage concerning the
development of Dh
S clusters: within the spirantization model, there is no need to introduce the
voiced aspirated sibilants at all.
2.6.3 The trajectories of clusters labial + t/s/dh
The development of both peripheral series differs in Indic and in Iranian since Iranian plosives
are spirantized in the context of a right standing plosive or sibilant, but preserved as plosives in
Indic.
The spirantization in Iranian is one of many spirantization processes we meet in the
development of the Indo-European languages in different stages (Celtic, Sabellian, Pre-Slavic,
Middle Greek etc., could serve as examples of this common, though necesarilly not mutually
related development).
51
It seems sure that the Iranian spirantization of the peripheral/grave block is a later
process than the development of the acute/central series, which could be demonstrated by the
Iranian development of Ks/Ps > xš/fs on the one hand but Ḱs/Ts > 0s/0s on the other hand. since
being both parallel developments; we could expect similar outcomes, similarly Kt/Pt > xt/ft38
on one hand but Ḱt/Tt > št/st of the central series has sibilants instead of spirants of the
peripheral development.
We model the development of clusters formed by non-aspirated labial plosives with
spirantization in Iranian, but without spirantization in Indic; the Nūristānī development mirrors
that of the Indic, but see the note below:
i. P + t > pt (Indic)
ii. P + t > φt > ft (Iranian)
iii. P + t > pt > tt > 0t (Nūristānī)
Note: The outcome of IE *Pt is *0t in Nūristānī. Such an outcome could be reached at least by two trajectories:
either ‘Indicʼ, attested in the development of MIA and NIA languages, i.e., through gemination and later
simplification of geminates: Pt > tt > 0t, or through the Iranian trajectory of lenition, with spirantization first,
followed by debuccalization and elision: Pt > φt > ht > 0t. At the moment we are not able to determine the
one more probable.
We have already seen above that the regular Avestan39
development is without spirantization
(cf. Av. hapta- ‛seven’; YAv. ppp. vipta- from √vaēp- ‛engage in homosexual activities’; OAv.
pr. haptī from √hap- ‛keep’). However, the Middle and New Persian has an ft cluster (cf. Av.
hapta- vs. Phl., NP haft ‘sevenʼ; YAv. ppp. xv
apta- from √xv
ap- ‛sleep’ vs. NP xuftan, Blažek
1999: 200; Cheung 2007: 145–146) and even in Avestan the parallel cluster of Ps is realized as
fs/fš.
The question then naturally arises whether the attested pt cluster in Avestan is either an
archaism or a later innovation (most probably due to the analogical leveling). We encounter in
Avestan forms of pitar- ‛father’: N. sg. OAv. ptā, tā, YAv. ptā, pita, D. sg. OAv. fə
δrōi, piϑrē,
YAv. piϑre, the form tā is easy to explain as the result of the development from *ftā, which
underwent lenition (debuccalization and subsequent elision) ft > ht > 0t,40
the form ptā being
then the secondary fortition of aspirant back to a plosive. Reichelt (1909: 40) assumes
spirantization and subsequent fortition, Kümmel (2007: 63–65, 147–148) assumes partial
restitution of the spirants *f, *θ, *χ in Later Iranian (cf. Lipp 2009a: 158–160 with further ref.).
38
With different development in Avestan, where Pt > ft, see above and below.
39
Unluckily, there are no secure Old Persian data.
40
Hoffmann/Forssman (1996: 94) assume direct simplification of initial pt- > t-, which seems to be too complex
a change if considered without at least one intermediate stage.
52
Beekes (1988: 73) and Hoffmann/Forssman (1996: 94) reckon with the preservation of the
inherited pt.
Our model presumes fricativization and later fortition on plosive, especially concerning
the development as mentioned earlier of Av. pitar-.
The development of clusters of voiced aspirates *bh
+ t- shares its outcome with the
development of clusters of any plosive (without regard to its aspiration or voice) + dh
-, since
Bartholomae’s Law afflicts it. For this development we state the following trajectory, based on
spirantization, followed by the assimilation of voice and finally by the subsequent fortition of
both plosives, according to either to Indic or Iranian peculiarities
i. bh
+ t > βδ > bdh
(Indic)
ii. bh
+ t > βδ > bd/βð (Iranian)
Note: Nūristānī has no secure data for this development.
The labial clusters Pdh
and bh
dh
are rarely attested (only the cluster of Pdh
is securely attested
in Avestan), but we can surely assume that the mechanism was even simpler, in accord with the
rule of the regressive assimilation triggered by the right member, and the result of the process
is shared with the preceding development of clusters of bh
t. An asterisk marks the reconstructed
patterns, Nūristānī trajectory is omitted for the lack of examples:
i. P + dh
> βδ > *bdh
(?) (Indic)
ii. P + dh
> βδ > bd/βð (Iranian)
i. bh
+ dh
> βδ > *bdh
(?) (Indic)
ii. bh
+ dh
> βδ > *bd/βð (?) (Iranian)
The clusters with non-aspirated plosives in the left position and a sibilant in the right position
followed a simple trajectory of the spirantization. The variant cluster of fš in Iranian has
probably arisen due to the analogy with the cluster of kš (see below), or as an example of a
unique extension of Pedersen’s Law/ruki-rule on any cluster of a peripheral plosive and sibilant,
which seems improbable. As in the case of the peripheral clusters of Pt-, we do not assume a
spirantization in Indic, but we assume an intermediate bilabial plosive in the Iranian trajectory:
i. P + s > ps (Indic)
ii. P + s > φs > fš/fs (Iranian)
Note: Nūristānī lacks secure examples of this development.
53
The development of clusters bh
s is similar to the development of the clusters with t-, i.e., the
result is affected by Bartholomae’s Law (including the assumed spirantization), with later
analogical leveling in Indic. Again, in Iranian labial clusters could be realized either as βz
(unattested!) or βž as an analogy to clusters of Ps/Pš, as described above (and also analogical
to the development of the clusters of K/gh
+ t/dh
/s). Analogical leveling to voiceless clusters is
attested in Iranian, similar to that in Indic, where it is, as we already know, a regular process:
i. bh
+ s > βz (→ ps) (Indic)
ii. bh
+ s > βz > βž/(βz ?) (Iranian)
Note: Again, Nūristānī has no secure data for this development.
The outcome of the clusters made of a labial aspirate plosive + sibilant as a spirant and a voiced
sibilant do not require the intermediate cluster with an unusual combination of the type Dzh
(where D is any voiced plosive), appearing (for example) in Burrow (1955: 94, who reconstructs
the intermediate cluster of bžh
as a first outcome of the process of clustering of bh
+s). As we
have noted above, the outcome ps/bz is limited to those clusters arising from original cluster of
labial + sḱ (inchoatives), the outcome pš/βž on labial + s, clusters reflecting in this way the
different origins of Avestan s (and the morphemic structure of suffixes).
2.6.4 The trajectories of clusters velar + t/s/dh
The development of the velar (including the original labiovelar and later palatalized velars) is
wholly parallel to that of the labial series in the same contexts: in Iranian, the plosive is
spirantized, but it is preserved as a plosive in Indic.
The development of the clusters of non-aspirated velar (either voiceless or voiced)
plosives + t could be then modelled simply as trajectories:
i. K + t > kt (Indic)
ii. K + t > xt (Iranian)
iii. K + t > kt > tt > 0t (Nūristānī)
Note: The Nūristānī development follows similar lines as that of the cluster of Pt mentioned above: we assume
that it in general features followed the Middle Indo-Aryan trajectory of the geminate and its simplification.
Again, it is even possible to accept an alternative trajectory of the spirantization and later debuccalization (Kt
> φt > ht > 0t). As in the preceding case, the solution is open at the moment, but the preferred solution is the
‘Indianʼ one, since it explains even the development of clusters of *TT > Nur. 0t (see below).
Similarly to the development of the Iranian initial cluster of #pt in forms of Avestan pitar‘fatherʼ
(see above), we meet a special development with the cluster of #kt- both in OIA and
Avestan, thus the development is securely already Indo-Iranian. It follows the trajectory: #kt-
54
> #xt- > #ht- > #0t-, attested in OIA turī́ya-,41
Av. tūiriia- ‘fourthʼ; the root is in the reduced
grade (*ktur-). We have to emphasize that Avestan has prefixed forms ā-xtūirim ‘four timesʼ
(cf. ā-ϑritīm ‘three timesʼ; cf. Blažek 1999: 201) with original velar preserved in the form of a
spirant. It seems that in the non-prefixed forms, spirantization was followed by a
debuccalization and later elision, which is valid not only for Avestan but also for Vedic numeral
turīya-. It seems that the spirantization already affected the grave anlaut clusters (but probably
not the inlaut clusters) in the common Indo-Iranian period.
The development of clusters of voiced aspirates *gh
+ *t- shares its outcome with the
development of clusters of any plosive (without regard to its aspiration or voice) + *dh
-, being
again affected by Bartholomae’s Law. For this development we state the following trajectory,
based on spirantization, followed by the assimilation of voice and finally by the subsequent
fortition of both plosives, according to either to Indic or Iranian peculiarities:
i. gh
+ t > γδ > gdh
(Indic)
ii. gh
+ t > γδ > gd/γð (Iranian)
Note: Old Persian has a regular, analogy based, leveling of clusters of gh
t to kt; cf. OP. duruxta- from √druj- ‘lieʼ,
though Bartholomae’s Law could be working in unattested examples. Nūristānī examples are not securely
attested.
The same outcomes are shared by regularly formed clusters with a dominance of the right
context of *dh
- with any quality of voice or aspiration of the left plosive preserved. The
development of clusters of gh
dh
either in Indic or Iranian is not attested; the outcome is based
on the analogy:
i. K + dh
> γδ > gdh
(Indic)
ii. K + dh
> γδ > gd/γð (Iranian)
i. *gh
+ dh
> γδ > gdh
(?) (Indic)
ii. *gh
+ dh
> γδ > gd/γð (?) (Iranian)
Note: Again, the Nūristānī development is not attested either for Kdh
or gh
dh
.
The clusters either of a voiceless non-aspirated plosive or a voiced non-aspirated plosive and sare
affected by Pedersen’s Law (the ruki-rule), hence the outcome (if not leveled by the
analogy) always has š, in Indic shifted to the cerebral ṣ, while the Iranian outcome is regularly
spirantized (as is the analogical cluster of Kt) in the velar:
41
If the meaning is ‘fourth part, quarter”, the accent shifts to the first syllable: túrīya- AV, cf. MacDonell 1910:
311. Blažek (1999: 200–201, 209) reconstructed *ku̯
tur(ī)yo-. In OIA, there is another term for ‘fourth”, which
is caturthá-, regularly derived from the cardinal catúr- ‘four”.
55
i. K + s > kš > kṣ (Indic)
ii. K + s > xš (Iranian)
iii. K + s > kš > tš > č (Nūristānī)
The development of clusters of *gh
+ *s is similar to the development of the clusters with *t-,
i.e., affected again by Bartholomae’s Law (including the assumed intermediate spirantization),
with later analogical leveling on kṣ in Indic. Similar leveling is also attested in Iranian (cf. YAv.
ppp. anādruxta- ‘undeceivableʼ vs. OAv. pr. pai
rii-aoγžā, both being derived from the same
root √aoj- ‘sayʼ). Again, in Iranian velar clusters could be realized either as γž according to
analogy to clusters of Kš, as described above:
i. gh
+ s > γž → kš > kṣ (Indic)
ii. gh
+ s > γž > γž (Iranian)
Note: Nūristānī has no secure data for this development.
Again, as with labials above, the outcome of the development of the clusters of velar voiced
aspirated plosive + sibilant do not require an intermediate cluster with a combination gzh
as
proposed by Burrow (1955: 94).
2.6.5 The trajectories of clusters palatovelar + t/s/dh
The development of the Indo-Iranian palatovelar obstruent clusters has parallels in other
languages having original palatovelars (i.e., all satəm-languages, the process in clusters of our
interest are well attested in Baltic and Slavic languages, rudimentarily also in Armenian and
Albanian).
Before we sketch the possible trajectories of the development of the two plosive clusters
of Ḱt, ǵh
t, Ḱdh
, ǵh
dh
, Ḱs, ǵh
s, we have to turn our attention the development of the original
palatovelar plosives in non-alternating contexts (which could be found in Indo-Iranian
especially before vowels), since there are different outcomes not only for Indic and Iranian
branches but also inside the Iranian branch, since the IE triad *ḱ, *ǵ, *ǵh
is realized in Vedic as
a triad ś, j1, h1 but in Iranian as a dyad (resulting from the merge of both voiced plosives) either
s, z (in Avestan) or ϑ, d in Old Persian.
Note: Nūristānī, as far as we can safely reconstruct, has a dyad of affricates *ċ, *j [= /dz/]. There are some cases
with *š instead of *ċ for IE *ḱ, generally considered to be attested in early loanwords (cf. Degener 2002: 104).
Again, our analysis will be based on old Indo-Iranian languages, and Nūristānī will be dealt with in
commentaries below.
56
The development of the original Indo-European palatovelars before vowels in the three old
Indo-Iranian languages leads either to sibilants in Avestan, to a spirant and a plosive in Old
Persian and to a sibilant affricates in Vedic.
The models of the reconstructed Indo-Iranian state could be classified into two groups:
the sibilant model and the affricate model.
As for examples of sibilant-based models we should mention that of Bartholomae (1895:
12), who uses the symbols *ś, *ź, *źh
. Similarly the purely sibilant nature of the Indo-Iranian
outcomes of original IE palatovelars was taken for granted at least by Leumann (1942: 2–3)
and Kuiper (1967: 103–105; 117–120) or later by Erhart (1980: 21) or Kobayashi (2017: 334).
The affricate-based model is preferred since Morgenstierne (especially 1945).42
Among
others supporting this point of view we can name Burrow (1955: 73), who assumes the
development from IE *ḱ, *ǵ,* ǵh
to earlier affricates *ć,* j́, *j́h
and later to Common IndoIranian
sibilants *ś, *ź, *źh
(i.e., to the sibilant phase), while Mayrhofer (1989: 6) reconstructs
the outcomes of IE palatovelars as *ć, *j, *jh
, as does Schmitt (1989: 27), who assumes for
Iranian development another intermediate stage *ts, *dz (for both voiced plosives), similarly
Hoffmann/Forssman (1996: 93)43
, Kobayashi (2004: 52–54); Lipp (2009a: 131–189, especially
139–149) and Beekes (2018: 1880–1881) and others. The reconstruction of the reflexes of the
original IE palatovelars as affricates was influenced by the discovery of Nūristānī, where the
palatovelars are realized as non-palatal affricates *ċ [ts], *j [dz] (< *ǵ(h)
), but alternatively
reconstructed as *š and *z (< *ǵ(h)
) (cf. Morgenstierne 1926: 23–53; Morgenstierne 1945: 225–
238; Nelson 1986: 72–74; Degener 2002: 104, 110; Cardona/Jain 2003: 22–23; Lipp 2009a:
153–170; Werba 2016: 354–356; Cantera 2017: 492–493). It should be noted that Sihler (1997)
criticizes the affricate trajectory with well-based arguments44
and returns to the traditional
sibilant-based model.
The sibilant model with a direct development of palatovelars to sibilants could be
schematically reconstructed as follows (note that the sibilant model constructs voiced aspirated
sibilants, otherwise constructed as outcomes of clusters of voiced aspirate + s according to
Bartholomae’s Law):
42
But it already was Bloomfield (1911) who doubted the traditional sibilant trajectory (assuming some occlusion
present).
43
The symbols used here are for simplicity, the same as those used by Bartholomae, Burrow, and Erhart.
44
Sihler (1997: 190) argues that the development of affricates and spirants from original sibilants is an attested
process in Burmese, Athapaskan, and Castilian, hence, according to his opinion, the Nūristānī affricate has
arisen from original sibilant.
57
i. *ḱ > ś > ś (Old Indo-Aryan)
ii. *ḱ > ś > s (Avestan)
iii. *ḱ > ś > ϑ (Old Persian)
i. *ǵ > ź > j (Old Indo-Aryan)
ii. *ǵ > ź > z (Avestan)
iii. *ǵ > ź > đ (Old Persian)45
i. *ǵh
> źh
> h1 (Old Indo-Aryan)
ii. *ǵh
> źh
> ź > z (Avestan)
iii. *ǵh
> źh
> ź > đ (Old Persian)46
A possible trajectory for the Indic developments is done by:
i. the preservation of the voiceless sibilant;
ii. the occluvisation of both voiced palatovelars;
iii. the debuccalization of the aspirated palatovelar;
Note: The debuccalization of *ǵh
> h1 is essentially similar to the development of the palatalized velars (gh
> jh
>
h2). It is interesting that there are some examples of the debuccalization of originally voiced aspirates in Vedic,
especially for the original dh
: cf. 2nd
imp. act. ending of athematic verbs -dhi vs. -hi; the 1st
pl. primary med.
ending -mahe (cf. Av. -maide), similarly the 1st
pl. secondary med. ending -mahi (cf. Av. maidī). This tendency
to debuccalization of dh
is attested with root phonemes too; cf. ppp. hitá- from √dhā- ‘putʼ. There is even an
example of the debuccalization of bh
to h in Vedic, attested with two variants of the single root √grabh-/grah‘seizeʼ.
This debuccalization of the voiced aspirates is regular in the development of MIA languages; cf. Pāli
bhoti vs. hoti: both parallel forms are from √bhū- ‘beʼ (cf. Lubotsky 1995 for a detailed analysis of OIA h <
dh
, bh
).
The sibilant trajectory for Iranian development is remarkable because of:
i. the depalatalization;
ii. the merging of voiced and aspirated members (as in all other series).
Note: Since the traditional sibilant model was developed before the discovery of Nūristānī, it does not solve the
development of palatovelars in Nūristānī. Only Sihler (1997) tried to reconstruct a possible trajectory for
Nūristānī according to the sibilant model, with the spirantization of the original palatovelars (a similar process
is assumed for Old Persian) and later affricativization of such spirants (as is modelled for OIA voiced
palatovelars anyway).
The affricativization trajectory could be modelled as follows (this reconstruction is based on
that by Lipp 2009a, esp. 131–189; but cf. Lubotsky 2018: 1884–1885); note that with this model
45
It seems that originally, the reflex of both voiced palatovelars was a voiced dental spirant, cf. Bartholomae
(1895: 166), Reichelt (1927: 41), similarly Schmitt (1989: 68), but even by Mayrhofer (1989: 6), as it is by
Erhart (1980: 24) and others, though Brandenstein/Mayrhofer (1964: 39) strictly reject the possibility that OP
d < *ǵ(h)
could be a spirant(synchronically). The intermediate spirant state is accepted by Lipp (2009a: 115,
144). Here the spirant symbol is used purely to distinguish the phoneme from the voiced dental plosive
originating from original *d(h)
.
46
The development in OP requires deaspiration before the merging of ź and z, otherwise it would be impossible to
accept the transformation of some of zs to d and the preservation of others. The development in Avestan is
hence analogous to that of Old Persian.
58
the voiced affricate (< *ǵ) is preserved in Indic, but the Old Persian shifts from affricates to
fricatives (plosives respectively47
):
i. *ḱ > ć > ś (Old Indo-Aryan)
ii. *ḱ > ć > c > s (Avestan)
iii. *ḱ > ć > c > ϑ (Old Persian)
iv. *ḱ > ć > ts
(Nūristānī)
i. *ǵ > j > j (Old Indo-Aryan)
ii. *ǵ > j > ź > z (Avestan)
iii. *ǵ > j > ź > đ (Old Persian)
iv. *ǵ > j > dz
(Nūristānī)
i. *ǵh
> jh
> źh
> h1 (Old Indo-Aryan)
ii. *ǵh
> jh
> źh
> ź > z (Avestan)
iii. *ǵh
> jh
> źh
> ź > đ (Old Persian)
iv. *ǵ > j > dz
(Nūristānī)
The affricativization trajectory could be considered a first phase of a wider, two-step model,
sibilantization being the second phase, but it avoids the affricativization of voiced palatovelar
in Indic, simply preserving the reconstructed affricate.
Note: Bloomfield (1911: 42–44) assumes that the development of palatovelars happened more ‘rapidlyʼ in Iranian
than in Indic and this different ‘speed of developmentʼ caused the merging of secondary voiced palatals (j, jh
<
*g, *gh
before e/i) in Indic, the Indic process being ‘slowʼ enough to be ‘caught upʼ by the secondary
palatalization, which was not the case of *ś. This developed, according to Bloomfield, rapidly, hence
spirantized before its voiced counterparts. We assume that this merging was not due to the slowness of the
development of voiced ‘palatovelarsʼ, but due to the later fortition in Indic, which affected even true sibilants,
especially those voiced (see below). There is no reason to assume (except for the development of clusters of
Ḱs) that the development of original palatovelars did not follow the same trajectory in Common Indo-Iranian
(cf. Lipp 2009a: 135).
The development of the left positioned original palatovelars in the contexts of *t-, *dh
-, *s- is
different from the development in the free context above in many aspects, given by the
alternations occurring in such contexts.
Note: For simplicity, we will use the traditional symbols (i.e., *ḱ, *ǵ, *ǵh
) for palatovelars to avoid the use either
of the affricate or the sibilant model. However, such ‘palatovelarʼ symbols are used here purely as symbols,
not as descriptions.
Here two possible trajectories for the description of the development of palatovelars will be
used, the first using the affricativization model, the second using the spirantization model,
which assumes the replacement of ‘palatovelarsʼ before obstruents by spirants.
47
As above we preserve for simplicity the marking of OP reflexes of *ǵ(h)
as d.
59
The development before *t-, according to the affricate model, could be modelled as
follows. This model is based on the trajectory given by Lipp (2009a: 139–142), who assumes
the development through affricates, later simplified by a loss of their plosive segment in all
three sub-branches:
i. *Ḱ + t > ćt [i.e., tš
t] > št > ṣṭ (Indic)
ii. *Ḱ + t > ćt [i.e., tš
t] > št (Iranian)
iii. *Ḱ + t > ćt [i.e., tš
t] > št (Nūristānī)
For the clusters formed by the voiced aspirated palatovelar + t- we can model the development
as follows ( is an approximand, causing either the lengthening of the preceding vowel or its
diphthongization, according to the nature of the preceding vowel):
i. *ǵh
+ t > jdh
[i.e., dž
dh
] > ždh
> ẓḍh
> ḍh
(Indic)
ii. *ǵh
+ t > jdh
[i.e., dž
dh
] > ždh
> žd (Iranian)
Note: Nūristānī has no secure data for this development.
The development of clusters of palatovelar + *s- as a second member could be modelled,
according to the ‘affricate trajectoryʼ as follows (cf. Lipp 2009a: 155; Lubotsky 2018: 1885),
with sibilantization of the whole cluster in Iranian:
i. *Ḱ + s > čs [= tš
s] > tš > tṣ > kṣ (Indic)
ii. *Ḱ + s > čs [= tš
s] > šš > 0š (Iranian)
iii. *Ḱ + s > čs [= tš
s] > tš > ts = c (Nūristānī)
Note: We will return below to the transformation of ḱs into tš and later into kš in Indic.
Analogously, the affricate trajectory for the voiced aspirated palatovelar *ǵh
+ s could be
modelled as follows (cf. Lipp 2009a: 172; Indian outcome is given by the analogy, the Nūristānī
outcome is again not attested), again with a sibilantization of the whole cluster in Iranian:
i. *ǵh
+ s > jžh
[= dž
žh
] > džh
> ḍẓ > (→kṣ) (Indic)
ii. *ǵh
+ s > jžh
[= dž
žh
] > žžh
> žž >0ž (Iranian)
The confusing point of the affricate trajectory is that it is of the opposite vector than the assumed
traditional (affricate) trajectory for the development of T + t/dh
/s dental clusters, since in Indic
such dental clusters, according to the traditional theory, lose the internal sibilant (ts
t > tt, dz
dh
> ddh
etc., see below), but according to the affricate theory for the palatovelar series the left
plosive would be lost but the sibilant preserved (tš
t > št, dž
dh
>ždh
), though both processes
should appear about the same time (this objection is not valid for Iranian, only for the Indian
60
development). There is a question concering the affricate development both of dental and
palatovelar series before a sibilant: why and how to lose a plosive before a sibilant, due to a
process tšš > 0šš > 0š etc., or due to tšš > tš > šš > 0š?48
For this reason, we dare to replace the affricativization model by the spirantization model,49
which could solve some of the problems sketched above.
According to this model, the trajectory for the left standing non-aspirated plosive Ḱ (either
originally voiceless or voiced) in the t-context produces a palatal spirant ç and is equal to the
outcome of the sibilant resulting from the ruki-rule (i.e., Pedersen’s Law) in the same context50
.
This spirant was later sibilantized, and in Indic it later underwent the typical Indo-Aryan shift
to cerebrals51
(secondarily affecting secondarily the following plosives), whereas in Iranian is
the sibilant preserved:
i. *Ḱ + t > çt > št > ṣṭ (Indic)
ii. *Ḱ + t > çt > št (Iranian)52
iii. *Ḱ + t > çt > št (Nūristānī)53
Note: The minor development of *Ḱt in Iranian xšt, mentioned by Kellens (1976) (attested widely for word-initials,
internally there are only attested forms: Av. paiti.fraxštar- ‘interrogatorʼ, yaoxšti ‘branchʼ, spaxšti- ‘visionʼ) is
considered a proof that clusters from Ḱ + t and š + t did not merge fully even in the Proto-Iranian. Lubotsky
(2018: 1884) explains this feature as a proof of a dialectal development in Eastern Iranian (similar reflexes are
found in Sogdian and Bactrian), and the different outcomes in Avestan are then proof of the dialectally mixed
origin of the attested Avestan texts, which is probably the most acceptable solution. However, the development
of the cluster of Ḱt would require the split of the fricative çt to xšt in some of the dialects. A similar process,
universal for all clusters of št in the word-initial (cf. YAv. xštāt̰ ‘stands, quoted by Bartholomae) as described
by Bartholomae (1895–1901: 36) could be a prothesis, as Bartholomae stated, though Kellens (1976: 68) rejects
the idea. However, the prosthesis explanation remains the most probable solution for the anlaut clusters. The
development of the internal clusters would follow the trajectory of a ‘split spirantʼ: *Ḱt > çt > xšt > št in all
dialects except those Eastern Iranian, which is possible, but it requires a parallel later development both of
Indic and Western Iranian. It would probably be easier to assume the dialectal split of the çt on xšt in Eastern
Iranian as a later independent process.
Similarly modelled are all clusters, including the cluster of *ǵh
t are affected by the
Bartholomae’s Law: in our spirant model we assume first the transformation of both plosives
into voiced spirants (ʝδ), second the sibilantization of the first spirant as ž and re-occlusivization
48
It seems very improbable to chart a trajectory like: tšš > ššš > šš > 0š.
49
A simple model of this type was proposed by Morgenstierne )1942: 81), who assumes ϑ´t as an intermediate
stage for Early Iranian, arising from the common Indo-Iranian t´t and this from the IE *ḱt).
50
About clusters št and ždh
, see more below.
51
Which is considered a regional development, cerebrals are present even in Nūristānī.
52
Wíth a secondary depalatalization on st in some cases, attested in Old Persian (without clear trigger, cf. OP
ufrašta- and ufrasta- both from √fraϑ- ‘ask”). Since the depalatalization lacks a trigger, it could be either
spontaneous, or another solution is simply analogical leveling with the cluster, either from original *s+t or
*Ḱ+t.
53
The same development as for Indic and Iranian is assumed for Nūristānī. Interestingly, Lipp (2009a: 335)
proposes the development ḱt > çt > št for Nūristānī only, not for other Indo-Iranian languages.
61
of the second member on the voiced aspirated plosive in Indic, though the intermediate
aspirated stage could be assumed even for the Iranian development. Avestan preserves this state
(when not leveled to št), while Indic underwent a typical cerebralization to ẓḍh
. The subsequent
Indian development could contain the debuccalization phase of the sibilant to the voiced
approximant and finally, the loss of this approximant. The mark of the lost approximant is the
lengthening of the preceding high vowel (ppp. gūḍhá-, gd. gūḍhvī́ from √guh- ‘hide; ppp. ūḍháfrom
√vah- ‘carryʼ), sometimes even of an original short a (ppp. sāḍhá- from √sah- ‘prevailʼ)
but with context with a usually forming diphthongs (pr. tṛṇédhi from √tṛh- ‘crushʼ54
; inf.
vóḍhum from √vah- ‘carryʼ). Note that we can probably assume three allophones of the voiced
approximant: ɦ for the pure lengthening, i̯ for forming e and u̯ for forming o, this allophone
clearly before the syllable containing a labial vowel55
.
i. *ǵh
+ t > ʝδ > žδ > ždh
> ḍh
(Indic)
ii. *ǵh
+ t > ʝδ > žδ > žd/žð (Iranian)
Note: The Nūristānī outcome is not attested, hence the trajectory is not reconstructed:
The developments of clusters of two aspirated plosives (i.e., the dh
-context) are similar to that
of *ǵh
t. The cluster was later cerebralized in Indian again; the sibilant underwent lenition and
elision with a compensatory lengthening of the same type (√vah- ‘carryʼ: voḍhvám, voḷhám).
Again, since the Nūristānī data are not attested, the reconstruction of its trajectory is omitted:
i. *ǵh
+ dh
> ʝð > žδ > žḍh
> ḍh
(Indic)
ii. *ǵh
+ dh
> ʝð > (žδ/žd ?) (Iranian)
Note: Since the Old Persian data are not fully attested, it seems that clusters of T + Dh
do not follow Bartholomaeʼs
Law in Old Persian, probably secondarily, hence the result would be št/st as with *ḱ/ǵ + t.
The development of clusters of Ḱdh
generally mirrors both above-described processes. The
trajectory leads again towards Indo-Iranian cluster of ždh
, attested in Iranian. In Indic, the
sibilant was re-occlusivized as ḍḍh
(√diś- ‘pointʼ: pr. didiḍḍhí), in contrast to the development
of previous clusters leading towards lenition of the sibilant to an approximant and to subsequent
elision and the compensatory lengthening. The reasons for this different outcome are not clear,
though not within the range of trajectories described here (again, Nūristānī is omitted for the
lack of secure data):
54
From the root √tṛh- ‘crush” there are even non-lengthened forms ppp. tṛḷhá-, tṛḍhá-, here the short syllabic
resonant has probably arisen due to analogical leveling instead of the expected ri (< ṝ).
55
These different outcomes could be demonstrated with the both afore- and below-mentioned form of the root
√vah- ‘carry”.
62
i. *Ḱ + dh
> ʝð > žδ > žḍh
> ḍḍh
(Indic)
ii. *Ḱ + dh
> > ʝð > žd/(žδ ?) (Iranian)
Note that outcomes of *Ḱdh
in OIA (i.e., ḍḍh
) are different from the outcomes of clusters of *ǵh
t
and *ǵh
dh
(realized in OIA as 0ḍh
, which is regular of *ǵh
t and demonstrated by numerous
examples, but attested by only a single secure example for *ǵh
dh
). We will meet a similar
difference between Vedic clusters from *dh
t on one side (resulting, according to Bartholomae’s
law, in regular ddh
) and clusters of *tdh
generally resulting not only in ddh
, but also in dh
(Vedic
daddhí vs. dehí) and dh
dh
, always resulting in dh
(see below). The outcomes of both central
series in this aspect strongly contrast with peripheral series, since the OIA clusters from *gh
t
and *Kdh
(*gh
dh
is not attested; cf. above) are both realized by gdh
; for OIA clusters from *bh
t
we meet bdh
(clusters from *bh
t and *bdh
are not attested; cf. above). Also note that in Avestan
the outcomes of IE clusters of *ǵh
t and *Ḱdh
(there are no data for the IE *ǵh
dh
in Avestan) are
same, žd.56
The development of clusters in the *s-context is, in many aspects, different from the
development of clusters within the *t-context. To the Indic development we have to emphasize
again the lack of Bartholomae’s Law for original clusters of ǵh
s, to the Iranian development is
needful to restate that not only clusters of Ḱs and ǵh
s, but even clusters of Ts and dh
s, having
arisen from dental + sibilant, underwent the elision of the left plosive to 0s; hence we have to
explain the loss of an acute plosive before a sibilant in Iranian in general.
Note: The Nūristānī development of the cluster of *Ḱs is regular, with an assumed loss of palatalization (modelled
as Ḱs > tš > ts > c) but again, we lack any solid proof of Bartholomae’s Law functioning in cluster of ǵh
s.
First, we will analyze the development of the clusters of Ḱs, leading in OIA to the cluster of kš;
this development is regularly modelled as a trajectory Ḱs > tš > kš (> kṣ). The process tš > kš
itself is generally considered to be a result of a leveling (cf. Kuryłowicz 1951/1973: 129–130;
Burrow 1955: 91–92; Kuryłowicz 1956: 373–374; Schindler 1967: 1999–200; Gunnarsson
1971: 38–42; Lipp 2009a: 150–152; Lipp 2009b: 12–18).
The reasons for the reconstruction of the intermediate state *tš is the twofold outcome of the
original *Ḱs (and *ǵh
s due to the absence of Bartholomae’s Law in Indic) are:
56
Old Persian and Nūristānī data are not satisfying enough to make any statements based on them, cf. given data
above.
63
i. the difference between OIA kṣ and Iranian 0š (and Nūristānī ts);
ii. the cerebral plosive as an allophone of a voiced palatovelar in left position before a plosive
(a cluster of ḍḍh
is attested for *Ḱdh
, for clusters of *ǵh
t/ǵh
dh
the outcome is 0ḍ, i.e., without
the first plosive).
Note: There are examples of the alternation of original palatovelars from the nominal flexion in Old Indo-Aryan,
namely the development of some nominal consonantal stems ending in an original palatovelar, though other
stems (sometimes even same stems) end in a plain velar. The forms where cerebrals could appear are: nom.
sg., dat.-abl. pl. and instr. pl., du. abl.-dat.-instr., in contrast to loc. pl., where only kṣ could appear, see
following lines (note that -j2 and -h2 are regularly realized as -k, -gbhyaḥ, -gbhiḥ, -gbhyam, -kṣu):
The OIA forms from -ś realized as a cerebral:
nom. sg. -ṭ: víṭ (< √viś- ‘settlementʼ); spáṭ (< √spaś- ‘seeʼ) is derived from -tš (final š is an ending) In other cases
the outcome is a plain velar: -dṛ́k (always present in a compound, < √dṛś- ‘seeʼ); dík (√diś- ‘directionʼ); spṛ́k
(in compounds, < spṛś- ‘touchʼ);57
Similarly, the OIA forms from -j1 realized as a cerebral:
nom. sg. -ṭ: rā́ṭ58
(< √rāj- ‘kingʼ); bhrā́ṭ59
(< √bhrāj- ‘shiningʼ);
And the OIA forms from -h1 realized as a cerebral:
nom. sg. -ṭ: sā́ṭ60
(< √sah- ‘prevailʼ); vā́ṭ61
(< √vah- ‘carryʼ); saráṭ TS ‘beeʼ;
The voiced cerebral plosive for *ž from original *ǵ is attested before bh
:
dat.-abl. pl. -ḍbhyaḥ: viḍbhyáḥ but a counterexample: digbhyáḥ;
inst. pl. -ḍbhiḥ: viḍbhíḥ; padbhíḥ, but a counterexample with a velar: -dṛ́gbhiḥ;
abl.-dat.-instr. du. -ḍbhyam: viḍbhyás
The voiced cerebral plosive for *ž from original *ǵh
before bh
:
dat.-abl. pl. -ḍbhyaḥ: saráḍbhyaḥ (no attested counterexample of -gbhyaḥ);
Nevertheless, note that OIA loc. pl. is uniformly with -kṣu: dikṣú AV, VS, vikṣú; srakṣú (< √sṛj- ‘emitʼ).
According to the spirantization model, the development could be modelled slightly differently,
since the Old Indo-Aryan data lead us toward *kš uniformly for verbs and nouns, i.e., towards
the original neutralization of palatovelars to a plain velar before original *s (cf. Šefčík 2017,
but already Bartholomae 1895: 12). On the contrary, the Iranian development has no traces of
this, supposedly archaic, state, since the Iranian outcome is 0š/0s. However, we can surely
assume that the Iranian development of the palatovelar series mirrored in its main feature the
development of the dental series (see below).
The development of the cluster of the aspirated voiced stop + s is even harder to describe
since, in OIA, the presumed cluster according to Bartholomae’s Law has been replaced by the
analogous cluster of kṣ (the cluster of *ǵh
s could not regularly be realized as kš). The Iranian
development preserves the result according to Bartholomae’s Law, but the plosive was lost as
with the voiceless cluster.
57
Final -s is regularly dropped after a plosive in nom. sg.
58
Including compounds and derivatives.
59
Again, including compounds and derivatives.
60
Including compounds and derivatives.
61
Again, including compounds and derivatives.
64
If the Iranian clusters underwent a spirantization, it had to be earlier or different in its
nature than that of clusters of *Ks/gh
s, otherwise the results of both development would merge.62
From Iranian development, we can be sure that there was a phase when there was a plain velar
present (due to the palatalization of the sibilant according to the ruki-rule). We dare to propose
that in Iranian there was a later replacement of the original cluster of *Ḱs by a newly created
analogously çš (due to the development of the original palatovelars in other positions), this
process appearing after the ruki-rule. Similarly, the cluster *ǵh
s was replaced by a newly created
ʝž. Both fricative clusters were assimilated later as fully sibilant clusters, and finally, the first
member was elided. The Iranian state is then innovative, and the Indic state is archaic.
Note: The Nūristānī development is hard to trace back, but we assume the spirantization of the palatovelars and
their later depatalization.
i. *Ḱ + s > kš > kṣ (Indic)
ii. *Ḱ + s → çš > šš > 0š/0s (Iranian)63
iii. *Ḱ + s → çs > ϑš > tš > ts = c (Nūristānī)
i. *ǵh
+ s > γz > γž →kṣ (Indic)
ii. *ǵh
+ s > γz > γž→ ʝž > žž > 0ž (Iranian)
Note: OIA anaḍ-vah- ‘oxʼ has irregular forms nom. sg. anaḍvā́n, dat. pl. anadúdbhyaḥ AV loc. pl. anaḍútsu
without cerebralization. The word is a compound from √vah- ‘carryʼ (cf. EWAi I: 69). Wackernagel (1896 I:
180) assumes the dissimilation of a cerebral after a preceding cerebral (ḍuḍh > ḍudh; ḍuṭs > ḍuts), but another
trajectory could be modelled çš > ϑs > ts and similarly ʝbh
(or ʝβ with two spirants?) > δbh
> dbh
as with regular
development of *Ts and *ddh
(see below for details).
2.6.6 The trajectories of clusters dental + t/s/dh
We can surely assume that the processes, attested in the development of clusters of dentals T/dh
+ t/dh
/s, are of a very ancient origin, at least in their oldest stage of development, since we meet
similar processes in other Indo-European branches outside the Indo-Iranian family. Such a wide
geographical distribution is a mark of an old process, originating, at least in its first phase, from
the Common Indo-European era.
To demonstrate the antiquity of the development of dental clusters, we will turn our
attention to the best-attested cluster of Tt, since this cluster is very well covered in source
languages.
The development of the cluster of Tt in IE languages generally has three different
outcomes64
in given branches of the Indo-European languages:
62
*Ks gives xš in Iranian, gh
s gives gž, see the section on the development of velars.
63
The form 0s is a result of the analogical leveling, seems to be limited on inchoatives.
64
For simplicity other outcomes such as Arm. u̯ t, Alb. 0s and Hittite and Tocharian ts
t(s)
are omitted here, cf.
given chapters for details.
65
i. the cluster is preserved as tt (OIA being the sole example);
ii. the left dental is sibilantized, the right t is preserved: st (e.g., in Iranian, Slavic, Baltic,
Greek);
iii. the whole cluster is sibilantized to ss (e.g., in Italic languages, Celtic languages, Germanic
languages).
The model trajectory of this development was first proposed by Kräuter (1877: 88) and accepted
especially by the most influential authority of the era, Brugmann (1880: 140–142; 1886: 347;
Brugmann 1896 Ib: 624) and it has been generally accepted since (cf. Kent 1932a; Burrow
1955: 90; Hill 2003: esp. 3–7; Kobayashi 2004: 37–38 etc.).
For the cluster of Tt we can arrange this trajectory of development, where three
outcomes are equated to stages of such a development:
i. ii. iii.
Tt > tt > st > ss
Within this model trajectory, OIA would represent the oldest phase, equal to the assumed early
IE state; Iranian, etc., the second phase; Italic, etc., the third phase of a single development.
Note: Since the development *tt > *ss does not appear in Indo-Iranian languages, we will pay no attention to this
development here, though we accept fully the fact that this outcome is related to the development of *tt > *st,
making its final stage.
The confusing fact is that though the Iranian development fits perfectly into the frame of a
rightfully old process, the Indian development seems to be an exception to this antiquity, since
cluster of Tt is in OIA always realized as tt, not as st as in Iranian (cf. OIA sattá- vs. Av. hasta<
*satta-). This unique feature of Old Indo-Aryan, its confusing and apparent exception from
the process otherwise affecting otherwise all Indo-European languages, is a puzzle with at least
two possible solutions: either the Indic state is an archaism (i.e., OIA was never affected by the
sibilantization of the left dental plosive or any possible preceding process) or the Indic state is
an innovation, in fact, a re-archaization of the same process we meet in Iranian and elsewhere.
Accepting the model assuming the archaicity of the Indic state, we would face a
problem: the current models generally accept the original unity of Indic and Iranian protolanguages
in a Common Indo-Iranian and for good reasons, and since both branches so strictly
differ in the preserving/innovation of such old feature as is (non)sibilatization of left dentals, it
is hard to accept the existence of the Common Indo-Iranian, though this common stage could
be safely reconstructed in many counter-examples. The single solution could be accepting that
Iranian development appeared after the split of both branches (the details of the Nūristānī
66
development are covered by too many eons to build any theory on it), but since all other IE
branches are subjected to the same development, it is hard to imagine that the sibilantization of
dentals affected all IE languages after the split of the whole Indo-Iranian branch off the IE
continuum. That the Indo-Aryan state is a ‘re-archaizationʼ is for these reasons the prevailing
opinion (cf. Brugmann 1880: 140–142; Brugmann 1886: 347, etc.; Wackernagel 1978: 177–
178; MacDonell 1910: 35; Leumann 1942: 13; Burrow 1955: 90 etc.).
However, if accepting that the fricativization of dentals already was, at least in its early
stage, a regular process in the IE stage, before the split of the Indo-Iranian branch, we have to
explain:
i. how the seemingly archaic state in Indo-Aryan has arisen from the more progressive state;
ii. whether we can trace this assumed older progressive stage in OIA, i.e., an internal proof of
the older fricativization phase;
iii. whether we can harmonize it with existing sibilantization preserved in Iranian65
.
The authors assuming the ‘archaizationʼ of OIA who are trying to find an internal proof of the
older ‘fricativeʼ stage in OIA are primarily focused on the developments of clusters of dental
plosive + dh
in OIA, since they result either regularly in ddh
(cf. √ad- ‘eatʼ: addhí; √vid- ‘findʼ:
viddhí) or exceptionally in dh66
(cf. √budh- ‘wakeʼ: bodhí; √yudh- ‘fightʼ: yódhi; √dhā- ‘putʼ:
dhehí). It is noteworthy is that from the root √dā- ‘giveʼ we have both forms: dehí and daddhí,67
and also remarkable that all ‘problematicʼ outcomes are results of the clustering with -dh
, but
clusters of dh
+t (according to Bartholomae’s law) result in regular ddh
(cf. √yudh- ‘fightʼ:
yuddhá-, yuddhv ; √idh- ‘kindleʼ: inddhé, iddhá-; √budh- ‘wakeʼ: buddhá-). This feature is
striking, especially since √yudh- and √budh- have in that way clearly distinguished outcomes
of two different clusters. This is in accord with the distinction between clusters of ǵh
t and Ḱdh
in OIA, the first resulting in 0ḍh
, the second in ḍḍh
(cf. above, it is interesting that the cluster of
ǵh
dh
realizes in accordance with ǵh
t, as far as we can judge from a single example). In contrast,
the OIA clusters of gh
t and Kdh
are both realized by gdh
, for OIA clusters of bh
t and Pdh
we have
bdh
attested only for bh
t (cf. above). Also note that in Avestan, the outcomes of clusters of dh
t,
tdh
and dh
dh
are always zd.68
65
The Nūristānī, having the outcome 0t is left aside for a moment once again.
66
The marks the approximant, causing the lengthening or the diphtongization of the preceding vowel.
67
Note that roots with an infix have forms -ndh
-: √ud- ‘wet”: undhí; √idh- ‘kindle”: indhvám, but here we can
assume either the simplification of -nddh
- on -ndh
- or the simplification -nddh
- > -nzdh
- > -ndh
-, the former
process is more straightforward and in the accordance to the Ockham’s razor.
68
Old Persian and Nūristānī data are not numerous enough for any persuasive statements based on them.
67
Note: Jasanoff (2002) does not connect yódhi and bódhi to original -ddh
- at all, but assumes them the analogical
formations based on the form jóṣi; hence both forms would be a product of the morphological reanalysis, not
of any regular phonemic development. This would leave yuddhá- and buddhá- a single and regular phonemic
outcome of the development of ddh
.
Wackernagel (1896: 178) considers dehí and daddhí both from *az and compares to Av. dazdi:
he assumes a similar process for dhehí (though *dh
addh
í is not attested). Wackernagel’s opinion
is generally accepted (cf. Meillet 1922: 59; Burrow 1955: 89–90; Renou 1996: 21–22; Görtzen
1998: 313–315). Marsh (1941: 45–46) has considered both forms to be rather sigmatic aorists
(i.e., forms †dh
asdh
i, †dasdh
i), not presents. Hoffmann (1956: 21) explained the forms dehí and
dhéhi as results of the dissimilation of original *daddh
i/dh
addh
i (cf. Lubotsky 1995: 10, who
presumes further dissimilations to attested forms). Insler (1972: 551–565) discusses some
irregular imperatives ending on -dh
i (for us yódhi, bodhí, randhí are relevant); he explains yódhi
and bódhi as secondarily formed based on subjunctives, randhí as a result of the analogy of the
assumed original *randha to the dh
i-imperatives. Tedesco (1968: 1–24) considers dehí and
dhehí to be ‘redactional substitutions for older Ved. *dādh
í, *dh
ādh
íʼ, taken from Middle Indo-Aryan69
,
as a regular counterpart of YAv. dazdi he accepts just daddhí, the mechanism he assumes is the
‘regularizationʼ derived from the reduced grades *didh
í, *dh
idh
í. Insler (1975: 4–5) takes
Tedesco’s opinion as highly probable, but he presumes that the e-vocalization has come due to
the analogy to the MIA optatives deyā- and dheyā-. Pisani (1976: 166) connects these forms
with Gr. δός and ϑές, i.e., as *das and *dh
as + hi, forming the original cluster of sdh
, not ddh
,
which is, as Hill (2003: 65) states, phonemically impossible, since both ο and ε are from original
vocalized laryngeals. Kobayashi (2004: 90) takes dehí and dhehí as just examples of a broader
process of deocclusion of dh
after a front vowel. Hill (2003: 65–69) brings a new hypothesis,
assuming that e is here a result of a regular process of the monophthongization of *ai̯, he follows
Jamison’s (1997: 78–79) observation that only exceptionally is this imperative accented and
equates OIA dhehí vs. YAv. dazdi to OIA neśa- vs. Av. nąsa-. Hill accepts the change of ddh
to i̯dh
, without changes ddh
> dz
dh
> zdh
, i.e., he assumes the lenition of the original root-initial.
69
The Pāli forms like demi, desi, deti, dema, detha, denti (Fahs 1989: 285) are results of a different process than
the process leaning towards OIA dehi since aforementioned Pāli forms arose from the stem dā-y(a)-, contracted
in MIA. The resulting single paradigm is the result of a later merging of stems of different origins. Similarly
Pāli -dheti, etc., have arisen (Fahs 1989: 136, 291); the original y-suffix is preserved in Pāli dhāyati without
contraction. The ya-suffix is preserved for √dā- in dajjati (= da-d-ya-). That these forms have nothing common
with the reduplicated stems could demonstrate the root √ṭhā- ‘stand” (cf. OIA √sthā-), which has the
reduplicated form tiṭṭhasi, athematic root form -ṭhāti and ya-form -ṭheti (besides -ṭhāyāmi). Fahs (1989: 137)
considers deti as a result of analogy to dehi, but his solution seems to be too complicated.
68
Any proposed trajectory for the development of the Indo-Iranian clusters of T/dh
+ t/dh
/s has
especially to set forth:
i. both the ‘preservationʼ (or ‘restitutionʼ) of the left dental in Indic and its transformation to a
sibilant in Iranian (as in all other IE branches);
ii. the variation d ~ 0 (i̯, u̯ in given cases) before dh
- in Indian;
iii. the loss of a dental plosive in Iranian before s- and the preservation of the plosive in the
same context in Indic.
Generally speaking, there could be more than one possible trajectory; at the moment, there are
two of them, though both have the same incomes and the same outcomes.
The first trajectory we can term the trajectory of affricativization since it assumes the
affricativization of the left plosive in its first phase and the subsequent fricativization of the first
member, generally due to the loss of the plosive segment of the affricate. The trajectory was
expressed first by Kräuter (1877: 88)70
and popularized by Brugmann (firstly 1880: 140–142,
used since), for Indo-Aryan we have to mention especially seminal works by Johansson (1903
and 1906). The disadvantage of this theory was expressed by Hill (2003: 4), who points out the
supposed and very problematic long lifespan of the ts
t stage (he expressly mentions the forming
of the Germanic weak preterite). The approach by Hammerich (1955: 127–128), who assumes
a later development of the dental clusters in the later phases of given branches/languages is too
overstated and without solid proofs. The existence of the affricate stage in Indo-European is
supported by Anatolian data, since Hitt. 3rd
sg. preterite <e-iz-ta> and 3rd
sg. imperative <e-izdu>
from √H1ed- ‘eatʼ shows the affricate form (probably first already noted by Götze 1928:
126; but put in the spotlight first by Sturtevant 1933a: 6–7 and Sturtevant 1933b: 129; later
especially cf. Oettinger 1979: 530–532 and Melchert 1994: 113, 151, 249).
The development in Iranian according to the affricativization trajectory for the original
cluster resulting from T + t is regular like that of Balto-Slavic or Greek: Tt > ts
t > st. The Indic
development could be described in two possible trajectories: either we can assume the ‘archaicʼ
development without any change at all (Tt > tt) or we can presume a more complicated model,
where the outcome tt is the result of a further process, following the common trajectory: Tt >
ts
t > st > tt. Schematically expressed both trajectories would be:
i. T + t > ts
t > st > tt (Indic)
ii. T + t > ts
t > st (Iranian)
70
It is interesting that Kräuter speaks about affricativization, but his description of the feature is that of a
spirantization! Verner (1878: 341–342) has a critical evaluation of the idea.
69
Note: The Nūristānī outcome is 0t, which could be probably traced to the same trajectory as in Indian, but with a
further simplification of the geminate: tt > 0t, similarly to the simplification of geminates in Indic (Late Middle
Indo-Aryan). A similar process is assumed for clusters of *Pt and *Kt in Nūristānī (cf. above). A variant
trajectory (‘Iranianʼ) following st > ht > 0t seems not to be probably since clusters of sibilant + plosive are
preserved in Nūristānī.
Note: Though the trajectory for the above mentioned third group of languages with the sibilantization of the whole
cluster (Italic, Celtic, Germanic) is of no direct relationship to the development of Indo-Iranian languages, we
can model the trajectory of the development, according to the affricativization theory as: Tt > ts
t > ts > ss just
to present the whole context of the development (cf. especially Hammerich 1955: 127–128, but already
Schwyzer 1934: 234–335).
Clusters from original dh
+ t are subjected to Bartholomae’s Law, their outcome in Indic being
ddh
and zd in Iranian. According to the traditional affricativization model, the trajectory should
be modelled as:
i. dh
+ t > dz
dh
> zdh
> ddh
(Indic)
ii. dh
+ t > dz
dh
> zd (Iranian)
Note: There is no secure example of the development of the cluster of dh
t in Nūristānī.
Note: Old Persian has a regular, analogy based, leveling of clusters of dh
t to st, i.e., there is no Bartholomae’s Law
working with this cluster; cf. OP. basta- from √band- ‘bindʼ.
As noted above, the outcomes of clusters of *Tdh
and *dh
dh
are both dh
in Indic, but zd in
Iranian. The Iranian outcome is same as for the development of clusters of *dh
t, in contrast to
the Indic development (probably requiring the change zdh
> i̯ dh
in the Indic development for
clusters resulting dh
; cf. Götzen 331–315):
i. T + dh
> dz
dh
> zdh
> ddh
(major) (Indic)
zdh
> i̯dh
(minor)
ii. dh
+ dh
> dz
dh
> zd (Iranian)
i. dh
+ dh
> dz
dh
> dh
(Indic)
ii. dh
+ dh
> dz
dh
> zd (Iranian)
Note: As far as we can state from a single attested example, Old Persian has the same development as Avestan for
the cluster a regular outcome zdfor the cluster of tdh
(resulting from the dominance of the right plosive). There
are no reliable data for this development attested in Nūristānī.
Now we have to turn our attention to clusters with a sibilant in the right position, either T + s
or dh
+ s.
According to the affricativization trajectory, the trajectory for the cluster of Ts requires
the affricativization of a left plosive71
, followed by a simplification to a diphthong and with a
subsequent elision of one sibilant in Indic (or a simple preservation of a plosive is more
71
The clusters affricate + s are typologically possible (cf. Old Czech čsti ‘of honour”, later simplified on cti).
70
probable?), but a sibilantization and subsequent elimination of sibilants to a single one in
Iranian.
i. T + s > ts
s > ts (Indic)
ii. T + s > ts
s > ss > 0s (Iranian)
Note: The affricate is preserved in Nūristānī, with the sibilant segment of the affricate merged with the original
sibilant, forming a final affricate (cf. Lipp 2009a: 169). For Nūristānī we could reconstruct a trajectory: T + s
> ts
s > ts > ċ, but based on a single example of Kati maċi ‘fish’ (OIA matsya-, YAv. masiia-; cf. Lipp 2009a:
169), we cannot rule out a secondary palatalization due to the following palatal approximant; cf. Pāli, Pkt.
maccha- (Turner 1966: 560).
Similarly, the affricativization trajectory could be modelled for the cluster of Dh
s. The Indic
state has no traces of Bartholomae’s law, being replaced by the same analogical forms as for
Ts. Nūristānī and Old Persian have no useful data, hence our reconstruction relies heavy on
Avestan, again with the affricativization of the plosive, sibilantization of the affricate and
simplifying of the whole cluster as 0z (cf. Lipp 2009a: 169):
i. dh
+ s > dz
z → ts (Indic)
ii. dh
+ s > dz
z > zz > 0z (Iranian)
The typical feature of the development of clusters dental plosive + s in Indo-Iranian is then a
simplification of the sibilant segment of the affricate with a following sibilant in Indic, but
sibilantization (and following simplification) of the cluster in Iranian.
The affricativization model requires, to explain the Indic development, a loss on the
sibilant segment between two plosives; cf. ts
t > OIA tt, dz
dh
> OIA ddh
.72
Such a process is well
attested for Indic development, but it is not a proof of the validity of the affricativization model,
since such a loss is attested for all clusters formed by the sequence: plosive–sibilant–plosive,
since such sequences have two full stops of the airflow, interrupted by a momentous opening
(which causes a change of sonority), which could be expressed schematically by a letter M.
Such a cluster is articulatory complicated and the elision of the sibilant forms much simpler
clusters of two plosives, schematically: II. We meet such alternation in examples like those of
the sigmatic aorist: patthās AV (< -d-s-th
, √pad- ‘flyʼ); ácchānta (< d-s-t, √chand- ‘seemʼ);
ábhakta (< -j-s-t, √bhaj- ‘eatʼ), ápṛkta (< c-s-t, √pṛc- ‘mixʼ), árabdha (< dh
-s-t, √rabh- ‘graspʼ
etc., It is known from other formations with s between two plosives; cf. ppp. -gdha TS. < gsdh
< gh
s-t-73
(√ghas- ‘eat’). Cf. Brugmann 1880: 140–142; Wackernagel 1896 I: 76, 131, 269;
72
Besides, if we assume the trajectoryts
s > OIA ts, but it is hard to be sure that the middle obstruent is lost and not
the last one.
73
Note Bartholomae’s law working over the sibilant.
71
Johansson 1903; Johansson 1009; Meillet 1922: 59; Kent 1932a: 24; Kent 1936: 241; Görtzen
1998: 308, 312–313, 317; Hill 2003: 4).
Another possible trajectory we propose is not based on the affricativization of the left dental
plosive, but on its spirantization, assuming that the original spirantization had already appeared
in Late Indo-European and that further developments (sibilantization in Iranian and reoccluvisation
in Indic) appeared independently later in further stages.
The spirantization trajectory was already first proposed for Indo-Iranian by
Bartholomae (1895: 16), who assumed *Tt > ϑt and *Tdh
/*dh
t > δd(h)
. The process is taken as
a possibility by Leumann (1942: 13). It is worth noting that Brugmann in one of the early
versions of his affricativization model (1886: 347) states that ‘Wir schreiben ts
t(h) und dz
d(h) und
geben gerne zu, dass vielleicht richtiger tþ
t(h) und dđ
d(h) gesetzt würde.’ Morgenstierne (1942: 80) proposes
*ϑt as an intermediate stage for Early Iranian (ts
t being reconstructed for the common IndoIranian
period, but *tt for the common Indo-European).
To sum up the outcomes, we assume that T was first spirantized as ϑ before t- or s- but
as δ before original dh
(also which turned δ), according to the Bartholomae’s Law. The clusters
of *Tdh
and *dh
dh
were also realized as δδ. The cluster of *dh
s would realize as δz.
Note: Alternatively, there appeared, in all cases, just the spirantization of the first plosive, and dh
- is preserved in
the right position (dh
-context). However, the model would be too complicated, especially since we assume that
δ was always just an allophone to dh
; hence we will use cluster of δδ, though readers can substitute it with δdh
,
if it is more convenient for them.
Summing up the whole development of the early Indo-Iranian clusters, the model of
spirantization could be represented as follows:
t- dh
- s-T
ϑt δδ ϑs
-dh
δδ δδ δz
In the following development in later stages, the Iranian spirants generally underwent a
sibilantization, and the sibilants were preserved before plosives. The Indic development was
more complicated: the voiceless spirants were fortitied into plosives in the left position, the
voiced spirant was fortified on d in the left position in clusters of dh
t, but on dh
in the right
position, since δ was always only a positional allophone of dh
(addhí, viddhí, daddhí, yuddhv ,
72
inddhé74
, buddhá-). In some cases, the δδ cluster underwent a lenition to i̯ δ/i̯ dh
(cf. Hill 2003:
68 and more below). In this case the sonant approximant formed a diphthong with a preceding
vowel (dehí, dhehí), or was lost (bodhí, yodhí, indhvám), as we have already stated above,
though the causes of the split could only be expressed by means of a sound law with difficulty,
it is noteworthy that Tdh
could be both realized as ddh
(a major trajectory; cf. daddhí) or as 0dh
(a minor trajectory; cf. dehí), all clusters of dh
dh
are realized as 0dh
. However, both the minor
and the major trajectories can be modelled on similar grounds, without the sibilant grade in the
Indo-Iranian stage as an intermediate state.
The trajectories of development according to the spirantization model for given IndoIranian
branches could be modelled for the original clusters resulting from T + t the trajectories
in both branches as:
i. T + t > ϑt > tt (Indic)
ii. T + t > ϑt > st (Iranian)
iii. T + t > ϑt > tt > 0t (Nūristānī)
Note: As has been already stated above, the Nūristānī development is similar in its outcome to that of *Kt, *Pt,
with the outcome being simple 0t. It has to be noted that original cluster of *st is preserved with a sibilant in
Nūristānī, as are št clusters both from *št or *ḱt, hence the outcome is not a product of a later elision of a
sibilant in such clusters (for original *st cf. Kati dušt, Waigali došt, Ashkun dōšt ‘handʼ vs. OIA hastá-, Av.
zasta, OP. dasta-; for original *št: Kat. uṣṭ, Waigali ūṣṭ, Ashkun ōṣt ‘mouthʼ vs. OIA óṣṭha-, Av. aošta,
‘mouthʼ; for original *Ḱt cf. Kati. uṣṭ, Waigali oṣṭ, Ashkun ōṣt ‘eightʼ vs. OIA aṣṭáu, Av. ašta, ‘eight). The
simplest solution seems to accept the Old, Middle and Early New Indic development of the same clusters:
spirantization of the first T, re-plosivation as tt followed by a later simplification of geminates (cf. MIA satta
> Hindi, Marathi sāt; cf. Bloch 1965: 93–96; Masica 1991: 187–188). In contrast with NIA, there are no signs
of the compensatory lengthening in Nūristānī.
For clusters resulting from dh
+ t, according to Bartholomae’s Law, the trajectories will be:
i. dh
+ t > δδ > ddh
(Indic)
ii. dh
+ t > δδ > zd (Iranian)
Note: There are no data on which the Nūristānī development could be based, as we also lack them for clusters of
dh
t and dh
dh
.
The development of the clusters from original T + dh
is similar to the preceding, but with a
variation in the Indic development. The major development is the same for clusters of dh
t, the
minor as for clusters of dh
dh
, Iranian sibilantization being a regular development:
i. T + dh
> δδ > ddh
(major) (Indic)
> i̯dh
(minor)
ii. T + dh
> δδ > zd (Iranian)
74
But some forms of imperatives from this root have a single plosive: indhvám, indhā́m (cf. MacDonell 1916:
371).
73
Moreover, the development for clusters from dh
+ dh
could be modelled, with an approximant
arising from the left spirant in Indic, the Iranian sibilantization being also a regular
development:
i. dh
+ dh
> δδ > i̯dh
(Indic)
ii. dh
+ dh
> δδ > zd (Iranian)
Data seems to lead us towards a projection that clusters of dh
t in Indic always realized as first
as δδ and preserved as such till the re-plosivation as ddh
, in contrast with clusters of dh
dh
,
realized first as i̯ δ and later as i̯ dh
, the distinction probably based on morphemic reasons, not
phonemic. The ‘mixedʼ realization of Tdh
is probably a result of analogy, the original state being
hard to determinate.
We assume spirantization even for clusters of dental plosive + s in similar lines.
The cluster of Ts is preserved in Indic, or better, the initial spirant was later re-
occlusivised;
in Iranian, the spirant was first assimilated to a sibilant, and the subsequent geminate was later
simplified to a simple 0s.
Note: In Nūristānī, the cluster is preserved as an affricate, but this development is not principally different from
the Indic development:
i. T + s > ϑs > ts (Indic)
ii. T + s > ϑs > ss > 0s (Iranian)
iii. T + s > ϑs > ts > c (Nūristānī)
The cluster of dh
s underwent an analogical leveling in Indic. The Iranian clusters developed
similarly to Ts clusters, i.e., through a spirantization, a sibilantization and finally simplification
i. dh
+ s > δz (→ ts) (Indic)
ii. dh
+ s > δz > zz > 0z (Iranian)
Note: There are no secure examples for the Nūristānī development of clusters Ts and dh
s.
The Iranian process of elision (or simplification of the geminate) for both clusters of Ts and dh
s
has to be a late process, since otherwise, a newly intervocalic VsV would have to regularly
become VhV in Iranian: at least clusters of VssV had to exist before the debuccalization of
Iranian intervocalic s into h (cf. OIA sapta- vs Av. hapta- < Indo-Iranian *sapta- ‘sevenʼ; OIA
sácate vs. YAv. hacaitē̆ < Indo-Iranian *√sac- ‘accompany, followʼ; OIA siňcáti vs. YAv.
hiṇcaiti < Indo-Iranian *√sai̯ c- ‘pourʼ). From the logic of the development, it is clear that the
74
debuccalization of Iranian VsV to VhV had to appear before the transformation of VśV to VsV
(otherwise the debuccalization would affect even the sibilants from the original palatovelars).
Generally speaking, since Iranian clusters of st/zd are not affected by the ruki-rule, this
proves that the final transition either from the first phase to the sibilant phase had to appear after
the application of Pedersen’s Law. Even the Indic clusters from ddh
resulting in i̯ dh
were not
transformed due to the ruki-rule (in contrast to those arising from ǵdh
; cf. above and ff.
Brugmann 1987: 637; Meillet 1922: 60).
Note: The singular Nūristānī example of ṛ from *dh
t (Ashkun būṛə ‘mind, spiritʼ, Waigali buṛā́, buṛṓk ‘meaning,
intentʼ (< *būḍhi < *būṣḍhi < *bh
udz
dhi < *bh
udh
-ti; cf. Turner 1964; Turner 1966: 525; Hill 2003: 44–45) is
probably a result of an analogy based leveling, not a counter-argument (cf. Budruss 1977: 24; Görtzen 1998:
310–312).
2.6.7 The trajectories of clusters sibilant + t/s/dh
Historically there are only two sibilants (S)75
in Indo-Iranian, each with two positional
allophones:
i. the primary sibilant, the old Indo-European *s (with its positional variant *z before voiced
plosives);
ii. the secondary sibilant *š (with its positional variant *ž before voiced plosives), arising from
the primary sibilant due to Pedersen’s Law (ruki-rule).
The development of clusters of S + t is simple; all clusters are preserved, except for the
subsequent cerebralization in Indic and Nūristānī of the clusters of št, which is a regular process:
i. s + t > st (Indic)
ii. s + t > st (Iranian)
iii. s + t > st (Nūristānī)
i. š + t > št > ṣṭ (Indic)
ii. š + t > št (Iranian)
iii. š + t > ṣṭ/0ṭ (Nūristānī)
Note: The twofold outcome of the cluster of *št in Nūristānī could be a result of original dialectal Nūristānī
developments and probably of the later dialect mixing. While Nūristānī ṣṭ would be a direct descendant of *št,
Nūristānī 0ṭ could be a result of the gemination št > ṣṭ > ṭṭ, similar to the Middle Indo-Aryan development,
with a later simplification of the geminate to 0ṭ. The presence of both outcomes in Kati, Waigali could be a
result of a secondary mixing of dialects originally differing either by ṣṭ or 0ṭ during the history of the Nūristānī
language area.
The trajectory of clusters of S + dh
is far more interesting and less trivial in its development
than the previous development of the clusters of St, since the sibilant voiced allophones were
75
The sibilants which arose from original palatovelars we treat as plosives, especially since their sibilant form is
probably an outcome of later developments, as we demonstrated above.
75
later regularly lost in Indic, though preserved in Iranian (the Nūristānī data are scarce and data
unreliable and doubtful).
Generally, the model assumes the voicing of the original sibilants before *dh
-, which is
fully proportional to the development of the clusters of T + dh
, as described above, with the
dominance of the right element. The Iranian developments fully follow these lines, but the
dental plosive was later deaspirated, as is a general rule of Iranian development. For the
development in Indic, we have to assume a far more complex trajectory, according to known
outcomes. The outcomes are two: the first is a moraic lengthening of the preceding vowel,
either by newly created vowel length (cf. OIA tāḍhí < √takṣ- + dh
i; the process is ‘invisibleʼ on
roots with the inherent length of a root: OIA ā́dhvam < √ās- ‘sitʼ + dh
vam; śādhí < √śās- ‘orderʼ
+ dh
i) or by its diphthongization (cf. OIA edhí < √as- ‘beʼ + dh
i); cf. a similar process with the
development of clusters with palatovelars above. Both variants of the development could be
traced through a lost approximant since both the lengthening and the diphthongization could be
explained through this approximant. The second outcome is a fortition of a sibilant to a plosive,
either dental (for an original dental sibilant) or cerebral (for an original palatal, later cerebral,
sibilant); cf. OIA vaddhvam S. < √vas- ‘dwellʼ, vidiḍḍhi < √viṣ- ‘be activeʼ + dh
i. However,
these processes are parallel to the development of original palatovelars in the same context; cf.
above. The transition of palatal clusters to cerebral is essentially the same. The trajectories we
reconstruct as follows:
i. s + dh
> zδ > i̯dh
(major) (Indic)
> ddh
(minor)
ii. s + dh
> zδ > zd (Iranian)
Note: There is no attested example for the cluster of *sdh
in Nūristānī.
i. š + dh
> žδ > ždh
> ẓḍh
> i̯ḍh
(major) (Indic)
> ḍḍh
(minor)
ii. š + dh
> žδ > žd (Iranian)
iii. š + dh
> žδ > žd(h)
> ẓḍ/0ḍ/0ṛ (Nūristānī)
Note: The data for the Nūristānī development are scarce (cf. Morgenstierne 1926: 61), but it seems that they also,
in general features, follow a similar development as Indic did, both in the older and younger development. The
‘mixedʼ outcomes could be a consequence for the dialect confusion in the later history of Nūristānī.
Alternatively, we can model a spirant intermediate stage, either dental or cerebral, between both
sibilant and plosive stages for Indic development. The spirant inter-stage would easily explain
the transformation of a sibilant to a plosive (a two-step process instead of a single step), but
alternatively, the spirant could be a subject of a further lenition and turned into an approximant,
76
which, when elided, would either cause a relict lengthening or be preserved as a part of an
unetymological diphthong. The trajectory would then be, in both variants for both sibilants:
LENITION RE-PLOSIVATION
s + dh
> zδ > δδ > i̯dh
~ *s + dh
> zδ > δδ > ddh
š + dh
> žδ > δ̣δ̣ > i̯ḍh
~ *š + dh
> žδ > δ̣δ̣ > ḍḍh
Extremely intricate developments are met with clusters of s + s and š + s, which in Indic result
in plosive + sibilant (ts, kṣ) clusters, but in Iranian as 0s/0š (no secure data are known for
Nūristānī).
The Iranian development is easy to model. We can only assume that the geminate was
later simplified to a single sibilant, or alternatively the first sibilant underwent lenition to h and
then it underwent an elision. However, we should keep in mind that the palatal sibilant in 0š is
a result of an earlier assimilation of *šs to *šš and that the simplification had to appear relatively
later, since intervocalic -s- underwent the change to -h-, but this process did not affect the dental
sibilant from the original geminate.
The Indic development was more complicated than that of Iranian and Lipp (2009a:
213–214) correctly relates it to the development of Ḱs clusters. The trajectory he reconstructs
for the cluster of *šš in Indic is: šs > šš > ṭṣ (Lipp 2009a: 213) and for the cluster of *ss as: ss
> ts (Lipp 2009a: 214).
Note: A special attention could be paid to clusters of *ss and *šš of s-stems:
i. The OIA as-stems end in nom. sg. with -as#, realized primarily as ḥ for an original *-s (one sibilant is lost; cf.
-māḥ ‘moonʼ, ájňāḥ, ápaḥ ‘waterʼ etc.,), Av. ah-stems are realized as -ā̊0# < Ir. *-ah# < IIr. *-s(s)# in nom. sg.
(cf. OAv. ušā̊ ‘dawnʼ etc.,) and OP. ah- stems are realized as -ā in nom. sg. (cf. OP tauviyah- ‘strongerʼ).
ii. The clusters of loc. pl. of as-stems are realized in Vedic as -ss- or -s-: OIA mássu (but másu TS), apsarássu,
áṃhassu AV (but áṃhasu in mss.), rájassu, vákṣassu etc., but ápasu (cf. MacDonell 1910: 221, 223;
Wackernagel I: 111); the geminate is probably a restoration due to analogy with other stems. Avestan has ahu(ua),
with a single phoneme: YAv. ązahu, rauōhu, raucōhuua, ušahuua (Hoffmann/Forsman 1996: 154,
156).
iii. Similarly, clusters of *šš# (arising due to the ruki-rule) are realized in Vedic as -ṭ if from a radical stem (cf.
dvíṭ ‘hatredʼ) but as -ḥ# representing the original *-š (again, one sibilant was lost; cf. OIA áhaviḥ ‘not offering
oblationsʼ, -jyotiḥ ‘shiningʼ etc.,). In Av. and OP such clusters are not attested for masculine and feminine.
iv. The clusters of loc. pl. of uṣ/iṣ-stems are realized in Vedic as -ṣṣ- (cf. havíṣṣu). The forms of locative pl. are
not attested in Vedic, as they are not attested in both Old Iranian languages.
We assume the following trajectory, similar to that of Lipp, but with an intermediate
spirantization: the cluster of ss merged in Indic with a cluster of ϑs (resulting from Ts according
to the spirantization trajectory) and the cluster šs (assimilated later to šš) similarly merged with
a cluster of xš (otherwise arising from the original cluster of Ḱs). In both cases, resulting clusters
underwent the same later fortition to ts for the original cluster of ss, and to kš (> kṣ) for the
original cluster of šs (see above).
77
i. s + s > ϑs > ts (Indic)
ii. s + s > hs > 0s (Iranian)
i. š + s > šš > xš > kš > kṣ (Indic)
ii. š + s > šš > γš > 0š/0s (Iranian)
In some cases, the original state with two sibilants is restored due to analogy (śāssi). The
development of the ss clusters in the second sg. from the root √as- ‘beʼ in Indo-Iranian is
different. In Old Indo-Aryan, we meet ási, in Avestan ahi (< *asi); both forms require an older
simplification of a geminate ss to 0s. This process is probably already Indo-European, since we
meet Gr. εἶ (< *esi), OLith. esi, OCS jesi also without a geminate, and forms with geminates
such as Gr. Hom./Dor. ἐσσί and L. ess (Plautus) are probably not archaisms, but results of the
later morphemic leveling (cf. Mayrhofer 1986: 120–121). This is the reason why this
morphologically affected ss cluster has developed differently from other such clusters.
The transition of the original dental plosive to a sibilant had to be younger than the effect
of Pedersen’s Law (the ruki-rule) on a sibilant since Iranian st/zd from original tt/ddh
was not
affected by the ruki-rule (cf. Hill 2003: 45–46).
2.7 Conclusions and final remarks
The development of clusters formed by any left-standing obstruent in the contexts of the right
standing t/dh
/s- could be split into minor blocks based either on the context of the right
obstruent, but also into blocks according to the centrality or peripherality of the given series.
The block of sibilants stands independently then, since its existence is based not on location but
on its sonority (sibilants vs. plosives of other blocks). The blocks we have recognized are:
i. the central (or acute) block formed by the dental and by the palatovelar series;
ii. the peripheral (grave) block, formed by the labial and velar/palatal series;
iii. the sibilant block, containing both sibilants, the old dental one, and the palatal one, a
result of the ruki-rule.
The perpendicular process affecting all three blocks in the Indo-Iranian languages is
Bartholomae’s Law, which causes the transition of cluster of voiced aspirated plosives (Dh
) + t
to DDh
(usually merging in this way the outcome of the process with outcomes of processes
Dh
+dh
and T+dh
) and of voiced aspirated plosives + s to D(h)
z.
The proposed solution presumes the spirantization of both segments in a cluster (Dh
+t
> ΔΔ), and later re-plosivation of both voiced spirants of the cluster on DDh
in Indic, dialectally
in Avestan. The spirantization could explain both voicings of the right segment, and the later
re-plosivation explains why only the right segment is aspirated in Indian. The voiced spirant is
78
considered an allophone of the given corresponding voiced aspirated plosive Dh
, not a phoneme
of its own (the voiced spirant in Iranian has also the same value). The development of cluster
of Dh
+s could be modelled, according to the spirantization model, as the spirantization of the
plosive, accompanied by the voicing of the sibilant, with later re-plosivation of the left spirant
(the sibilant is not affected): Dh
+s > ΔZ > DZ. Within this model, there is no need to introduce
an intermediate stage †D(h)
zh
, especially since the existence of the aspirated sibilant it is not
attested (though still typologically possible, albeit extremely rare).
The oldest development affected the clusters formed by a dental in the left position. This process
is usually modelled as affricativization of the left dental (the affricativization trajectory). The
opinion can be traced back to Kräuter (1877: 88) and especially to Brugmann (1880 and
passim), who established it as a leading model. An alternative trajectory, which we prefer, at
least for Indo-Iranian, leads us to spirantization of the left dental; this trajectory can be traced
back to Bartholomae (1880) (the spirantization trajectory).76
The affricativization model assumes the transition of the dental to an affricate and later
loss of the sibilant segment of the affricate in Indic, but the loss of the dental segment of the
affricate in Iranian (e.g., Ts
t > In. Tt, Ir. st; dz
dh
> In. ddh
, Ir. zd;). It should be noted that for
some Indic clusters resulting from *ddh
we can safely reconstruct the outcome i̯ dh
.
The spirantization explains the (not only) Iranian transition to st/zd as a further lenition
on the trajectories t > ϑ > s and dh
> Δ > z , but fits as well into the frame of later Indic
developments, since it assumes the fortition of the spirant to plosives, but it even could explain
the transition of *δδ to *i̯ dh
as a minor process, but still within the frame of a lenition.
The assumed existence of dental spirants in clusters of ϑs could explain the otherwise
difficult to explain the fortition of *ss to ts in Indic: we assume that the left sibilant in the
cluster merged with clusters of *ϑs (from *ts) and the resulting ϑs was fortified in Indic to ts
without regard to its origin. In Iranian, the trajectory would be: *ϑs > ss > 0s/*δz > zz > 0z.
However, the process of final sibilantization of dentals before a dental plosives + sibilant
in Iranian was finished after the operationality of the ruki-rule, since such sibilants were not
affected by it and the existence of ϑt (< *Tt) and δδ (< *dh
t/dh
dh
/Tdh
) could easily explain why
the first segment of such a cluster was never affected by Pedersen’s Law: there was no sibilant
yet to be subdued by the process, though otherwise the very old one.
76
De Saussure (1877), Cocchia (1883) and Bartholomae himself (1887) assumed this trajectory for Italic
development, modelling it as: Tt > ϑt > ϑϑ > ss, we will return to this model below.
79
Note: For the development of the Nūristānī dental series we have to assume the ‘Indicʼ model, which, according
to the spirantization model, would be: Tt > ϑt > tt > 0t; Ts > ϑs > ts (?)77
.
The second oldest development is that of the original palatovelar series. The clusters of
palatovelar + plosive in the satəm-languages result in clusters of sibilant + plosive, Albanian
being a single exception (see below).
Though the rise of palatovelars is highly probably a dialectal feature (the centum
languages were not affected by this process), the process is still ancient.
The rise of palatovelars and their transition to (generally) sibilants was modelled with
two different trajectories (similar to those of the development of the analogous clusters with
dentals), first being the affricativization trajectory, the second the spirantization trajectory.
According to the affricate trajectory, palatovelars turned to affricates in general and later lost
the plosive segment before t/dh
/s both in Indic and Iranian (Ḱt > tš
t > št; ǵh
t > dž
dh
> ždh
, etc.)
The problematic point of the affricativization model is hidden in the Indic
development: if the interstage for both of dental and palatovelar series was an affricate (and we
have to reconstruct the same state for both branches before the split), why in Indic was it the
intermediate sibilant segment with dental clusters but the initial plosive segment with
palatovelar clusters that was lost? The Indic development hence has an innate contradiction: the
affricate clusters of ts
t and dz
dh
both lose the sibilant segment of an affricate, but the affricate
clusters of tš
t and dž
dh
lose the plosive segment of a plosive.
For this reason we prefer the spirantization model for the development of both series,
hence Indo-Iranian clusters of ϑt and δδ for dentals and çt and ʝδ for original palatovelars: the
Iranian process would lead towards the sibilantization of both series due to the general lenition
of the acute plosives to fricatives; Indian towards the re-plosivation of the dentals (in some
cases with a minor lenition to i̯ δ) and sibilantization for voiceless palatovelars and similarly replosivation
or lenition for voiced clusters. The Indian development of palatovelars was affected
(as was Iranian) by the already existing clusters of št and ždh
, resulting from original clusters of
st and zdh
according to Pedersen’s Law (ruki-rule).
According to the affricativization model, the palatovelar clusters with a right-standing
sibilant would be realized as tš
s and dž
z in Common Indo-Iranian. The Iranian development
would first simplify the clusters to tš/dž and later šš/žž, to finally be degeminated to 0š/0ž
77
Since based on single example maċi ‘fish’ (OIA matsya-, YAv. masiia-), we cannot rule out a secondary
palatalization due to following palatal, cf. Pāli, Pkt. maccha- (Turner 1966: 560).
80
(analogously to the development of clusters of Ts and dh
s). The Indic development would be:
tš
s/ dž
z > tš/dž > kṣ (the outcome for dž leveled on tš due to analogy).
The spirantization model assumes the transformations the patalatovelar clusters as: Ḱs
> kš in Indic (with neutralization of a palatal marker in the neutralization position), but in
Iranian with leveling to çš, followed by a later sibilantization and simplification to šš > 0š, with
the development of the original ǵh
s as: ʝž > žž > 0ž.
Note: For the development of the Nūristānī palatovelar series we model with the spirantization and the
sibilantization of the palatovelars, which, according to the spirantization model, would be: Ḱt > çt > št; Ḱs >
çš > ϑs > ts (= c).
The development of both series of the peripheral block (i.e., velars and labials) is simpler than
that of the central block: the left plosives underwent spirantization in all contexts in Iranian, but
plosives are preserved as such in Indic in all contexts. Clusters with originally voiced plosives
contained voiced spirants, later re-plosived, according to our model of Bartholomae’s Law.
Note: For the development of the Nūristānī clusters, we assume the gemination and later simplification (as with
the Tt); the model is similar to the development of the Middle Indo-Aryan languages: Kt/Pt > tt > 0t.
Note: The existence of clusters of pt in Avestan seems to be a result of a later re-archaization specific for Avestan;
cf. above.
Both phonemes of the sibilant block are voiceless, and they are not subjected to any alternation
before t-; the clusters are preserved. Clusters of Sdh
have a more complicated development,
which are realized as Zdh
in Iranian (the simple assimilation of voice), but the sibilant is lost in
Indic, being replaced either by a plosive (analogically to clusters of dentals or palatovelars +
dh
) or by an approximant (changing a preceding vowel either to a diphthong or lengthening it).
Clusters of sibilant + s underwent an occlusion to ts (with the dental sibilant) and kṣ (with the
palatal sibilant) in Indian, probably through a spirant intermediate stage; cf. similar
development of given series, but in Iranian such clusters were simplified to 0s/0š, as were
clusters with secondary sibilants (from original dentals or palatovelars) + s. Again, this process
was later than the debuccalization on the intervocalic s to h in Iranian.
Speaking about general tendencies, the development followed in its earliest phase the lenition
of both central series in all contexts. If we accept the spirantization model of Bartholomae’s
Law, the lenition affected even clusters of voiced aspirates + t/dh
/s, or any plosive + dh
of all
series. However, this statement is valid both for for the affricate and the spirantization strategy,
since both lead from plosives towards sibilants, though even for the late Common Indo-Iranian
phase we cannot assume the sibilantization yet (otherwise we would hardly explain the lack of
81
ruki-affection on sibilants from dentals and the different outcomes of dentals in Indic and
Iranian). The sibilantization (at least of original dentals) is a matter of later dialectal
development in Iranian, though the sibilantization of original palatovelars could be already
Indo-Iranian since it affected all branches.
The Iranian development followed the lenition trajectory even with both peripheral
series: the spirantization of velars (of different origin, including palatals) and labials
mechanically mirrors the original development of the central series.
The Indian development later abandoned lenition to such extent that even original
lenited dentals were re-occlusivized (the fortition trajectory). This re-plosivation affected
even spirants, which had arisen from original sibilants before sibilants (SS) since even such
fricatives turned into being newly created plosives (ss > ϑs > ts, šs > çš > kš). Both grave
series, according to the new trajectory, were hence not affected by a spirantization but preserved
as plosives.
However, even opposite trajectories share the same feature: the tendency to leveling:
Iranian lenited all clusters of our interest, Indic undid most of the lenited clusters, and turned
towards fortition. However, in both branches, the structure of clusters was finally more uniform
than that in the transition stage of Late Common Indo-Iranian. The leveling in Indic was
finished in Middle Indo-Aryan, since all clusters of plosive + s or s + plosive are leveled on two
plosive clusters; cf. Pāli √sak- ‘canʼ + si = ao. asakkhi and √as- ‘beʼ + ti = pr. atthi.
83
3 The development of the two-obstruent clusters from Indo-European into
Baltic
3.0 Baltic languages
Though relatively late attested, the Baltic languages, due to their relatively archaicity, play a relative role in the
reconstruction of Indo-European phonology.
Lithuanian, beside older glosses, is attested since 1547 by (Mažvydas’ Catechismvsa), Latvian securely
since 1585 (Casinius’ Catechismus Catholicorum,78
Prussian is attested since 14th
century (if we accept this age of
the Elbing vocabulary) until 16th
century (three catechisms) (cf. Larson/Bukelskytė-Čepelė 2018).
Note: We will use the term Prussian for the same language otherwise named Old Prussian since there is no chance
to confuse it with German dialects (Low Prussian and High Prussian) in the context we use since we exclusively
speak about Baltic languages, not Germanic.
3.1 On the reconstruction of the trajectory of Baltic development
The development of the IE clusters of plosive + t, plosive + dh
, plosive + s, sibilant + t and
sibilant + sibilant in Baltic languages is in its many aspects highly conservative, as the whole
phonemic system of Baltic languages often is (cf. Smoczyński 2001: 167–178 for examples of
archaicity of Baltic phonemic systems).
Note: This tendency to be phonemically conservative is not shared with the morphemic system, especially that of
verbs.
The innovative features of the consonantal phonemes of the Baltic languages are often shared
with other satəm-languages, and we can list especially:
i. Baltic languages, as all satəm-languages, merged original IE labiovelars with plain velars (IE
K, Ku̯
> Balt. K);
ii. similarly to other satəm-languages, again, the reconstructed palatovelars were sibilantized,
the new sibilants are preserved as palatal sibilants in Lithuanian, but secondarily
depalatalized in other Baltic languages (as they were in Slavic or Iranian);
iii. the original IE *s changed to *š according to Pedersen’s Law (ruki-law), though the contexts
and extension of this process are debatable79
; again this process is well attested in Slavic and
Indo-Iranian languages;
iv. the reconstructed IE voiced non-aspirates and voiced aspirates merged into voiced plosives
(IE D, Dh
> Balt. D); the process is shared with Iranian and Slavic80
;
78
A paradox is that the oldest Lithuanian text is Lutheran and the oldest Latvian Catholic catechisms, contrary to
prevailing Christian denominations of the given countries.
79
See below for clusters *K(u̯)
s, *Ḱs, *št, *šs and their assumed development in Baltic languages.
80
Since a similar process also appeared in Celtic, it is probably not a dialectal feature of the satəm- or Balto-Slavic
languages but a universal phenomenon of a common drift (especially since the loss of aspiration affected even
Sinhalese, though Indo-Aryan languages usually preserve this feature). The original distinction between both
modal classes could be traced due to Winter’s Law, which causes the lengthening of vowels before original IE
voiced non-aspirated plosives (Winter 1976; Kortlandt 1978c; Kortlandt 1978d; Kortlandt 1985a; Korlandt
1985b; Kortlandt assumes the original glottal nature of IE voiced non-aspirates; cf. also Sukač 2013, here an
especially detailed overview of the given literature; note that Winter 1979 and Winter 2011 reject any glottalic
explanations). For similar processes in Latin (Lachmann’s Law), Slavic and Tocharian see given chapters.
84
As a primary source, we use Lithuanian, as a language both well attested (in contrast to
Prussian) and phonemically more conservative than Latvian (and Prussian). Latvian and
Prussian data will be used below to illustrate the wider state of Baltic languages in the way of
commentary on the Lithuanian developments.
3.2 The development of two-obstruent clusters in Lithuanian
In the formation of clusters in Baltic, following tendencies are observed:
i. the final voicedness/voicelessness of the cluster is given by the quality of the right obstruent,
there are no signs of Bartholomae’s Law;
ii. the two sibilant clusters, either original Ss or formed by a palatovelar or dental plosive + s,
are simplified to 0s/0š, according to the original left sibilant.
While using Lithuanian, we focused primarily on synchronic alternation (especially those of
the verbal system), the ʽliving fleshʼ of synchronic alternations. The most used forms are:
infinitive or supine for C + t (that Baltic infinitive is of old IE instrumentality is beyond doubt:
cf. Slavic infinitive on -ti, OIA -taye; both old datives of ti-stem; similarly Baltic supine81
has
the related supine forms in Slavic -tъ and L. -tum, OIA infinitive -tum, being old accusatives of
tu-stem (cf. Erhart 1980: 181–182; Erhart 1984: 147–148); both related to verbal abstracta on
ti/tu- (cf. Stang 1966: 394–397), if possible, the old athematic present 3rd
sg. ending *-ti is used.
For sigmatic forms, the future (derived from the infinitive stem) is used (related in its
instrumentality to the sya-future of OIA and sigmatic future of Greek; these sigmatic forms are
related to the sigmatic desideratives and sigmatic aorists; cf. Stang 1966: 397–399; Erhart 1980:
173; Erhart 1984: 142), on sta-iteratives (Stang 1966: 338–349). Additionally, if possible, the
old athematic present 2nd
sg. ending *-si is used.
Note: The old suffix on *dh
- is used to form Baltic imperfect (-davau); this form could be related to reconstructed
dh
-presents (LIV 20, 717). However, the relationship of davau-imperfect to IE forms is debatable.
When needed, the synchronic data are supported by those ʽetymologicalʼ data, i.e., from
unproductive clusters with clear and sure etymology, usually from numerals, if possible, since
they usually have a reliable etymology.
Note: We need to keep in mind that the alternation of voice is not reflected in Lithuanian orthography.
81
Which is later a base for Baltic conditional.
85
3.2.1 The development of clusters labial + t/dh
/s
The labial series has no specific features besides the (de)voicing of the left plosive according
to its right context; its development shares the same general features with the development of
the velar series:
P + t = Lith. pt:
inf. kir̃pti (cf. pr. kerpù ʽcut, shearʼ; < IE *√(s)kerp-; cf. L. carpō ‘grab, graspʼ; cf. Pokorny
IEW: 944–945; Fraenkel LEW: 257–258; LIV2
: 559; Smoczyński 2007: 289; Derksen
2015: 246–247);
inf. lìpti (cf. pr. limpù ʽglueʼ; < IE *√lei̯p-; cf. OIA limpáti ‘smearʼ, OCS -lěpiti ‘stickʼ; cf.
Pokorny IEW: 670–671; Fraenkel LEW: 375–376; LIV2
: 408–409; Smoczyński 2007:
357–358; NIL 453–454; Derksen 2015: 288);
Lith. neptė̃ ʽgranddaughterʼ (< IE *neptī; cf. OIA naptī, L. neptis, OHG nift; cf. Pokorny
IEW: 764; Fraenkel LEW: 494; Smoczyński 2001: 172, 183, 240; Smoczyński 2007:
420; NIL 520–524; Derksen 2015: 332);
Lith. septynì, Latv. septiņi ʽsevenʼ, Pruss. septmas ʽseventhʼ (< IE *septm̥; cf. OIA saptá-,
L. septem; cf. Pokorny IEW: 909; Fraenkel LEW: 776; Stang 1966: 279, 283; Comrie
1992: 756–759; Blažek 1999: 246–250; Smoczyński 2007: 543; Derksen 2015: 393–
394);
inf. dìrbti (cf. pr. dìrbu ʽworkʼ; < IE *√derbh
-; cf. OIA pr. part. dṛbhánti- ‘forming tuftsʼ;
cf. Pokorny IEW: 211–212, 257; Fraenkel LEW: 82; LIV2
: 121; Smoczyński 2007:
114–115; Derksen 2015: 131);
inf. del̃bti (cf. pr. delbiù ʽput lowerʼ; < IE *√dh
elbh
-; cf. OE delfan ‘digʼ; cf. Pokorny IEW:
246; Fraenkel LEW: 81; LIV2
: 143; Smoczyński 2007: 112–113; Derksen 2015: 120–
129–130);
inf. grė́bti (cf. pr. grė́biu ʽrakeʼ; < IE *√grebh
-; cf. Gr. γράφω ‘scratch, grazeʼ; cf. Pokorny
IEW: 392; Fraenkel LEW: 165–166; LIV2
: 187; Smoczyński 2007: 196–197; Derksen
2015: 186);
P + s = Lith. ps:
fut. kir̃psiu (cf. pr. kerpù ʽcut, shearʼ; < IE *√(s)kerp-; cf. L. carpō ‘grab, graspʼ; cf.
Pokorny IEW: 944–945; Fraenkel LEW: 257–258; LIV2
: 559; Smoczyński 2007: 289;
Derksen 2015: 246–247);
fut. lìpsiu (cf. pr. limpù ʽglueʼ; < IE *√lei̯p-; cf. OIA limpáti ‘smearʼ, OCS -lěpiti ‘stickʼ;
cf. Pokorny IEW: 670–671; Fraenkel LEW: 375–376; LIV2
: 408–409; Smoczyński
2007: 357–358; NIL 453–454; Derksen 2015: 288);
fut. dìrbsiu (cf. pr. dìrbu ʽworkʼ; < IE *√derbh
-; cf. OIA pr. part. dṛbhánti- ‘forming tuftsʼ;
cf. Pokorny IEW: 211–212, 257; Fraenkel LEW: 82; LIV2
: 121; Smoczyński 2007:
114–115; Derksen 2015: 131);
fut. del̃bsiu fut. dìrbsiu (cf. pr. dìrbu ʽworkʼ; < IE *√derbh
-; cf. OIA pr. part. dṛbhánti‘forming
tuftsʼ; cf. Pokorny IEW: 211–212, 257; Fraenkel LEW: 82; LIV2
: 121;
Smoczyński 2007: 114–115; Derksen 2015: 131);
fut. grė́bsiu grė́bti (cf. pr. grė́biu ʽrakeʼ; < IE *√grebh
-; cf. Gr. γράφω ‘scratch, grazeʼ; cf.
Pokorny IEW: 392; Fraenkel LEW: 165–166; LIV2
: 187; Smoczyński 2007: 196–197;
Derksen 2015: 186);
86
P + dh
= Lith. bd:
impf. kir̃pdavau (cf. pr. kerpù ʽcut, shearʼ; < IE *√(s)kerp-; cf. L. carpō ‘grab, graspʼ; cf. Pokorny IEW:
944–945; Fraenkel LEW: 257–258; LIV2
: 559; Smoczyński 2007: 289; Derksen 2015: 246–247);
impf. lìpdavau (cf. pr. limpù ʽglueʼ; < IE *√lei̯p-; cf. OIA limpáti ‘smearʼ, OCS -lěpiti ‘stickʼ; cf. Pokorny
IEW: 670–671; Fraenkel LEW: 375–376; LIV2
: 408–409; Smoczyński 2007: 357–358; NIL 453–454;
Derksen 2015: 288);
impf. dìrbdavau (cf. pr. dìrbu ʽworkʼ; < IE *√derbh
-; cf. OIA pr. part. dṛbhánti- ‘forming tuftsʼ; cf. Pokorny
IEW: 211–212, 257; Fraenkel LEW: 82; LIV2
: 121; Smoczyński 2007: x; Derksen 2015: 131);
impf. del̃bdavau (cf. pr. delbiù ʽput lowerʼ; < IE *√dh
elbh
-; cf. OE delfan ‘digʼ; cf. Pokorny IEW: 246;
Fraenkel LEW: 81; LIV2
: 143; Smoczyński 2007: 112–113; Derksen 2015: 120; Smoczyński 2007: x;
Derksen 2015: 120);
impf. grė́bdavau grė́bti (cf. pr. grė́biu ʽrakeʼ; < IE *√grebh
-; cf. Gr. γράφω ‘scratch, grazeʼ; cf. Pokorny IEW:
392; Fraenkel LEW: 165–166; LIV2
: 187; Smoczyński 2007: 196–197; Derksen 2015: 186);
3.2.2 The development of clusters (labio)velar + t/dh
/s
The plain velars and labiovelars merged in plain velars; as in all satəm-languages; voiced
phonemes merged in all positions.
The development could be expressed by formulae (where K = IE. *k/g/gh
/ ku̯
/gu̯
/gu̯h
):
K(u̯ )
+ t = Lith. kt:
inf. lė̃kti (cf. pr. lekiù ʽflyʼ; < IE *√lek-; cf. MHG lecken ‘knock out with feetʼ, OCS letěti
‘flyʼ; cf. Pokorny IEW: 673; Fraenkel LEW: 353–354; LIV2
: 411; Smoczyński 2007:
343–344; Derksen 2015: 278);
inf. sèkti (cf. pr. senkù diminish, fall, sinkʼ; < IE *√sek-; cf. OIA ásaścant- ‘dry upʼ; OCS
i-sȩčetъ ‘dry outʼ ; cf. Pokorny IEW: 894–895; Fraenkel LEW: 772–773; LIV2
: 523–
524; Smoczyński 2007: 541; Derksen 2015: 392);
inf. jùngti (cf. pr. jùngiu ʽyoke upʼ; cf. OIA yunákti ‘harnessʼ, L. iungō ‘joinʼ; < IE *√i̯ung;
cf. Pokorny IEW: 508–510; Fraenkel LEW: 196–197; LIV2
: 316; Smoczyński 2007:
237–238; NIL 397–404; Derksen 2015: 214);
inf. dérgti, drė́gti (cf. pr. dérgia ʽ, moist, rainʼ; < IE *√dh
reH2gh
-; cf. Gr. θράσσω, Att.
θράττω ‘trouble, disturbʼ, OCS raz-dražǫ ‘provoke angerʼ; cf. Pokorny IEW: 251, 273;
Fraenkel LEW: 103; LIV2
: 154–155; Smoczyński 2007: x; Derksen 2015: 12382
);
inf. sir̃gti (cf. pr. sergù, OLith. sérgmi ʽbe illʼ; < IE *√su̯ergh
-; cf. OIA sūrkṣata ‘careʼ (?),
OHG sorgēn ‘careʼ; cf. Pokorny IEW: 1051; Fraenkel LEW: 776–777; LIV2
: 613–
614; Smoczyński 2007: 550; Derksen 2015: 399);83
OLith. pr. liekti, inf. lìkti (cf. pr. liekù ʽhold, keepʼ; < IE *√lei̯ku̯
-; cf. OIA rikthās ‘protrude
beyondʼ, L. līquī ‘leaveʼ; cf. Pokorny IEW: 669–670; Fraenkel LEW: 372; LIV2
: 406–
407; Smoczyński 2007: 355–356; Derksen 2015: 287);
inf. sèkti (cf. pr. sekù ʽfollowʼ; < IE *√seku̯
-; cf. OIA sáścati ‘accompanyʼ, L. sequor
‘followʼ; cf. Pokorny IEW: 896–897; Fraenkel LEW: 773; LIV2
: 525–526;
Smoczyński 2007: 540–541; Derksen 2015: 392);
Lith. peñktas, Latv. pìektais, Pruss. penckts ʽfifthʼ (< IE *penku̯
-to-; cf. OIA paktháḥ, Gr.
πέμπτος; Stang 1966: 283; Comrie 1992: 752–754; Blažek 1999: 221, 224; Pokorny
IEW: 808; Fraenkel LEW: 570; Smoczyński 2007: 450; Derksen 2015: 351);
inf. bė́gti (cf. pr. bė́gu, OLith. bė́gmi ʽrunʼ; < IE *√bh
egu̯
-; cf. Gr. φέβομαι ‘put to glight,
fleeʼ, OCS -běgnǫti ‘runʼ; cf. Pokorny IEW: 116; Fraenkel LEW: 38; LIV2
: 67;
Smoczyński 2007: 52–53; Derksen 2015: 85–86);
82
Derksen reconstructs IE *√dh
erg- and considers the relationship between Lith. and Gr. verbs problematic (l.c.).
83
Fraenkel reconstructs IE *√sergh
- (l.c.).
87
inf. dègti (cf. pr. degù ʽburnʼ; < IE *√dh
egu̯h
-; cf. OIA dáhati ‘burnʼ, Alb. djeg ‘burntʼ; cf.
Pokorny IEW: 240–241; Fraenkel LEW: 85–86; LIV2
: 133–134) ; Smoczyński 2007:
97–98; Derksen 2015: 119);
inf. snìgti (cf. pr. sniẽga ʽsnowʼ; < IE *√snei̯gu̯h
-; cf. Gr. νείφω, L. ningit ‘snowʼ; cf.
Pokorny IEW: 974; Fraenkel LEW: 853; LIV2
: 573; Smoczyński 2007: 591; NIL 622–
625; Derksen 2015: 416);
K(u̯ )
+ s = Lith. ks:
fut. lė̃ksiu (cf. pr. lekiù ʽflyʼ; < IE *√lek-; cf. MHG lecken ‘knock out with feetʼ, OCS letěti
‘flyʼ; cf. Pokorny IEW: 673; Fraenkel LEW: 353–354; LIV2
: 411; Smoczyński 2007:
343–344; Derksen 2015: 278);
fut. sèksiu (cf. pr. senkù diminish, fall, sinkʼ; < IE *√sek-; cf. OIA ásaścant- ‘dry upʼ; OCS
i-sȩčetъ ‘dry outʼ ; cf. Pokorny IEW: 894–895; Fraenkel LEW: 772–773; LIV2
: 523–
524; Smoczyński 2007: 541; Derksen 2015: 392);
fut. jùngsiu (cf. pr. jùngiu ʽyoke upʼ; cf. OIA yunákti ‘harnessʼ, L. iungō ‘joinʼ; < IE *√i̯ung;
cf. Pokorny IEW: 508–510; Fraenkel LEW: 196–197; LIV2
: 316; Smoczyński 2007:
237–238; NIL 397–404; Derksen 2015: 214);
fut. dérgs (cf. pr. dérgia ʽ, moist, rainʼ; < IE *√dh
reH2gh
-; cf. Gr. θράσσω, Att. θράττω
‘trouble, disturbʼ, OCS raz-dražǫ ‘provoke angerʼ; cf. Pokorny IEW: 251, 273;
Fraenkel LEW: 103; LIV2
: 154–155; Smoczyński 2007: x; Derksen 2015: 12384
);
fut. sir̃gsiu (cf. pr. sergù, OLith. sérgmi ʽbe illʼ; < IE *√su̯ergh
-; cf. OIA sūrkṣata ‘careʼ
(?), OHG sorgēn ‘careʼ; cf. Pokorny IEW: 1051; Fraenkel LEW: 776–777; LIV2
: 613–
614; Smoczyński 2007: 550; Derksen 2015: 399);
fut. lìksiu (cf. pr. liekù ʽhold, keepʼ; < IE *√lei̯ku̯
-; cf. OIA rikthās ‘protrude beyondʼ, L.
līquī ‘leaveʼ; cf. Pokorny IEW: 669–670; Fraenkel LEW: 372; LIV2
: 406–407;
Smoczyński 2007: 355–356; Derksen 2015: 287);
fut. sèksiu (cf. pr. sekù ʽfollowʼ; < IE *√seku̯
-; cf. OIA sáścati ‘accompanyʼ, L. sequor
‘followʼ; cf. Pokorny IEW: 896–897; Fraenkel LEW: 773; LIV2
: 525–526;
Smoczyński 2007: 540–541; Derksen 2015: 392);
fut. bė́gsiu (cf. pr. bė́gu, OLith. bė́gmi ʽrunʼ; < IE *√bh
egu̯
-; cf. Gr. φέβομαι ‘put to glight,
fleeʼ, OCS -běgnǫti ‘runʼ; cf. Pokorny IEW: 116; Fraenkel LEW: 38; LIV2
: 67;
Smoczyński 2007: 52–53; Derksen 2015: 85–86);
fut. dègsiu (cf. pr. degù ʽburnʼ; < IE *√dh
egu̯h
-; cf. OIA dáhati ‘burnʼ, Alb. djeg ‘burntʼ; cf.
Pokorny IEW: 240–241; Fraenkel LEW: 85–86; LIV2
: 133–134) ; Smoczyński 2007:
97–98; Derksen 2015: 119);
fut. snìgs (cf. pr. sniẽga ʽsnowʼ; < IE *√snei̯gu̯h
-; cf. Gr. νείφω, L. ningit ‘snowʼ; cf.
Pokorny IEW: 974; Fraenkel LEW: 853; LIV2
: 573; Smoczyński 2007: 591; NIL 622–
625; Derksen 2015: 416);
K(u̯ )
+ dh
= Lith. gd:
impf. (cf. pr. lekiù ʽflyʼ; < IE *√lek-; cf. MHG lecken ‘knock out with feetʼ, OCS letěti ‘flyʼ; cf. Pokorny
IEW: 673; Fraenkel LEW: 353–354; LIV2
: 411; Smoczyński 2007: 343–344; Derksen 2015: 278);
impf. sèkdavau (cf. pr. senkù diminish, fall, sinkʼ; < IE *√sek-; cf. OIA ásaścant- ‘dry upʼ; OCS i-sȩčetъ
‘dry outʼ ; cf. Pokorny IEW: 894–895; Fraenkel LEW: 772–773; LIV2
: 523–524; Smoczyński 2007:
541; Derksen 2015: 392);
impf. jùngdavau (cf. pr. jùngiu ʽyoke upʼ; cf. OIA yunákti ‘harnessʼ, L. iungō ‘joinʼ; < IE *√i̯ung-; cf.
Pokorny IEW: 508–510; Fraenkel LEW: 196–197; LIV2
: 316; Smoczyński 2007: 237–238; NIL 397–
404; Derksen 2015: 214);
84
Derksen reconstructs IE *√dh
erg- and considers the relationship between Lith. and Gr. verbs problematic (l.c.).
88
impf. dérgdavo (cf. pr. dérgia ʽ, moist, rainʼ; < IE *√dh
reH2gh
-; cf. Gr. θράσσω, Att. θράττω ‘trouble,
disturbʼ, OCS raz-dražǫ ‘provoke angerʼ; cf. Pokorny IEW: 251, 273; Fraenkel LEW: 103; LIV2
: 154–
155; Smoczyński 2007: x; Derksen 2015: 123);
impf. sir̃gdavau (cf. pr. sergù, OLith. sérgmi ʽbe illʼ; < IE *√su̯ergh
-; cf. OIA sūrkṣata ‘careʼ (?), OHG
sorgēn ‘careʼ; cf. Pokorny IEW: 1051; Fraenkel LEW: 776–777; LIV2
: 613–614; Smoczyński 2007:
550; Derksen 2015: 399);
impf. lìkdavau (cf. pr. liekù ʽhold, keepʼ; < IE *√lei̯ku̯
-; cf. OIA rikthās ‘protrude beyondʼ, L. līquī ‘leaveʼ;
cf. Pokorny IEW: 669–670; Fraenkel LEW: 372; LIV2
: 406–407; Smoczyński 2007: 355–356; Derksen
2015: 287);
impf. sèkdavau (cf. pr. sekù ʽfollowʼ; < IE *√seku̯
-; cf. OIA sáścati ‘accompanyʼ, L. sequor ‘followʼ; cf.
Pokorny IEW: 896–897; Fraenkel LEW: 773; LIV2
: 525–526; Smoczyński 2007: 540–541; Derksen
2015: 392);
impf. bė́gdavau (cf. pr. bė́gu, OLith. bė́gmi ʽrunʼ; < IE *√bh
egu̯
-; cf. Gr. φέβομαι ‘put to glight, fleeʼ, OCS běgnǫti
‘runʼ; cf. Pokorny IEW: 116; Fraenkel LEW: 38; LIV2
: 67; Smoczyński 2007: 52–53; Derksen
2015: 85–86);
impf. dègdavau (cf. pr. degù ʽburnʼ; < IE *√dh
egu̯h
-; cf. OIA dáhati ‘burnʼ, Alb. djeg ‘burntʼ; cf. Pokorny
IEW: 240–241; Fraenkel LEW: 85–86; LIV2
: 133–134) ; Smoczyński 2007: 97–98; Derksen 2015:
119);
impf. snìgdavo (cf. pr. sniẽga ʽsnowʼ; < IE *√snei̯gu̯h
-; cf. Gr. νείφω, L. nīuit ‘snowʼ; cf. Pokorny IEW: 974;
Fraenkel LEW: 853; LIV2
: 573; Smoczyński 2007: 591; NIL 622–625; Derksen 2015: 416);
3.2.3 The development of clusters palatovelar + t/dh
/s
The original palatovelars are realized as palatal sibilants in Lithuanian, both voiced phonemes
merged. Before *t- both sibilants are realized as -š (i.e., devoiced), before *s- the palatovelars
are lost, and an original sibilant has become palatalized (i.e., Ḱ+s > 0š).
The development could be expressed by formulae (where Ḱ = IE. *ḱ/ǵ/ǵh
):
Ḱ + t = Lith. št:
inf. nèšti (cf. pr. nešù ʽcarryʼ; < IE *√Hneḱ-; cf. Toch. B eṅtär ‘graspʼ, OCS nošǫ ‘carryʼ;
cf. Pokorny IEW: 316–318; Fraenkel LEW: 497–498; LIV2
: 250–251; Smoczyński
2007: 423; Derksen 2015: 334);85
inf. pèšti (cf. pr. pešù ʽpluck, pickʼ; < IE *√peḱ-; cf. Gr. πέκω ‘combʼ, L. pectō ‘comb,
shearʼ; cf. Pokorny IEW: 797; Fraenkel LEW: 580–581; LIV2
: 467; Smoczyński 2007:
453; Derksen 2015: 353);
Lith. aštuonì, Latv. astoņi ʽeightʼ (< IE *oḱtō-ni-; cf. OIA aṣṭau, L. octō; cf. Fraenkel
LEW: 19–20; Stang 1966: 279, 283–284; Comrie 1992: 758–760; Blažek 1999: 267;
Pokorny IEW: 775; Smoczyński 2007: 27; Derksen 2015: 64–65);
inf. réižti (cf. pr. reižiù ʽstretchʼ; < IE *√rei̯ ǵ-; cf. OIr. rigid ‘stretch out, ruleʼ; cf. Pokorny
IEW: 862; Fraenkel LEW: 715; LIV2
: 503; Smoczyński 2007: 512–513; Derksen
2015: 380–381);86
inf. mìlžti, mélžti (cf. pr. mélžu ʽmilkʼ; < IE *√H2melǵ-; cf. Gr. ἀμέλγω, L. mulgeō ‘milkʼ;
cf. Pokorny IEW: 722–723; Fraenkel LEW: 434–435; LIV2
: 279; Smoczyński 2007:
387–388; Derksen 2015: 310–311);
inf. liẽžti (cf. pr. liežiù ʽlickʼ; < IE *√lei̯ǵh
-; cf. OIA réḍhi, Gr. λείχω ‘lickʼ; cf. Pokorny
IEW: 668; Fraenkel LEW: 369; LIV2
: 404; Smoczyński 2007: 353; Derksen 2015:
285);
85
LIV reconstructs either *H1neḱ- or *H2neḱ-, Derksen H1neḱ- (l.c.), since this question does not concert the final
plosive, we use a neutral symbol.
86
Derksen reconstructs IE *√H3reǵ- (l.c.).
89
inf. mỹžti (cf. pr. mȩžù, OLith. minžu ʽurinateʼ; < IE *√H3mei̯ǵh
-; cf. OIA méhati, L. mingō
‘urinateʼ; cf. Pokorny IEW: 713; Fraenkel LEW: 461–462; LIV2
: 301–302; NIL 384–
385; Smoczyński 2007: 407; Derksen 2015: 322);
Ḱ + s = Lith. 0š:
fut. nèšiu (cf. pr. nešù ʽcarryʼ; < IE *√Hneḱ-; cf. Toch. B eṅtär ‘graspʼ, OCS nošǫ ‘carryʼ;
cf. Pokorny IEW: 316–318; Fraenkel LEW: 497–498; LIV2
: 250–251; Smoczyński
2007: 423; Derksen 2015: 334);
fut. pèšiu (cf. pr. pešù ʽpluck, pickʼ; < IE *√peḱ-; cf. Gr. πέκω ‘combʼ, L. pectō ‘comb,
shearʼ; cf. Pokorny IEW: 797; Fraenkel LEW: 580–581; LIV2
: 467; Smoczyński 2007:
453; Derksen 2015: 353);
ašìs ʽaxle, axisʼ (< IE *H2eḱs-i-; cf. OIA ákṣa-, L. axis; cf. Pokorny IEW: 6; Fraenkel
LEW: 19; Smoczyński 2007: 26; NIL 259–262; Derksen 2015: 63);
fut. réišiu (cf. pr. reižiù ʽstretchʼ; < IE *√rei̯ ǵ-; cf. OIr. rigid ‘stretch out, ruleʼ; cf. Pokorny
IEW: 862; Fraenkel LEW: 715; LIV2
: 503; Smoczyński 2007: 512–513; Derksen
2015: 380–381);
fut. mìlšiu (cf. pr. mélžu ʽmilkʼ; < IE *√H2melǵ-; cf. Gr. ἀμέλγω, L. mulgeō ‘milkʼ; cf.
Pokorny IEW: 722–723; Fraenkel LEW: 434–435; LIV2
: 279; Smoczyński 2007: 387–
388; Derksen 2015: 310–311);
fut. liẽšiu (cf. pr. liežiù ʽlickʼ; < IE *√lei̯ǵh
-; cf. OIA réḍhi, Gr. λείχω ‘lickʼ; cf. Pokorny
IEW: 668; Fraenkel LEW: 369; LIV2
: 404; Smoczyński 2007: 353; Derksen 2015:
285);
fut. mỹšiu (cf. pr. mȩžù, OLith. minžu ʽurinateʼ; < IE *√H3mei̯ǵh
-; cf. OIA méhati, L. mingō
‘urinateʼ; cf. Pokorny IEW: 713; Fraenkel LEW: 461–462; LIV2
: 301–302;
Smoczyński 2007: 407; NIL 384–385; Derksen 2015: 322);
Ḱ + dh
= Lith. žd:
impf. nèšdavau (cf. pr. nešù ʽcarryʼ; < IE *√Hneḱ-; cf. Toch. B eṅtär ‘graspʼ, OCS nošǫ ‘carryʼ; cf. Pokorny
IEW: 316–318; Fraenkel LEW: 497–498; LIV2
: 250–251; Smoczyński 2007: 423; Derksen 2015: 334);
impf. pèšdavau (cf. pr. pešù ʽpluck, pickʼ; < IE *√peḱ-; cf. Gr. πέκω ‘combʼ, L. pectō ‘comb, shearʼ; cf.
Pokorny IEW: 797; Fraenkel LEW: 580–581; LIV2
: 467; Smoczyński 2007: 453; Derksen 2015: 353);
impf. réiždavau (cf. pr. reižiù ʽstretchʼ; < IE *√rei̯ ǵ-; cf. OIr. rigid ‘stretch out, ruleʼ; cf. Pokorny IEW: 862;
Fraenkel LEW: 715; LIV2
: 503; Smoczyński 2007: 512–513; Derksen 2015: 380–381);
impf. mìlždavau (cf. pr. mélžu ʽmilkʼ; < IE *√H2melǵ-; cf. Gr. ἀμέλγω, L. mulgeō ‘milkʼ; cf. Pokorny IEW:
722–723; Fraenkel LEW: 434–435; LIV2
: 279; Smoczyński 2007: 387–388; Derksen 2015: 310–311);
impf. liẽždavau (cf. pr. liežiù ʽlickʼ; < IE *√lei̯ǵh
-; cf. OIA réḍhi, Gr. λείχω ‘lickʼ; cf. Pokorny IEW: 668;
Fraenkel LEW: 369; LIV2
: 404; Smoczyński 2007: 353; Derksen 2015: 285);
impf. mỹždavau (cf. pr. mȩžù, OLith. minžu ʽurinateʼ; < IE *√H3mei̯ǵh
-; cf. OIA méhati, L. mingō ‘urinateʼ;
cf. Pokorny IEW: 713; Fraenkel LEW: 461–462; LIV2
: 301–302; Smoczyński 2007: 407; NIL 384–
385; Derksen 2015: 322);
3.2.4 The development of clusters dental + t/dh
/s
The development of the dental series has a typical (and not specifically Baltic) development of
Tt > st. The clusters of Ts developed into 0s; i.e., the plosive was probably first sibilantized and
lost after due to degemination, similarly to the development of palatovelars.
The development could be expressed by formulae (where T = IE. *t/d/dh
):
90
T + t = Lith. st:
inf. kir̃sti (cf. pr. kertù ʽhew, string, hit, strikeʼ; < IE *√(s)kert-; cf. OIA kṛntáti ‘cutʼ; cf.
Pokorny IEW: 941–942; Fraenkel LEW: 258; LIV2
: 559–560; Smoczyński 2007: 289–
290; Derksen 2015: 247–248);
inf. ver̃sti (cf. pr. verčiù ʽfell, turn overʼ; < IE *√u̯ert-; cf. OIA vártate, OCS vraštǫ, L.
uertor ‘turnʼ; cf. Pokorny IEW: 1156–1158; Fraenkel LEW: 1228; LIV2
: 691–962;
Smoczyński 2007: 739–740; Derksen 2015: 498);
inf. ė́sti (cf. pr. ėdù ʽeatʼ; < IE *√H1ed-; cf. Hitt. ēdmi, OIA átti, L. edō ‘eatʼ; cf. Pokorny
IEW: 287–289; Fraenkel LEW: 124–125; LIV2
: 230–231; Smoczyński 2007: 149; NIL
208–220; Derksen 2015: 157–158);
pr. OLith. duostí (cf. pr. dúodu ʽgiveʼ; < IE *√deH3-; cf. OIA ádāt, OCS dati ‘giveʼ; cf.
Pokorny IEW: 223–225; Fraenkel LEW: 111–112; LIV2
: 105–106; Smoczyński 2007:
134–135; Derksen 2015: 146–147);
inf. sė́sti (cf. pr. sė̀dù ʽsitʼ; < IE *√sed-; cf. OIA sī́dati, L. sīdō ‘sitʼ; cf. Pokorny IEW: 884–
887; Fraenkel LEW: 777; LIV2
: 513–515; Smoczyński 2007: 538–539; NIL 590–600;
Derksen 2015: 395);
inf. bùsti (cf. pr. bundù ʽwake upʼ; < IE *√bh
eu̯dh
-; cf. OIA bódhati ‘noticeʼ, Gr. πεύθομαι
‘give noticeʼ; cf. Pokorny IEW: 150–152; Fraenkel LEW: 62; LIV2
: 82–83;
Smoczyński 2007: 78–79; NIL 36–37; Derksen 2015: 83, 107);
T + s = Lith. 0s:
fut. kir̃siu (cf. pr. kertù ʽhew, string, hit, strikeʼ; < IE *√(s)kert-; cf. OIA kṛntáti ‘cutʼ; cf.
Pokorny IEW: 941–942; Fraenkel LEW: 258; LIV2
: 559–560; Smoczyński 2007: 289–
290; Derksen 2015: 247–248);
fut. ver̃siu (cf. pr. verčiù ʽfell, turn overʼ; < IE *√u̯ert-; cf. OIA vártate, OCS vraštǫ, L.
uertor ‘turnʼ; cf. Pokorny IEW: 1156–1158; Fraenkel LEW: 1228; LIV2
: 691–962;
Smoczyński 2007: 739–740; Derksen 2015: 498);etymological development: Lith.
giesmė̃, Lat. dziêsma but Lith. giedóti, Latv. dziêdat, Pruss. waisei but waidimai (Stang
1966: 107);
fut. ė́siu (cf. pr. ėdù ʽeatʼ; < IE *√H1ed-; cf. Hitt. ēdmi, OIA átti, L. edō ‘eatʼ; cf. Pokorny
IEW: 287–289; Fraenkel LEW: 124–125; LIV2
: 230–231; Smoczyński 2007: 149; NIL
208–220; Derksen 2015: 157–158);fut. sė́siu (cf. pr. sė̀dù ʽsitʼ; < IE *√sed-);
fut. bùsiu (cf. pr. bundù ʽwake upʼ; < IE *√bh
eu̯dh
-; cf. OIA bódhati ‘noticeʼ, Gr. πεύθομαι
‘give noticeʼ; cf. Pokorny IEW: 150–152; Fraenkel LEW: 62; LIV2
: 82–83;
Smoczyński 2007: 78–79; NIL 36–37; Derksen 2015: 83, 107);
T + dh
= Lith. zd:
impf. kir̃sdavau (cf. pr. kertù ʽhew, string, hit, strikeʼ; < IE *√(s)kert-; cf. OIA kṛntáti ‘cutʼ; cf. Pokorny
IEW: 941–942; Fraenkel LEW: 258; LIV2
: 559–560; Smoczyński 2007: 289–290; Derksen 2015: 247–
248);
impf. ver̃sdavau (cf. pr. verčiù ʽfell, turn overʼ; < IE *√u̯ert-; cf. OIA vártate, OCS vraštǫ, L. uertor ‘turnʼ;
cf. Pokorny IEW: 1156–1158; Fraenkel LEW: 1228; LIV2
: 691–962; Smoczyński 2007: 739–740;
Derksen 2015: 498);-);
impf. ė́sdavau (cf. pr. ėdù ʽeatʼ; < IE *√H1ed-; cf. Hitt. ēdmi, OIA átti, L. edō ‘eatʼ; cf. Pokorny IEW: 287–
289; Fraenkel LEW: 124–125; LIV2
: 230–231; Smoczyński 2007: 149; NIL 208–220; Derksen 2015:
157–158);
impf. sė́sdavau(cf. pr. sė̀dù ʽsitʼ; < IE *√sed-; cf. OIA sī́dati, L. sīdō ‘sitʼ; cf. Pokorny IEW: 884–887;
Fraenkel LEW: 777; LIV2
: 513–515; Smoczyński 2007: 538–539; NIL 590–600; Derksen 2015: 395);
impf. bùsdavau (cf. pr. bundù ʽwake upʼ; < IE *√bh
eu̯dh
-; cf. OIA bódhati ‘noticeʼ, Gr. πεύθομαι ‘give
noticeʼ; cf. Pokorny IEW: 150–152; Fraenkel LEW: 62; LIV2
: 82–83; Smoczyński 2007: 78–79; NIL
36–37; Derksen 2015: 83, 107);
91
3.2.5 The development of clusters sibilant + t/dh
/s
Both original Baltic sibilants (*s and *š, the second developed from IE *s due to Pedersen’s
Law/ruki-rule) are preserved in Lithuanian before t- and lost before s-, the outcome of the š+s
cluster is 0š.
The development could be expressed by formulae:
s + t = Lith. st:
pr. ẽsti (cf. pr. OLith esmì ʽbeʼ; < IE *√H1es-; cf. OIA ásti, L. est ʽbeʼ; cf. Pokorny IEW:
340–341; Fraenkel LEW: 124; LIV2
: 241–242; Smoczyński 2007: 148; NIL 235–238;
Derksen 2015: 157);
inf. júosti (cf. pr. júosiu ʽgirdʼ; < IE *√i̯eH3s-; cf. Gr. ζώννῡμι ʽgirdʼ, OCS po-jašǫ ʽgird upʼ;
cf. Pokorny IEW: 513; Fraenkel LEW: 198; LIV2
: 311; Smoczyński 2007: 239;
Derksen 2015: 214–215);
inf. kláusti (cf. pr. kláusiu ʽaskʼ; < IE *√ḱleu̯s-; cf. cf. OIA śróṣan ʽobeyʼ, Toch. A klyoṣäs,
B klyauṣäm’listenʼ; cf. Pokorny IEW: 606–607; Fraenkel LEW: 265–267; LIV2
: 336–
337; NIL 432–434; Smoczyński 2007: 294–295; Derksen 2015: 249);
Note: The palatal sibilant š, which has arisen due to Pedersen’s Law, is well attested after r (Stang 1966: 95), but
scarcely after other original triggers, cf. sunsù, sùsti; kláusiu, kláusti without palatalization.
Note: The forms from √klaus- ‘askʼ we would expect to be under Pedersen’s law.
š + t = Lith. št:
inf. kar̃šti (cf. pr. karšiù ʽcomb, cardʼ; < IE *√(s)kers-; cf. L. carrō ‘cardʼ cf. Pokorny IEW:
552–553; Fraenkel LEW: 224; LIV2
: 559; Smoczyński 2007: 258–259; Derksen 2015:
228);
inf. pur̃kšti (cf. pr. purškiù ʽsplashʼ; < IE *√pres-87
; cf. OIA pṛ́ṣant- ‘sprinkledʼ, Toch. B
prantsäṃ ‘sprinkleʼ; cf. Pokorny IEW: 823; Fraenkel LEW: 673; LIV2
: 492–493
Smoczyński 2007: 490);
Lith. pir̃štas (cf. Latv. pìrst, Pruss. pirsten ʽfingerʼ; < IE *prstH2o-; cf. Pokorny IEW: 813;
Fraenkel LEW: 598; Smoczyński 2007: 464; NIL 637–659; Derksen 2015: 358);
inf. aũšti ʽdawn, break dayʼ (< IE *H2u̯es-; cf. Lith. aušrà ʽdawnʼ, OIA uccháti ‘shineʼ; cf.
Pokorny IEW: 86–87; Fraenkel LEW: 27; LIV2
: 292–293; Smoczyński 2007: 35–36;
Derksen 2015: 72);
s + s = Lith. 0s:
pr. esì (cf. pr. OLith esmì ʽbeʼ; < IE *√H1es-; cf. OIA ásti, L. est ʽbeʼ; cf. Pokorny IEW:
340–341; Fraenkel LEW: 124; LIV2
: 241–242; Smoczyński 2007: 148; NIL 235–238;
Derksen 2015: 157);
fut. júosiu (cf. pr. júosiu ʽgirdʼ; < IE *√i̯eH3s-; cf. Gr. ζώννῡμι ʽgirdʼ, OCS po-jašǫ ʽgird
upʼ; cf. Pokorny IEW: 513; Fraenkel LEW: 198; LIV2
: 311; Smoczyński 2007: 239;
NIL 391–392; Derksen 2015: 214–215);
fut. kláusiu (cf. pr. kláusiu ʽaskʼ; < IE *√ḱleu̯s-; cf. cf. OIA śróṣan ʽobeyʼ, Toch. A klyoṣäs,
B klyauṣäm’listenʼ; cf. Pokorny IEW: 606–607; Fraenkel LEW: 265–267; LIV2
: 336–
337; Smoczyński 2007: 294–295; NIL 432–434; Derksen 2015: 249);
87
The plosive k is inserted, cf. Lith. ankštas ʽcloseʼ but OIA aṁhú-, OCS ǫzъkъ, L. angustus; Lith. áuksas ʽgoldʼ
but. Pruss. ausis, L. aurum, cf. Stang (1966: 108–109)
92
š + s = Lith. 0š:
fut. kar̃šiu (cf. pr. karšiù ʽcomb, cardʼ; < IE *√(s)kers-; cf. L. carrō ‘cardʼ cf. Pokorny
IEW: 552–553; Fraenkel LEW: 224; LIV2
: 559; Smoczyński 2007: 258–259; Derksen
2015: 228);
fut. pur̃kšiu (cf. pr. purškiù ʽsplashʼ; < IE *√pres-88
; cf. OIA pṛ́ṣant- ‘sprinkledʼ, Toch. B
prantsäṃ ‘sprinkleʼ; cf. Pokorny IEW: 823; Fraenkel LEW: 673; LIV2
: 492–493
Smoczyński 2007: 490);
s + dh
= Lith. zd:
impf. júosdavau (cf. pr. júosiu ʽgirdʼ; < IE *√i̯eH3s-; cf. Gr. ζώννῡμι ʽgirdʼ, OCS po-jašǫ ʽgird upʼ; cf. Pokorny
IEW: 513; Fraenkel LEW: 198; LIV2
: 311; Smoczyński 2007: 239; NIL 391–392; Derksen 2015: 214–
215);
impf. kláusdavau (cf. pr. kláusiu ʽaskʼ; < IE *√ḱleu̯s-; cf. cf. OIA śróṣan ʽobeyʼ, Toch. A klyoṣäs, B
klyauṣäm’listenʼ; cf. Pokorny IEW: 606–607; Fraenkel LEW: 265–267; LIV2
: 336–337; NIL 432–434;
Smoczyński 2007: 294–295; Derksen 2015: 249);
š + dh
= žd:
impf. kar̃šdavau (cf. pr. karšiù ʽcomb, cardʼ; < IE *√(s)kers-; cf. L. carrō ‘cardʼ cf. Pokorny IEW: 552–553;
Fraenkel LEW: 224; LIV2
: 559; Smoczyński 2007: 258–259; Derksen 2015: 228);
impf. pur̃kšdavau (cf. pr. purškiù ʽsplashʼ; < IE *√pres-89
; cf. OIA pṛ́ṣant- ‘sprinkledʼ, Toch. B prantsäṃ
‘sprinkleʼ; cf. Pokorny IEW: 823; Fraenkel LEW: 673; LIV2
: 492–493 Smoczyński 2007: 490);
3.2.6 The overview of the Lithuanian alternations
Summing up the developments in Lithuanian, we should remark that clusters of K + s are
realized as ks, surprisingly not as kš, according to the ruki-rule, though the cluster of Ḱs is
realized as 0š, hence reflecting *kš with the ruki-rule operating. A plausible solution is the
levelling of paradigms: the original palatovelars were considered sibilants (phonetically being
sibilants!), even having the same outcome as clusters on -š, the velar paradigm was levelled on
s, as were many other outcomes of the original š: cf. Lith. ausìs ʽearʼ (but OCS uxo), teisùs
ʽsilenceʼ (but OCS tixъ).
IE Lith. t- d(h)
- s-
-K(ṷ)
-k kt gd ks
-Ḱ -š šṭ žd 0š
-T -t st zd 0s
-P -p pt bd ps
-s -s st zd 0s
-š -š št žd 0š
88
The plosive k is inserted, cf. Lith. ankštas ʽcloseʼ but. OIA aṁhú-, OCS ǫzъkъ, L. angustus; Lith. áuksas ʽgoldʼ
but. Pruss. ausis, L. aurum, cf. Stang (1966: 108–109)
89
The plosive k is inserted, cf. Lith. ankštas ʽcloseʼ but. OIA aṁhú-, OCS ǫzъkъ, L. angustus; Lith. áuksas ʽgoldʼ
but. Pruss. ausis, L. aurum, cf. Stang (1966: 108–109)
93
3.3 Main features of development in Latvian
Latvian development of clusters with t- follows in its main features known from Lithuanian.
The development leads to non-palatal clusters, resulting from the depalatalization of the original
palatal clusters (preserved otherwise in Lithuanian).
We have to add to clusters of plosive + s in Latvian that future stems formed from the
peripheral plosives (labial and velar) regularly, the future stems from the central plosives (dental
and palatovelar) are always with the anaptyctic vowel ī, which prohibits any alternation; in
contrast, the (present) sta-stems (diachronically iteratives) are directly attached to the sibilant
(either original or from a palatal or a dental) – but for such stems we cannot exclude analogical
forms90
.
The examples are selective, and only illustrative of the main features of the
development; the etymologies are based on Karulis (1992a; 1992b) and Derksen (2015), while
the Indo-European reconstruction is based on LIV.
3.3.1 The development of clusters obstruent + t/s
The old IE peripheral clusters are fully preserved, the palatovelars are realized as s before t-,
lost before s-; old IE dentals have the same outcomes in the same contexts:
P + t = Latv. pt:
inf. diȓtb ʽwalk fastʼ (< IE *√dh
r̥ bh
-< IE *√derbh
-; cf. OIA pr. part. dṛbhánti- ‘forming
tuftsʼ; cf. Pokorny IEW: 211–212, 257; Karulis 1992a: 199–200; LIV2
: 121; Derksen
2015: 131);
num. septiņi ʽsevenʼ (< IE *septm̥; cf. OIA saptá-, Pruss. septmas ʽseventhʼ, L. septem
‘sevenʼ; cf. Pokorny IEW: 909; Stang 1966: 279, 283; Karulis 1992b: 171; Comrie
1992: 756–759; Blažek 1999: 246, 249; Derksen 2015: 393–394);
P + s = Latv. ps:
fut. teps ‘smearʼ (cf. pr. tepju; < IE *√tep-; cf. OCS teti ‘flog, beatʼ; cf. Karulis 1992b:
392; LIV2
: 630; Derksen 2015: 464);
sta-pr. kvēpstu ʽsmoke, smellʼ (cf. pr. kvēpu; < IE *√ku̯ ep-; cf. OIA kúpyati ‘get angryʼ,
Goth. afƕapjan ‘smother, wipeʼ; cf. Pokorny IEW: 596; Karulis 1992a: 453; LIV2
:
376; Derksen 2015: 268);
apse, epse ʽaspenʼ (< IE *aps-; cf. Lith. āpušė, ẽpušė, OHG aspa ‘aspenʼ; cf. Pokorny
IEW: 55; Karulis 1992a: 73–74; Derksen 2015: 154);
lapsa ʽfoxʼ (< IE *H2lōp-s; cf. Lith. lãpė, Gr. ἀλώπηξ ‘foxʼ; cf. Pokorny IEW: 1179;
Karulis 1992a: 501; Derksen 2015: 274);
K(u̯ )
+ t = Latv. kt:
90
For details on sta-stems in general see Endzelin (1923: 580-588), Stang (1966: 338–349), Forssman (2001: 178–
181), Holst (2001: 175–177).
94
inf. lèkt (cf. pr. lęcu ʽjump, flyʼ; < IE *√lek-; cf. MHG lecken ‘knock out with feetʼ, OCS
letěti ‘flyʼ; cf. Pokorny IEW: 673; Karulis 1992a: 513; LIV2
: 411; Derksen 2015: 278);
inf. júgt (cf. pr. júdzu ʽjokeʼ; (cf. pr. jùngiu ʽyoke upʼ; < IE *√i̯ung-; cf. OIA yunákti
‘harnessʼ, L. iungō ‘joinʼ; cf. Pokorny IEW: 508–510; Karulis 1992a: 360; LIV2
: 316;
NIL: 397–404; Derksen 2015: 214);
num. pìektais ʽfifthʼ (< IE *penku̯-to-;91
cf. OIA paktháḥ, Gr. πέμπτος; cf. Stang 1966:
283; Karulis 1992b: 42–43; Comrie 1992: 752–754; Blažek 1999: 221, 224; Pokorny
IEW: 808; Smoczyński 2007: 450; Derksen 2015: 351);
inf. diêgt ʽthread, beat, sew, stickʼ (cf. pr. diêdziu; < IE *√dh
ei̯(H)gu̯
-; cf. L. fīgere ‘insert,
pierce, fix, attachʼ; cf. Pokorny IEW: 243–244; Karulis 1992a: 214–215; LIV2
: 142;
Derksen 2015: 127);
K(u̯ )
+ s = Latv. ks:
fut. liks (cf. pr. liekt ʽleave, lay, putʼ; < IE *√lei̯ku̯
-; cf. OIA rikthās ‘protrude beyondʼ, L.
līquī ‘leaveʼ; cf. Pokorny IEW: 669–670; Karulis 1992a: 535–536; LIV2
: 406–407;
Derksen 2015: 287);
sta-pr. dîgstu ʽthread, beat, sew, stickʼ (cf. pr. diêdziu; < IE *√dh
ei̯(H)gu̯
-; cf. L. fīgere
‘insert, pierce, fix, attachʼ; cf. Pokorny IEW: 243–244; Karulis 1992a: 214; LIV2
: 142;
Derksen 2015: 127);
Ḱ + t = Latv. st:
inf. nest ʽcarry, bringʼ (cf. pr. nęsu; < IE *√Hneḱ-; cf. Toch. B eṅtär ‘graspʼ, OCS nošǫ
‘carryʼ; cf. Pokorny IEW: 316–318; Karulis 1992a: 624–625; LIV2
: 250–251; Derksen
2015: 334);
inf. riêzt ‘stick out, become warpedʼ (cf. pr. riêžu < IE *√rei̯ ǵ-; cf. OIr. rigid ‘stretch out,
ruleʼ; cf. Pokorny IEW: 862; Karulis 1992b: 101–102; LIV2
: 503; Derksen 2015: 380);
astuôņi ʽeightʼ (< IE *oḱtō-ni-; cf. OIA aṣṭau, L. octō; cf. Stang 1966: 279, 283–284;
Karulis 1992a: 81; Comrie 1992: 758–760; Blažek 1999: 267; Pokorny IEW: 775;
Derksen 2015: 64–65);
Ḱ + s = Latv. 0s:
sta-pr. lūstu ʽbreakʼ (inf. laûzt ʽbreakʼ; < IE *√leu̯ ǵ-; cf. OIA rujáti ‘breakʼ, OE tōlūcan
‘destroyʼ; cf. Pokorny IEW: 686; Karulis 1992a: 510–511; LIV2
: 415–416; Derksen
2015: 276);92
ass93
ʽaxle, axisʼ (< IE *H2eḱs-i-; cf. OIA ákṣa-, L. axis; cf. Pokorny IEW: 6; Karulis
1992a: 79–80; NIL 259–262; Derksen 2015: 63);
lasis94
ʽsalmonʼ (< IE *loḱsi; cf. Tocharian B laks ‘fishʼ; OHG lahs, Lith. la͂ šiš ‘salmonʼ;
cf. Pokorny IEW: 653; Karulis 1992a: 503; Derksen 2015: 274);
Note: Latvian future is regularly formed with an anaptyctic vowel from roots ending on an original palatovelar, cf.
fut. nesīs (cf. pr. nęsu ʽcarry, bringʼ; < IE *√Hneḱ-) and fut. lauzīs (cf. pr. laûžu ʽbreakʼ; < IE *√louǵ-).
Note: Latv. seši ʽsixʼ (cf. Lith. šeši) has š is from *si̯, cf. Stang 1966: 277; Karulis 1992b: 174; Derksen 2015:
446). If IE was formed by a palatovelar + s *(K)seḱs/(K)su̯eḱs (cf. Blažek 1999: 234–245 for details), the Baltic
form could be traced to *seḱs-i; Lith. šẽstas, Latv. sȩstaĩs, Pruss. wuschts ʽsixthʼ are derived directly from the
stem by to-.
91
Latvian has lost the root nasal.
92
LIV reconstruct *√leu̯ g-.
93
The assumed development is: aḱšis > ašis > asis > ass after a syncope (Karulis 2001: 79).
94
The assumed development is: -ḱs- > -ḱš- > šš > 0š > 0s.
95
T + t = Latv. st:
inf. mest (cf. pr. mȩtu ʽthrowʼ; < IE *√met- ; cf. OCS metǫ, mesti ʽthrowʼ; cf. Pokorny
IEW: 703–704; Karulis 1992a: 584–585; LIV2
: 442; Derksen 2015: 313–314);95
inf. êst (pret. êdu ʽeatʼ; < IE *√H1ed-; cf. Hitt. ēdmi, OIA átti, L. edō ‘eatʼ; cf. Pokorny
IEW: 287–289; Karulis 1992a: 273–274; LIV2
: 230–231; NIL 208–220; Derksen
2015: 157–158);
inf. bust ʽawake, wake upʼ (cf. pr. bùdu; < IE *√bh
eu̯dh
-; cf. OIA bódhati ‘noticeʼ, Gr.
πεύθομαι ‘give noticeʼ; cf. Pokorny IEW: 150–152; Karulis 1992a: 112–114; LIV2
:
82–83; NIL 36–37; Derksen 2015: 83, 107);
inf. brìst (cf. pr. brìedu ʽwadeʼ; < IE *√bh
redh
; cf. OCS bredǫ ‘wadeʼ; cf. Pokorny IEW:
164; Karulis 1992a: 146; LIV2
: 91; Derksen 2015: 101);
T + s = Latv. 0s:
sta-pr. bilstu ʽsayʼ (pret. bil̃da, < IE *√bh
el-t-st-; cf. OE bellan ‘roarʼ;cf. Pokorny IEW:
123–124; Karulis 1992a: 127; LIV2
: 74; Derksen 2015: 90);
Note: Similarly to stems ending on a palatovelar (see above), Latvian future is regularly formed with an anaptyctic
vowel from roots ending on an original dental, cf. fut. [fut. metīs (cf. pr. mętu ʽthrowʼ; < IE *√met-) or fut.
vedīs (inf. vest ʽleadʼ; < IE *√u̯ edh
-) or fut. edīs (inf. ēst ʽeatʼ; < IE *√H1edh
-).
s + t = Latv. st:
juôsta ʽgirdleʼ (< IE *√i̯eH3s-; cf. Gr. ζώννῡμι ʽgirdʼ, OCS po-jašǫ ʽgird upʼ; cf. Pokorny
IEW: 513; Karulis 1992a: 358–359; LIV2
: 311; NIL 391–392; Derksen 2015: 214–
215);
s + s = Latv. 0s:
sta-pr. dzièstu ʽextinguish, put outʼ (pr. dzešu; < IE *√gu̯
es-; cf. Lith. gèsti, pret. gẽso, OIA
jásate ‘be exhaustedʼ, Gr. σβέννῡμι ‘quenchʼ, OCS -gasiti ‘extinguishʼ; cf. Pokorny
IEW: 479–480; Karulis 1992a: 250; LIV2
: 541543; Derksen 2015: 173);
š + t = Latv. st:
inf. klàust (pret. klàusu ʽaskʼ; < IE *√ḱleu̯ s-; < IE *√ḱleu̯s-; cf. cf. OIA śróṣan ʽobeyʼ, Toch.
A klyoṣäs, B klyauṣäm ‘listenʼ; cf. Pokorny IEW: 606–607; Karulis 1992a: 401; LIV2
:
336–337; NIL 432–434; Derksen 2015: 249);
inf. kā̃rst (prêt. kā̃rsu ʽcard, combʼ; < IE *√(s)kers-; cf. L. carrō ‘cardʼ cf. Pokorny IEW:
532–533; Karulis 1992a: 385; LIV2
: 559; Derksen 2015: 228);
pìr(k)sts ʽfingerʼ (< IE *prstH2o-; cf. Lith. pir̃štas; cf. Pokorny IEW: 813; Karulis 1992b:
54; NIL 637–659; Derksen 2015: 358);
inf. àust ʽdawn, break dayʼ (< IE *√H2u̯es-; cf. Lith. aušrà ʽdawnʼ, OIA uccháti ‘shineʼ; cf.
Pokorny IEW: 86–87; Karulis 1992a: 91; LIV2
: 292–293; Derksen 2015: 72);
inf. aízmìrst ʽforgetʼ ( < IE *√mers-; cf. Lith. mir̃šti, OIA mṛ́ṣyate ‘forgetʼ; cf. Pokorny
IEW: 737–738; Karulis 1992a: 61; LIV2
: 440–441; Derksen 2015: 320);
š + s = Latv. 0s:
sta-pr. àustu ʽdawn, breakʼ (< IE *√H2u̯es-; cf. Lith. aušrà ʽdawnʼ, OIA uccháti ‘shineʼ;
cf. Pokorny IEW: 86–87; Karulis 1992a: 91; LIV2
: 292–293; Derksen 2015: 72);
sta-pr. aizmirstu ʽforgetʼ ( < IE *√mers-; cf. Lith. mir̃šti, OIA mṛ́ṣyate ‘forgetʼ; cf. Pokorny
IEW: 737–738; Karulis 1992a: 61; LIV2
: 440–441; Derksen 2015: 320);
95
Not securely attested outside Balto-Slavic. Derksen (l.c.) proposes as cognates L. metere “mow, harvest” and
W. medi “reap”. LIV2
(l.c.) proposes as a cognate Gr. μέτρον “measure”.
96
Note: Endzelin (1923: 178) brings several dialectal examples of the Latvian future with geminate sibilants – we
consider these forms analogous, not preserving the older stage of development.
3.3.2 Overview of the Latvian alternations
The development of clusters of our interest in Latvian is similar in its main features with that
of Prussian, and remotely to Lithuanian. The peripheral series clusters are preserved, as are
clusters of a sibilant + t/s, but both ruki-clusters and old palatovelars are depalatalized. A
remarkable feature is the loss of a plosive (from the synchronic point of view a total elision)
before s-affixes of both central series.
IE Latvian t- s-
-K(ṷ)
-k kt ks
-Ḱ -s sṭ 0s
-T -t st 0s
-P -p pt ps
-s -s st 0s
-š -s st 0s
Note: The palatal phonemes š, ž in Latvian are results of secondary, specifically Latvian developments of clusters.
3.4 Main features of development in Prussian
Prussian is a dead language, in contrast to Lithuanian and Latvian, documented by a very
limited corpus of texts, hence we are forced to use few attested examples instead of the wider
possibilities while dealing with Lithuanian or Latvian.
Note: Prussian data were written by German-speaking (and writing) authors and the orthography of Prussian is
that of locally used German, hence we see <ckt> or <ct> for /kt/, <x> for /ks/, geminates could mark the length
of the preceding syllable, etc. The problem of the phonetic reality of Prussian is important, especially in cases
of clusters of sibilant + t, since attested <st> could be a realization of /št/ as well as /st/ (cf. Gerullis 1922:
221). Similarly, the phonetic value of <s> could as well be either /š/ or /s/ or /z/.
Regarding our clusters of interest, the general features of clustering follow the general Baltic
course (clusters of peripheral plosives + t/s are preserved; Tt realizes as st, clusters formed by
a sibilant of any origin + s are realized by a single sibilant), in fact in many features are more
similar to Latvian than Prussians geographical neighbour Lithuanian.
3.4.1 The development of clusters obstruent + t/s
The development of clusters of our interest is generally well attested, though the literary sources
we have at our disposal are not numerous. The data are written with a Low German orthography
97
and by writers without a native knowledge of the language, hence with varying orthography,
which negatively affects the outcomes.
P + t = Pruss. pt:
inf. trapt ʽstepʼ (< IE *√trep-; cf. cf. cf. Gr. τρᾰπέω ʽtread grapesʼ, OS thrabōn ʽtrotʼ; cf.
Pokorny IEW: 1094; LIV2
: 650; Smoczyński 2005: 370; Mažiulis 2013: 915; Derksen
2015: 469–470);
septmas ʽseventhʼ (< IE *septm̥; cf. OIA saptá-, L. septem; cf. Pokorny IEW: 909; Comrie
1992: 756–759; Blažek 1999: 246, 249; Mažiulis 2013: 843; Derksen 2015: 393–394);
dalptan ʽchiselʼ (< IE *√dh
elbh
-; cf. Lith. dálba ʽleverʼ, OS delfan ‘digʼ; cf. Mažiulis
2013: 103; cf. ; cf. Pokorny IEW: 246; LIV2
: 143; Mažiulis 2013: 103; Derksen 2015:
113, 120);
P + s = Pruss. ps:
abse ʽaspenʼ (< IE *aps-; cf. Lith. āpušė, ẽpušė, OHG aspa ‘aspenʼ ; cf. Pokorny IEW:
55; Mažiulis 2013: 4-5; Derksen 2015: 154);
wobse ‘waspʼ (< *(H1)u̯ obh
-s-; cf. Lith. vapsvà, Latv. vapsene, vapsine, L. vespa, OHG
wafsa ‘waspʼ; cf. Pokorny IEW: 1179; Mažiulis 2013: 963; Derksen 2015: 488);
K(u̯ )
+ t = Pruss. kt:
deickton, deicktan ʽplace, somethingʼ (< IE *√dh
ei̯Hgu̯
-; cf. Lith. dáiktas ʽthing, objectʼ,
Latv. daîkts ʽobject, thing, toolʼ; cf. Pokorny IEW: 243–244; LIV2
: 142; Mažiulis 2013:
111; Derksen 2015: 111);
penckts, pyienckts, piēncts ʽfifthʼ (< IE *penku̯
-to-; cf. Lith. peñktas, OIA paktháḥ ‘fifthʼ;
Comrie 1992: 752–754; Blažek 1999: 224; cf. Pokorny IEW: 808; Mažiulis 2013: 690;
Derksen 2015: 351);
duckti ʽdaughterʼ (< PBalt.*duktē < IE *dh
ugH2
-ter-96
; cf. Litv. duktė̃, OIA duhitár‘daughterʼ;
cf. Pokorny IEW: 277; NIL 126–130; Mažiulis 2013: 146; Derksen 2015:
145);
nacktin ʽnightʼ (< IE *neku̯
-t-; cf. Lith. naktìs, Latv. nakts, Hitt. nekuz gen sg. ʽin the
eveningʼ; etc.; cf. Pokorny IEW: 762–763; NIL 504–513; Mažiulis 2013: 626; Derksen
2015: 327–328);
K(u̯ )
+ s = Pruss. ks:
various personal names: Paxis (Lith. Pakšys); Lixa (Lith. Likšas); Kixe (Lat. Kiksis) (cf.
Trautmann 1974: 178–179);
Note: Other examples on Ks-clusters than proper names are hard to identify.
Ḱ + t = Pruss. st:
instixs ʽthumbʼ (< IE *H1enḱ-t-/H2neḱ-t- (?);97
cf. Lith. nykštỹs98
, Latv. îkstis; cf. LIV2
:
282–283; Mažiulis 2013: 304; Derksen 2015: 335);
96
Alternatively < IE *dh
uǵH2
-ter-, then belonging to Ḱt-clusters.
97
But Fraenkel (LEW: 188) proposes IE *√H2id- “swell”.
98
Lith. n- is regarded as secondary, cf. Derksen 2015: 335. The etymology is not clear, but the internal cluster is
reconstructed always with ḱt.
98
pistwis ʽdogs flies (= 4th
of Plagues of Egypt)ʼ99
(< IE *√peḱ-; cf. OCS pьsъ ʽdogʼ; cf.
Pokorny IEW: 795; Mažiulis 2013: 716; Derksen 2015: 431);
Ḱ + s = Pruss. 0s/ss (?):100
assis101
ʽaxleʼ (< IE *H2eḱs-i-; cf. OIA ákṣa-, L. axis; cf. Pokorny IEW: 6; NIL 259–262;
Mažiulis 2013: 49–50; Derksen 2015: 63);
lalasso ʽsalmonʼ (< IE *loḱsi; cf. Toch. B laks ‘fishʼ; OHG lahs, Lith. la͂ šiš ‘salmonʼ; cf.
Pokorny IEW: 653; Derksen 2015: 274);
wuschts, usts, uschts etc. ʽsixthsʼ (< IE *(s)uḱs-tH2-102
; cf. Lith. šẽstas, OIA. ṣaṣṭhá-; cf.
; cf. Pokorny IEW: 1044; Comrie 1992: 754–755; Blažek 1999: 238; Mažiulis 2013:
924–925; Derksen 2015: 446);
T + t = Pruss. st:
inf. istwei, īst ist ʽeatʼ (< IE *√H1ed-; cf. Pruss. īdis ʽfoodʼ, Hitt. ēdmi, OIA átti, L. edō
‘eatʼ; cf. Pokorny IEW: 287–289; LIV2
: 230–231; Smoczyński 2005: 184; NIL 208–
220; Mažiulis 2013: 316–317; Derksen 2015: 157–158);
ppp. pomests ʽsubjectʼ (< IE *√met-; cf. Latv. pr. mȩtu, OCS metǫ, mesti ʽthrowʼ; cf.
Pokorny IEW: 703–704; LIV2
: 442; Smoczyński 2005: 279–280; Mažiulis 2013: 741–
743; Derksen 2015: 313–314);103
inf. dāst, dast, pr. 3rd
sg. athem. dāst ʽgiveʼ (inf. dātwei, datwei; < IE *de-dH3-ti; cf. Lith.
dúoti, OCS dastъ, OIA dádāti; cf. Pokorny IEW: 223–225; LIV2
: 105–106;
Smoczyński 2005: 72–80; Mažiulis 2013: 106–107; Derksen 2015: 146–147);
inf. waist ʽknowʼ (cf. pr. 1st
pl. waidimai; < IE *√u̯ e d-; cf. OIA véda, Gr. (ϝ)οῖδα ‘knowʼ;
cf. Pokorny IEW: 1125–1127; LIV2
: 665–667; Smoczyński 2005: 382–388; Mažiulis
2013: 928–929; Derksen 2015: 566);
T + s = Pruss. 0s:
pr. 2nd
sg. athem. waisei, waisse ʽknowʼ104
(cf. pr. 1st
pl. waidimai; < IE *√u̯ e d-; cf. OIA
véda, Gr. (ϝ)οῖδα ‘knowʼ; cf. Pokorny IEW: 1125–1127; LIV2
: 665–667; Smoczyński
2005: 382–388; Mažiulis 2013: 928–929; Derksen 2015: 566);
sta-pr. poprestemmai ʽfeelʼ (cf. inf. issprestun; < IE *√pret-; cf. Goth. fraþjan
‘understandʼ; cf. Pokorny IEW: 845; LIV2
: 493; Smoczyński 2005: 281–282; Mažiulis
2013: 315–316, 746–747; Derksen 2015: 369);
sta-pr. wīrst ʽbecomeʼ (< IE *√u̯ert-; cf. Lith. vir̃sta, OIA vártate ʽturnʼ OCS vraštǫ, L.
uertor ‘turnʼ; cf. Pokorny IEW: 1156–1158; LIV2
: 691–962; Smoczyński 2005: 409–
411; Mažiulis 2013: 953–954; Derksen 2015: 498);
99
Since the word is a translation of German huntfliege of the same meaning, there is no doubt of its relationship
to CS pьsъ ́dogʼ.
100
The forms of wuschts, uschts, uschtai, uschtan ʽsixthsʼ offer only a solution with the št-cluster (cf. Mažiulis
2013: 924–925), but this is probably an outcome of the cluster Ḱst only.
101
The ʽgeminateʼ <ss> marks the shortness of the preceding vowel (as in German, cf. Endzelin 1944: 23–24).
102
The Prussian form we reconstruct: šukš-to- > (š)ušta- (Stang 1966: 279, 283, Blažek 1999: 238, Derksen 2015:
446)
103
Not securely attested outside Balto-Slavic. Derksen (l.c.) proposes as cognates L. metere “mow, harvest” and
W. medi “reap”. LIV2
(l.c.) proposes as a cognate Gr. μέτρον “measure”.
104
The assumed development here is: -d-s- > -ss- > -0s-.
99
s + t = Pruss. st:
pr. 3rd
sg. athem. ast, æst, est ʽbeʼ (< IE *√H1es-; cf. OIA ásti, L. est, OLith. ẽsti ʽbeʼ; cf.
Endzelin 1944: 161; Pokorny IEW: 340–341; LIV2
: 241–242; Smoczyński 2005: 24–
28; Mažiulis 2013: 47–48; Derksen 2015: 157);
inf. tiēnstwei ʽprovokeʼ (imp. tenseiti; < IE *√tens-; cf. OIA taṁsayethe ‘tugʼ, Goth. þinsan
‘pullʼ, Lith. tȩ̃sti ʽcontinue, proceedʼ; cf. Pokorny IEW: 1068–1069; LIV2
: 629;
Smoczyński 2005: 366–367; Mažiulis 2013: 910; Derksen 2015: 464);
inxcze ʽkidneyʼ (< IE *H2id-st-; cf. Lith. ìnkstas, Latv. îkstis, OCS istesa ‘kidneyʼ, ON eista
‘testicleʼ; cf. Pokorny IEW: 774; LIV2
: 258; Mažiulis 2013: 304; Derksen 2015:
202);105
š + t = Pruss. st:
pirsten ʽfingerʼ (< IE *prstH2o-; Lith. pir̃štas, Latv. pìrst; cf. Pokorny IEW: 813; NIL 637–
659; Mažiulis 2013: 715–716; Derksen 2015: 358);
austo, āustin ʽmouthʼ (< IE *Heu̯s-t-; cf. Lith. úostas, Latv. uōsta ʽport, harbor, mouth of
a riverʼ, OCS usta ʽmouthʼ, OIA óṣṭha- ʽ(upper) lipʼ; cf. Pokorny IEW: 784–785;
Mažiulis 2013: 64; Derksen 2015: 481–482);
s + s = Pruss. 0s:
pr. 2nd
sg. athem. assei, essei, assai, asse, esse ʽbeʼ (< IE *√H1es-; cf. Lith. esi, OIA ási
ʽbeʼ; cf. Endzelin 1944: 161; Pokorny IEW: 340–341; LIV2
: 241–242; Smoczyński
2005: 24–28; Mažiulis 2013: 47–48; Derksen 2015: 157);
fut. postāsei ʽbecomeʼ (√steH2-s-sei, cf. Endzelin 1944: 176; Stang 1966: 397;
Pokorny IEW: 1004–1108; LIV2
: 590–592; Smoczyński 2005: 284–287, 468; NIL
637–659; Mažiulis 2013: 749–750; Derksen 2015: 430);
š + s = Pruss. [0s]:
not attested (?). The reconstructed form is based on analogous Ḱs, assuming that both
clusters could have a common outcome, as in other satəm-languages.
3.4.2 The overview of the Prussian alternations
The development of clusters of our interest in Prussian is similar to that of Latvian. The
peripheral series clusters are preserved, as are clusters of a sibilant + t/s, but note that the rukiclusters
are depalatalized, and what is more, even clusters of old palatovelars + t are
depalatalized too. A remarkable feature is the loss of a plosive (from the synchronic point of
view a total elision) before s-affixes.
IE Prussian t- s-
-K(ṷ)
-k kt ks
-Ḱ -s st 0s
-T -t st 0s
-P -p pt ps
-s -s st 0s
-š -s st (0s)
105
In our opinion, it is possible to assume the connection of Baltic *i(n)st- (reconstructed in the accord to Derksen
l.c.) with IE *√H2e d- (cf. Gr. οἰδέω, Arm. yatnowm “swell”, ON eitr “pus”, OCS jadъ).
100
3.5 Trajectories of the Baltic developments
The development of clusters of obstruent + t/s(/dh
) can be split into three blocks: the first block
is that of acute/central plosives (i.e., dentals and original palatovelars); the second block is that
of the grave/peripheral plosives (i.e., original plain velars/labiovelars and labials); the third
block is that of sibilants. The distinction between blocks can be listed as:
i. based on the opposition between plosives (the first and second block) and sibilants (the third
block);
ii. based on the opposition between the central and the peripheral series, this distinction is given
by the fricativization of the first series before t/s.
3.5.1 The development of the peripheral series
The development of both peripheral/grave series has no special outcomes besides the expected
alternation of voice; the clusters were neither fricativized nor simplified.
For the development of the labial series, the trajectory could be modelled as:
i. P + t > pt (Common Baltic)
i. P + s > ps (Common Baltic)
For the development of the (labio)velar series (labiovelars merged with plain velars in all satəmlanguages),
the trajectory could be modelled as:
i. K(u̯ )
+ t > kt (Common Baltic)
i. K(u̯ )
+ s > ks (Common Baltic)
The interesting point is whether the IE K(u̯ )
s was affected by Pedersen’s Law (the ruki-rule) as
it was in Indo-Iranian (cf. OIA √vac- ʽspeakʼ: ft. vakṣyáti, ds. vívakṣati; Av. √vac- ʽspeakʼ:
OAv. fut. vaxšiiā, both from IE *√u̯ eku̯
-) and Slavic (ao. těxъ ʽflowʼ from IE *√teku̯
-). If PreBaltic
*s was affected by the ruki-rule after *k (of different origin), the expected outcome would
be *kš, the same outcome as *Ḱs already has (and is attested in Lithuanian). We have to bear in
mind that Indic, Iranian and Baltic register the same outcome for *Ks and *Ḱs; the Baltic
situation remarkably deviates from its parallels. The possible solution we propose is: the cluster
palatovelar + s was soon realized as çš, but original Kš was preserved as kš. The further
development of the original cluster of Ḱš was: çš > šš > 0š (see above) but of the original cluster
of Kš had a development: kš > ks, i.e., with the later depalatalization of the sibilant. We assume
similar depalatalizations as a general process in Latvian (and similarly for Prussian).
101
3.5.2 The development of the central series I: original palatovelar series
The second oldest development is that of the original palatovelar series, which is attested for
all satəm-languages.
As with the development of clusters of Tt/Ts, there are two possible trajectories, first the
affricativization trajectory, second the spirantization trajectory, both leading to the sibilant
outcome, hence again both are two variants of the more general fricativization model.
The affricate model assumes the affrication of original IE palatovelars to palatal
affricates (in all positions, but we are interested only in the context of t/s), and the later loss of
the plosive segment of an affricate. The clusters with sibilants were hence formed by two
sibilants, one of which was elided. For Latvian (and possibly for Prussian) we assume the later
depalatalization:
i. Ḱ + t > tš
t > št (Lithuanian)
ii. Ḱ + t > tš
t > št > st (Latvian)
iii. Ḱ + t > tš
t > št > st/št (?) (Prussian)
i. Ḱ + s > tš
š > šš > 0š (Lithuanian)
ii. Ḱ + s > tš
š > šš > 0š > 0s (Latvian)
iii. Ḱ + s > tš
š > šš > 0s/0š (?) (Prussian)
We model the spirantization trajectory with a spirantization of IE palatovelar to palatal spirant
ç, later sibilantized on š, later depalatalized in Latvian (and Prussian ?). The cluster of Ḱt is then
preserved as št in Lithuanian, as st in Latvian. The cluster of Ḱs first underwent palatalization
of *s to š due to Pedersen’s Law, but later the first segment was also replaced by the ç due to
analogy. This cluster turned into a sibilant geminate, which was later simplified (and
depalatalized in Latvian – we have no firm data on Prussian):
i. Ḱ + t > çt > št (Lithuanian)
ii. Ḱ + t > çt > št > st (Latvian)
iii. Ḱ + t > çt > št > st/št (?) (Prussian)
i. Ḱ + s > çš > šš > 0š (Lithuanian)
ii. Ḱ + s > çš > šš > 0š > 0s (Latvian)
iii. Ḱ + s > çš > šš > 0s/0š (?) (Prussian)
However, palatovelar + s clusters were simplified in a similar way to Ts clusters and palatovelar
+ t cluster was preserved as sibilant + t as the Tt cluster was (see above).
102
Balto-Finnish languages, or more precisely, loanwords from Baltic into them bring external
proof that the old palatovelars were originally palatal sibilants . The old palatal sibilants *š and
*ž are realized as h in Balto-Finnish, cf. Finn. hammas, Est. hammas ‘toothʼ (cf. Lith. žam̃bas
‘sharp edgeʼ, Latv. zûobs ‘toothʼ < IE *ǵombh
o-), Finn. herne, Est. hernes ‘peaʼ (cf. Lith. žìrnis,
Latv. zir̃nis < IE *ǵṛH2no-), Finn. lohi, Est. lõhe ‘salmonʼ (cf. Lith. lašišà, lašis, Latv. lasis
ʽsalmonʼ (< IE *loḱso-) (Young 2017: 498). It is hardly imaginable that Balto-Finnish forms
traceable to palatal sibilants came through another channel than the North Baltic dialects, i.e.
the ancestors of Modern Latvian, hence the original existence of palatal sibilants (from original
palatovelars) should be taken for granted for the whole Baltic area.
3.5.3 The development of the central series II: dental series
The oldest stratum of the development is the development of the dental series of the central
block, which is shared with all other Indo-European branches. Its input is Tt/Ts; its output is
st/0s.
To model this development, we can use two possible strategies, that of affricatization
and that of spirantization, though both end with the same final outcome.
The classic affrication model assumes the affricatization of the first dental, the later loss of the
plosive section of the affricate, and simplification (for clusters of Ts). This model is based on
the affricatization model for IE languages, developed initially by Kräuter (1877: 88)106
and
popularized by Brugmann (firstly 1880: 140–142, used since). This model is supported by the
fact that affricates are outcomes of T in clusters of *Tt attested in Anatolian (cf. Hitt. 3rd
sg.
preterite <e-iz-ta> and 3rd
sg. imperative <e-iz-du> from √H1ed- ʽeatʼ); this feature of Anatolian
was first noted by Götze (1928: 126), but put in the light first by Sturtevant (1933a: 6–7; 1933b:
129); later especially see Oettinger (1979: 530–532) and Melchert (1994: 113, 151, 249). It has
to be emphasized that attested affricates in Anatolian do not prove that this process was
universal for all Indo-European languages. The trajectory as such is expressed by Otrȩbski
(1958: 338–339).
The affricate model could be modelled as:
i. T + t > ts
t > st (Common Baltic)
i. T + s > ts
s > ss > 0s (Common Baltic)
106
Interesting is that Kräuter speaks about affrication, but his description of the feature is that of a spirantization!
Verner (1878: 341–342) has a critical evaluation of the idea.
103
The alternative spirantization model assumes the spirantization of the left plosive and later
sibilantization of the spirant (and the subsequent simplification of clusters from Ts – this feature
is the same for both models). This model is based on the ideas of Bartholomae (1895: 16 and
later), who assumed Tt > ϑt > st, and the model is taken as a possibility by Leumann (1942:
13). It is also interesting that Brugmann (1886: 347), otherwise the popularizer of the
affricatization theory, in his first version assumes the development: Tt > tþ
t (i.e., tϑ
t [sic!]).
i. T + t > ϑt > st (Common Baltic)
i. T + s > ϑs > ss > 0s (Common Baltic)
The spirantization model has clear advantages for the development of Indo-Iranian languages.
However, no internal data are leading us to prefer it over the traditional affrication model within
the development of Baltic languages, though, on the other hand, there is similarly no reason to
prefer the affricativization trajectory over that of spirantization. Both, however, lead towards
the same outcome (a dental sibilant), and both could be generalized as fricativization
trajectories.
3.5.4 The development of the sibilants
The development of the Baltic sibilants is simple: the clusters of the sibilant + *t are preserved
as such, the clusters of sibilant + *s are simplified, and these processes are parallel to the
development of clusters of palatovelars, as described above.
It should be noted that cluster st is merged with original cluster Tt; similarly, clusters of
ss and clusters of Ts. A similar process is attested in Iranian and Slavic.
The processes are fully preserved in Lithuanian; Latvian later depalatalized clusters št/0š
of any origin (Prussian dates are inconclusive, hence the Latvian outcome is simplified it its
outputs.
Here we have to emphasize that clusters of št and Ḱt are merged (as they are in other
satəm-languages), as are outcomes of clusters of šs and Ḱs.
i. s + t > st (Common Baltic)
i. š + t > št (Lithuanian)
ii. š + t > št > st (Latvian)
iii. š + t > št > st (Prussian)
i. s + s > ss > 0s (Lithuanian)
ii. s + s > ss > 0s (var. → sīs) (Latvian)
104
iii. s + s > ss > 0s (Prussian)
i. š + s > šš > 0š (Lithuanian)
ii. š + s > šš > ss > 0s (var. → sīs) (Latvian)
iii. š + s > šš > ss/0š (?) (Prussian)
To shed some light on the development of clusters of št, šs, we have to make an excursion on
the problem of the ruki-rule in Baltic (see the excursion).
3.5.5 An excursion on the ruki-rule in Baltic
Clusters with *š, resulting from the earlier *s due to Pedersen’s Law deserve special attention.
This law operates in Indo-Iranian and Slavic after triggers107
, represented by the shortcut ruki. The rule as
a mechanism is attested directly only in Lithuanian (if this rule was limited to dialects preceding Lithuanian is also
a matter of debate).
The problematic points are:
i. the context of the rule;
ii. the dialectal extension of the rule.
To this, we have to add that št/0š are securely preserved in Lithuanian but not in Latvian (the Prussian state was
commented on above).
The (Lithuanian) examples (and contra-examples) of Pedersen’s rule:
r + s: viršùs ʽsummitʼ (Latv. vìrsus, OCS vrъxъ); širšuō ʽwaspʼ (Latv. sir̂sins, Pruss. sirsilis, OCS sъrъšenь
ʽhornetʼ); pir̃štas ʽfingerʼ (Latv. pìrst, Pruss. pirsten, OCS prъstъ);
k + s:108
ašìs ʽaxle, axisʼ (Latv. ass, Pruss. assis, OCS osь); lašišà, lašìs etc. ʽsalmonʼ (Latv. lasis, Pruss. lalasso
[instead of lasasso?109
], Cz. losos, ON lax, Toch B laks ʽfishʼ);
i + s: maĩšas ʽbag, sackʼ (Latv. màiss, Pruss. moasis ʽbellowsʼ, OCS měxъ); ríešutas ʽnutʼ (Latv. riēks, OCS
orěxъ);
but:
visas ʽallʼ (Latv. viss, Pruss. wissa-, OCS vьsь); lysė ʽ(garden)bedʼ (Pruss. lyso, OCS lěxa ʽrow, furrowʼ;
u + s: aušrà ʽdawnʼ (Latv. àustra, OCS utro); ẽpušė, ãpušė ʽaspenʼ (Latv. apse, Ru. osína, OHG aspa); vẽtušas
ʽold, archaicʼ (OCS vetъxъ);
but:
ausìs ʽearʼ (Latv. àuss, Pruss. āusins, OCS uxo, L. auris); saũsas ʽdryʼ (Latv. sàuss, Pruss. sausā, OCS suxъ).
It was already Pedersen himself (Pedersen 1895) who first assumed that the rule affected even the Baltic
continuum to the same extent in the whole area, though he was aware of discrepancies and a general lack of
regularity of the rule in Baltic (in fact, in Lithuanian). Similarly, Otrȩbski (1954: 32) considers the development
to be originally developed in all four contexts if *s was followed by i̯ , but later partially undone after i/u (in
contrast, the universality of the law is assumed in Otrȩbski 1958: 301–302, 309). Rozwadowski (1961: 100–101)
and Szemerényi (1957: 106–107) argue for all-contexts change, later re-archaized (a case of a partial regression),
and they strongly argue for the universality of the ruki-rule in Baltic, not only Lithuanian, as later does Andersen
(1968), who emphasizes the structural reasons to assume original satəm unity of the process called Pedersen’s Law
and brings models of the assumed levelling due to morphology (Andersen 1968: 183–185). The old ʽuniversalʼ
application of the rule is supported by Kortlandt (1980: 245) or Smoczyński (2001: 22). Petit (2018: 1649)
surprisingly gives no statement on this question.
107
Surprisingly, in Iranian, the trigger could even be p, cf. YAv. drafša- “flag” (cf. OIA drapsá-), OAv. diβžaidiiāi
(cf. OIA dípsati, √dabh
- “harm”). That this process is a later analogy is betrayed by the fact that it affects even
sibilants from original palatovelar in given contexts, cf. Av. fšu- “cattle” (cf. Av. pasu-, OIA paśu-). However,
this innovation is far from universal since we meet: pr. YAv. tafsat̰ (√tap- “heat”); pr. YAv. xv
afsa, xv
afsata
(√xv
ap- “sleep”). The original extension covered only original *s, not clusters with it.
108
All examples are on original cluster Ḱs, not on Ks (cf. Stang 1966: 95–96). The outcome of Ḱs is in Lithuanian
0š, in Latvian 0s. The cluster Ks has outcome: ks
109
Derksen 2015: 274–275
105
In contrast, Endzelin (1923: 110; Endzelin 1939: 107–115), Fraenkel (1950a: 113–114), defended the thesis
that Baltic/Lithuanian extension of the law was limited to the context of r, k only and that Lithuanian examples on
the effectuation in context i u have other etymologies. Pisani (1947) even restricts the extension of the law to the
context of r; in this, he is followed later by Senn (1966: 82).
Karaliūnas (1966) produced the idea that Pedersen’s law affected Lithuanian110
*s only in unproductive
words, while *s was unchanged in productive contexts. Karaliūnas hence tried to resolve why there Lith. maĩšas
is ʽsackʼ but Lith. ausìs ʽearʼ, but though his observations on productivity have value, it is an example of an improper
extension of modern synchronic productivity to diachronic development and should be rejected.
One of the most influential was Stang’s ʽdialectal solutionʼ (Stang 1966: 94–100, especially see 99–100),
which assumes that the ruki-rule was not unexceptional and did not cover all possible incomes, i.e., it never
happened in all positions as it was realized in other languages affected by it (Indo-Iranian, Slavic) and that the
ʽcoreʼ of the rule was the context of r/k: in the context of i/u original s was preserved, especially in the productive
affixes (cf. Karaliūnas above), with the cases of š in this context being secondary. Stang does not even recognize
š after i/u as being an outcome of a single phenomenon and that current state in Lithuanian is a result of a mixing
of dialects with various degrees of the application of the rule. Kümmel (2007: 406) and Young (2017: 497) follow
the Stang’s ideas.
Our opinion could be stated fully within the aforementioned ‘universalist approachʼ:
i. Pedersen’s Law was fully operational on whole Pre-Baltic area, without any early dialectal differences (hence
we reject Stang’s ʽdialectal solutionʼ, in all four ruki-contexts;
ii. the later appearance of s in ruki-contexts in Lithuanian is a result of two processes: either of a levelling due to
morphological processes, or of a later depalatalization resulting from the same process which fully affected
Latvian (and probably also Prussian);
iii. this depalatalization in Latvian (and partially in Lithuanian) has its parallel outside the Baltic area, in Middle
Indo-Aryan, where OIA ś, ṣ, s (< IE *ḱ, š, s) merged, according to the given dialect, either to ʽWesternʼ s or
ʽEasternʼ ś (cf. OIA su- ʽwellʼ, daśa- ʽtenʼ, puruṣa- ʽmanʼ and Pāli su-, dasa, purisa or Māgadhī śu-, daśa-,
puliśa; Bubenik 1996: 34). In Middle Indo-Aryan, as in Latvian, the seemingly archaic state of sibilants is a
result of later processes, not the actual old state – in Latvian and Prussian, even prevocalic sibilants from
original palatovelars were affected and merged with s – again the same situation we meet in MIA.
3.6 Conclusive remarks to the Baltic development
The oldest stratum of the development is the development of the dental series clusters. There
are two possible trajectories for this development: the traditional Kräuter/Bruggmann
affricativization trajectory and the de Saussure/Cocchia/Bartholomae spirantization trajectory,
both being variants of a wider fricativization trajectory. The preferred trajectory is that of
spirantization, since it fully explains the loss of the plosive in the IE cluster *Ts.
Similar alternative trajectories could be proposed for the development of the IE
palatovelar clusters. Again, since they better explain the development of the *Ḱs-clusters, we
prefer the spirantization over the affricativization.
The development of the peripheral clusters is uniformly conservative, preserving
unchanged plosives in all three contexts, the development of the (labio)velar series is
extraordinary (in the comparison with other satəm-languages), which, probably due to the
analogy, lost the presumed ruki-sibilant, replacing it by s. However, the analogical restorations
of the non-palatal sibilant are frequent in the wider Baltic continuum (it is total in Latvian and
Prussian).
110
He seems to be not willing to accept the Pan-Baltic extension of the rule.
106
Similarly conservative is the development of the clusters with a sibilant: the *St clusters
are preserved (those afflicted by the ruki-rule are depalatalized in Latvian and Prussian), the
*Ss clusters are degeminated (again, the ruki-clusters are depalatalized in Latvian and Prussian).
107
4 The development of the two-obstruent clusters from Indo-European into
Slavic
4. 0 Slavic languages
Slavic people came into the light of history first in the sixth century AD; however their languages enter history
first in the ninth century in the earlier forms of Old Church Slavonic, the dialect based on the South Slavic dialect,
which was spread later over the Slavic territory as a cultural language. Other written languages arose over the
whole area later (Langston 2017).
The original Slavic language continuum has undergone rapid development, resulting in dialectal split,
later languages often following parallel trajectories (c. f. the loss of yers is a universal process, but hardly a single
process; it is rather independent processes following similar trajectories, i.e., result of a drift).
4.1 Common Slavic and Old Church Slavonic
On the following lines, we will focus on the development of the Indo-European two-obstruent
clusters into Common Slavic. The development of the Indo-European plosives into Common
Slavic is well described if we focus exclusively on plosives in prevocalic positions (either in
anlaut or inlaut positions). It is the development of two-obstruent clusters which still has some
uncertainty if following trajectories, as we will see below.
Slavic languages are usually considered a sub-branch of the Balto-Slavic family, but the
observations of similarities between both IE sub-branches are not valid for the developments
of the clusters of our interest. Baltic developments of the peripheral series are more conservative
than those of Slavic; hence we deal with both branches independently.
4.2 Common Slavic and Indo-European
The typical features separating the Common Slavic obstruent system from that of IndoEuropean,
relevant for our analysis, are:
i. the merging of the reconstructed voiced aspirates and non-aspirated voiced plosives in a
single modal class;111
ii. the loss of labiovelars (the merge with plain velars), the satəm-series preserved;
iii. the development of the IE cluster Tt into a cluster of st;
iv. palatalization of old IE plain velars and labiovelars;
v. the ruki-rule (=Pedersen’s Law), i.e., the shift of IE *s to Common Slavic x (or š before
palatal vowels).
111
The original distinction between both modal classes could be traced due to Winter’s Law, which causes the
lengthening of vowels before original IE voiced non-aspirated plosives (Winter 1976; Kortlandt 1978c; Kortlandt
1978d; Kortlandt 1985a; Korlandt 1985b; Kortlandt 1994a. Kortlandt assumes the original glottal nature of IE
voiced non-aspirates; cf. also Sukač 2013, here especially detailed overview of the given literature; note that
Winter 1979 and Winter 2011 reject any glottalic explanations). For similar processes in Latin (the Lachmann’s
Law), Slavic and Tocharian see given chapters.
108
Baltic, Iranian and Celtic have the same outcome in the development of the modal classes . We
can securely assume a direct connection between processes in the first case and a probable
connection in the second case; the Celtic development is on the other hand securely
independent. The loss of labiovelars is present in all satəm-languages; however, a similar
development is attested in the centum-languages as well, often in the later stages of the same
branch (cf. Greek development as a good example). The Pre-Slavic palatovelars were, as usual
in satəm-languages, later sibilantized (and depalatalized), the same process as we know from
the Iranian. The development of clusters of two dentals is universal in the Indo-European
family, the Indic development forming not an exception, but a restoration. Similarly, the
palatalization of velar plosives is known from Indo-Iranian, but we cannot state it being
anything more than a parallel process. The ruki-shift is securely attested in Baltic (though not
with such a wide scope as in Slavic), and Indo-Iranian, but the examples of such process in
Armenian and Albanian are scarce and insecure.
4.3 Slavic clusters and their IE origins
As seems to be regular outside the Indo-Iranian languages, Slavic clusters of plosive + t/s- are
not subjected to Bartholomae’s Law, otherwise the development of clusters formed by -Dh
+ twould
have the output †Ddh
(DD after deaspiration), similarly -Dh
+ s- in †Dz (after
deaspiration). The final voicedness of all clusters is hence given by the quality of voice of the
right obstruent.
Old Church Slavonic, as the most archaic attested Slavic language, will serve as the
primary source of Slavic data; only in cases of etymologies when OCS data are not attested will
other Slavic languages provide data – the OCS data will not be marked in the following lines.
We will focus again on productive examples of clusters with t or s from synchronic data,
supported when needed by purely etymological data (especially in cases of ‘minorʼ etymologies
of clusters of velar or labial plosives + t-).
When speaking about diachrony and synchrony, we have to point out that the clusters
of labials + t and (labio)velars + t have two different outcomes each. One of the outcomes is a
result of the synchronic productive process on the contact of two morphs; this will be termed a
major development. The other outcome is detectable only by means of etymological analysis
and is not synchronically productive, and it will be termed a minor development.
109
Note: In the lines below, the examples of verbal flexion are given concerning the Indo-European phonemes in the
left position of the cluster. Pokorny IEW, LIV2
, ESJS and Derksen (2008) were used as reference sources;
other sources will be quoted when used, especially Arumaa (1976: 54–184).
4.3.1 The development of clusters of labials + t/s
The clusters of labials + t/s show both a productive (major) development and an etymological
(minor) development.
The major development of the cluster *Pt results in 0t, and the development of the
cluster *Ps in 0s (there is, again, no minor development of such clusters):
P + t = OCS pt:
inf. teti ‘beatʼ (pr. tepǫ < IE*√tep-; cf. Lith. tepù ‘smearʼ; cf. van Wijk 1931: 52; ESJS:
957; LIV2: 630; Derksen 2008: 491–492);
inf. -črěti, sup. črětъ ‘scoop, drawʼ (pr. -črьpǫ < IE*√(s)kerp-; cf. L. carpere ‘pick,
pluckʼ, Lith. kerpù ‘chop, cutʼ; cf. Pokorny IEW: 944–945; ESJS: 668; LIV2: 559;
Derksen 2008: 84);
inf. greti, sup. gretъ ‘rowʼ (< *‘dig, buryʼ,112
pr. grebǫ < IE*√gh
rebh
-; cf. OIA gṛbhnā́ ti
‘seizeʼ, Lith. grė́bti ‘rake, seizeʼ; cf. van Wijk 1931: 52; Pokorny IEW: 455–456;
ESJS: 201; LIV2: 201–202 ; Derksen 2008: 186);
inf. zęti ‘tearʼ (pr. zębǫ < IE*√ǵembh
-; cf. OIA jambháyati ‘crushʼ; cf. Pokorny IEW: 369;
ESJS: 1128; LIV2: 162–163; Derksen 2008: );
P + s = OCS 0s:
ao. gresъ ‘row, buryʼ (< *‘dig, buryʼ, pr. grebǫ < IE*√gh
rebh
-; cf. OIA gṛbhnā́ ti ‘seizeʼ,
Lith. grė́bti ‘rake, seizeʼ; cf. van Wijk 1931: 52; Pokorny IEW: 455–456; ESJS: 201;
LIV2: 201–202 ; Derksen 2008: 186);
CS nom. *osina/osika (R. osína, Pol., Cz. osika) ‘aspenʼ (< IE *ap-s-; cf. Arm. opʼi ‘white poplarʼ, Lith.
ēpušė, āpušė (dial.); Latv. apse, epse; cf. Vasmer 2: 282; Pokorny IEW: 55; Shevelov 1964: 188;
Arumaa 1976: 138; Derksen 2008: 378);
CS nom. *osa (Ru. osá, Cz. vosa, Pol. osa) ‘waspʼ (< IE *u̯ obh
-s-; cf. L. uespa, OHG wafsa; cf. Vasmer 2:
280; Pokorny IEW: 1179; Shevelov 1964: 188; Derksen 2008: 377);
The minor development of the cluster *Pt is *st. The factor determining this development is not
clear at the moment, and some of the examples are doubted. Note that not all are attested in
OCS:
P + t = OCS st:
CS nom. *stryjь/strъjь/strycь (SerbCS strъi, ORu. strъi, stryi, Uk. stryj, Cz. strýc, Pol.
stryj, SCr. strȋc, Sln. stríc, etc.) ‘uncle (lit. father’s brother)’ is connected to OIA
pítṛya-, Av. tūirya- L. patruus, Gr. πάτρως OHG fatureo ‘fatherly’ and shows the
initial str- < ptr- < *pH2tr- (cf. Vasmer 3: 29; Vey 1931a: 65–66; Kortlandt 1982: 26;
Patri 2003: 121; NIL: 554–562; Derksen 2008: 470). It should be noted that Gippert
(2002) rejects any connection between *stryjь and *pH2ter-,but assumes a connection
with Ir. struith.
Similarly, the deity name ORu. Stribogъ can be derived from pH2tr- as well, if this name is considered as
a compound with the meaning ‛Father/fatherly god’, reflecting probably PIE voc. *di̯ éṷ pH2tér ‘father
112
This meaning is attested in OCz. hřésti ; ‘buryʼ, Sln. grébsti ‘digʼ, etc. The meaning ‘rowʼ is attested in Ru.
grestí, B. grebá (beside ‘spoon, scoop, rakeʼ).
110
heaven!’ > Gk. Ζεῦ πάτερ, Lat. Iuppiter (cf. Vey 1958; Schmidt 1973: 75ff., 79f.; 82f.; Pohl 1980: 62;
NIL: 554–562; but again, for a contrary view, see Kortlandt 1982: 26; for further ref. cf. Vasmer 3: 27;
Hock 2004: 12f.).
And from the same root cf. CS nom. *pastorъkъ ‘stepson’ (Uk. pastorok, SCr. pastorak, Cz. pastorek etc.)
is allegedly derived from *pa-pH2tr̻ - (Vey 1931: 65–66; NIL: 554–562). However, this etymology was
strongly rejected by Zubatý (1891) and by Kortlandt (1982: 26), cf. Vasmer (2: 322).
CS nom. *nestera ‛niece’ (RuCS. nestera, OPol. nieściora, SCr. nèstera), *netьjь
‛nephew’ (ORu. netii, OPol. nieć, Sln. nečák etc.) . It is remarkable that we find here
two variants of the realization of the hypothetical cluster *pt, i.e., either with regular -
0t- or with minor -st , as both forms are derived from IE *nep(ō)t- (cf. OIA naptī, L.
neptis, OIr. necht; cf. Vasmer 2: 215–216; Pokorny IEW: 764; Schmidt 1973: 78ff.
with n. 254; NIL: 520–524; Derksen 2008: 351; ESJS 538). Shevelov (1964: 192) links
the different developments to the context of e or i respectively. We should note that
Meillet (1902: 167) supposes an original form *nept-terā, which would regularly give
-st- < *-tt-. Vey (1931b: XV) considers the pt-clusters to have been realized as -0tacross
morpheme boundaries, but within a single morpheme as -st-. Fraenkel (1950b:
63–64) favors analogical contamination with sestra ‛sister’.
CS nom. *(j)as(ъ)trębъ ‘hawk’ (Ru. jastrjabь, jástreb, SCr. jȁstrijeb, Cz. jestřáb, etc.)
was reconstructed by Vey as deriving from *HeHḱú- + ptr̻ - ‘fast flyer’ (Vey 1953,
supported by Pohl 1980; NIL: 200–201;, strictly rejected by Kortlandt 1982: 26; cf.
Patri 2003: 121–122; cf. Vasmer 3: 497–498; ESSJ 1995: 5, 274–275; Derksen 2008:
29).
There are other possible etymologies, but items 1 and 2 (including given sub-items) seem to be
the most reliable and promising (for a list of possible etymologies, see especially Pohl 1980;
for a strongly opposing view see Kortlandt 1982, but note Arumaa 1976: 139–141). Shevelov
(1964: 192) proposes a shift Pt > st before i (cf. the major development of the clusters of velar
+ t below), while Patri (2003: 132) states the minor formula as: #Pt- > #tt- > #st-. It seems that
the fragmentary nature of the attested etymologies following the minor development could not
give us a definitive statement on the causes of the minor development, should there be any
(again, cf. Kortlandt 1982). Darden (1978) proposes two independent processes: #pt- > #st- in
the word-initial and -pt- > 0t in other positions, but this does not explain both nestera and
jastrębъ with st in the middle of the word. The minor development *Pt > st is also accepted by
Greenberg (2017: 528). It seems most probable that *Pt > CS st in the context of the following
r. Note that three examples above are on *Ptr; *nestera could be the result of a later levelling
from *neptr- - (a reduced grade), which could explain why CS *netьjь ‛nephew’ is with 0t
outcome – here was no r within the context of *pt).
4.3.2 The development of clusters (labio)velar + t/s
There are two possible developments of clusters of IE (labio)velars + t-, the first one regular
and fully productive in OCS (a major development), the second attested only etymologically (a
minor development).
111
The regular (major) development of the IE clusters of *Kt results in OCS št, the output
of the development of IE *Ks is regularly 0x/0š (the second being a palatal variant of the first
one). The development of labiovelars mirrors that of plain velars since Slavic is a satəmlanguage,
hence the outcomes are necessarily the very same:
Note: There is a dialectal difference between main Slavic sub-branches. The outcome št for IE *K(u̯ )
t is attested
for Old Church Slavonic, and closely related East South Slavic, West South Slavic and East Slavic have the
outcome č and West Slavic the outcome c.
K + t = OCS št (major development):
inf. vlěšti, sup. vlěštъ ‘dragʼ (cf. pr. vlěkǫ < IE*√H2u̯ elk-; cf. YAv. *-vərəciṇta, Lith. velkù
‘haul, pullʼ; cf. Pokorny IEW: 1145; ESJS: 1069–1070; LIV2
: 289–290; Derksen 2008:
514);
inf. -lęšti, sup. leštъ ‘bendʼ (cf. pr. -lękǫ < IE*√lenk-; cf. Latv. lùocu ‘bend repeatedlyʼ;
cf. Pokorny IEW: 676–677; ESJS: 417; LIV2
: 413; Derksen 2008: 277);
inf. rešti, sup. reštъ ‘sayʼ (cf. pr. rekǫ < IE*√rek-; cf. OIA racayati ‘effectʼ (?); cf.
Pokorny IEW: 863; ESJS: 761–762; LIV2
: 506; Derksen 2008: 433);
inf. sěšti ‘cut, mowʼ (cf. pr. sěkǫ < IE*√sek-; cf. L. secō; cf. Pokorny IEW: 896–897;
ESJS: 809–810; LIV2
: 524; NIL: 604–605; Derksen 2008: 446);113
inf. tlěšti ‘poundʼ (cf. pr. tlъkǫ < IE*√telk-; cf. Pokorny IEW: 1062; ESJS: 699; LIV2
:
623; Derksen 2008: 490);114
inf. -sęšti ‘touchʼ (cf. pr. -segǫ < IE*√seg-; cf. OIA sájati, Lith. segù ‘attach, fastenʼ;cf.
Pokorny IEW: 887–888; ESJS: 141; LIV2
: 516; Derksen 2008: 449);
inf. strěšti ‘keepʼ (cf. pr. strěgǫ < IE*√sterg-; cf. Gr. στέργω ‘loveʼ; cf. Pokorny IEW:
1032; ESJS: 890; LIV2
: 598; Derksen 2008: 467);
inf. strišti ‘cutʼ (cf. pr. strigǫ < IE*√stre g-; cf. Lith. striégiu ‘bait, pinʼ; cf. Pokorny IEW:
1036; ESJS: 890–891; LIV2
: 604; Derksen 2008: 469);
inf. -brěšti ‘take care ofʼ (cf. pr. -brěgomъ < IE*√bh
ergh
-; cf. OE borgian ‘lendʼ; cf.
Pokorny IEW: 658–659; ESJS: 79; LIV2
: 79; Derksen 2008: 36);
inf. lešti ‘lieʼ (cf. pr. lȩgǫ < IE*√legh
-; cf. Goth. ligan, Gr. λέχομαι; cf. Pokorny IEW:
658–659; ESJS: 408–410; LIV2
: 398–399; Derksen 2008: 270);
inf. mošti, sup. moštъ ‘be ableʼ (cf. pr. mogǫ < IE*√magh
-; cf. Goth. mag ‘be ableʼ; cf.
Pokorny IEW: 695, 697; ESJS: 492–493; LIV2
: 422; Derksen 2008: 321);
Ku̯
+ t = OCS št:
inf. pešti, sup. peštъ ‘bakeʼ (cf. pr. pekǫ < IE*√peku̯
-; cf. OIA pácati, L. coquō; cf.
Pokorny IEW: 798; ESJS: 636–637; LIV2
: 468; NIL: 548–552; Derksen 2008: 393);
inf. tešti, sup. teštъ ‘flowʼ (cf. pr. tekǫ < IE*√teku̯
-; cf. OIA takti ‘goes throughʼ, Lith.
tekù ‘run, flowʼ; cf. van Wijk 1931: 53; Pokorny IEW: 1059–1060; ESJS: 956–957;
LIV2
: 620–621; Derksen 2008: 489);
inf. vrěšti ʽthrowʼ (cf. pr. vrъgǫ < IE*√u̯ ergu̯
-; cf. Goth. wairpan ‘throwʼ; cf. Pokorny
IEW: 1153; ESJS: 1088-1090; LIV2
: 689; Derksen 2008: 515);
inf. žešti ‘burnʼ (cf. pr. žegǫ < IE*√dh
egu̯ h
-; cf. OIA dáhati, Lith. degù ‘burnʼ ; cf. Pokorny
IEW: 240–241; ESJS: 1150–1151; LIV2
: 133–134; Derksen 2008: 554);
K + s = OCS 0x (/0š):
113
LIV2
: 524 reconstructs √sekH- based on L. pf. secuī, but Slavic form would be **sex-.
114
No secure cognates outside Slavic.
112
ao. -vlěxъ ʽdragʼ (cf. pr. vlěkǫ < IE*√H2u̯ elk-; cf. YAv. *-vərəciṇta, Lith. velkù ‘haul,
pullʼ; cf. Pokorny IEW: 1145; ESJS: 1069–1070; LIV2
: 289–290; Derksen 2008: 514);
ao. -lęxъ ‘bendʼ (cf. pr. -lękǫ < IE*√lenk-; cf. Latv. lùocu ‘bend repeatedlyʼ; cf. Pokorny
IEW: 676–677; ESJS: 417; LIV2
: 413; Derksen 2008: 277);
ao. rěxъ ‘sayʼ (cf. pr. rekǫ < IE*√rek-; cf. OIA racayati ‘effectʼ (?); cf. van Wijk 1931:
52; Pokorny IEW: 863; ESJS: 761–762; LIV2
: 506; Derksen 2008: 433);
ao. tlěxъ ‘poundʼ (cf. pr. tlъkǫ < IE*√telk-; cf. Pokorny IEW: 1062; ESJS: 699; LIV2
:
623; Derksen 2008: 490);
Ku̯
+ s = OCS 0x (/0š):
ao. těxъ ʽflowʼ (cf. pr. tekǫ < IE*√teku̯
-; cf. OIA takti ‘goes throughʼ, Lith. tekù ‘run,
flowʼ; cf. Pokorny IEW: 1059–1060; ESJS: 956–957; LIV2
: 620–621; Derksen 2008:
489);
ao. žaxъ, žašę ʽburnʼ (cf. pr. žegǫ < IE*√dh
egu̯ h
-; cf. OIA dáhati, Lith. degù ‘burnʼ; cf.
van Wijk 1931: 52; Pokorny IEW: 240–241; ESJS: 1150–1151; LIV2
: 133–134;
Derksen 2008: 554);
The minor (etymological) development of the cluster of velar + t leads to the loss of the velar
plosive, the examples are not numerous, but also not insignificant:
K(u̯ )
+ t = OCS 0t:
letěti but Lith. lėkti ‘fly̕, the Slavic form is extended by the suffix -t- (< IE *√lek-; cf.
Meillet 1902: 180; Vasmer 2: 35; Pokorny IEW: 673; ESJS 410; LIV2
: 411; Derksen
2008: 271);
netopyrъ ‘batʼ is considered to be related to OCS nošti, L. nox, noctis, Hit. gen. nekuz
‘nightʼ (with the e-grade as in *netopyrъ!) (< IE *√neku̯
-; cf. Pokorny IEW: 762–763;
Vasmer 2: 216; ESJS 540; NIL: 504–513; Derksen 2008: 350);
pętь ‘five’, the Slavic form going back to the abstract *pénku̯
-ti- ‘unity of five’,115
which
also explains the construction with gen. pl. (cf. Vasmer 3: 471; Comrie 1992: 752–
754; Blažek 1999: 225–226; ESJS 643–644; Derksen 2008: 400);
potъ ‘sweat’ is an o-grade allomorph of the same root-morpheme as in pekǫ, pešti ‘bake’,
i.e., from the IE root *√pekṷ
-, again extended by the -t-suffix (cf. Zupitza 1899: 266;
Vasmer 2: 417; ESJS 689; NIL: 548–552; Derksen 2008: 415). This development is
of a special importance, since we have attested a major productive development as
well.
Note: There is no special minor development of the cluster of *K(u̯ )
s.
The different major and minor outcomes were, it seems, initially determined by the context;
however the assumption of a twofold development of the clusters of velar + t is sometimes
rejected, as by Hujer (1913), who strongly argues for merely a single development: Kt > 0ti̯
>
št/c/č, according to the dialect (via gemination and simplification).
115
Or it is a backward form of the ordinal pȩtъ cf. Lith. peñktas, Latv. pìektais, OPruth. penckts ʽfifthʼ, OIA
paktháḥ, Gr. πέμπτος; cf. Blažek 1999: 224);
113
The prevailing opinion is, however, that the different outcomes are caused by the palatal
context, before i̯ /i (but some scholars suppose this development even before e): Kti̯
> št, in
other contexts: Kt > 0t, the palatal outcomes being later extended to all productive contexts due
to analogy. This opinion was first formulated by Fortunatov (1888: 566–568) and later by
Uhlenbeck (1894: 519), and it was accepted by Vaillant (1950: 83), Shevelov (1964: 191),
Mareš (19691
: 75; 19992
: 67), Lamprecht (1987: 51), Arumaa (1976: 111–113), Rejzek (2008:
169) and others, accepting a trajectory through gemination for both variants. Kortlandt (1994a:
101) presupposes the development kt > kć
> ti̯ (> št/c/č) for the palatal context and does not
mention other contexts, and in rejecting gemination, he differs from the other authors listed
above.
It is worth noting that the reflexes of the Kti̯
-clusters have merged with the development
of the ti̯ -clusters, and these dialectally differ among the Slavic languages. Old Church Slavonic
and the East South Slavic languages have št; the West Slavic languages have c, East Slavic and
West South Slavic have č, SCr. ć, cf. OCS noštь, B. nošt, Cz., Pol. noc, Ru. noč’, Sln. nȏč, SCr.
nȏć ‘night’ (similarly for *ti̯ : OCS svěšta, B. svešta, Cz. svíce, Pol. świeca, Ru. sveča, Sln.
sveča, SCr. sveća/svijeća ‘candle’).
4.3.3 The development of the clusters palatovelar + t/s
All clusters formed by an original palatovelar and t or s follow the regular formulae. The
development seems to correspond to the behaviour of *st/ss-clusters, with which the original
Indo-European Ḱt/Ḱs-clusters have merged. The original palatovelar is depalatalized before tand
lost before s-, surprisingly there is no palatalization of this suffixal s- according to
Pedersen’s Law (ruki-rule), known from Indo-Iranian and securely attested for Slavic clusters
from IE *K(u̯ )
s > CS 0x/0š (see above):
Ḱ + t = OCS st:
inf. nesti, sup. nestъ ‘carryʼ (cf. pr. nesǫ < IE*√H1neḱ-; cf. Lith. nešù ‘carryʼ, Toch. B.
eṅtār ‘grab, carryʼ; cf. Pokorny IEW: 316–318; ESJS: 538–540; LIV2
: 250; Derksen
2008: 350);
nom. pьstrъ ‘colorfulʼ (inf. pьsati ‘drawʼ < IE*√pe ḱ-; cf. OIA piśāná- ‘drawʼ, Toch. B
piṅken ‘draw, colorʼ; cf. Pokorny IEW: 794–795; ESJS: 740–741; LIV2
: 465–466;
NIL: 546–548; Derksen 2008: 430–431);
nom. mastъ ‘anointʼ (pr. mazati < IE*√maǵ-; cf. Gr. μάσσω, Att. μάττω ‘knead, wipeʼ;
cf. Pokorny IEW: 696–697; ESJS: 459–460; LIV2
: 421; Derksen 2008: 304–305);
RuCS inf. mlěsti ‘bring downʼ (RuCS pr. mъlzǫ < IE*√H2melǵ-; cf. Gr. ἀμέλγω ‘milkʼ; cf. Pokorny IEW:
722–723; ESJS: 482; LIV2
: 279; Derksen 2008: 307–308);
inf. -lěsti, sup. lěstъ ʽcrawlʼ (pr. -lězǫ < IE*√leH1ǵh
-; cf. Pokorny IEW: 660; ESJS: 259–
260; LIV2
: 400; Derksen 2008: 275–276);
114
inf. vesti, sup. vestъ ʽcart, leadʼ (pr. vezǫ < IE*√u̯ eǵh
-; cf. OIA váhati ʽcart, leadʼ, L. uehō
ʽdriveʼ; cf. Pokorny IEW: 1118–1119; ESJS: 1047; LIV2
: 661; Derksen 2008: 517);
inf. -vrěsti ʽboundʼ (pr. -vrьzǫ < IE*√u̯ erǵh
-; cf. Pokorny IEW: 1154–1155; ESJS: 693;
LIV2
: 688; Derksen 2008: 515);
Ḱ + s = OCS 0s:
ao. nesъ ʽcarryʼ (cf. pr. nesǫ < IE*√H1neḱ-; cf. Lith. nešù ‘carryʼ, Toch. B. eṅtār ‘grab,
carryʼ; cf. Pokorny IEW: 316–318; ESJS: 538–540; LIV2
: 250; Derksen 2008: 350);
ao. lěsъ ʽcrawlʼ (pr. -lězǫ < IE *√leH1ǵh
-; cf. OPruss. līse ‘crawlʼ; cf. Pokorny IEW: 660;
ESJS: 259–260; LIV2
: 400; Derksen 2008: 275–276);
ao. -věsъ ʽcart, leadʼ (pr. vezǫ < IE*√u̯ eǵh
-; cf. OIA váhati ʽcart, leadʼ, L. uehō ʽdriveʼ;
cf. Pokorny IEW: 1118–1119; ESJS: 1047; LIV2
: 661; Derksen 2008: 517);
ao. -vrěsъ ʽboundʼ (pr. -vrьzǫ < IE *√u̯ erǵh
-; cf. OE wyrgan ‘strangleʼ; cf. Pokorny IEW:
1154–1155; ESJS: 693; LIV2
: 688; Derksen 2008: 515);
adj. desnъ ʽrightʼ (< IE *√deḱs-n; cf. OIA dákṣiṇa-, L. dexter, Lith. de͂ šinas; cf. Pokorny
IEW: 190–191; ESJS: 127; Derksen 2008: 100–101);
nom. osь ʽaxleʼ (< IE *√H2eḱs-i; cf. OIA ákṣa-, L. axis, Lith. ašìs; cf. Pokorny IEW: 6;
Arumaa 1976: 100; ESJS: 603; NIL: 259–263; Derksen 2008: 380);
*CS lososъ ʽsalmonʼ (cf. Cz. losos, Ru. losós, Pol. łosoś; < IE *loḱs-os-; cf. Lith. lašišà, ON lax ʽsalmonʼ,
TochB laks ʽfishʼ; cf. Pokorny IEW: 653; Vasmer 2: 61; Derksen 2008: 285);
4.3.4 The development of clusters dental + t/s
All clusters of dental + t and dental + s follow regular formulae of the sibilantization of the
dental before an obstruent; both sibilants are later simplified for clusters of Ts:
T + t = OCS st:
inf. gnesti ‘kneadʼ (pr. gnȩtǫ < IE *√gnet-; cf. OHG knetan ‘kneadʼ; cf. Pokorny IEW:
371; ESJS: 182; LIV2
: 191; Derksen 2008: 168);
inf. cvьsti ʽflourishʼ (pr. cvьtǫ < IE *√ḱu̯e t-; cf. OIA cetati ‘glow, shineʼ; cf. Pokorny IEW:
916–917; ESJS: 97; LIV2
: 347; Derksen 2008: 258–259);
inf. čisti ‘count, readʼ (pr. čьtǫ < IE *√ku̯
e t-; cf. OIA cikéta ‘beware, know, understandʼ;
cf. Pokorny IEW: 637; ESJS: 108–109; LIV2
: 382; Derksen 2008: 89–90);
inf. mesti ‘throwʼ (pr. metǫ < IE *√met-; cf. Lith. mèsti ‘throwʼ; cf. Pokorny IEW: 703–
704; ESJS: 462–463; LIV2
: 442; Derksen 2008: 308–309);
pr. dastь, daste, dasta, sup. dastъ ‘giveʼ (inf. dati < IE*√deH3-; cf. OIA dattá-; cf. Pokorny
IEW: 223–225; ESJS: 123–124; LIV2
: 105; Derksen 2008: 96);
inf. jasti, pr. jastъ ‘eatʼ (nom. jadъ ‘foodʼ < IE*√H1ed-; cf. OIA átti, L. edō; cf. Pokorny
IEW: 287–288; ESJS: 273–274; LIV2
: 230; NIL: 208–210; Derksen 2008: 154);
inf. pasti, sup. pastъ ‘fallʼ (pr. padǫ < IE*√ped-; cf. OIA pádyate ‘fallʼ; cf. Pokorny IEW:
790–792; ESJS: 628–629; LIV2
: 458; NIL: 526–540; Derksen 2008: 392);
inf. sěsti ‘sitʼ (pr. sedǫ < IE*√sed-; cf. OIA sádathas, L. sēdī; cf. Pokorny IEW: 884–887;
ESJS: 808–809; LIV2
: 513–514; NIL: 590–600; Derksen 2008: 445, 447);
inf. bosti ‘pierce, stabʼ (pr. bodǫ < IE*√bh
edh
H2-; cf. L. fodiō, Lith. bedù ‘digʼ; cf.
Pokorny IEW: 113–114; ESJS: 74; LIV2
: 66; Derksen 2008: 59);
inf. bljusti ʽpay attentionʼ (pr. bljudǫ < IE*√bh
eu̯ dh
-; cf. OIA bódhayati ‘observeʼ, Gr.
πεύθομαι ‘give noticeʼ; cf. Pokorny IEW: 150–152; ESJS: 69; LIV2
: 82–83; NIL: 36–
37; Derksen 2008: 46);
115
inf. žlěsti ʽcompensateʼ (pr. žlědǫ < IE*√gh
eldh
-; cf. Goth. -gildan ‘repayʼ; cf. Pokorny
IEW: 436; ESJS: 1157–1158; LIV2
: 197; Derksen 2008: 556–557);116
inf. gręsti ʽgoʼ (pr. grędǫ < IE*√gh
re dh
-; cf. Goth. grid ‘stepʼ, OIr. in:greinn ‘persecute,
followʼ; cf. Pokorny IEW: 457–458; ESJS: 202–203; LIV2
: 203; Derksen 2008: 188);
inf. klasti ‘putʼ (pr. kladǫ < IE*√k(u̯)
leH2-dh
-; cf. Lith. klóju ‘spread, coverʼ; cf. Pokorny
IEW: 599; ESJS: 310–311; LIV2
: 362; Derksen 2008: 222–223);
inf. vesti, sup. vestъ ‘lead, conductʼ (pr. vedǫ < IE*√u̯edh
-; cf. OIr. fedid, Lith. vèsti; cf.
Pokorny IEW: 1115–1117; ESJS: 1046–1047; LIV2
: 659; Derksen 2008: 517);
T + s = OCS 0s:
ao. -cvisę ʽflourishʼ (pr. cvьtǫ < IE *√ḱu̯e t-; cf. OIA cetati ‘glow, shineʼ; cf. Pokorny IEW:
916–917; ESJS: 97; LIV2
: 347; Derksen 2008: 258–259);
ao. čisъ ‘count, readʼ (pr. čьtǫ < IE *√ku̯
e t-; cf. OIA cikéta ‘beware, know, understandʼ;
cf. van Wijk 1931: 52; Pokorny IEW: 637; ESJS: 108–109; LIV2
: 382; Derksen 2008:
89–90);
pr. dasi ‘giveʼ (inf. dati < IE*√deH3-; cf. OIA dátsvá, dítsant-; cf. Pokorny IEW: 223–
225; ESJS: 123–124; LIV2
: 105; Derksen 2008: 96);
ao. jasъ, pr. jasi ‘eatʼ (nom. jadъ ‘foodʼ < IE*√H1ed-; cf. OIA átti, L. edō; cf. Pokorny
IEW: 287–288; ESJS: 273–274; LIV2
: 230; NIL: 208–210; Derksen 2008: 154);
ao. basъ ‘pierce, stabʼ (pr. bodǫ < IE*√bh
edh
H2-; cf. L. fodiō, Lith. bedù ‘digʼ; cf. Pokorny
IEW: 113–114; ESJS: 74; LIV2
: 66; Derksen 2008: 59);
ao. -bljusъ ʽpay attentionʼ (pr. bljudǫ < IE*√bh
eu̯ dh
-; cf. OIA bódhayati ‘observeʼ, Gr.
πεύθομαι ‘give noticeʼ; cf. van Wijk 1931: 52; Pokorny IEW: 150–152; ESJS: 69;
LIV2
: 82–83; NIL: 36–37; Derksen 2008: 46);
ao. žlěsъ ʽcompensateʼ (pr. žlědǫ < IE*√gh
eldh
-; cf. Goth. -gildan ‘repayʼ; cf. Pokorny
IEW: 436; ESJS: 1157–1158; LIV2
: 197; Derksen 2008: 556–557);
ao. klasъ ‘putʼ (pr. kladǫ < IE*√k(u̯ )
leH2-dh
-; cf. Lith. klóju ‘spread, coverʼ; cf. Pokorny
IEW: 599; ESJS: 310–311; LIV2
: 362; Derksen 2008: 222–223);
ao. věsъ ‘leadʼ (pr. vedǫ < IE*√u̯ edh
-; cf. OIr. fedid, Lith. vèsti; cf. Pokorny IEW: 1115–
1117; ESJS: 1046–1047; LIV2
: 659; Derksen 2008: 517);
4.3.5 The development of clusters sibilant + t/s
All clusters of a sibilant (with either s or š) + t or + s have regular development: the clusters
with t- are preserved, the clusters of two sibilants are simplified:
s + t = OCS st:
inf. trȩsti ‘shakeʼ (pr. trȩsǫ < IE *√tre(m)s-; cf. OIA trásati, Gr. τρέμω, L. tremō ‘shiverʼ;
cf. Pokorny IEW: 1095; ESJS: 982–983; LIV2: 650–651; Derksen 2008: 497);
pr. jestъ, jeste ‘beʼ (pr. jesmь < IE *√H1es-; cf. OIA ásti, L. est, Goth. ist; cf. Pokorny
IEW: 340–341; ESJS: 283–284; LIV2: 241–242; Derksen 2008: 146);
inf. pasti ‘pasture, herdʼ (pr. pasǫ < IE *√peH2-s-; cf. Pokorny IEW: 787, 839; ESJS:
629–630; LIV2: 460; Derksen 2008: 392);
nom. gostь ‘guestʼ (< IE *gh
osti-; cf. L. hostis ‘enemy, strangerʼ, Goth. gasts ‘guestʼ; cf.
Vasmer 1: 300; Pokorny IEW: 453, 540; ESJS: 193; NIL: 173;Derksen 2008: 180–
181);
116
Derksen (2008: 597) considers a borrowing from Germanic, however if valid, it still fully participates on
developments Tt > st and Ts > 0s.
116
š + t = OCS st:
ao. -kryste ‘coverʼ (ao. kryxomъ, pr. kryjǫ < IE*√kreu̯ H-; cf. Lith. kráuju; cf. Pokorny
IEW: 616; ESJS: 372–373; LIV2
: 371; Derksen 2008: 254);
nom. prьstъ ‘fingerʼ (< IE *√pr̥ -stH2-; cf. OIA pṛ-ṣṭhám ‘back, ridgeʼ, Lith. pir̃štas; cf.
Vasmer 2: 344; Pokorny IEW: 813; ESJS: 732–733; Derksen 2008: 428–429);
nom. pьrstь ‘dust, earthʼ (< IE *√pr̥ s-t-; cf. OIA pṛṣant- ‘dottedʼ, Lith. dial. pir̃kšnys; cf.
Vasmer 2: 344; Pokorny IEW: 823; ESJS: 733; Derksen 2008: 429);
usta ‘mouthʼ (< IE *Heu̯s-t; cf. OIA óṣṭha- ‘lipʼ; cf. Vasmer 3: 191–192; Pokorny IEW:
499, 739; Arumaa 1976: 43; ESJS: 1025–1026; NIL: 390–391; Derksen 2008: 509;);
s + s = OCS 0s:
pr. jesi ‘beʼ (pr. jesmь < IE *√H1es-; cf. OIA ási, L. es, Gr. Aeol. ἔσσι, Ep., Dor. ἐσσί; cf.
van Wijk 1931: 52; Pokorny IEW: 340–341; ESJS: 283–284; LIV2: 241–242; Derksen
2008: 146);
ao. -gasъ ‘extinguishʼ (pr. -gasiti < IE*√(s)gu̯
esH2-; cf. OIA jajāsa ‘goneʼ, Toch. A
ksalune ‘extinguishʼ; cf. Pokorny IEW: 479–480; ESJS: 1017; LIV2
: 541–542;
Derksen 2008: 161);
š + s = OCS 0š:
ao. -sъxъ ‘dryʼ (pr. -sъxnǫti117
< IE*√H2seu̯ s-; cf. OIA śúṣyati; cf. Pokorny IEW: 880–
881; ESJS: 900; LIV2
: 285; Derksen 2008: 473–474, 479);
The cluster š is regularly depalatalized to st (cf. Martinet 1955: 240; Andersen 1968: 176–177),
similarly to clusters of Ḱt (cf. examples above and more on it below). The alternation is visibly
attested for the sigmatic aorist endings: 1st
sg. -sъ/xъ, 1st
pl. -somъ/xomъ vs. 2nd
pl. -ste, 2nd
du.
-sta, 3rd
du. -ste.
4.3.7 Overview of Old Church Slavonic development
The peripheral series in the t-context tend to be realized either as a sibilant cluster or as 0t. Both
central series are realized as st within the t-context. All clusters formed in the s-context are
realized as a simple sibilant:
IE OCS t- s-
-ḱ/ǵ/ǵh
s/z st 0s
-k/g/gh
k/g118
št/0t 0š
-kṷ
/gṷ
/gṷh
k/g7
št/0t 0š
-t/d/dh
t/d st 0s
-p/b/bh
p/b 0t/st 0s
-s s st 0s
-š x/š st 0š
117
The aorist could be otherwise asigmatic.
118
Or the palatalized forms, valid also for labiovelars.
117
4.4 Trajectories of Slavic development
The whole development could be split into three blocks: the development of two central series
(dental and palatalovelar); the development of the peripheral series (plain velar + original
labiovelar and labial series); and sibilants (though this group is not based on its location
properties but its sonority)
The most archaic is the development of the central block, especially that of dental series,
which underwent similar processes as in other IE languages (cf. OCS sěsti, Lith. sėsti, Av.
hasta- < *sed-to-, etc. the same process *Tt > st; similarly to that in Italic, Germanic and Celtic,
too, where the development is *Tt > ss, cf. L. -sessus, OIr. sess, ON. sess- < *sed-to- etc.).
Similarly, the original Indo-European palatovelar series was sibilantized in an
independent process, with parallels in other Indo-European satəm-languages. Here again, the
first stage of the whole transformation of the Late Indo-European palatovelars into Slavic
sibilants started very early, being a part of the complex and shared development of all satəm-
languages.
The old plain velar and labiovelar series are merged, but a remarkable feature of their
development is twofold outcomes, similar to that of the labial series. Arumaa (1976: 56)
proposed two possible strategies for the Pre-Slavic development of the peripheral series of
clusters of plosive + t and plosive + s: either the strategy of spirantization of the left plosive (he
explicitly points to Iranian as an example of this process) or the strategy of gemination (Arumaa
gives the development of Italian as an example). It must be added that some authors seem to
prefer the ‘directʼ strategy of simplification due to the law of open syllables.
The sibilant block preserved the sibilant before t- (with depalatalization of *š, see
below); the two sibilant clusters are simplified.
4.4.1 The strategies of the development of peripheral series – (labio)velars and labials
As we have already stated, there are three possible strategies for the development of the
peripheral series (the simplification trajectory, the gemination trajectory and the spirantization
trajectory). These developments are later than the developments of the dental clusters (which
are already Late-Indo-European) or the palatovelar clusters (which are common satəm-areal at
least in their first phases), but specifically Pre-Slavic, since they are not shared with Baltic
languages and (in the case of the development of velars) even dialectally split between different
sub-branches of the Slavic languages.
118
4.4.1.1 The strategy of simplification
The ‘law of open syllablesʼ is an assumed tendency in the development of Slavic; however the
development of plosives in clusters of plosive + t and plosive + s does not necessarily have a
direct relation to any ‘law of open syllablesʼ since the whole process can be split into a series
of independent developments,119
and does not represent a single process appearing at a single
moment.
Different versions of the simplification strategy are present in van Wijk (1931: 39, 52–
53), Martinet (1952); Mareš (19691
: 28; 19992
: 34), Lamprecht (1987: 42), Carlton (1991: 100),
and Schenker (2002: 67–68). In some respect, the simplification strategy is a convenient
descriptive shortcut; however, it can hardly be a proper description of the trajectory itself. On
the other hand, the simplification of clusters of two syllables (SS) is within the model of the
‘simplificationʼ trajectory.
The trajectories for all three peripheral series would be (note the dialectal distinction
between given Slavic languages in the case of the major development of *K(u̯ )
t):
K(u̯ )
t > št (major) (East South Slavic)
K(u̯ )
t > c (major) (West Slavic)
K(u̯ )
t > č (major) (West South, East Slavic)
K(u̯ )
t > 0t (minor) (Common Slavic)
Pt > 0t (major) (Common Slavic)
Pt > st (minor) (Common Slavic)
K(u̯ )
s > 0x/0š (Common Slavic)
Ps > 0s (Common Slavic)
The strategy of simplification would mean that the input plosives were changed into the output
elements (including zero) in a single stroke (i.e., directly from the input to the output). Tempting
though this strategy could be, it is hardly a workable model: why would part of the clusters of
*K(u̯ )
t be realized as št/c/č and another part as 0t – the explanation needs at least one intermediate
stage between the input and the output. Similarly, the twofold development of the cluster of *Pt
needs at least one intermediate stage for the same reasons. The development of K(u̯ )
s also
119
This is strongly supported by the fact that some of the partial developments considered to be part of the PreSlavic
‘law of open syllablesʼ are attested in other Indo-European branches as well: the loss of final plosives
is known from Ancient Greek and Lithuanian, the loss of the final -s is attested in numerous branches, the
nasalization of vowels and monophthongizations of diphthongs from Middle Indo-Aryan, etc.
119
requires an intermediate stage affected by Pedersen’s Law; again, the outcome could not be a
result of a single process.
4.4.1.2 The strategy of gemination
The strategy of gemination is preferred not only by Arumaa himself but also by other scholars
(cf. Meillet 1924: 111; Mikkola 1942: 162; Vaillant 1950: 73–74, 82; Shevelov 1964: 188;
Townsend/Janda 1996: 52). Arumaa (1979: 56) mentions the development of Italian as an
external typological example.
Note: The gemination in Italian can be documented in the following examples:
Vulgar Latin Kt-clusters in factus, octo, cocta, directus, frigidus are realized in Italian fatto, otto, cotta, diritto,
freddo (L. Kt > It. tt); Vulgar Latin Pt-clusters in scriptus, raptus, subtus are realized in Italian scritto, ratto, sotto
(L. Pt > It. tt); Vulgar Latin Ks-clusters in rixa, saxum, lapsus are realized in Italian ressa, sasso, lasso (L. Ks > It.
ss); the Vulgar Latin Ps-clusters in gypsum, capsa, ipse are realized as Italian gesso, cassa, esso (L. Ks > It. ss);
The central Tt/Ts-cluster had shifted to ss already before. Note that the gemination product always reflects the right
obstruent both in location and sonority.
As we have noted, in Slavic, the development of the clusters of (labio)velar + t has, as noted
above, two different outcomes: either the major št (or its replacement c/č according to the given
dialect) or the minor 0t. It seems that the context initially governed the outcomes: palatalized
Kti̯
gave št while a non-palatal context gave 0t. The palatal context seems to be limited to the
position before a high palatal vowel only. The palatal variant later superseded the non-palatal
variant in all productively formed clusters, even those where the palatal form could not be
original, as in the supine forms where the supine suffix was -tъ (from IE *-tum, cf. OIA dātum,
L. datum, OCS otъ-datъ). This process was probably caused by analogy with numerous forms
in the palatal contexts. It has to be noted that Hujer (1913) rejects the two-way development of
the clusters of (labio)velar + t and assumes only one universal development Kt > tt > ti
(>
št/c/č).
As mentioned above, the theory of the context-based parallel development was
originally presented by Fortunatov (1888: 566–568) and Uhlenbeck (1894: 83) and later, in
minor variants, it has been generally accepted by Mikkola (1942: 162),120
Vaillant (1950: 83)
and others, becoming the prevailing view. Shevelov (1964: 191) assumes the development kt >
tt > ti̯ (> št/c/č) before i, otherwise the simple development kt > tt > 0t. Martinet (1955: 353)
120
Mikkola (1942: 162) uses the development of the Romance cluster kt before a palatal vowel to demonstrate the
two-way development of Slavic kt. Mikkola assumes the trajectory kti̯
> ćt > št. Independently, Kortlandt
presupposes a development kti̯
> kć > ti̯ .
120
suggests that the left plosive became implosive121
and later formed a geminate with the right
plosive (kt > ɠ̊ t > tt). Mareš (19691
: 75; 19992
: 67) prefers the trajectory: kt > kti̯
> ki̯
ti̯
> ti̯
(>
št/c/č) for the palatal version, and his idea was accepted later by Lamprecht (1984: 51). Different
trajectories, both for palatal and non-palatal clusters, are assumed by Arumaa (1976: 111–113).
Rejzek (2008: 169) described the trajectory as: kt > ki̯
t > ti̯
t > ti̯
ti̯
> (št/c/č). Collins (2017: 1451)
assumes a secondary spread of the palatal variant to the original non-palatal (productive)
contexts. We prefer the trajectory as described by Mareš for the velar clusters in the palatal
context, Shevelov’s for the non-palatal context (see below).
Similarly, the clusters of labials + t have two outcomes too, probably governed by
contexts, though the contexts of both variants could not be specified as the same as the clusters
of (labio)velar + t. The gemination theory seems to be a prevailing explanation (cf. Meillet
1924: 111; Vaillant 1950: 73–74, 82; Shevelov 1964: 188; Townsend/ Janda 1996: 52). For the
major development, the trajectory could be stated as Pt being first geminated to tt, and later this
geminate being simplified to 0t. The minor development shares the gemination stage, but
instead of simplification, it is subjected to dissimilation, hence the trajectory Pt > tt > st (Vey
1931: 64–65; Patri 2003: 124–127). Patri assumes that the dissimilation was essentially the
same as that of the ‘acuteʼ labial clusters, and this provides an explanation for why this
dissimilation would affect only parts of the geminates (especially since even Pt gave tt,
according to the gemination strategy).
All the clusters of the peripheral plosive + s underwent, according to the geminate
strategy, development through universal gemination and simplification: Cs > ss > 0s (cf.
Shevelov 1964: 188–190).
The gemination trajectory could be modelled as follows:
K(u̯ )
t > tt > št (major) (East South Slavic)
K(u̯ )
t > tt > c (major) (West Slavic)
K(u̯ )
t > tt > č (major) (West South, East Slavic)
K(u̯ )
t > tt > 0t (minor) (Common Slavic)
Pt > tt > 0t (major) (Common Slavic)
Pt > ? > st (minor) (Common Slavic)
K(u̯ )
s > kš > šš > 0x/0š (Common Slavic)
Ps > ss > 0s (Common Slavic)
121
It seems that Martinet by the term meant ‘un-explodedʼ stop, not the true implosive stop.
121
What is questionable is the minor development of the cluster of *Pt, resulting in st since it
would be the same as that of IE *Tt > CS st, but since all other peripheral clusters do not share
this development, it has to be later than the development of the dental series!
4.4.1.3 The strategy of spirantization
The strategy of spirantization is considered as another possibility by Arumaa (1976: 56), who
offers the Iranian development as a parallel. Arumaa himself rejects the spirantization
trajectory, but it is present in some works by other scholars. Martinet (1955: 353, 365–366)
works at least in some cases with spirantization as a model for the development of the
palatalized Kti̯
-cluster, since he assumes the trajectory kti̯
> çť, at least for a part of the Slavic
dialectal continuum (otherwise he operates within the limits of gemination/simplification
strategy for kti̯
> ti̯
ti̯
> 0ti̯
). Similarly, Mikkola (1942: 162) reconstructs a cluster of ćť. A general
spirantization of the velar in the Kt-cluster is also considered by Rejzek (2008: 169), but
afterwards, he prefers the traditional gemination strategy. Vey (1931b: XV) mentions a personal
communication with Maurice Grammont, who favored the idea of a spirantization of the
clusters Pt > ft (and later > st).122
The spirantization of the peripheral series is well attested for
Sabellian languages, Celtic, Iranian and Middle Greek, being the common strategy of
development.
The spirantization model can be viewed as only a part of a broader lenition model, as
we will see in the following. For the Pt-clusters, we suppose the following development: Pt was
spirantized to φt, this cluster was in the second phase debuccalized to ht, and this cluster was in
the third phase subjected to the elision of h, resulting in 0t in the major development.
The minor development of the clusters of Pt was the same in the first phase, i.e., in the
spirantization Pt > φt, but in the second phase instead of debuccalization, the cluster was
sibilantized to st.
The non-palatal Kt-clusters developed similarly to the development of the cluster of
labial + t: Kt was spirantized to xt. This cluster was later debuccalized to ht and elided to 0t,
which is in accord with the minor development as described above.
The palatalized velar clusters of Kti̯
had a slightly different trajectory of development:
Kti̯
was spirantized to xti
in the first phase, but there appear different developments according
122
By ft is highly probably meant any cluster formed by a labial spirant, either bilabial or labiodental.
122
to the later dialectal split, because the Kti̯
-clusters are realized as št in Old Church Slavonic and
Bulgarian (East South Slavic), but in West Slavic they appear as c, in East Slavic and Slovenian
as č and in West Balkan languages as ć; cf. OCS (pekǫ ~) pešti ‘bakeʼ, but OCz. (peku ~) péci;
Pol. (piekę ~) piec; Ru. (pekú ~) peč̕; Sln. péči; SCr. pèći, etc. In all branches of Slavic, the
result of this process merged with the result of the development of the ti̯ -cluster (cf. OCS svěšta
‘candleʼ but OCz. sviecě, Pol. świeca, Ru. svěčá, Sln. svẹ́ča, SCr. svijèća). According to the
spirantization/lenition trajectory, the Kti̯
-cluster in the dialect preceding Old Church
Slavonic/Bulgarian was spirantized first to xti̯
and later to çt, and the cluster later became
sibilantized to št. In other dialects, where the output is c/č, the cluster of çt was realized in the
second phase as a cluster of i̯ t, and in the third phase turned into an affricate according to the
specific (and later) parallel developments in given dialects (cf. Schenker 2002: 76; but
previously Belić 1921).
For the development of the labial plosive + t, we assume the spirantization of the labial
plosive, later debuccalized and subjected to elision according to the major trajectory. In the
minor trajectory we assume a sibilantization of the labial spirant (probably valid for ptr-clusters
only).
All peripheral plosives were spirantized before s-. Later this spirant was uniformly
sibilantized, and the cluster was even later degeminated:
K(u̯ )
t > xt > çt > št (major) (East South Slavic)
K(u̯ )
t > xt > t > c (major) (West Slavic)
K(u̯ )
t > xt > t > č (major) (West South, East Slavic)
K(u̯ )
t > xt > ht > 0t (minor) (Common Slavic)
Pt > φt > ht > 0t (major) (Common Slavic)
Prt > φtr > st(r) (minor) (Common Slavic)
K(u̯ )
s > xx/šš > 0x/0š (Common Slavic)
Ps > φs > ss > 0s (Common Slavic)
Note that the null minor outcome of *Kt is realized as the major outcome of *Pt and there is a
similar parallel between the sibilant major outcome of *Kt and the minor outcome of *Pt.
4.4.2 The development of the central series I: the palatovelar series
The development of IE reconstructed palatovelars is a matter of the whole satəm-area, in that
at least the oldest stage was common for all later satəm-languages and for later stages of
development we can assume at least a common drift in the similar direction towards sibilants.
123
The second phase of the development of IE palatovelars was earlier assumed to be a
sibilant. Inside this model, we can only assume a depalatalization both before t- and s- (and
later simplification of *śs on 0s, as with old IE cluster *ss).
However, in the last decades the prevailing opinion, based on Nūristānī evidence, is that
old Indo-Europeans were realized as affricates in the common satəm-phase, and these were later
sibilantized. Within this model, the trajectory of the development could be modelled with
affrication in the first phase, later loss of the plosive segment of the affricate and later
depalatalization. That in Slavic the IE palatovelars were depalatalized could be demonstrated
by the parallel development in Iranian, where, though palatovelars were depalatalized in
general, palatalization was preserved before a plosive, cf. YAv. ppp. vašta- vs YAv. pr. vazaiti,
both from the root √vaz- ‘driveʼ < IE *√u̯ eǵh
-. We assume a similar process for Slavic,
especially since even a cluster of *št resulting from Pedersen’s Law was probably also
depalatalized (cf. Andersen 1968: 175–177, 188–190). For IE *Ḱt > CS *st we model the
following trajectory (a similar trajectory for Indo-Iranian was modelled by Lipp 2009 I: 139–
140):
Ḱt > tš
t > št > st (Common Slavic)
The development of IE clusters *Ḱs, according to the affricate model could be modelled with
an affricatization, and later a loss of the plosive segment and simplification (cf. Andersen 1968:
175–177 and Lipp 2009 I: 155; Lubotsky 2018: 1885 for the parallel Indo-Iranian
development):
Ḱs > tš
s > šs > ss > 0s (Common Slavic)
The problematic point of this development is why *šs (< *Ḱs) would lose its palatal segment
when *šs (from the ruki-cluster *šs) is realized as CS 0š. The solution could be to assume that
*Ḱs was subjected to another development than that of affricatization, but this argument just
brings another variable to the list of possible trajectories.
We propose a different variant, assuming that the old palatovelars were realized before
obstruents neither as sibilants nor affricates, but as spirants, before an t/s- as ç (the spirantization
model). This palatal fricative was later depalatalized to ϑ both before t- and s-. This dental
spirant was sibilantized and preserved before t-, the two-sibilant cluster (from *Ḱs) was
simplified (cf. Andersen 1968: 189):
Ḱt > çt > ϑt > st (Common Slavic)
124
Ḱs > çs > ϑs > ss > 0s (Common Slavic)
4.4.3 The development of the central series II: the dental series
The traditional model, popularized by Brugmann (1880 and since, but originally by Kräuter
1877) assumes the affricatization of the first dental plosive and a later loss of the plosive
segment of this affricate. This affricatization model is usually used for the Slavic development
as well123
(cf. Vaillant 1950: 80–81; Arumaa 1976: 79–80). The popularity of this model was
supported by the fact that in Hittite the outcome of IE *Tt is ts
t (cf. Hitt. pr. ezši ‘eatʼ < IE
*√H1ed-; Pokorny 1959: 287–289; Friedrich 1990: 44; Melchert 1994: 97, 109; HED 1–2: 315–
321; Kloekhorst 2008: 26, 261–263).
The affricatization trajectory for the development of IE *Tt then could be modelled as:
Tt > ts
t > st (Common Slavic)
On the other hand, the same affricatization trajectory for the development of IE *Tt then would
be modelled with an affricatization first, followed by the loss of the plosive segment of the
affricate and later simplification:
Ts > ts
s > ss > 0s (Common Slavic)
Note: The universal strategy both for *Tt and *Ts is usually abandoned, and authors usually assume the ‘directʼ
assimilation outside the ‘affrication trajectoryʼ, cf. Vaillant (1950: 80); Arumaa (1976: 78): Ts > ss > 0s.
Another trajectory was proposed for the development of IE cluster *Tt , assuming spirantization
instead of affrication. For Italic languages, it was proposed by Cocchia (1883: 16–58), for IndoIranian
by Bartholomae (1895: 16 and later works), and later taken as a possibility by Leumann
(1942: 13). Within this model, the trajectory for both Tt and Ts will contain first the
spirantization of the plosive, later sibilantized (and the first sibilant being degeminated for *ss
as it was with all other sibilants-only clusters):
Tt > ϑt > st (Common Slavic)
Ts > ϑs > ss > 0s (Common Slavic)
The advantage of this model is its simplicity and universality for both clusters; it is worthy of
mention that Armenian development (see below), as described by Winter (1962: 261): *Tt >
123
The development of IE *Tt is often omitted in general overviews on the phonemic development of Slavic; it is
not a subject of analysis in Meillet 1924; Shevelov 1964; Mareš 1969 (1999); or Townsend/Janda (1996) at
all.
125
tϑ > ϑt > u̯ t124
(Arm. giwt ‘findʼ < IE *u̯ id-ti-) assumes spirantization and could serve as a
counter-proof to the Hittite example.
4.4.4 The development of the sibilant series
The development of the cluster of *st is conservative; clusters are fully preserved. The
development of the ruki-cluster of št is more interesting, which has undergone a
depalatalization, as did clusters from IE *Ḱt (cf. Martinet 1955: 240; Andersen 1968: 176–177,
188–190). The alternative solution could be that Pedersen’s Law was not operating before *t(cf.
Shevelov 1964: 127; Arumaa 1976 II: 43; Pedersen himself had the idea that his law did
not operate in Slavic before a plosive, cf. Pedersen 1895: 74).
We prefer the later depalatalization of a cluster for we have attested the aforementioned
similar development *Ḱt > st (in Iranian, where IE palatovelars were depalatalized as in Slavic,
the palatalization was preserved before a plosive, cf. YAv. ppp. vašta- vs YAv. pr. vazaiti, both
from the root √vaz- ‘driveʼ < IE *√u̯ eǵh
-):
st > st (Common Slavic)
št > st (Common Slavic)
Note: The development of the Pre-Slavic ruki-sibilant could be even more complicated if the outcome of the
Pedersen’s rule was originally a palatal non-sibilant spirant ç (depalatalized universally to x later, š being its
later palatal sibilant variant before front vowels). In this case, the input would be çT, with later palatalization
of a second sibilant, assimilation and degemination: çt > št > st, alternatively çt > ϑt > st (as in the case of the
spirantization model of the development of the IE cluster *Tt).
The development of the two-sibilant clusters is simple; the cluster was degeminated (the rukicluster
was first assimilated):
ss > 0s (Common Slavic)
šs > šš > 0š (Common Slavic)
Note: Again, if the outcome of Pedersen’s Law was originally a palatal non-sibilant spirant ç (later x, except before
front vowels, where š), the input would be çs, with later palatalization of a second sibilant, assimilation and
degemination: çs > çš > šš > 0š, alternatively çs > çš > ϑš > sš > 0š (as in the case of the spirantization model
of the development of the IE cluster *Tt).
4.5 Concluding remarks
The Pre-Slavic development is remarkably different from that of Baltic, since almost all the
clusters of our interest were totally or partially remodelled.
The oldest part is the development of the clusters of dental + t/s. Instead of the traditional
affricativization trajectory of Kräuter and Brugmann we prefer its spirantization variant, since
124
We assume the variant trajectory: Tt > ϑt > ht > u̯ t
126
it could better explain the transition of the IE *Ts into Slavic 0s. The spirant is more probably
to be sibilantized instead of the affricate in the s-context.
The second oldest is the development of the clusters of palatovelar + t/s. Here we also
prefer the spirantization trajectory, again because of the *Ḱs cluster and for similar reasons as
we do with the assumed development of the *Ts cluster. The output clusters were later both
depalatalized.
For the development of peripheral series, we have to reject any trajectory assuming the
simple loss of the plosive via ‘the law of open syllablesʼ: this model could only be a shorthand
explanation, nothing more. If we have to choose from other two proposed trajectories, we have
to prefer the spirantization trajectory, since it fits the known outcomes more fully, especially if
considering the explanation of the minor developments, inexplicable as results of a gemination
process. Spirantization is a very common development, attested at least for Iranian (but not for
Indic), Sabellian (but not for Latin) and Celtic, and even as a later development for the Middle
Greek and Italian.
The *st cluster is preserved, but the ruki-cluster *št is depalatalized, as is the cluster *Ḱt;
both clusters were merged before this development.
Both two-sibilant clusters were simplified, as were all s-context clusters with the
original plosive in the left position.
127
5 The development of the two-obstruent clusters from Indo-European into
Armenian
5.0 Armenian language
Armenian is a satəm-language, forming its own branch within the Indo-European family, related, as many suggest,
especially to Greek. This is often as a part of a wider branch of IE languages, often including Phrygian and relic
languages of Balkan peninsula (cf. Pedersen 1904; Solta 1960; Džaukjan 1967: 9-31; Hamp 1976; Martirosyan
2013; but rejected by Clackson 1994 or Kim 2018). More recently, the closeness of Armenian and Albanian was
put forward (‘Proto-Albano-Armenianʼ); there is at least of strong parallelism in their development (cf. Kortlandt
1980b; Kortland 1986).
Armenian is literally attested since 5th
century AD, with its own script. Armenian was influenced in its
development by its extensive contacts with the Urartian substrate, by Middle-Eastern adstrates, by the Caucasian
languages, by the Iranian superstrate, Greek adstrate and cultural superstrate and many other influences both in
the lexicon and grammatical features, resulting in a very complex and intricate language, in many aspects
extremely transformed from the reconstructed Indo-European state (cf. Olsen 2017: 421–423).
5.1 Armenian and Indo-European
The typical features separating Armenian obstruent system from that of Indo-European are:
i. a partial merging of old labiovelars and plain velars125
and the existence of palatovelars
(again, in the modified form);
ii. a shift in the modality of plosives, shortly described as transition of voiceless non-aspirated
plosives to voiceless aspirated (T > Tʿ; cf. Winter 1954; Winter 1955 – Winter assumes an
original spirantization, followed by a fortition on an aspirate; in this he follows Meillet 1903:
7–8, 12–15; Meillet 1936: 25–26, 31–34; similarly Kortlandt 1980b: 28; Kümmel 2007:
370–371; Kim 2016); of voiced non-aspirated plosives to voiceless non-aspirated (D > T);
and of voiced aspirates to voiced non-aspirates (Dh
> D) (cf. Kortlandt 1978a: 24; Kortlandt
1980a: 100; Macak 2018: 1047–1048);
iii. a probable split of IE *s according to Pedersen’s Law (ruki-rule), at least after IE r (see
below). IE *s in other positions than before a voiceless plosive or after r/n underwent
numerous processes: #s- > #h- before i, -s# > -x# > -kʿ#126
, s > h > 0 in all other positions
(cf. Winter 1955: 7; Beekes 2003: 169–170).
The first feature is shared with Albanian since Albanian has etymologically preserved at least
partially the distinction between old labio- and plain velars in similar conditions as Armenian;
the second process has a parallel in the development127
of the Germanic consonantal shift
(Lautverschiebung), Pedersen’s Law/the ruki-rule is securely attested for Indo-Iranian, BaltoSlavic
and is also possible (but not securely proven) for Albanian.
The development of given obstruents is far from being simple when covering the main
tendencies (for general overviews and internal chronology, cf. esp. Solta 1963; Džaukjan 1967,
125
The labiovelars were palatalized before *e, i, cf. Stempel (1994); Job (1995); Beekes (2003: 177–179); Schmitt
(2007: 62–65, 78–79); Martirosyan (2010: 711); Macak (2018: 1056).
126
Note that IE -s# also gave -ḥ# in OIA, this spirant visarga was later elided in MIA.
127
Though we cannot accept the idea that Armenian and Germanic languages are more close to the reconstructed
IE triad of modal classes, as proposed by Griffen (1988: 162–189; Griffen 1989).
128
especially 313–332; Godel 1975: 61–91; Kortlandt 1976; Kortlandt 1980a; Job 1995; Beekes
2003; Schmitt 2007: 56–79; Martirosyan 2010: 705–747; Kim 2016; Olsen 2017: 423–434;
Macak 2018). An overview of the Armenian historical phonology by Mann (1963) is very
unreliable, and his findings are not shared or quoted by other authors; hence we should use them
with the greatest caution.
5.2 Armenian clusters and their IE origins
Since Armenian morphology is a result of deep and significant structural changes, the attested
system remarkably differs both from the reconstructed Indo-European morphology and from
that of Greek or Indo-Iranian, forcing us to use the etymological data exclusively.
Note: For an overview of Armenian morphology from the Indo-European point of view, cf. especially a short
overview of the verbal system by Kortlandt (1996), for a wider description of the same Klingenschmitt (1982),
for the noun morphology Olsen (1999; especially 815–856); an overview of the historical morphology of
Armenian in the complex was given by Godel (1975, especially 92–129) or later by Olsen (2017: 434–447;
2018).
Since Armenian does not show any signs of Bartholomae’s Law, all clusters with a left voiceless
obstruent are also voiceless in their particular outcome; however, such clusters are often
subsequently modified in their later developments.
5.2.1 The development of the cluster of labial + t/s
The development of the IE clusters of labial + t has the output u̯ tʿ in the inlaut and 0tʿ in the
anlaut. The IE cluster *Ps has the output 0s, but the output pʿ is also attested, resulting probably
from original clusters of *sp after metathesis:
P + t = Arm. u̯ tʿ:
ewtʿn ‘sevenʼ (< IE *septḿ̥ -; cf. OIA saptá-, Gr. ἑπτά, L. septem; cf. Hübschmann 1897:
445; Winter 1955: 6; Pokorny IEW 909; Džaukjan 1967: 96; Godel 1975: 80; Winter
1992a: 350; Kortlandt 1994b: 254; Görtzen 1998: 344; Blažek 1999: 247; Beekes
2003: 172; Schmitt 2007: 57, 59; Martirosyan 2010: 270–271; Kim 2016: 151, 155);
kartʿ ‘fish hook; legʼ (< PArm. *kar(p)ti; < IE *gr̥ b-ti; OIA grapsa- ‘bunchʼ, MHG krëbe
‘basketʼ; cf. Džaukjan 1967: 95; Pokorny IEW: 387; Görtzen 1998: 338; Olsen 1999:
81; Martirosyan 2010: 354, 725);
#pt = Arm. tʿ:128
tʿer ‘side, leafʼ (< IE *ptero-; derived from IE *√pet- ‘flyʼ; Gr. πτερόν ‘wing, featherʼ;
cf. Bugge 1893: 40; Winter 1955: 5; Džaukjan 1967: 95; Pokorny IEW: 826; Godel
128
For this alternation cf. Bugge 1893: 39–40. However, the loss of the approximant in the word-initial is not
surprising.
129
1975: 80; Klingenschmitt 1983: 99; Olsen 1999: 51–52; Martirosyan 2010: 286–287;
Kim 2016: 152; Macak 2018: 1019);
tʿekʿem ‘twist, warp, weaveʼ (< IE *√tek-; cf. Hitt. takkeszi ‘undertake, prepareʼ, L. texō
‘weaveʼ; cf. Pokorny IEW: 1058; LIV2
619–620; Kim 2016: 152);
P + s = Arm. 0s:
eres ‘faceʼ (< IE *prep-s-; cf. Gr. πρέπω ‘be clearly seenʼ; cf. Pokorny IEW: 845; LIV2
492; Beekes 2003: 198; Beekes 2016b: 1663–1664);
sut ‘lieʼ (< IE *√pseu̯ d-/ku̯
seu̯ d- (?); cf. Gr. ψεύδω ‘cheat, lieʼ, Slk. šudiť ‘deceiveʼ; cf.
Bugge 1893: 25–26; Meillet 1903: 18; Meillet 1936: 39; Pokorny IEW: 1058; Görtzen
1998: 339; Beekes 2003: 198; Martirosyan 2010: 587–588; Macak 2018: 1057);
P + s = Arm. pʿ:
epʿem ‘ boilʼ (< IE *(s)eps-; cf. Gr. ἕψω; no other IE cognate, cf. Hübschmann 1883: 69; Mann 1963:165;
Pokorny IEW: 325; Beekes 2003: 198, who suspects it to be a borrowing from non-IE;);
Arm. kapʿankʿ ‘enclosure, lid, trapʼ, kapʿnum ‘cover, shutʼ (< IE *√keH2p-; cf. Gr. κά(μ)ψα ‘basket, caseʼ,
L. Plautus capsō ‘I will takeʼ, W. caffio ‘to cachtʼ, OE haespian, haepsian ‘fastenʼ, Lith. kapsiù ‘peck
atʼ; cf. Mann 1963: 165; Pokorny IEW: 527–528; LIV2
: 344–345);
Arm. opʿi ‘white poplarʼ (< IE *ap-s-; cf. OE æpse, Ru. osína; cf. Pokorny IEW: 55; Beekes 2003: 198–
199 suspects it to be a borrowing);
Pst = Arm. st:
stin ‘woman’s breastʼ (< IE *psteno- (?); cf. Gr. στήνιον Hsch., Av. fštāna- ‘breastʼ; cf. Hübschmann 1897:
493;Pokorny IEW: 990; Olsen 1999: 135–136; Beekes 2003: 198; Martirosyan 2010: 584–585; Macak
2018: 1057);
5.2.2 The development of the cluster velar + t/s
There are only a few examples for the development of the IE clusters *Kt. Godel (1975: 80)
even assumed there are none and he relates such pairs as ałačʿem ‘I prayʼ vs aławtʿkʿ ‘prayerʼ;
čanačʿem, aor. caneay ‘I knowʼ vs canawtʿ ‘noticeʼ (in canawtʿs tam ‘I give notice); amačʿem
‘I am ashamedʼ vs amawtʿ ‘shameʼ; he traced the stem morpheme to *ak-i̯e- (cf. Gr. ἀλλάσσω
‘change, alterʼ) and the above-mentioned action nouns on -awtʿ as being from *ak-ti-.
Nevertheless, other authors propose more etymologies, with the same outcomes:
Note: Mann (1963: 119) gives a list of other possible examples on *Kt, not accepted by other research, though
fitting to the assumed pattern: butʿ ‘bluntʼ; latʿ ‘rag, clothʼ, satʿ ‘amberʼ. Similarly, Mann (1963: 120) sees the
outcome of IE *Kst in laxt ‘stick, cudgelʼ.
K + t = Arm. (u̯)tʿ:
katʿn ‘milk’ (< IE *gl̥kt-; cf. Gr. γάλα, ‑κτος; L. lac, -tis; cf. Džaukjan 1967: 95; Pokorny
IEW 400–401; Beekes 2010: 256; Martirosyan 2010: 345);
but ‘foodʼ (< IE *bh
eu̯g-ti; cf. OIA bhunákti ‘create enjoymentʼ, L. fungor ‘enjoyʼ; cf.
Hübschmann 1897: 430; Pokorny IEW 153; LIV2
: 84–85; Martirosyan 2010: 187)
Note: There is ‘non-etymologicalʼ utʿ in owtʿ ‘eightʼ from -ḱt- (cf. OIA aṣtáu, Lith. aštuonì etc.). The Armenian
form is explained from PArm. *optṓ resulting from the analogy to *septm̥ (cf. Hübschmann 1897: 483–484;
Winter 1955: 6; Solta 1960: 111–112; Pokorny IEW: 775; Godel 1975: 80; Kortlandt 1994b: 255; Blažek 1999:
265; Schmitt 2007: 59, 75; Martirosyan 2010: 631; Kim 2016: 151), only if we assume the neutralization of
130
the palatovelar before *t (to kt, otherwise irregular in Armenian) and a regular development: Kt > 0tʿ, followed
by its assimilation to pt.
Similarly poorly attested is the development of K + s, limited at the moment to a single example:
K + s = Arm. 0š:
uši either ‘storax-treeʼ or ‘holm-oakʼ (< PArm. *hoši < *hošíya < IE *H3ek-s-i̯eH2; cf.
Lith. úosis ‘ashtreeʼ (?); cf. Džaukyan 1967: 255; Martirosyan 2010: 641–642, 710, but
cf. other etymologies he mentions l.c.);
5.2.3 The development of the cluster labiovelar + t/s
There are no secure etymologies for the development of the IE clusters of *Ku̯
t/s, but since we
assume that the distinction between IE plain- and labiovelars was neutralized in all satəmlanguages
in all contexts, we can assume the same outputs as with the IE clusters of *Kt/s
mentioned above. It seems, that there is a single (and doubtful) example of *Ku̯
+ ti̯, resulting
in cʿ, which could be assumed to be a palatalized version of a regular cluster *Kt:
Ku̯
+ t = Arm. *u̯ t:
not attested
Ku̯
+ ti̯ = Arm. cʿ:
hacʿ ‘breadʼ (< IE *√peku̯
-ti̯a- (?); etymology disputed, cf. Lith. kėptas ‘cookedʼ, L.
coctor ‘cookʼ; cf. Pokorny IEW 798; LIV2
468; NIL 548–552; Martirosyan 2010:
396–397, especially for further literature);129
Ku̯
+ s = Arm. *0š:
not attested
Note: Mann (1963: 176) proposes IE *Ku̯
s > Arm. x (as he does for *sku̯
), but the examples he gives are both few
and not persuasive (xotor ‘awry, obliqueʼ; xul ‘deafʼ)
5.2.4 The development of the cluster palatovelar + t/s
The development of IE cluster *Ḱt is relatively securely attested:
Ḱ + t = Arm. st:
erastankʿ ‘buttocksʼ (< IE *prHḱto- an ablaut variant of *prōḱtós, cf. Gr. πρωκτός, ‘anusʼ;
cf. Hübschmann 1897: 443; Bugge 1889: 12–13; Pokorny IEW: 846; Winter 1962a:
256; Olsen 1999: 320; Schmitt 2007: 57; Martirosyan 2010: 258; Kim 2016: 152;
Macak 2018: 1019);
dustr ‘daughterʼ (PArm. *dust(i)r; < IE *dh
uḱtḗr < -ǵh
t-/-ǵHt-; cf. Lith. duktė̃, Goth.
daúhtar; cf. Hübschmann 1897: 440; Mann 1963: 75; Pokorny IEW: 277; Godel 1975:
80; Olsen 1999: 148; Beekes 2003: 173; Schmitt 2007: 61; NIL 126–130; Martirosyan
2010: 244–245; Kim 2016: 152; Macak 2018: 1019);
129
This etymology is often disputed; cf. Olsen 1999: 83, 827, who relates to parallel to ti̯ > Arm. cʿ.
131
hast ‘firm, hard, solidʼ (< IE *pH2ḱt-; cf. Gr. πηκτός ‘stuck inʼ, L. pāctus ‘fixedʼ, OHG
festi, fasti ‘firm, steadyʼ; cf. Hübschmann 1897: 464; Mann 1963: 75; Olsen 1999:
201; Martirosyan 2010: 390–391);130
Clusters of Ḱs are also realized either as Arm. cʿ or čʿ:
Ḱ + s = Arm. cʿ:
vecʿ ‘sixʼ (< IE *su̯ eḱs-; cf. Gr. ἕξ, Lat. sex, W. chwech; cf. Hübschmann 1897: 495;
Meillet 1903: 19; Meillet 1936: 40; Mann 1963: 101, 155; Pokorny IEW: 1044; Winter
1992a: 349–350; Kortlandt 1994b: 254; Blažek 1999: 236; Beekes 2003: 201; Schmitt
2007: 74; Martirosyan 2010: 594; Macak 2018: 1019);
aycʿ ‘goatʼ (< IE *aiǵ-s; cf. Gr. αἴξ; cf. Hübschmann 1897: 417; Olsen 1999: 816–817;
Martirosyan 2010: 58);
Ḱ + s = Arm. čʿ:
čʿir, čʿor ‘dried fruitʼ is reconstructed from IE *ḱsēro-; this etymology is often considered doubtful (cf. Gr.
ξερόν ‘dry landʼ, OHG serawēn ‘become dryʼ; cf. Hübschmann 1897: 485; Pokorny IEW: 625;
Džaukjan 1967: 257; Kortlandt 1995: 15; Martirosyan 2010: 546);
ḱst = Arm. št: veštasan ‘sixteenʼ (< IE *sweḱs-deḱm- (a result of the ruki-law?131
); cf. Meillet 1903: 19; Meillet
1936: 40; Winter 1992a: 350; Blažek 1999: 236; Beekes 2003: 201; Martirosyan 2010: 709);
5.2.5 The development of the cluster dental + t/s
Persuasive and secure examples of this development (otherwise a common process in all IndoEuropean
languages) are hard to find since they are limited to a singly commonly accepted
example (we add another example, though of a limited acceptance, but within the limits of the
first one):
T + t = Arm. u̯ t:
giwt ‘findʼ (cf. Arm. gitem; < IE *u̯ id-ti-m < √u̯ e d-; cf. OAv. vīnastī ‘findʼ, L. uīdī, uīsum
‘seeʼ; cf. Hübschmann 1897: 435; Pokorny IEW: 1125; Winter 1962a: 261; Schmidt
1980: 43; Peters 1997; Görtzen 1998: 337, 344–345; Olsen 1999: 851; LIV2:
665–667;
Schmitt 2007: 52, 134; NIL: 717–722; Martirosyan 2010: 211, 723);
hat ‘grain, seed, pieceʼ (< PArm. *hawt-i- < IE *H2ed-ti-; cf. L. ador ‘coarse grain, speltʼ,
Goth. atisk ‘cornfieldʼ; cf. Pokorny IEW: 3; Martirosyan 2010: 392–393, 723132
);
Note: Another example brought by Martirosyan (2010: 451–452, 723–724) is Arm. √mat- ‘approach, come closeʼ,
mawt ‘near, closeʼ (Martirosyan does not give any details of the development, relating this to ON mót
‘meetingʼ). Klingenschmitt (1982: 70–71) explains these forms from *mədu-.
Nawtʿi ‘hungry, fastingʼ is often quoted as another example of this process, if from *n̥-H1d-tii̯o(cf.
Gr. νῆστις ‘not eating, fastingʼ; Klingenschmitt 1982: 501; Martirosyan 2010: 501), but in
contrast to giwt, there is an aspirated plosive, hence others are sceptical (cf. Olsen 1999: 437),
130
But cf. LIV2
: 536–537, where the root is related to IE √*ses-, NIL (637–660) relates to *√steH2-.
131
Cf. analogy in Arm. harcʿ “question” < IE *pṛḱsḱā (OIA pṛcchā; Schmitt 2007: 71).
132
See especially the other possible etymologies listed there.
132
though it is semantically plausible. Another etymology for nawtʿi is related to Gr. νῆψις
‘sobrietyʼ (< IE *nagu̯h
-tii̯o-; Bugge 1889: 22; Pokorny IEW: 754; NIL: 208–220), which fits
better phonemically, but worse semantically – we leave the question open. Similarly, małtʿ
‘prayerʼ is often related to Lith. maldýti ‘imploreʼ (Bugge 1889: 15) or OCS moliti ‘ask, prayʼ
etc., is derived, according to Martirosyan (2010: 445–446) from IE *ml̥dh
-ti-. Again the
outcome does not accord with the expected one.
T + s = Arm. cʿ:
kʿacʿax ‘vinegarʼ (< IE *ku̯ atH2-so-; cf. OCS kvasъ; cf. Pokorny IEW 627–628; Džaukjan
1967: 229; Olsen 1999: 949; LIV2
: 384; Martirosyan 2010: 659–660)
5.2.6 The development of the cluster sibilant + t/s
The outcome of an IE cluster *St is a simple st, the outcome of IE clusters of *Ss is 0s:
S + t = Arm. st:
sterj ‘sterileʼ (< *steri̯ a- < IE *stér-iH2- ~ str̥ -yéH2; cf. Gr. στεῖρα ‘barren cowʼ, L. sterilis
‘unfruitfulʼ; cf. Meillet 1903: 18; Meillet 1936: 39; Winter 1955: 6–7; Winter 1962a:
256; Pokorny IEW: 1031; Godel 1975: 80; Beekes 2003: 169, 198; Kim 2016: 152;
Macak 2018: 1019);133
stêp ‘quickly, oftenʼ, stipem ‘urge, compelʼ (< *stei̯ b-; cf. Gr. στείβω ‘tread, stamp onʼ;
cf. Pokorny IEW: 1015; LIV2
: 592; Beekes 2003: 167, 198; Kim 2016: 152);
astł ‘starʼ (< IE *Hsstḗr; cf. Gr. ἄστρον, L. stēlla ‘starʼ; cf. Hübschmann 1897: 421;
Pokorny IEW: 1027–1028; Godel 1975: 80; Olsen 1999: 159–161, 843; Beekes 2003:
169, 198; NIL 348–354; Martirosyan 2010: 120–122; Kim 2016: 152; Macak 2018:
1019)
zgest ‘cloth(es), dressʼ (< IE *u̯ es-ti; cf. Lat. vestis, Goth. wasti ‘garmentʼ; cf.
Hübschmann 1897: 446; Pokorny IEW: 1172–1173; Godel 1975: 80; Görtzen 1998:
337; Beekes 2003: 169, 198; Martirosyan 2010: 274; Kim 2016: 152);
Note: Bugge (1893: 43–46) gives possible examples of the development *st > Arm. c.
S + s = Arm. 0s:
es ‘thou artʼ (< IE *H1es-si; cf. OIA ási, L. es ; cf. Hübschmann 1897: 442; Meillet 1903:
18; Meillet 1936: 39; Godel 1975: 40–41, 72, 112, 116–117, 124; Schmitt 2007: 65, 139;
Olsen 1999: 159–10, 44; Martirosyan 2010: 255; Macak 2018: 1057). But Klingenschmitt
(1982: 278) considers a secondary analogical building, re-archaizing the older simple
sibilant, cf. OIA asi as realization of as-si etc.). Similarly loc. Pl. -s < *-s-su (under the
assumption of an analogical reconstitution in the s-stems; Macak 2018: 1057);
S + s = Arm. cʿ is assumed by Klingenschmitt (1982: 278) as a regular outcome. This opinion is based on a sigmatic
aorist, to this cf. Kortlandt 1995; Kortland 1996.
5.2.7 The overview of the Armenian development
133
But Mann (1963: 101) gives another etymology; he considers the initial s- being from IE *eḱs-!
133
The Armenian outcomes are given in the reconstructed forms of spirants (not aspirates, since
we assume that spirants, not aspirates, were regular outcomes of the Proto-Armenian shift),
given here in the form of archiphonemes (marked by capitals):
IE Armenian t- s-
-k/g/gh
-X utʿ š
-kṷ
/gṷ
/gṷh
-X (utʿ š)
-ḱ/ǵ/ǵh
-Ç st cʿ/čʿ
-t/d/dh
-Θ u̯t cʿ
-p/b/bh
-P utʿ 0s/pʿ
-s134
-S st 0s
5.3 Trajectories of the development
There are at least two remarkable features: the almost universal loss of plosives before a
consonant (cf. Kortlandt 1980: 29; Beekes 2003: 204; Schmitt 2007: 56–65; Martirosyan 2010:
723; Kim 2016: 154), the exception being palatovelars before t-, and the lack of a fricative
outcome for cluster Tt (otherwise almost universal in Indo-European languages).
5.3.1 Development of the clusters labial + t/s
In the development of the IE cluster labial + t we assume, as for clusters of (labio)velar + t,
the spirantization of both plosives in the first phase, followed by the debuccalization of the first
segment and later by the elision (and with the shift of the spirant of the aspirated plosive) (cf.
Winter 1955: 549–553; Winter 1962: 554–562; Kümmel 2007: 371):
P + t > φϑ > hϑ > u̯ tʿ
Note: For all word-initial clusters with two plosives, we assume the loss of the debuccalized approximant.
Similarly, the development of clusters of *Ps are parallel to clusters of Ks. We assume the
spirantization of the plosive, later either a debuccalization (or sibilantization), followed by the
simplification:
P + s > φs > hs > 0s / P + s > φs > ss > 0s
5.3.2 Development of the clusters velar + t/s
The cluster velar + t (including labiovelars) was, similarly to clusters of labials + t, first
spirantized in both plosives, later the first was debuccalized, and elided and the right obstruent
134
Including *š according to Pedersen’s Law.
134
became a voiceless aspirated plosive (cf. Winter 1955: 549–553; Winter 1962: 554–562;
Kümmel 2007: 371):
K(ṷ)
+ t > kt > xϑ > hϑ > u̯ tʿ ʿ
To model a trajectory for the development of clusters of *K(ṷ)
is harder, especially since there
is a single example we have to depend on. We assume that the plosive was first spirantized,
later debuccalized and then elided, and that the sibilant was afflicted by the ruki-law (a variant
could be sibilantization instead of debuccalization):
K(ṷ)
+ s > xš > hš > 0š / K(ṷ)
+ s > xš > šš > 0š
5.3.3 Development of the clusters labiovelar + t/s
A remarkable feature of Armenian (shared with Albanian) is that it partially distinguishes the
original plain velars and labiovelars, since IE labiovelars *ku̯
, gu̯
, gu̯h
realize as Arm. čʿ, č, j/ž
before *e, *i, but have the same outputs as IE plain velars in other positions (i.e. as kʿ, k, g) (cf.
Stempel 1994; Job 1995; Beekes (2003: 177–179); Schmitt (2007: 62–65, 78–79); Martirosyan
(2010: 711).
Unfortunately, we have no secure data for the development of clusters of *Ku̯
t, but since this
position is not the one in which the old distinction between plain and labiovelars could be
preserved, we can assume it was the same development as with the plain velars (see above).
Similarly, we have no secure data for the development of the IE clusters of *Ku̯
s into Armenian.
Note: Beekes (2003: 201) assumes the outcome cʿ, based on the analogue with clusters of Ḱs and sk(u̯)
but gives no
examples to support his model.
5.3.4 Development of the clusters palatovelar + t/s
The development of clusters of palatovelar + t followed a similar trajectory as in other satəmlanguages
since their outcome is a sibilant + t. A remarkable feature, shared with the
development of dental series and contrary to the development of the (labio)velar and labial
series, is the lack of aspiration of the outcome.
The palatovelars are usually expected to be later palatal affricates (ć) in the later
development of satəm-languages. Accepting that such an affricate was present in the position
before t- would lead us to the trajectory: *Ḱt > ćt > śt > st, assuming the loss of the plosive
segment of the affricate and later depalatalization.
135
Here a problem arises within the affricativization trajectory: if this trajectory is applied
to the development of both of the central clusters, even the output for both clusters should be
similar in nature, since the affricates in both clusters have to undergo the essentially the same
developments (the assumed loss of the plosive segment of the palatovelar cluster has no
counterpart with the development of the dental cluster).
We assume as more probable the spirantization of the palatovelar before t-, later sibilantized
and depalatalized. However, we assume that the aspiration of the outcome was blocked by the
existence of the cluster fricative (either a sibilant or a spirant) + t, as it was blocked in the cases
of clusters of st and Tt:
Ḱ + t > çt > śt > st
Note: Beekes (2003: 201) assumes for cluster Ḱst the development: ḱst > ḱšt > ćšt > št.
Again, the development of a cluster with a sibilant is worse attested, though usually the
development *Ḱs > cʿ is accepted (cf. Olsen 1999: 965), though Godel (1975: 81) assumes a
neutralization of *Ḱs on *Ks. This development is not in accord with the attested development
of Ks, which results in 0š (see above).
A remarkable feature is that the outcome is the same as with the development of clusters
of *Ts, both resulting in cʿ. The trajectory is, again, hard to establish, but we propose
spirantization, followed by a transition to dental clusters and later affricatization and aspiration:
Ḱ + s > çs > ϑs > cʿ
5.3.5 Development of the clusters dental + t/s
What can be taken for granted is that IE *Tt is realized in Armenian without the aspiration of
the right plosive, similarly to the outcome of IE *Ḱt and *st and unlike to outcomes of *Kt and
*Pt and that the development, unlike in all other IE languages, is not a cluster with a sibilant.
Even the earlier sibilantization of the dental plosive is impossible, considering that the cluster
st is preserved. The lack of aspiration of the right plosive excludes any chance of an earlier
preservation of the left dental (‘archaicʼ or ‘re-archaizedʼ *Tt).
The traditional affricativization model is probably impossible: it would have the
trajectory: *Tt > ts
t > tt > 0t, which is at odds with the assumed trajectory of the other central
series (the palatovelars, cf. above) and it does not explain the frequent approximant before the
suffixed t-, the lack of aspiration excludes the existence of the second plosive in a cluster at all.
136
Görtzen (1998: 342–343) assumes re-archaization to tt (as in Old Indo-Aryan) and later
simplification of a geminate (in contrast to the development of non-geminate: *t > Arm. tʿ), but
there is no reason why assumed central geminates (without aspiration) should behave
differently from the peripheral geminates (with the aspiration).
Martirosyan (2010: 723) and Kim (2016: 154) assume *t > Arm. u̯ after vowels as a
universal process (cf. also Kümmel 2007: 371), hence this is independent from any Common
IE process otherwise attested in the whole IE area.
For these reasons we prefer, as we do with the development of the palatovelars, the
spirantization/lenition trajectory.
Winter (1962a: 261) assumes the trajectory: *Tt > tϑ > ϑt > u̯ t135
, i.e. with a
spirantization of the right plosive, with later metathesis and lenition.
Our model is based, similarly to that of Winter, on assumed spirantization, but of the
left dental plosive, its later debuccalization and replacement of an approximant by a labial one.
In Armenian the cluster fricative + t is never subjected to aspiration (as we can see in the
development of clusters of *st and *Ḱt); the u̯ is a regular continuation of a plosive, not an
inserted vowel (cf. Görtzen 1998: 346), thus:
T + t > ϑt > ht > u̯ t
The development of clusters of *Ts are harder to establish: we propose the spirantization of the
dental plosive, and that later the whole cluster became affricate and was aspirated (in other
words: the two fricative clusters became an aspirated affricate within the same process which
turned all reconstructed spirants into the affricates):
T + s > ϑs (?) > cʿ
Note: Martirosyan (2010: 719–720) documents a development of cC > sC. It is hard to establish how old this
process is, but if IE *Tt had an outcome in PArm. *ct (according to the affrication trajectory), the outcome in
Armenian would be st, which is not attested; hence the sibilantization of an affricate never affected clusters
originating from IE Tt, not being clusters of affricate + plosive. Olsen (2017: 431) assumes plosive + plosive>
0tʽ. On the other hand, Martirosyan (2010: 724) mentions the alternation -c ~ -wt (arac- ‘browse, grazeʼ ~ arawt
‘pasturelandʼ and two more highly questionable attested).
5.3.6 Development of the clusters of sibilant + t/s
A remarkable feature of the Armenian development is that though IE *t usually gives Arm. tʿ
(but neither in a word-initial before a consonant nor before/after a resonant or between vowels)
135
Since the only example Winter could depend on is from dt, he uses *dt; we generalize it here.
137
it is realized as t in the cluster st, arising either from IE *st or from IE clusters (*Ḱt and *Tt)
(Beekes 2003: 173).
The IE cluster *rs has a unique development, resulting in Armenian *r̄:136
Arm. or̄ ‘back
(body part)ʼ < IE *H1orsos (Hitt. arras, Gr. ὄρρος, OHG ars; Meillet 1903: 19; Meillet 1936:
40; Beekes 2003: 196); Arm. tʿar̄amim, tʿaršamim ‘witherʼ (OIA tṛ́ṣyati, Gr. τέρσομαι, L.
torreō, Goth. gaÞaursan; Meillet 1903: 19; Meillet 1936: 40; Beekes 2003: 196; Schmitt 2007:
72; Macak 2018: 1057). How much this process is related to the ruki-rule is questionable.
Martirosyan (2013: 89) takes tʿar̄amim vs tʿaršamim for a proof of the validity of Pedersen’s
Law in Pre-Armenian (especially for the alternation r̄ ~ rs, but cf. already Meillet 1903: 19;
Meillet 1936: 40). Martirosyan (2010: 709–710) asserts that the ruki-rule was also applied to
clusters of *r/k + s, following Meillet (1903: 19; 1936: 40) again, though Godel (1975: 77)
limits the ruki-rule to cluster *rs. Macak assumes the ruki-rule after *r/k/ḱ 2018: 1057–1058).
S + t > st
As far as we can judge from poorly attested examples, the trajectory of the development of
two-sibilant clusters is straightforward: the geminate is simplified:
S + s > 0s
Klingenschmitt (1982: 287) assumes the trajectory Ss > cʿ (i.e. the dissimilation of the left
fricative, similar to OIA a-vās-sam > avātsam, followed by an aspiration). This marker was
later used as a marker of the aorist.
If we accept the existence of PArm. š resulting from Pedersen’s Law, the possible (and purely
speculative) trajectory would be similar to that of st/ss-clusters and the reader could simply
derive them.
5.4 Conclusion and final remarks
The modelling of possible trajectories of the development of the Indo-European clusters of our
interest into Armenian faces many complications given by the complex changes affecting
Armenian phonology in general, hence it will necessarily be very sketchy and with many
undisclosed variables inside the ‘black boxʼ.
136
It should be noted that the same process affected the cluster *sr: Arm. kʿer̄, gen. sg. to kʿoyr “sister” < IE
*su̯ esrós; jer̄n “hand” < IE *ǵh
ésrm̥ etc., cf. Schmitt 2007: 72).
138
We can postulate the following main points of the development of clusters of IE plosive
+ t into Armenian:
The oldest development is that of the dental series, as it is present in the all other
branches of IE languages (with a virtual exception of OIA), resulting either in st (Iranian, BaltoSlavic,
etc.) or ss (Italic, Germanic, Celtic). Neither of outcomes fits for Armenian (as much as
we can depend on poorly attested examples) since if the outcome were at some point of
development *st, such a cluster would be preserved as clusters of IE *st are, attested as Arm. st
(but cf. *IE Tt > Arm. u̯ t); similarly, the outcome ss (including that from IE *Ts) would merge
with the outcome of IE *ss, which is not valid, since IE *ss > Arm. 0s but IE *Tt > Arm. u̯ t.
We can securely conclude that there was no sibilant as an intermediate stage in the
Armenian development of the IE cluster *Tt (cf. the development of IE cluster Ḱt, which results
in a sibilant + plosive). Another remarkable feature of the Armenian development of the central
series is that the right plosive is not aspirated, unlike all clusters of the peripheral plosive + t.
The proposed trajectory assumes that all clusters of fricative + t (of any origin) were not
subjected to the shift voiceless plosive > (voiceless spirant >) voiceless aspirate (in the
intermediate fricative stage cf. Pisani 1951: 68–71; Winter 1954: 200; Winter 1955: 7;
Kortlandt 1980: 28; Kim 2016: 157–159). However the spirant was later debuccalized.
The second oldest development is that of palatovelar clusters, as in other satəmlanguages.
As in the case of IE clusters of *Tt and *st, the left plosive was not (first spirantized
and later) aspirated, and the original palatovelar was, similarly to other satəm-languages,
sibilantized. The model trajectory requires a spirant intermediate stage and assumes that all
clusters of spirant + t were not subjected to the shift of IE voiceless plosives to (spirants and
later) aspirates.
The (labio)velars and labials clusters were not spirantized at the moment of the shift;
hence the right t was (spirantized and later) aspirated to Arm tʿ. However, the loss of the right
plosive could be attributed to spirantization, probably happening at the same time as that of the
right plosive (Kt > xϑ, Pt > φϑ). The left spirants were later debuccalized to u̯ in word-internal
clusters; the first spirant was elided in word-initials: #x/ϑ/φC- > #0C-, and here we can assume
an intermediate stage with a debuccalization. The process is similar to debuccalization of *s
before resonants (cf. Arm. now ‘daughter in lawʼ < IE *snusós; Arm. kʿoyr ‘sisterʼ < IE *su̯ ésōr;
Arm. gen.-dat. sg. hawr ‘fatherʼ < IE *pətrós). The loss of a plosive before any consonant is a
standard and universal development in Armenian (cf. Kortlandt 1980: 29; Schmitt 2007: 56–
65; Martirosyan 2010: 723; Kim 2016: 154).
139
The trajectories of clusters of IE plosive + s are even harder to reconstruct. We can surely
presume that it was parallel to the development of clusters of plosive + t; however, we should
keep in mind that data are usually even worse attested.
The oldest stratum was the development of the cluster *Ts (as with Tt), resulting in cʿ,
merging it with that of the IE cluster *Ḱs. We assume in both cases the spirantization of plosives
and later the merging of both clusters (via ϑs?) and their later affrication and aspiration at once.
The spirantization of both peripheral series was a later process, but in this case, both
spirants were later elided, either through debuccalization or through sibilantization and later
simplification.
The IE cluster st is fully preserved (as are all clusters of s + plosive), the IE cluster ss is
simplified due to elision.
141
6 The development of the two-obstruent clusters from Indo-European into
Albanian
6.0 Albanian language
Albanian is a satəm-language, forming an independent branch within the Indo-European family, related usually to
Illyrian or Thracian/Dacian languages (cf. Pedersen 1900b; for an overview of the debate see Jokl 1963;
Ködderitzsch 1991; Matzinger 2009; Matzinger 2012). The relationship of Albanian to Armenian shows traces if
not of a common descent (‘Proto-Albano-Armenianʼ), then at least of strong parallelism in their development (cf.
Kortlandt 1980b; Kortlandt 1986). How much this assumed ‘closenessʼ of both IE branches is present in the
development of the clusters of obstruent + *t/s will be demonstrated below.
6.1 Albanian and Indo-European
The development of Albanian from Indo-European is a trajectory with stages: Indo-European
> Pre-Proto-Albanian (PPAlb.; before contact with Latin) > Proto-Albanian (PAlb.; affected by
contact with Latin and Early Slavic) > Old Albanian (OAlb.; after the Tosk–Geg split) >
Modern Albanian (Alb.; since 19th
century) (cf. Hock 2005137
; Rusakov 2017: 539–560;
Schumacher 2006: 23, 85–86; de Vaan 2018: 1732–1733).
The typical features separating the Albanian obstruent system from that of IndoEuropean
are:
i. merging of voiced and voiced aspirated plosives;
ii. merging of old labiovelars and plain velars138
and the existence of palatovelars;
iii. palatalization of velars;
iv. a probable split of IE *s139
according to Pedersen’s Law (the ruki-rule).
The first feature is shared with Iranian, Balto-Slavic and Celtic, the second and fourth with all
the satəm-languages, the third has its reflexes in the Indo-Iranian and Slavic, but hardly could
be a result of a single process. More probably we face parallel processes on similar grounds;
the existence of Pedersen’s sibilant *š in Albanian is questionable, though the feature is
common within the satəm-area.
137
Hock uses the set of terms: Urindogermanisch, Vorlateinisches Voruralbanisch, Vorslavisches Voruralbanisch,
Uralbanisch, (Albanisch), since he points out the influence of various adstrates on the development of Albanian
phonology.
138
It seems that old plain velars and labiovelars merged everywhere except before front vowel – Albanian hence
uniquely preserving the IE triad palatovelar – plain velar – labiovelar, cf. Pedersen (1900a: 305–307); Jokl
(1937); Kortlandt (1980: 246).
139
Later, the IE *s underwent palatalization to *š as a default development, hence the effects of the ruki-rule, if
there were any at all, are indistinguishable (cf. Jokl 1963: 127; Kortlandt 1987; Demiraj 1997: 56; Kortlandt
1998; Schumacher 2013: 258–265, de Vaan 2018: 1746). However, Orel (2000: 61–62) assumes that the
regular development was IE *s > Alb. ɟ <gj> and assumes the ruki-rule after i, u (examples dash “ram” <
PPAlb. *dausa < IE *dh
ou̯so-; lesh “wool” < PPAlb. *laiša).
142
The development of given obstruents is far from being straightforward, and has to be
reconstructed with great carefulness and with a lot of dark spots and trajectories not always
unarguably established. Mann (1952); Kortlandt (1998); Demiraj (1997: 56–58, 61–97), Orël
(1998: xvii–xxii), Orël (2000: 60–101), Hock (2005); Schumacher (2006: 68–73, 77–79, 87–
92; 2013: 233–244, 258–264); Rusakov (2017: 569–570) and de Vaan (2018: 1745–1746) give
variously detailed overviews of the developments of obstruents, covering the main tendencies.
Note: The Albanian etymologies are often insecure, since many details of the etymologies and developments are
only partially described. Such questionable examples, without the wider support of other research, will be listed
in square brackets.
6.2 Albanian clusters and their IE origins
Since Albanian morphology is a result of deep structural changes (cf. the development of
Albanian morphology in Camaj 1966; the historical morphology of Albanian in Demiraj 1993;
or specifically of the Albanian verbal system in Schumacher/Matzinger 2013: 25–198), we
cannot use productive examples on the formation of the clusters of our interest and we are
forced to use exclusively the etymological data, with all disadvantages and difficulties such
source has.
Note: Since Albanian does not show any traces of the operability of the Bartholomae’s Law, all clusters formed
with a right voiceless obstruent are also voiceless in their respective output. More to that, the clusters are often
simplified in the development.
6.2.1 The clusters labial + t/s
A remarkable feature of the Albanian development is the labial plosive + t-, results in 0t (a
development shared with the development of the plain velar and labiovelar plosives). The
output of the IE cluster of *Ps is, as far we can judge, fš, though f (as we can see in examples)
was proposed as another output:
P + t = Alb. 0t:
shtatë ‘sevenʼ (< *š(ē)tátë < IE *septm̥ -tā-; cf. OIA saptá-, L. septem; cf. Meyer 1891:
415; Pokorny IEW: 909; Kortlandt 1988: 221; Hamp 1992: 914; Demiraj 1997: 370;
Orël 1998: 436; Blažek 1999: 248; Schumacher 2013: 56);
P + s = Alb. fš:
fshij ‘wipe, cleanʼ < *bh
si(H)-i̯ o- (< IE *bh
si(H)- o-; cf. OIA psā́ ti ‘devourʼ, Gr. ψάω ‘rub
smoothʼ, OHG bes(a)mo ‘brush, wipeʼ; cf. Mann 1952: 40; Pokorny IEW: 145–146;
Demiraj 1997: 66, 173; LIV2
: 82)140
;
140
But Orel (2000: 104, 414) considers it as a borrowing (Lat. exigere) + a labial prefix (on such prefixes see Mann
1952: 40)
143
P + s = Alb. 0f (?):
[afër ‘nearʼ < PPAlb. *apsera (contamination of *aps-, a variant of IE *apo-, cf. Orël 1998: 1–2; Orël 2000:
9), but this etymology is not widely accepted, for other etymologies cf. Meyer (1891: 3); Jokl (1923:
271); Demiraj 1997: 70–71);]
6.2.2 The clusters plain velar and labiovelar + t/s
The ancient IE clusters of *Kt and *Ku̯
t are both realized in Albanian as 0t, i.e., the velar plosive
is totally elided. This process has an exact analogy with the development of the labial plosives
(and in some sense even with dental clusters). Similarly, the cluster of *Ks, which is rarely
attested, develops with the same output as the cluster of *Ku̯
s, both resulting in 0š:
K + t = Alb. 0t:
butë ‘soft, smoothʼ (< IE *bh
eu̯ gh
-to- (?); cf. OHG biugan ‘bendʼ ; cf. Pedersen 1900b:
341; Pokorny IEW: 152–153; Demiraj 1997: 114; Orël 1998: 43; Orël 2000: 101;
LIV2
: 84–85);
fletë ‘wing, leafʼ(< PPAlb. *awa-lekta- < IE *√lek-; cf. Lith. lekiù ‘flyʼ; cf. Pokorny IEW:
673; Orël 1998: 100; Orël 2000: 101; LIV2
: 411)141
;
gjatë, glatë ‘longʼ < PPAlb. *dlata- < IE *dln̥ gh
-to-; cf. L. longus ‘longʼ, Goth. lag
‘longʼ,OCS dlъgъ ‘longʼ; cf. Pedersen 1900a: 308; Pokorny IEW: 197; Demiraj 1997:
184–185; Orël 1998: 130; Orël 2000: 101)
Ku̯
+ t = Alb. 0t:
natë ‘nightʼ (< IE *noku̯
ti; cf. OIA naktam ‘at nightʼ, Goth. nahts ‘nightʼ; cf. Meyer 1891:
298; Mann 1952: 35; Pokorny IEW: 762–763; Hamp 1961c; Demiraj 1997: 283–284;
Orël 1998: 282–283; Orël 2000: 101; LIV2
: 449; NIL: 513–515; Schumacher 2013:
243);
Note: The possible output could also be 0s in pesë ‘fiveʼ if derived from *penku̯
-ti (Meyer 1891: 329), but this
could be directly from *penku̯
e-ās/om (cf. Pedersen 1900a: 307; Jokl 1937: 157–158; Orël 1998: 316). The
‘mixedʼ form of both variants is another variant, proposed by (Huld 1984: 102–103; cf. Demiraj 1997: 315–
316; Blažek 1999: 221). We prefer that pesë is a direct result of an antevocalic form of this numeral.
K + s = Alb. 0š:
[shesh ‘plain, plane, flatness, squareʼ < IE *ksesi̯ o- (Mann 1952: 40)];142
Ku̯
+ s = Alb. 0š:
shoh ‘see, showʼ either (< PPAlb. *(V)kśē-(sḱ)- < IE *H3(e)ku̯
-s-; cf. OIA ī́kṣate ‘seeʼ, Gr.
ὄσσομαι ‘see (in spirit)ʼ; cf. Pokorny IEW: 775–777; Demiraj 1997: 57; LIV2
: 297)
or (< PPAlb. *sāskska, < IE *√seku̯
-; OIA sácate ‘followʼ, Gr. ἕπομαι ‘followʼ, L.
sequor ‘followʼ, Goth. saiƕan ‘seeʼ; cf. Meyer 1892: 411–412; Pokorny IEW: 896–
897; Orël 1998: 100, 148, 184; Orël 2000: 425–426143
; LIV2
: 552); 144
[shore ‘rash of the skin, eruptionʼ < IE *ku̯
sēros (Mann 1952: 40)];
141
But Meyer (1891: 108) considers it to be from It. foglietta.
142
But Orel (1998: 412) considers it a borrowing from L. sessus “seat”, following in this Meyer (1891: 402).
143
Orel reconstructs PPAlb. *sāksa- with a dissimilation of sibilants, otherwise to the same root as Demiraj (l.c.)
144
However, both reconstructions lead towards the cluster of a labiovelar and s-.
144
[shale ‘pair of trousers, saddleʼ < IE *ku̯
salis (Mann 1952: 40)145
];
6.2.3 The clusters palatovelar + t/s
IE clusters of Ḱt and Ḱs are realized in Albanian as 0t and 0š, similarly to all velar clusters,
unlike prevailing tendencies in the satəm-languages, where the outputs of palatovelars are
usually different from the outputs of the plain- and labiovelars:
Ḱ + t = Alb. 0t:
tetë ‘eightʼ (< *oḱtō+tā; < IE *oḱtō; cf. Gr. ὀκτώ, Goth. ahtau ‘eightʼ; cf. Meyer 1891:
428; Mann 1952: 34; Pokorny IEW: 775; Hamp 1992: 915–916; Demiraj 1997: 385;
Blažek 1999: 266; Orël 1998: 453; Orël 2000: 64, 101);
dritë f. pl. ‘light, lustre, pupil of an eyeʼ (< *driktā; < IE *derḱ-to- ; OIA darśam ‘seeL,
Gr. δέρκομαι ‘seeʼ; cf. Meyer 1891: 74; Pokorny IEW: 213; Demiraj 1997: 145; Orël
1998: 75; Orël 2000: 101; LIV2
: 122–123; Schumacher 2013: 243);
Ḱ + s = Alb. 0š:
the intensive prefix sh- (< IE *H1eǵh
s-; cf. L. ex- ‘out, fromʼ, OIr. ess- ‘outʼ, Gr. ἐξ ‘fromʼ,
OCS iz ‘outʼ; cf. Pokorny IEW: 292–293; Demiraj 1997: 57);
Note: Schumacher (2018: 238) assumes the development of IE Ḱs > Alb. ϑ, without a specification of a trajectory,
but this model also is based just on a single example (IE *deḱsaH2 > Alb. djathlë).
Note: IE cluster *Ḱst, has a different output realized as Alb. 0ϑ:
Ḱst = Alb. 0ϑ:
gjashtë ‘sixʼ (< PPAlb. *šeštā < IE *seḱs-ti-; cf. OIA ṣaṣṭhá- ‘sixthʼ, L. sextus, sestus ‘sixthʼʼ; cf. Meyer
1891: 138; Pokorny IEW: 1044; Kortlandt 1988: 221; Hamp 1992: 913; Demiraj 1997: 184; Orël 1998:
130; Blažek 1999: 236; Orël 2000: 101, 248; Schumacher 2013: 238);
djáthtë, djáthë ‘rightʼ (< PPAlb. *detsa < IE *deḱs-t-; cf. L. dexter ‘rightʼ; cf. Meyer 1891: 69; Pokorny
IEW: 190–191; Demiraj 1997: 58, 137–138; Orël 1998: 67–68; Orël 2000: 64, 100; Schumacher 2013:
238);146
jashtë ‘outʼ (< PPAlb. *ekšta < IE *eǵh
s-to; cf. L. extrā; cf. Mann 1952: 40; Pokorny IEW: 292–293;
Demiraj 1997: 42; Orël 1998: 158);
6.2.4 The clusters dental + t/s
As far as we can judge from scarce data, the regular output of IE cluster *Tt is 0s in Albanian,
the output of IE cluster *Ts is Albanian 0š:
T + t = Alb. 0s:
besë ‘pledge, truce, trustʼ (cf. bind ‘conviceʼ; < PPAlb. *baitšā; < IE *bh
ei̯ dh
-to-147
; cf.
Gr. πείθω ‘persuadeʼ, L. fīdō ‘trustʼ; cf. Pedersen 1900a: 308; Mann 1952: 34; Pokorny
145
But Orel (1998: 407; 2000: 26) considers it a borrowing of L. sella “saddle, seat”, cf. also Meyer (1892: 398)
and Demiraj (1997: 118).
146
Kortlandt (1998: 36) reconstructs *deḱs-no-, the loss of -s- attributed to following -n-.
147
But Meyer (1891: 93) etymologizes from IE *bh
endh
-ti, similarly Pedersen (1900a: 308), which was rejected
by Hamp (1961b). However, the solution of the clusters is the same.
145
IEW: 117; Hamp 1961a; Demiraj 1997: 96–97; Orël 1998: 22; Orël 2000: 101; LIV2
:
71–72; NIL 12–13; Schumacher 2013: 244);
OGeg. pasë ‘haveʼ148
(< PPAlb. pat-ta- < IE *pot-tó-; cf. OIA pátyate ‘ruleʼ, L. potior
‘become masterʼ; cf. Pedersen 1900a: 308; Pokorny IEW: 842; de Vaan 2008: 484–
485; Schumacher 2013: 244);149
Note: Johannson (1903: 268 and 1906: 115) assumes IE *Tt > Alb. št: bisht ‘tailʼ < *bhid-t-; bushtër ‘female dogʼ
< *bhid-trī; gisht ‘fingerʼ < *gh
l̥ t-t- (cf. Jokl 1923: 261; Demiraj 1997: 103–104, 178; Orël 2000: 27, 43, 117–
118 for alternative etymologies).150
It should be noted that such an output is possible (since IE *st > Alb. št,
but valid only if IE *Tt is merged with *st, which is not the case).
T + s = Alb. 0š:
[láshë, lashtë ‘old, early, prematureʼ (either < PPAlb *ladśa; < IE *lH1u̯ ds-m̥ ; cf. Demiraj
1997: 57; or < PPAlb *lauša; < IE *leudh
-s-; cf. Orël 1998: 214–215);]
përposh ‘below, underneathʼ (< IE *-pēd-su; cf. Pedersen 1900a: 290; Jokl 1937: 32–33;
Demiraj 1997: 329–330; Orël 1998: 322, 340) ;
The IE cluster *dh
st is realized as Alb. 0ϑ, similarly to the cluster *Ḱst (see above):
dh
st = Alb. 0ϑ:
ethe ‘feverʼ (< IE *ai̯dhstis; cf. OIA óṣati ‘burn, Gr. εὕω ‘singeʼ, αἰθός ‘burntʼ; cf. Mann 1952: 40; Pokorny
IEW: 348; Demiraj 1997: 168–169; Orël 1998: 91; LIV2
: 245);
6.2.5 The clusters sibilant + t/s
All clusters of IE sibilant + t are realized regularly as št in all positions, including the clusters
possibly affected by Pedersen’s Law (the ruki-rule), therefore we can assume the merging of
IE s with *š (cf. Huld 1984: 147-148; Demiraj 1997: 56; Matzinger 2006: 77; Kümmel 2007:
372):
S + t = Alb. št:
shteg ‘path, roadʼ (< IE *√ste gh
-; cf. OIA stighnóti ‘climbʼ, Gr. στείχω ‘walkʼ; cf. Meyer
1891: 415; Pokorny IEW: 1017–1018; Kortlandt 1988: 221; Demiraj 1997: 371–372;
Orël 1998: 437; Orël 2000: 96; LIV2
: 593–594; NIL 660–661; Schumacher 2013: 260;
Rusakov 2017: 571);
shton ‘addʼ (< IE *st-né-H2-; cf. Arm. stanam ‘ariseʼ, OCS stanǫ ‘stepʼ; cf. Pokorny IEW:
1004–1006; Demiraj 1997: 378; Orël 1998: 440; LIV2
: 590–591; NIL 637–659;
Schumacher/Matzinger 2013: 998);
është ‘beʼ (< IE *es-t; cf. OIA ásti, L. est; cf. Pokorny IEW: 241; Demiraj 1997: 207–
208; Orël 2000: 156; LIV2
: 241–242; Schumacher/Matzinger 2013: 972–973);
asht ‘boneʼ (< IE *H2ost(i)-; cf. OIA ásthi ‘boneʼ, L. oss ‘boneʼ; cf. Meyer 1891: 19;
Mann 1952: 39; Pokorny IEW: 783; Kortlandt 1988: 221; Demiraj 1997: 82–83; Orël
1998: 11; Orël 2000: 96; Schumacher 2013: 260);
148
A suppletive participle of kā “have”, 3rd
sg. ao. (cf. Schumacher 2013: l.c.).
149
But Demiraj (1997: 313–314) is very sceptical about this explanation.
150
For simplicity, Johannsonʼs examples are present here in the modern Albanian orthography, not in the
orthography he actually used.
146
The output of IE cluster of two sibilants *Ss is realized by a single palatal sibilant, including
the possible clusters arising according to the Pedersen’s law:
S + s = Alb. 0š:
kush ‘whoʼ (< PPAlb. *kuśśa < IE *ku̯
ós só; cf. Pokorny IEW: 694–648; Schumacher
2013: 264)151
;
thóshe ‘sayʼ (< IE *ḱeH1s-si; cf. OAv. sīšā ‘showʼ; cf. Pokorny IEW: 533; Demiraj 1997:
57, 399–400; Orël 1998: 480; LIV2
: 318–319);
6.2.6 The overview of the Albanian development
The outputs of the Albanian development listed here are limited to the ‘secureʼ ones; the variant
outputs are entirely omitted here. The Albanian outputs of IE phonemes are listed here, for
simplicity as archiphonemes.
IE Albanian t- s-
-kṷ
/gṷ
/gṷh
-K152
0t 0š
-k/g/gh
-K 0t 0š
-ḱ/ǵ/ǵh
-Θ 0t 0š153
-t/d/dh
-T 0s 0š
-p/b/bh
-P 0t fš154
-s155
-š št 0š
6.3 Trajectories of the development
There are at least two remarkable features of the development of IE clusters of plosive + t:
i. all clusters of plosive + t (except those with dentals) are realized as 0t, even those with original
IE palatovelars, a feature unknown in any other satəm-languages;
ii. dental clusters of *Tt are realized as 0s – it is clear that this output is not from the intermediate
*st, since original IE *st has the Albanian output št, i.e., the Pre-Albanian never merged
outputs of the IE *Tt and *st, as Iranian, Baltic and Slavic. More than that, since the output
of the IE cluster of *ss is 0š in Albanian (at least if we can judge from scarce data), it is
impossible to assume a merging of the outputs of IE *Tt with those of *ss, attested in Italic
and Celtic (this process is also otherwise unknown in the satəm-languages).
On the contrary, the development of clusters of plosive + s is very simple: the Albanian output
is always 0š, except with IE *Pš (a single example has an output fš, but this could be a result of
151
But Orel (2000: 207) assumes this to be from *ku̯
u-so, and the reduced form of the first root is also assumed by
Demiraj (1997: 228).
152
The primary output is *K (a plain velar), the secondary is *S (due to palatalization before *e/i/i̯ ), cf. Schumacher
2013: 241. Since we deal with the original labiovelar in the positions before t/s only, the difference of both
outputs is irrelevant.
153
Schumacher (2013: 238) brings another output ϑ.
154
We will return below to the output f seen in afër “near” < PPAlb. *apsera brought by Orel (1998: 1–2; Orel
2000: 95).
155
Including *š according to Pedersen’s Law.
147
the development of clusters in a word-initial; we have no secure data for the development of
world-internal clusters of *Ps).
6.3.1 The development of the IE clusters of labial plosive + t/s
The clusters resulting from IE labial + t are realized according to the pattern: P + t > 0t,
similarly to all velar clusters (see below). The trajectory we propose assumes a spirantization156
of the labial, a debuccalization and later simplification (cf. Huld 1984: 142, 148; Orël 2000:
35–36; Kümmel 2007: 372):
P + t > pt > φt > ht > 0t
Note: Schumacher (2013) does not propose his own trajectory for the development of this cluster, but we can
model his trajectories for labials as follows, with a gemination inserted between a debuccalization stage and a
simplification stage: Pt > φt > xt > tt > 0t.
The ‘double outputʼ of the IE cluster *Ps either as Albanian fš or just f. In our opinion, the
‘regularʼ output is just fš-, at least in the word-initial (there are no secure data for the wordinternal
position). The trajectory we model (with a high degree of uncertainty) as:
#P + s > #φs > #fs > #fš157
(P + s > φs > hs > 0s > 0š) (?)
The output f, seen by Orël in afër ‘nearʼ (< PPAlb. *apsera; Orël 1998: 1–2; Orël 2000: 95)
should be probably rejected (for other etymologies cf. Meyer (1891: 3); Jokl (1923: 271);
Demiraj 1997: 70–71), being a result of a shift of *ps to PPAlb. *sp (if related to Orël’s
*apsera), since *sp was realized as 0f, at least in a world-initial (Schumacher 2013: 233, 260
brings word-initial examples: farë ‘seed, breedʼ < PPAlb. *ph
arā < IE *sporáH2; fier ‘fernʼ <
PPAlb. *ph
era < IE *(s)perHom; there are attested no secure examples on intervocalic *sp (cf.
Huld 1984: 149; Orël 2000: 95; Matzinger 2006: 78; Matzinger 2007: 78, 88; Kümmel 2007:
372).
The trajectory we dare to propose is based on metathesis, spirantization and
debuccalization and simplification (cf. Huld 1984: 142; Orël 2000: 36; Kümmel 2007: 372):
P + s > ps → sp > hf > 0f
156
Here, we prefer a bilabial spirant φ instead of a labiodental f.
157
Matzinger (2007: 78, 88) proposes the trajectory of the IE inverse cluster *sp as: sp > hφ > 0f, which could
serve as an analogy for the development of IE *Ps.
148
Note: For the development of the cluster of sp Schumacher (2013: 233) gives a slightly different trajectory
(exclusively for the word-initial cluster!); assuming aspiration before a spirantization of a plosive, we modify
it for cluster of ps by adding the metathesis stage: Ps > sp > hp > ph > ph
> 0f.
6.3.2 The development of the IE clusters of velar plosive + t/s
In the original IE clusters of velar + t, as in all clusters of plosive + t (except dental + t), the
left plosive is lost: *K + t > 0t.
The trajectory we dare to propose has the following stages: spirantization of a velar,
debuccalization of the spirant, simplification due to elision of h (cf. Huld 1984: 142, 148; Orël
2000: 35–36; Kümmel 2007: 372):
K + t > kt > xt > ht > 0t
Note: Schumacher (2013: 243) proposes a trajectory (for both plain and labiovelars): K(u̯
t > xt > tt > 0t, i.e, with a
gemination instead of debuccalization.
The trajectory of the cluster *Ks is similar in its main features; the sibilant is later palatalized
(cf. Huld 1984: 142; Orël 2000: 36; Kümmel 2007: 372):
K + s > ks > xs > hs > 0s > 0š
Note: If Pedersen’s rule were valid for Pre-Proto-Albanian, the trajectory would be: Ks > kš > hš > 0š.
6.3.3 The development of the IE clusters of labiovelar plosive + t/s
For the IE clusters of *Ku̯
t/Ku̯
s, we assume the loss of labial markers before an obstruent,
similarly to the development of Latin clusters with a labiovelar. For this reason, we can also
surely assume the full merging of clusters of *Ku̯
t with *Kt and of *Ku̯
s with *Ks. The
trajectories of development of clusters of *Ku̯
t and *Ku̯
s are hence essentially the same as for
clusters of *Kt and *Ks, described above:
K(u̯ )
+ t > kt > xt > ht > 0t
Note: For the trajectory proposed by Schumacher (2013: 243) see above.
A remarkable feature is that Pre-Albanian seems to preserve the old neutralization of a
palatovelar before a sibilant, otherwise known from OIA:
K(u̯ )
+ s > ks > xs > hs > 0s > 0š
Note: As said above, in the case of the validity of the ruki-rule for Proto-Albanian, the development would be: Ku̯
s
> kš > hš > 0š.
149
6.3.4 The development of the IE clusters of palatovelar plosive + t/s
As in preceding cases, the original IE palatovelar is lost before t-: *Ḱt + t > 0t. This output is
remarkable in the context of all satəm-languages since it indicates that Proto-Albanian had no
merging of the original IE cluster of *Ḱt with the cluster of sibilant + t since cluster of sibilant
+ t is realized in Albanian as št. Since we have no signs of a loss of any sibilant (original or
secondary) before t-, we have to accept that Proto-Albanian never had a sibilant from an original
palatovelar (or dental, as we will see below) before t-.
A palatal affricate as an intermediate stage is usually reconstructed for the trajectory
from IE palatovelars to sibilants in given satəm-languages. A similar affricate is assumed for
the development of the cluster of *Ḱt into Albanian by Schumacher (2013: 243); the trajectory
he reconstructs is: Ḱt > ćt > ct > tt > 0t, i.e, with affrication, depalatalization, gemination and
elision.
Our spirantization model is slightly different, assuming the spirantization of a
palatovelar before a plosive, its depalatalization, debuccalization and elision:
Ḱ + t > çt > xt > ht > 0t
Note: An alternative model, assuming the depalatalization of a palatovelar as the earlier phase (Ḱt > xt > ht > 0t),
is possible, but unknown (in contrast to a position before s-) from other satəm-languages.
The trajectory of the development of the cluster of palatovelar + s could be analogically
modelled as a sequence of a spirantization, depalatalization, debuccalization and simplification:
Ḱ + s > çs > xs > hs > 0š
Note: An alternative model, assuming the depalatalization of a palatovelar as the earlier phase (Ḱs > xs > hs > 0s
> 0š), is also possible, since it has a parallel in the development of Old Indo-Aryan, where the cluster of Ḱs is
preserved as kš.
Note: Again, as with (labio)velars, if Pre-Albanian was affected by ruki-rule, the development would be: Ḱs > kš
> xš >hš > 0š.
Note: Schumacher (2007: 79; 2013: 238) proposes a trajectory: Ḱts > st > št. for the cluster of Ḱst He even
assumes the development of Ḱs > Alb. ϑ, without a specification of a trajectory, but this model is based on a
single example (IE *deḱsaH2 > Alb. djathlë). Demiraj (1997: 58) limits this output to word-final only (cf. Orël
2000: 96).
6.3.5 The development of the IE clusters of dental plosive + t/s
The development of IE clusters of *Tt and *Ts are rightly considered the oldest layer of all the
processes concerning clusters of plosive + t/s since it is shared with all Indo-European
150
branches.158
It was Pedersen (1900a: 308; cf. also Jokl 1923: 261–262; Hamp 1961a: 252;
Ködderitzsch 1991) who proved that the output of the IE *Tt in Albanian is 0s.
Note: On the contrary, we have to reject Johannson’s opinion that IE *Tt > Alb. št, though phonetically plausible159
(Johannson 1903: 268 and 1906: 115; cf. esp. Jokl 1923: 261).
To model the trajectory we have to deal with many sub-questions, especially because of the fact
that the IE dental clusters of *Tt are realized as 0s, an output which cannot be a result of the
intermediate (Johannson’s model) *st, since original IE *st has the Albanian output št. We can
thus be sure that the Pre-Albanian never merged outputs of Tt and st (cf. Görtzen 1998: 372),
as Iranian, Baltic and Slavic did. What is more, since the output of the IE cluster *ss is 0š in
Albanian (at least if we can judge from scarce data), it is impossible to assume a merging of the
outputs of IE *Tt with *ss, attested in Italic and Celtic. Hence both outputs known from the
development of IE languages (outside of Indo-Aryan) are impossible as intermediate stages in
the Albanian development, which makes Albanian development unique.
Though both the input and the output are clear, the details of the trajectory stay foggy160
with many details unknown. We can be sure that the output 0s is later than the palatalization of
old IE *s to Alb. š, which could otherwise affect this sibilant, being older than the palatalization
process, hence the output is later than the palatalization.
Traditionally, the first step of the whole process of development of the cluster of Tt in
Indo-European is considered the affrication of the left dental plosive; this idea was brought up
first by Kräuter (1877: 88) and popularized by Brugmann (1880: 140–142 and used since). We
dare to propose the model, assuming as a second step the affrication of the whole cluster and
its later simplification and the merging of the outputs of the cluster of *ti̯ (which results in c;
Ködderitzsch 1991: 121; Schumacher 2013: 234–235). The very sketchy trajectory we dare to
propose is then:
T + t > ts
t > ts
ts
> 0ts
> 0s
Note: This trajectory does not seem very probably, if we accept that Alb. ethe ‘feverʼ is related to L. aestus ‘heatʼ
and derived from IE *a dh
-sti- and Alb. dríthë ‘feverʼ related to L. hordeum ‘barleyʼ < IE *ǵrīdh
-st- (Mann
1952: 40; Demiraj 1997:145–146, 168–169; Orël 1998: 75, 91; Görtzen 1998: 372–378). In this case *Ts
t would
give Alb. ϑ, not the otherwise attested 0s. The development could not be related to Bartholomae’s Law, since
in such case the voicedness would be preserved as it is in other Albanian developments. If given etymologies
are correct, they entirely exclude the affricate trajectory for the Albanian development.
158
Indo-Aryan being a false exception since re-archaized later (c.f. the chapter on Aryan development of the
present study.).
159
“Johannson’s trajectory” would be easier to model as the affrication/sibilantization/palatalization trajectory:
Tt > ts
t > st > št, similar to those of Baltic, Slavic and Iranian.
160
As has Schumacher (2013: 244) noted: “die Zwischenstufen bleiben unklar.”
151
If we try to sketch a spirantization/lenition trajectory within the similar lines as de Saussure
(1877); Cocchia (1883: 16–58) (both for Italic) and Bartholomae (1895: 16; for Iranian), we
replace propose a spirant stage instead of the affricate; if we accept this for the development of
Albanian, the trajectory could be modelled as a sequence of the subsequent spirantization of
both dentals, followed by the later dissimilation on an affricate, which finally lost its plosive
segment:
T + t > ϑt > ϑϑ > 0ts
> 0s
Both trajectories are within the wider fricativization trajectory. It should be noted that the
sibilant outcome is late, since it is not a subject of the palatalization of the sibilant. We can in
both cases reject the possibility that IE *Tt ever was realized as ss (as in Italic, Celtic and
Germanic), since IE *ss is realized as Albanian 0š (see below).
In stark contrast with the cluster of *Tt, the development of the cluster of *Ts seems to be
very simple, we propose the model trajectory, assuming the affrication of the dental plosive:
T + s > ts
s > ss > 0s > 0š
The alternative model, based on the spirantization in the first phase, is otherwise similar to the
preceding one, again, both trajectories are within the frame of the wide fricativization trajectory.
The fricativization model is simpler, since sibilantization of a fricative is a more probably
process than de-occlusivization of an affricate, hence more probable:
T + s > ϑs > ss > 0s > 0š
6.3.6 The development of the IE clusters of sibilant + t/s
Clusters of original IE sibilant + t are regularly realized as št. Since Alb. š is an output of the
all IE sibilants (cf. Kortlandt 1987; Demiraj 1997: 56; Kortlandt 1998; Orël 2000: 96;
Schumacher 2013: 258–265, de Vaan 2018: 1746), we are not able to resolve whether ProtoAlbanian
ever was subjected to Pedersen’s Law. The cases assumed to be originally affected
by the ruki-rule could be those following (according to Orël 2000: 62): dash ‘ramʼ < PPAlb.
*dausa < IE *dh
ou̯so-; lesh ‘woolʼ < PPAlb. *lai̯ša, etc. However, Orël is probably mistaken in
his premise that a regular development of IE *s with an Albanian output is a palatal affricate ɟ
152
(<gj>)161
, since this output is limited to the word-initial position before an accented vowel (cf.
Demiraj 1997: 56; Schumacher 2007: 77–79, 87–88, 92; Schumacher 2013: 258–260).
The trajectory of the IE clusters of *st is then straightforward and simple, with just a
palatalization of a sibilant (Matzinger 2006: 78; Matzinger 2007: 78):
S + t > st > št
The development of the IE cluster of *Ss could be reconstructed as a simple trajectory of
simplification and palatalization:
S + s > ss > 0s > 0š
Note: If the palatalization happened earlier than the gemination, the trajectory would be: Ss > šš > 0š.
6.4 Conclusion and final remarks
A remarkable feature of the development of the Indo-European clusters of plosive + t into
Albanian is that all such clusters, with the exception of the IE *Tt, have an output 0t. The
development of dental series hence had to be dealt with separately, on a different trajectory,
that those of other series. From this we can deduce that even the cluster of palatovelar + t had
a similar development as the peripheral series, unlike to the situation of the same series in other
satəm-series.
The development of the cluster of dental + t/s underwent a fricativization trajectory,
either within an affricativization or spirantization frame.
For the development of the IE cluster of *Ḱ+t/s, we can also model either an
affricativization or a fricativization trajectory, but the second seems to be more probable.
However, the most atypical feature is depalatalization of the original palatovelar, either older
of the plain velar plosive (which would be the unique feature in the development of the satəmlanguages)
or later of the spirant.
The development of the peripheral series follows the general lines of spirantization as
the first step, followed by the debuccalization and later elision of the former plosive (except of
the cluster of fš).
Note: An alternative possibility could be gemination, with the trajectory for velars: Kt >tt > 0t, for labials: Pt >tt
> 0t, for palatovelars: Ḱt >tt > 0t. Such a trajectory could be even possible for the dental series: Tt >ts
t > ts
ts
> ss > 0s.
161
This output merged with an output of #i̯- (Schumacher 2013: 258).
153
The development of the IE cluster of *st is trivial, since the only process affecting it is the
palatalization of the sibilant. The cluster of *ss was affected in its development, besides
palatalization, by the simplification of the cluster (technically degemination).
155
7 The development of the two-obstruent clusters from Indo-European into
Greek
7.0 Greek language
The Greek language is the oldest attested Indo-European language of Europe, Mycenaean already in the last
centuries of the second millennia BC, Homeric Greek from the second half of the first millennia BC, classical and
post-classical Greek from the first half of the same millennia, with Greek being used until the present day not only
in modern Greece and Cyprus but even inside a world-wide diaspora (for the short overview of Greek from IndoEuropean
point of view, cf. Thompson 2017: 287–291). The Greek language, due to the developments of Greek
culture/science/arts, affected all other European languages, especially Latin and through Latin the rest of the world,
being a source of terminology until today.
Our focus in on the development of Ancient Greek tongue, the Mycenaean data will be used, if at hand,
and especially if bringing different outcomes than the later language. The later Middle-Greek developments are
omitted, though offering interesting parallels to the development in other IE languages (cf. Horrocks 2017).
Note: Quoted Greek data are from Attic dialect; if not, the dialect will be mentioned.
7.1 Greek and Indo-European
The typical features separating the Greek obstruent system from that of Indo-European, related
to our field of interest, are:
i. preservation of three modal classes of plosives (the third modal class preserved as a voiceless
aspirate);
ii. preservation of old IE labiovelars in Mycenaean, but labiovelars are merged with other series in Classical
language (cf. Lejeune 1972: 34–37, 43–53; Bartoněk 2003: 137–139);
iii. the loss of the IE *s (preserved next to a plosive) in the post-Mycenaean period162
;
iv. Grassmann’s law of deaspiration of the left aspirate (Th
_Th
> T_Th
).
The first feature is shared with Italic (where the third class was spirantized), with Germanic and
Armenian (where the number of classes is preserved, but classes themselves were subjected to
the shift) and with Indo-Aryan (where there is a fourth class).
Labiovelars are almost fully preserved in Latin (but not in any other Italic languages)
and in Germanic.
The IE *s underwent some of the similar developments (especially *sV- > hV-) like in
Iranian; the loss of the intervocalic sibilant is reflected in the rhotacism in Germanic and Italic.
Grassmann’s Law is operational in Indo-Aryan (in the form: Dh
–Dh
> D–Dh
), though it
is hard to establish if both processes are accidentally parallels or reflecting an old inherited
pattern163
(cf. Lejeune 1972: 56–58; Sihler 1995: 142–144).
162
But at least intervocalic -s- was preserved in Mycenaean, cf. Myc. fut. do-so-si (= dōsonsi, cf. δίδωμι), Myc.
ao. e-re-u-te-ro-se (= eleuth
erōse,; cf. ἐλευθερόω); Bartoněk 2003: 114.
163
If the IE triad was *T – D – Dh
, the expected outcome of the Grassmannʼs Law in Greek should be: †D_Th
. For
this reason, we more readily assume the secondarily remodelling of the older mechanism both in Greek and
Indo-Iranian, especially since the process affects even Greek h arisen from IE *s, which hardly could already
be an Indo-European feature.
156
Note: The operationality of Bartholomae’s Law is doubtful, cf. Brugmann (1987: 658–659); Rix (1992: 32);
Görtzen 1998: 355–356.
7.2 Classical Greek clusters and their IE origins
For the Greek cluster the assimilation of the modal properties of the left plosive to phonemic
properties of the right obstruent is typical. In Greek, as in most IE languages (but not in IndoIranian)
there is no mechanism like Bartholomae’s Law (cf. Sihler 1995: 200), hence this
process is always dominated by the right obstruent.
Since Greek is well attested with numerous examples, we focused, again, on the
productive examples, especially those of verbal derivation/flexion. The purely etymological
examples are limited to well-known items with a wider Indo-European validity and background.
7.2.1 The clusters labial + t/dh
/s
Indo-European clusters of *Pt, *Ps, *Pdh
are fully preserved (IE *dh
becoming Gr. th
); the left
plosive is assimilated to the right obstruent both in voice and aspiration:
P + t = Gr. pt:
pr. δρέπτω, verb.adj. δρεπτός (cf. pr. δρέπω ‘pluckʼ; < IE *√drep-; cf. Cz. drápati
‘scratchʼ, Sln. dŕpati ‘tearʼ; cf. Pokorny IEW: 211; Frisk 1960: 417; Liddell/Scott 1996:
449; LIV2
: 128; Beekes 2016: 353);
pr. κλέπτω, pf. κέκλεπται, nom. κλέπτης (cf. ao. ἐκλάπην ‘stealʼ; < IE *√klep-; cf. L.
clepō, Goth. hlifan ‘stealʼ; cf. Pokorny IEW: 604; Frisk 1960: 870–871; Liddell/Scott
1996: 958; LIV2
: 363–364; Beekes 2016: 713–714);
pf. τέτραπται, verb.adj. τρεπτός (cf. pr. τρέπω ‘turn, directʼ; < IE *√trep-; cf. Hitt. teripzi
‘plowʼ; cf. Pokorny IEW: 1094; Frisk 1970: 923–925; Liddell/Scott 1996: 1813; LIV2
:
650; Beekes 2016: 1503–1504);
num. ἑπτά ‘sevenʼ (< IE *septm̥ ; cf. OIA saptá, L. septem cf. Pokorny IEW: 909; Frisk
1960: 545; Liddell/Scott 1996: 677; Waanders 1992: 373, 380; Blažek 1999: 247;
Beekes 2016: 446);
pf. γέγραπται, verb.adj. γραπτός (cf. pr. γράφω ‘scratch, grazeʼ; < IE *√gerbh
-; cf. OE
ceorfan ‘cut offʼ; cf. Pokorny IEW: 392; Frisk 1960: 324–325; Liddell/Scott 1996:
359–360; LIV2
: 187; Beekes 2016: 285–286);
pf. (κατέ-)στραπται, verb.adj. στρεπτ-ός (cf. pr. στρέφω ‘turn aboutʼ; < IE *√strebh
-; cf.
Pokorny IEW: 1025; Frisk 1970: 808–809; Liddell/Scott 1996: 915, 1653–1654; LIV2
:
603; Beekes 2016: 1413–1414);164
P + s = Gr. ps:
ao. ἔδρεψα, fut. Dor. δρεψεῦμαι (cf. pr. δρέπω ‘pluckʼ; < IE *√drep-; cf. Cz. drápati
‘scratchʼ, Sln. dŕpati ‘tearʼ; cf. Pokorny IEW: 211; Frisk 1960: 417; Liddell/Scott 1996:
449; LIV2
: 128; Beekes 2016: 353);
164
Only in Greek, without cognates from other Indo-European languages.
157
ao. ἔκλεψα, fut. κλέψω (cf. ao. ἐκλάπην ‘stealʼ; < IE *√klep-; cf. L. clepō, Goth. hlifan
‘stealʼ; cf. Pokorny IEW: 604; Frisk 1960: 870–871; Liddell/Scott 1996: 958; LIV2
:
363–364; Beekes 2016: 713–714);
ao. ἔτρεψα, fut. τρέψω (cf. pr. τρέπω ‘turn, directʼ; < IE *√trep-; cf. Hitt. teripzi ‘plowʼ;
cf. Pokorny IEW: 1094; Frisk 1970: 923–925; Liddell/Scott 1996: 1813; LIV2
: 650;
Beekes 2016: 1503–1504);
ao. ἔγραψα, fut. γράψομαι (cf. pr. γράφω ‘scratch, grazeʼ; < IE *√gerbh
-; cf. OE ceorfan
‘cut offʼ; cf. Pokorny IEW: 392; Frisk 1960: 324–325; Liddell/Scott 1996: 359–360;
LIV2
: 187; Beekes 2016: 285–286);
ao. ἔστρεψα, fut. στρέψω (cf. pr. στρέφω ‘turn aboutʼ; < IE *√strebh
-; cf. Pokorny IEW:
1025; Frisk 1970: 808–809; Liddell/Scott 1996: 915, 1653–1654; LIV2
: 603; Beekes
2016: 1413–1414);
P + dh
= Gr. ph
th
:
ao. ἐδρέφθην (cf. pr. δρέπω ‘pluckʼ; < IE *√drep-; cf. Cz. drápati ‘scratchʼ, Sln. dŕpati
‘tearʼ; cf. Pokorny IEW: 211; Frisk 1960: 417; Liddell/Scott 1996: 449; LIV2
: 128;
Beekes 2016: 353);
ao. ps. ἐκλέφθην (cf. ao. ἐκλάπην ‘stealʼ; < IE *√klep-; cf. L. clepō, Goth. hlifan ‘stealʼ;
cf. Pokorny IEW: 604; Frisk 1960: 870–871; Liddell/Scott 1996: 958; LIV2
: 363–364;
Beekes 2016: 713–714);
ao. ps. τραφθείς (cf. pr. τρέπω ‘turn, directʼ; < IE *√trep-; cf. Hitt. teripzi ‘plowʼ; cf.
Pokorny IEW: 1094; Frisk 1970: 923–925; Liddell/Scott 1996: 1813; LIV2
: 650;
Beekes 2016: 1503–1504);
pf. imp. γέγραφθω (cf. pr. γράφω ‘scratch, grazeʼ; < IE *√gerbh
-; cf. OE ceorfan ‘cut offʼ;
cf. Pokorny IEW: 392; Frisk 1960: 324–325; Liddell/Scott 1996: 359–360; LIV2
: 187;
Beekes 2016: 285–286);
ao. ps. ἐστρέφθην (cf. pr. στρέφω ‘turn aboutʼ; < IE *√strebh
-; cf. Pokorny IEW: 1025;
Frisk 1970: 808–809; Liddell/Scott 1996: 915, 1653–1654; LIV2
: 603; Beekes 2016:
1413–1414);
7.2.2 The clusters velar + t/dh
/s
Indo-European clusters of *Kt, *Ks, *Kdh
are fully preserved (IE *dh
becoming Gr. th
); the left
plosive is assimilated to the right obstruent both in voice and aspiration:
K + t = Gr. kt:
verb.adj. ἑλκτέον, ἑλκτικός (cf. pr. ἕλκω ‘drag, drawʼ; < IE *√selk-; cf. Toch. B sälkāte
‘dragʼ; cf. Pokorny IEW: 901; Frisk 1960: 497–498; Liddell/Scott 1996: 534–535;
LIV2
: 530‒531; Beekes 2016: 412);
verb.adj. τηκτός, τηκτέον (cf. pr. τήκω ‘meltʼ; < IE *√teH2k-; cf. OCS tajǫ ‘meltʼ; cf.
Pokorny IEW: 1053; Frisk 1970: 891; Liddell/Scott 1996: 1786–1787; LIV2
: 617;165
Beekes 2016: 1477);
plqpf. ἔζευκτο (cf. pr. ζεύγνῡμι, ζεύγνῡσι ‘yokeʼ; < IE *√ eu̯ g- cf. OIA yunákti ‘yokeʼ, L.
iungō ‘bindʼ; cf. Pokorny IEW: 505‒510; Frisk 1960: 609–610; Liddell/Scott 1996:
754; LIV2
: 316; NIL: 397–404; Beekes 2016: 497–498);
165
Not attested outside Greek, LIV2 (
l.c.) has a variant with a final palatovelar as well.
158
verb.adj. τακτός (cf. ao. part. ps. ταγεὶς ‘draw up, form, arrayʼ; < IE *√tag-; cf. OP hamataxšatā
‘try to keep in orderʼ; cf. Pokorny IEW: 1055; Frisk 1970: 845–846;
Liddell/Scott 1996: 1759–1760, 1753; LIV2
: 615; Beekes 2016: 1444, 1454–1455);
pf. ἦρκται, verb.adj. ἀρκτέον (cf. pr. ἄρχω ‘rule, beginʼ; < IE *√regh
-; cf. MHG regen
‘erect, irritateʼ; cf. Pokorny IEW: 854, 863; Frisk 1960: 159; Liddell/Scott 1996: 242,
254; LIV2
: 498; Beekes 2016: 145–146);
K + s = Gr. ks:
fut. ἕλξω (cf. pr. ἕλκω ‘drag, drawʼ; < IE *√selk-; cf. Toch. B sälkāte ‘dragʼ; cf. Pokorny
IEW: 901; Frisk 1960: 497–498; Liddell/Scott 1996: 534–535; LIV2
: 530‒531;
Beekes 2016: 412);
ao. ἔτηξα, fut. τήξω (cf. pr. τήκω ‘meltʼ; < IE *√teH2k-; cf. OCS tajǫ ‘meltʼ; cf. Pokorny
IEW: 1053; Frisk 1970: 891; Liddell/Scott 1996: 1786–1787; LIV2
: 617;166
Beekes
2016: 1477);
ao. ἔζευξα, fut. ζεύξω, nom. -ζυξ (cf. pr. ζεύγνῡμι, ζεύγνῡσι ‘yokeʼ; < IE *√ eu̯ g- cf. OIA
yunákti ‘yokeʼ, L. iungō ‘bindʼ; cf. Pokorny IEW: 505‒510; Frisk 1960: 609–610;
Liddell/Scott 1996: 754; LIV2
: 316; NIL: 397–404;Beekes 2016: 497–498);
ao. ἔταξα, fut. τάξω (cf. ao. part. ps. ταγεὶς ‘draw up, form, arrayʼ; < IE *√tag-; cf. OP
ham-ataxšatā ‘try to keep in orderʼ; cf. Pokorny IEW: 1055; Frisk 1970: 845–846;
Liddell/Scott 1996: 1759–1760, 1753; LIV2
: 615; Beekes 2016: 1444, 1454–1455);
ao. ἦρξα, fut. ἄρξω (cf. pr. ἄρχω ‘rule, beginʼ; < IE *√regh
-; cf. MHG regen ‘erect,
irritateʼ; cf. Pokorny IEW: 854, 863; Frisk 1960: 159; Liddell/Scott 1996: 242, 254;
LIV2
: 498; Beekes 2016: 145–146);
ao. inf. θρᾶξαι (cf. pr. θράσσω, Att. θράττω ‘confuseʼ; < IE *√dh
reH2gh
-; cf. OCS -dražǫ
‘bewilderʼ; cf. Pokorny IEW: 251, 273; Frisk 1960: 679–670; Liddell/Scott 1996: 804;
LIV2
: 154‒155; Beekes 2016: 553);
K + dh
= Gr. kh
th
:
fut. ps. ἑλχθήσομαι (cf. pr. ἕλκω ‘drag, drawʼ; < IE *√selk-; cf. Toch. B sälkāte ‘dragʼ;
cf. Pokorny IEW: 901; Frisk 1960: 497–498; Liddell/Scott 1996: 534–535; LIV2
:
530‒531; Beekes 2016: 412);
ao. ps. ἐτήχθην (cf. pr. τήκω ‘meltʼ; < IE *√teH2k-; cf. OCS tajǫ ‘meltʼ; cf. Pokorny IEW:
1053; Frisk 1970: 891; Liddell/Scott 1996: 1786–1787; LIV2
: 617;167
Beekes 2016:
1477);
ao. ps. ἐζεύχθην, fut. ps. ζευχθήσομαι (cf. pr. ζεύγνῡμι, ζεύγνῡσι ‘yokeʼ; < IE *√ eu̯ g- cf.
OIA yunákti ‘yokeʼ, L. iungō ‘bindʼ; cf. Pokorny IEW: 505‒510; Frisk 1960: 609–610;
Liddell/Scott 1996: 754; LIV2
: 316; Beekes 2016: 497–498);
ao. ps. ἐτάχθην, fut. ps. ταχθήσομαι (cf. ao. part. ps. ταγεὶς ‘draw up, form, arrayʼ; < IE
*√tag-; cf. OP ham-ataxšatā ‘try to keep in orderʼ; cf. Pokorny IEW: 1055; Frisk 1970:
845–846; Liddell/Scott 1996: 1759–1760, 1753; LIV2
: 615; Beekes 2016: 1444,
1454–1455);
ao. ps. ἤρχθην, ἀρχθῆναι (cf. pr. ἄρχω ‘rule, beginʼ; < IE *√regh
-; cf. MHG regen ‘erect,
irritateʼ; cf. Pokorny IEW: 854, 863; Frisk 1960: 159; Liddell/Scott 1996: 242, 254;
LIV2
: 498; Beekes 2016: 145–146);
ao. ps. ἐθράχθη (cf. pr. θράσσω, Att. θράττω ‘confuseʼ; < IE *√dh
reH2gh
-; cf. OCS -dražǫ
‘bewilderʼ; cf. Pokorny IEW: 251, 273; Frisk 1960: 679–670; Liddell/Scott 1996: 804;
LIV2
: 154‒155; Beekes 2016: 553);
166
Not attested outside Greek, LIV2 (
l.c.) has a variant with a final palatovelar as well.
167
Not attested outside Greek, LIV2 (
l.c.) has a variant with a final palatovelar as well.
159
7.2.3 The clusters palatovelar + t/dh
/s
Since assumed IE palatovelars are indistinguishable in Greek from IE plain velars, the
mechanism is the same as with Indo-European clusters of *Kt, *Ks, *Kdh
:
Ḱ + t = Gr. kt:
pf. δέδεικται, pf.ps. δείδεκτο, verb.adj. δεικτέον (cf. pr. δείκνυμι ‘showʼ; < IE *√de ḱ-;
cf. OIA ádiṣṭa ‘showʼ, L. dīcō ‘sayʼ; cf. Pokorny IEW: 188‒189; Frisk 1960: 355–
356; Liddell/Scott 1996: 373; LIV2
: 108‒109; Beekes 2016: 309);
pf. δέδηκται (cf. pr. δάκνω ‘biteʼ; < IE *√denḱ-; cf. OIA dáśati ‘biteʼ; cf. Pokorny IEW:
201; Frisk 1960: 343; Liddell/Scott 1996: 367; LIV2
: 117‒118; NIL: 82–83; Beekes
2016: 399);
pr. πεκτέω, nom. πεκτήρ (cf. pr. πέκω ‘combʼ; < IE *√peḱ-; cf. L. pectō ‘comb, shearʼ,
Lith. pešù ‘pluckʼ; cf. Pokorny IEW: 797; Frisk 1970: 492–493; Liddell/Scott 1996:
1356, 1432; LIV2
: 467; Beekes 2016: 1164);
num. ὀκτώ ‘eightʼ (< IE *oḱtṓ; cf. OIA aṣṭáu, L. octō; cf. Pokorny IEW: 775; Frisk 1970:
374–375; Liddell/Scott 1996: 1213; Waanders 1992: 373; Blažek 1999: 247; Beekes
2016: 1066);
verb.adj. ἀκτέον, ἀκτή (cf. pr. ἄγω ‘drive, carry, fetchʼ; < IE *√H2eǵ-; cf. OIA ájati
‘driveʼ, Arm. acem ‘leadʼ; cf. Pokorny IEW: 4‒5; Frisk 1960: 18; Liddell/Scott 1996:
17–18, 58; LIV2
: 255‒256; NIL: 267–277; Beekes 2016: 18–19);
pf. λέλεκται, ao. λέκτο, verb.adj. λεκτός (cf. pr. λέγω ‘pick up, choose, sayʼ; < IE *√leǵ;
cf. L. legō ‘pick up, readʼ; cf. Pokorny IEW: 658; Frisk 1970: 94–96; Liddell/Scott
1996: 1033–1034, 1037; LIV2
: 397; Beekes 2016: 841);
nom. ἀγκτήρ (cf. pr. ἄγχω ‘squeeze, hugʼ; < IE *√H2emǵh
-; cf. Hitt. hamanki ‘tie upʼ, L.
angō ‘restrictʼ; cf. Pokorny IEW: 42‒43; Frisk 1960: 17–18; Liddell/Scott 1996: 10,
17; LIV2
: 264‒265; Beekes 2016: 18);
nom. λείκτης ‘cunnilingusʼ (cf. pr. λείχω ‘lickʼ; < IE *√le ǵh
-; cf. OIA réḍhi, OCS ližǫ
‘lickʼ; cf. Pokorny IEW: 668; Frisk 1970: 102–103; Liddell/Scott 1996: 1035, 1037;
LIV2
: 404; Beekes 2016: 846–847);
verb.adj. ἑκτός, nom. ἑκτωρ (cf. pr. ἔχω ‘have, possesʼ; < IE *√seǵh
-; cf. OIA sáhate
‘captureʼ; cf. Pokorny IEW: 888‒889; Frisk 1960: 602–604; Liddell/Scott 1996: 523,
749–750; LIV2
: 515‒516; NIL: 600–604; Beekes 2016: 490–491);
Ḱ + s = Gr. ks;
ao. ἔδειξα, fut. δείξω (cf. pr. δείκνυμι ‘showʼ; < IE *√de ḱ-; cf. OIA ádiṣṭa ‘showʼ, L.
dīcō ‘sayʼ; cf. Pokorny IEW: 188‒189; Frisk 1960: 355–356; Liddell/Scott 1996: 373;
LIV2
: 108‒109; Beekes 2016: 309);
ao. ἔδηξα, fut. δήξομαι (cf. pr. δάκνω ‘biteʼ; < IE *√denḱ-; cf. OIA dáśati ‘biteʼ; cf.
Pokorny IEW: 201; Frisk 1960: 343; Liddell/Scott 1996: 367; LIV2
: 117‒118; NIL:
82–83;Beekes 2016: 399);
ao. ἔπεξα (cf. pr. πέκω ‘combʼ; < IE *√peḱ-; cf. L. pectō ‘comb, shearʼ, Lith. pešù ‘pluckʼ;
cf. Pokorny IEW: 797; Frisk 1970: 492–493; Liddell/Scott 1996: 1356, 1432; LIV2
:
467; Beekes 2016: 1164);
nom. ἄξων ‘axleʼ, ἄμαξα ‘wagonʼ (< IE *sm-H2ḱs-iH2-; cf. OIA akṣa-, L. axis, OCS osь,
OHG. ahsa ‘axleʼ; cf. Pokorny 1959: 6; Frisk 1960: 85–86, 116; Liddell/Scott 1996:
172; NIL: 259–262; Beekes 2016: 81–82, 111);
160
num. ἕξ, Dor. Crete ϝέξ ‘sixʼ (< IE *su̯ eḱs; cf. L. sex, Goth. saihs; cf. Pokorny IEW: 1044;
Frisk 1960: 527; Waanders 1992: 372–673; Liddell/Scott 1996: 529; Blažek 1999:
236; Beekes 2016: 433–434);
ao. inf. ἄξαι, fut. ἄξω (cf. pr. ἄγω ‘drive, carry, fetchʼ; < IE *√H2eǵ-; cf. OIA ájati ‘driveʼ,
Arm. acem ‘leadʼ; cf. Pokorny IEW: 4‒5; Frisk 1960: 18; Liddell/Scott 1996: 17–18,
58; LIV2
: 255‒256; NIL: 267–277; Beekes 2016: 18–19);
ao. ἔλεξα, fut. λέξω (cf. pr. λέγω ‘pick up, choose, sayʼ; < IE *√leǵ-; cf. L. legō ‘pick up,
readʼ; cf. Pokorny IEW: 658; Frisk 1970: 94–96; Liddell/Scott 1996: 1033–1034,
1037; LIV2
: 397; Beekes 2016: 841);
ao. inf. ἄγξαι, fut. ἄγξω (cf. pr. ἄγχω ‘squeeze, hugʼ; < IE *√H2emǵh
-; cf. Hitt. hamanki
‘tie upʼ, L. angō ‘restrictʼ; cf. Pokorny IEW: 42‒43; Frisk 1960: 17–18; Liddell/Scott
1996: 10, 17; LIV2
: 264‒265; Beekes 2016: 18);
aor. ἔλειξα (cf. pr. λείχω ‘lickʼ; < IE *√le ǵh
-; cf. OIA réḍhi, OCS ližǫ ‘lickʼ; cf. Pokorny
IEW: 668; Frisk 1970: 102–103; Liddell/Scott 1996: 1035, 1037; LIV2
: 404; Beekes
2016: 846–847);
fut. ἕξω (cf. pr. ἔχω ‘have, possesʼ; < IE *√seǵh
-; cf. OIA sáhate ‘captureʼ; cf. Pokorny
IEW: 888‒889; Frisk 1960: 602–604; Liddell/Scott 1996: 523, 749–750; LIV2
: 515‒
516; NIL: 600–604; Beekes 2016: 490–491);
Ḱ + dh
= Gr. kh
th
:
ao.ps. δειχθήσομαι (cf. pr. δείκνυμι ‘showʼ; < IE *√de ḱ-; cf. OIA ádiṣṭa ‘showʼ, L. dīcō
‘sayʼ; cf. Pokorny IEW: 188‒189; Frisk 1960: 355–356; Liddell/Scott 1996: 373;
LIV2
: 108‒109; Beekes 2016: 309);
fut. ps. δηχθήσομαι, ao. ps. ἐδήχθην (cf. pr. δάκνω ‘biteʼ; < IE *√denḱ-; cf. OIA dáśati
‘biteʼ; cf. Pokorny IEW: 201; Frisk 1960: 343; Liddell/Scott 1996: 367; LIV2
: 117‒
118; NIL: 82–83;Beekes 2016: 399);
ao. ps. ἐπέχθην (cf. pr. πέκω ‘combʼ; < IE *√peḱ-; cf. L. pectō ‘comb, shearʼ, Lith. pešù
‘pluckʼ; cf. Pokorny IEW: 797; Frisk 1970: 492–493; Liddell/Scott 1996: 1356, 1432;
LIV2
: 467; Beekes 2016: 1164);
ao. ps. ἤχθην. ps. fut. ἀχθήσομαι (cf. pr. ἄγω ‘drive, carry, fetchʼ; < IE *√H2eǵ-; cf. OIA
ájati ‘driveʼ, Arm. acem ‘leadʼ; cf. Pokorny IEW: 4‒5; Frisk 1960: 18; Liddell/Scott
1996: 17–18, 58; LIV2
: 255‒256; Beekes 2016: 18–19);
ao. ps. ἐλέχθην (cf. pr. λέγω ‘pick up, choose, sayʼ; < IE *√leǵ-; cf. L. legō ‘pick up,
readʼ; cf. Pokorny IEW: 658; Frisk 1970: 94–96; Liddell/Scott 1996: 1033–1034,
1037; LIV2
: 397; Beekes 2016: 841);
ao. ps. part. ἐκλειχθέν (cf. pr. λείχω ‘lickʼ; < IE *√le ǵh
-; cf. OIA réḍhi, OCS ližǫ ‘lickʼ;
cf. Pokorny IEW: 668; Frisk 1970: 102–103; Liddell/Scott 1996: 1035, 1037; LIV2
:
404; Beekes 2016: 846–847);
7.2.4 The clusters labiovelar + t/dh
/s
The Indo-European labiovelars are realized either as labials or velars in the position before
t/th
/s-. In the given contexts there is no realization as t, possible in other contexts. The outcome
is k for an old labiovelar in the case of close contact of a labiovelar with u (cf. pr. ηὖκται, ao.
εὐξάμην and pr. εὔχομαι < IE *√H2eu̯gu̯h
-, this change is Pre-Mycenaean, since Myc. e-u-keto;
cf. Lejeune 1972: 44; Aura Jorro/Adrados 1985: 261–262; Sihler 1995: 156; Bartoněk 2003:
139):
161
Ku̯
+ t = PGr. *ku̯
t > Gr. pt/kt:
pf. λέλειπτο (ep.), ao. ἔλειπτο (ep.), verb.adj. λειπτέον (cf. pr. λείπω ‘leaveʼ; < IE *√le ku̯
;
cf. OIA riṇákti, L. līquī ‘leaveʼ; cf. Pokorny IEW: 669‒670; Frisk 1970: 99–100;
Liddell/Scott 1996: 1035–1036; LIV2
: 406‒407; NIL: 600–604; Beekes 2016: 844–
845);
pr. βλάπτω (cf. part. βλᾰβείς ‘disable, hinderʼ; < IE *√melku̯
-; cf. OIA marcáyati
‘damageʼ; cf. Pokorny IEW: 737; Frisk 1960: 239–240; Liddell/Scott 1996: 317; LIV2
:
434‒435; Beekes 2016: 217);
pf. πέπεπται (cf. pr. πέσσω ‘ripen, cookʼ; < IE *√peku̯
-; cf. OIA pácati ‘cook, ripeʼ, L.
coquō ‘cookʼ, OCS pekǫ ‘bakeʼ; cf. Pokorny IEW: 798; Frisk 1970: 519–520;
Liddell/Scott 1996: 1396; LIV2
: 468; NIL: 548–552; Beekes 2016: 1180–1181);
num. πέμπτος ‘fifthʼ (< IE *penku̯
to-; cf. OIA pakthá-, L. quīnctus ‘fifthʼ; cf. Pokorny
IEW: 808; Frisk 1970: 506–507; Liddell/Scott 1996: 1359; Waanders 1992: 372, 375,
379; Blažek 1999: 221; Beekes 2016: 1172–1173);
pf. ἤμειπται, plpf. ἄμειπτο (cf. pr. ἀμείβω ‘changeʼ; < IE *√H2me gu̯
-; cf. L. migrāre
‘migrateʼ (?); cf. Pokorny IEW: 713; Frisk 1960: 90; Liddell/Scott 1996: 79–80; LIV2
:
279; Beekes 2016: 85–86);
verb.adj. τριπτέον, τριπτήριον (cf. pr. τρίβω ‘rubʼ; < IE *√tre gu̯
-; cf. Pokorny IEW: 1071;
Frisk 1970: 930–931; Liddell/Scott 1996: 1817, 1822; LIV2
: 648; Beekes 2016: 1508–
1509);168
ps. ηὖκται, verb.adj. εὐκταῖος (cf. pr. εὔχομαι ‘pray, wishʼ; < IE *√H1u̯egu̯h
-; cf. OIA
óhate ‘speak solemnlyʼ, L. uoueō ‘vowʼ; cf. Pokorny IEW: 348; Frisk 1960: 595–596;
Liddell/Scott 1996: 719, 739; LIV2
: 253; Beekes 2016: 485–486);
Ku̯
+ s = PGr. *ku̯
s > Gr. ps/ks:
ao. ἔλειψα, fut. λείψω (cf. pr. λείπω ‘leaveʼ; < IE *√le ku̯
-; cf. OIA riṇákti, L. līquī ‘leaveʼ;
cf. Pokorny IEW: 669‒670; Frisk 1970: 99–100; Liddell/Scott 1996: 1035–1036;
LIV2
: 406‒407; NIL: 600–604; Beekes 2016: 844–845);
ao. ἔβλαψα, fut. βλάψω (cf. part. βλᾰβείς ‘disable, hinderʼ; < IE *√melku̯
-; cf. OIA
marcáyati ‘damageʼ; cf. Pokorny IEW: 737; Frisk 1960: 239–240; Liddell/Scott 1996:
317; LIV2
: 434‒435; Beekes 2016: 217);ao, ἔπεψα, fut. πέψω (cf. pr. πέσσω ‘ripen,
cookʼ; < IE *√peku̯
-; Beekes 2016: x);
ao. ἤμειψα, fut. ἀμείψα (cf. pr. ἀμείβω ‘changeʼ; < IE *√H2me gu̯
-; cf. L. migrāre
‘migrateʼ (?); cf. Pokorny IEW: 713; Frisk 1960: 90; Liddell/Scott 1996: 79–80; LIV2
:
279; NIL: 548–552; Beekes 2016: 85–86);
ao. ἔτριψα, fut. τρίψω (cf. pr. τρίβω ‘rubʼ; < IE *√tre gu̯
-; cf. Pokorny IEW: 1071; Frisk
1970: 930–931; Liddell/Scott 1996: 1817, 1822; LIV2
: 648; Beekes 2016: 1508–
1509);
ao. εὐξάμην, fut. εὔξομαι (cf. pr. εὔχομαι ‘pray, wishʼ; < IE *√H1u̯egu̯h
-; cf. OIA óhate
‘speak solemnlyʼ, L. uoueō ‘vowʼ; cf. Pokorny IEW: 348; Frisk 1960: 595–596;
Liddell/Scott 1996: 719, 739; LIV2
: 253; Beekes 2016: 485–486);
Ku̯
+ dh
= PGr. *ku̯ h
th
> Gr. ph
th
/kh
th
:
168
Not securely attested outside Greek.
162
ao. ps. ἐλείφθην (cf. pr. λείπω ‘leaveʼ; < IE *√le ku̯
-; cf. OIA riṇákti, L. līquī ‘leaveʼ; cf.
Pokorny IEW: 669‒670; Frisk 1970: 99–100; Liddell/Scott 1996: 1035–1036; LIV2
:
406‒407; NIL: 600–604; Beekes 2016: 844–845);
ao. ps. ἐβλάφθην (cf. part. βλᾰβείς ‘disable, hinderʼ; < IE *√melku̯
-; cf. OIA marcáyati
‘damageʼ; cf. Pokorny IEW: 737; Frisk 1960: 239–240; Liddell/Scott 1996: 317; LIV2
:
434‒435; Beekes 2016: 217);
ao, ps. ἐπέφθην, fut. ps. πεφθήσομαι (cf. pr. πέσσω ‘ripen, cookʼ; < IE *√peku̯
-; cf. OIA
pácati ‘cook, ripeʼ, L. coquō ‘cookʼ, OCS pekǫ ‘bakeʼ; cf. Pokorny IEW: 798; Frisk
ao. ps. ἠμείφθην, fut. ps. ἀμειφθήσεται (cf. pr. ἀμείβω ‘changeʼ; < IE *√H2me gu̯
-; cf. L.
migrāre ‘migrateʼ (?); cf. Pokorny IEW: 713; Frisk 1960: 90; Liddell/Scott 1996: 79–
80; LIV2
: 279; NIL: 548–552; Beekes 2016: 85–86);
ao. ps. ἐτρίφθην, fut. ps. τριφθήσομαι (cf. pr. τρίβω ‘rubʼ; < IE *√tre gu̯
-; cf. Pokorny
IEW: 1071; Frisk 1970: 930–931; Liddell/Scott 1996: 1817, 1822; LIV2
: 648; Beekes
2016: 1508–1509);
pf. inf. ps. ηὖχθαι (cf. pr. εὔχομαι ‘pray, wishʼ; < IE *√H1u̯egu̯h
-; cf. OIA óhate ‘speak
solemnlyʼ, L. uoueō ‘vowʼ; cf. Pokorny IEW: 348; Frisk 1960: 595–596; Liddell/Scott
1996: 719, 739; LIV2
: 253; Beekes 2016: 485–486);
7.2.5 The clusters dental + t/dh
/s
The IE cluster of *Tt is realized as Gr. ss, *Ts as Gr. 0s, *Tdh
as Gr. sth
:
T + t = Gr.st:
verb.adj. κεστός, κέστρος (cf. pr. κεντέω ‘prick, goadʼ; < IE *√ḱent-; cf. L. sīts ‘hunting
spearʼ,169
OHG hantag ‘pointedʼ; cf. Pokorny IEW: 567; Frisk 1960: 820–821;
Liddell/Scott 1996: 939, 944; LIV2
: 326‒327; Beekes 2016: 672–673);170
verb.adj. λιστός (cf. pr. λίτομαι Hom., aor. inf. λῐτέσθαι ‘beg, prayʼ; < IE *√le t-; cf. Lith.
liečiù ‘touchʼ (?); cf. Pokorny IEW: 664; Frisk 1970: 130; Liddell/Scott 1996: 1054;
LIV2
: 410‒411; Beekes 2016: 866);
plqpf. ἐπέπαστο, verb.adj. παστός (cf. pr. πάσσω, Att. πάττω ‘sprinkleʼ; < IE *√(s)ku̯ eH2t
-; cf. L. quatiō ‘shake, tossʼ, OHG scutten ‘pour, tossʼ; cf. Pokorny IEW: 632, 957‒
958; Frisk 1970: 478; Liddell/Scott 1996: 1346; LIV2
: 563; Beekes 2016: 1155–1156);
pf. ἴστε, ἴστω, verb.adj. ἰστέον (Boeot. ἴττω; cf. pf. οἶδα ‘knowʼ; < IE *√u̯e d-; cf. OIA
vindáti ‘findʼ, L. uīsō ‘visitʼ; cf. Pokorny IEW: 1125‒1127; Frisk 1970: 357;
Liddell/Scott 1996: 483, 841; LIV2
: 665‒667; NIL: 717–722; Beekes 2016: 577,
1053);
pf. πέπεισται, verb.adj. πειστέον (cf. pr. πείθω ‘persuadeʼ; < IE *√bh
e dh
-; cf. L. fīdō
‘trustʼ; cf. Pokorny IEW: 117; Frisk 1970: 487–488; Liddell/Scott 1996: 1353–1354,
1356; LIV2
: 71‒72; NIL: 12–13; Beekes 2016: 1161–1162);
pf. πέπυσται (cf. pr. πεύθομαι, πυνθάνομαι ‘learnʼ; < IE *√bh
eu̯dh
-; cf. OIA bódhati
‘awakeʼ, OCS bljudǫ ‘wake upʼ; cf. Pokorny IEW: 150‒152; Frisk 1970: 625–626;
Liddell/Scott 1996: 1398, 1554; LIV2
: 82‒83; NIL: 36–37; Beekes 2016: 1258);
T + s = Gr. 0s:
169
If from *ḱnt-o-.
170
Not securely attested outside Greek.
163
ao. inf. κένσαι (cf. pr. κεντέω ‘prick, goadʼ; < IE *√ḱent-; cf. L. sīts ‘hunting spearʼ,171
OHG hantag ‘pointedʼ; cf.Pokorny IEW: 567; Frisk 1960: 820–821; Liddell/Scott
1996: 939, 944; LIV2
: 326‒327; Beekes 2016: 672–673);
ao. ἐλῐσάμην, λίσαι (cf. pr. λίτομαι Hom., aor. inf. λῐτέσθαι ‘beg, prayʼ; < IE *√le t-; cf.
Lith. liečiù ‘touchʼ (?); cf. Pokorny IEW: 664; Frisk 1970: 130; Liddell/Scott 1996:
1054; LIV2
: 410‒411; Beekes 2016: 866);
ao. ἔπᾰσα, fut. πάσω (cf. pr. πάσσω, Att. πάττω ‘sprinkleʼ; < IE *√(s)ku̯ eH2t -; cf. L.
quatiō ‘shake, tossʼ, OHG scutten ‘pour, tossʼ; cf. Pokorny IEW: 632, 957‒958; Frisk
1970: 478; Liddell/Scott 1996: 1346; LIV2
: 563; Beekes 2016: 1155–1156);
ao. ἧσα, ἡσάμην (cf. pr. ἥδω, ἥδομαι ‘enjoyʼ; < IE *√su̯eH2d-; cf. OIA svādate ‘make
palatableʼ, L. suādeō ‘rate someoneʼ; cf. Pokorny IEW: 1039‒1040; Frisk 1960: 622–
623; Liddell/Scott 1996: 764; LIV2
: 606‒607; Beekes 2016: 509–510);
pf. ἴσασι (Boeot. ἴττω; cf. pf. οἶδα ‘knowʼ; < IE *√u̯e d-; cf. OIA vindáti ‘findʼ, L. uīsō
‘visitʼ; cf. Pokorny IEW: 1125‒1127; Frisk 1970: 357; Liddell/Scott 1996: 483, 841;
LIV2
: 665‒667; NIL: 717–722; Beekes 2016: 577, 1053);
ao. ἔπεισα, fut. πείσω (cf. pr. πείθω ‘persuadeʼ; < IE *√bh
e dh
-; cf. L. fīdō ‘trustʼ; cf.
Pokorny IEW: 117; Frisk 1970: 487–488; Liddell/Scott 1996: 1353–1354, 1356; LIV2
:
71‒72; NIL: 12–13; Beekes 2016: 1161–1162);
fut. πεύσομαι (cf. pr. πεύθομαι, πυνθάνομαι ‘learnʼ; < IE *√bh
eu̯dh
-; cf. OIA bódhati
‘awakeʼ, OCS bljudǫ ‘wake upʼ; cf. Pokorny IEW: 150‒152; Frisk 1970: 625–626;
Liddell/Scott 1996: 1398, 1554; LIV2
: 82‒83; NIL: 36–37; Beekes 2016: 1258);
T + dh
= Gr. sth
:
ao. ps. ἐπάσθην (cf. pr. πάσσω, Att. πάττω ‘sprinkleʼ; < IE *√(s)ku̯ eH2t -; cf. L. quatiō
‘shake, tossʼ, OHG scutten ‘pour, tossʼ; cf. Pokorny IEW: 632, 957‒958; Frisk 1970:
478; Liddell/Scott 1996: 1346; LIV2
: 563; Beekes 2016: 1155–1156);
pr. ἐσθίω, ps. ἐσθίομαι (cf. pr. ἔδω ‘eat, devourʼ; < IE *√H1ed-; cf. OIA átti, L. edō ‘eatʼ;
cf. Pokorny IEW: 287–289; Frisk 1960: 444–445; Liddell/Scott 1996: 478; LIV2
: 230–
231; Beekes 2016: 375);
ao. ἥσθην, fut. ἡσθήσομαι (cf. pr. ἥδω, ἥδομαι ‘enjoyʼ; < IE *√su̯eH2d-; cf. OIA svādate
‘make palatableʼ, L. suādeō ‘rate someoneʼ; cf. Pokorny IEW: 1039‒1040; Frisk 1960:
622–623; Liddell/Scott 1996: 764; LIV2
: 606‒607; Beekes 2016: 509–510);
pf. ἴσθι ((Boeot. ἴττω; cf. pf. οἶδα ‘knowʼ; < IE *√u̯e d-; cf. OIA vindáti ‘findʼ, L. uīsō
‘visitʼ; cf. Pokorny IEW: 1125‒1127; Frisk 1970: 357; Liddell/Scott 1996: 483, 841;
LIV2
: 665‒667; NIL: 717–722; Beekes 2016: 577, 1053);
pf. πέπεισϑε (cf. pr. πείθω ‘persuadeʼ; < IE *√bh
e dh
-; cf. L. fīdō ‘trustʼ; cf. Pokorny IEW:
117; Frisk 1970: 487–488; Liddell/Scott 1996: 1353–1354, 1356; LIV2
: 71‒72; NIL:
12–13; Beekes 2016: 1161–1162);
pf. inf. πεπύσθαι (cf. pr. πεύθομαι, πυνθάνομαι ‘learnʼ; < IE *√bh
eu̯dh
-; cf. OIA bódhati
‘awakeʼ, OCS bljudǫ ‘wake upʼ; cf. Pokorny IEW: 150‒152; Frisk 1970: 625–626;
Liddell/Scott 1996: 1398, 1554; LIV2
: 82‒83; NIL: 36–37; Beekes 2016: 1258);
7.2.6 The clusters sibilant + t/dh
/s
The IE cluster of *st is preserved, IE cluster of *ss is simplified as 0si:
171
If from IE *ḱnt-o-.
164
s + t = Gr. st:
pr. ἐστί, verb.adj. ἐστόν (cf. pr. ἐσμέν ‘beʼ; < IE *√H1es-; cf. Hitt. ēszi, OIA ásti, L. est
‘beʼ; cf. Pokorny IEW: 340–341; Frisk 1960: 463; Liddell/Scott 1996: 487–489; LIV2
:
241–242; Beekes 2016: 389);
verb.adj. μύστης, μυστήριον (cf. pr. μύω ‘closeʼ; < IE *√meu̯ s-; cf. Pokorny IEW: 752;
Frisk 1970: 279–280; Liddell/Scott 1996: 1156–1157; LIV2
: 444; Beekes 2016:
988);172
pr. ἵστημι (‘standʼ; < IE *√steH2-; cf. OIA ástāt, L. sistō ‘standʼ; cf. Pokorny IEW: 1004–
1008; Frisk 1960: 739; Liddell/Scott 1996: 841; LIV2
: 590–592; Beekes 2016: 601);
verb.adj. τρέστης (cf. pr. τρέω ‘flee, fearʼ; < IE *√tres-; cf. OIA trásati ‘trembleʼ, OCS
trȩsǫ ‘shakeʼ; cf. Pokorny IEW: 1095; Frisk 1970: 929–930; Liddell/Scott 1996: 1813,
1815; LIV2
: 650–651; Beekes 2016: 1507–1508);
s + s = Gr. 0s:
fut. ἔσομαι, Aeol. ἔσσι173
(cf. pr. ἐσμέν ‘beʼ; < IE *√H1es-; cf. Hitt. ēszi, OIA ásti, L. est
‘beʼ; cf. Pokorny IEW: 340–341; Frisk 1960: 463; Liddell/Scott 1996: 487–489; LIV2
:
241–242; Beekes 2016: 389);
ao. ἔδησα, fut. δήσω (cf. pr. δέω ‘bindʼ; < IE *√deu̯ s-; cf. Pokorny IEW: 219; Frisk 1960:
374–375; Liddell/Scott 1996: 383; LIV2
: 125; Beekes 2016: x);174
ao. ἔθραυσα, fut. θραύσω (cf. pr. θραύω, pf. τέθραυσμαι ‘break, shatterʼ; < IE *√dh
reu̯ s;
cf. Goth. driusan ‘fall (down)ʼ; cf. Pokorny IEW: 274–275; Frisk 1960: 680–681;
Liddell/Scott 1996: 805; LIV2
: 157–158; Beekes 2016: 553);
ao. ἔμυσα, fut. μύσω (cf. pr. μύω ‘closeʼ; < IE *√meu̯ s-; cf. Pokorny IEW: 752; Frisk
1970: 279–280; Liddell/Scott 1996: 1156–1157; LIV2
: 444; Beekes 2016: 988);
ao. ἔτρεσα (cf. pr. τρέω ‘flee, fearʼ; < IE *√tres-; cf. OIA trásati ‘trembleʼ, OCS trȩsǫ
‘shakeʼ; cf. Pokorny IEW: 1095; Frisk 1970: 929–930; Liddell/Scott 1996: 1813, 1815;
LIV2
: 650–651; Beekes 2016: 1507–1508);
Note: The ending 2nd
pr. εἶ (= es-si) shows that the IE cluster of *ss was simplified to *0s before the loss of the
intervocalic -s-. Aeol. ἔσσι, Ep., Dor. ἐσσί, Hom. Pi. ἐσί from the same Greek root √es- are morphological
restorations. Similarly, Dor. fut. ἐσσῇ, σσοῦνται are morphologically restored.
s + dh
= Gr. sth
:
pf. inf. δεδέσθαι (cf. pr. δέω ‘bindʼ; < IE *√deu̯ s-; cf. Pokorny IEW: 219; Frisk 1960:
374–375; Liddell/Scott 1996: 383; LIV2
: 125; Beekes 2016: x)
ao. ps. ἐθραύσθην, fut. ps. fut. θραυσθήσομαι (cf. pr. θραύω, pf. τέθραυσμαι ‘break,
shatterʼ; < IE *√dh
reu̯ s-; cf. Goth. driusan ‘fall (down)ʼ; cf. Pokorny IEW: 274–275;
Frisk 1960: 680–681; Liddell/Scott 1996: 805; LIV2
: 157–158; Beekes 2016: 553);
7.2.7 The overview of the Greek development
Classical Greek is a very conservative language, preserving all peripheral plosives as plosives
(the old IE labiovelars realized either as labials or velars). The dentals are realized as sibilants
in the t-context, lost in the s-context:
172
Not attested outside Greek.
173
This cluster is more probably a result of the re-archaization than a relict.
174
Attested only in Greek, without IE cognates.
165
IE Greek t- dh
- s-
-kṷ
/gṷ
/gṷh
-p/b/ph
-k/g/kh
-
t/d/th
pt
kt
ph
th
kh
th
ps
ks
-k/g/gh
-k/g/kh
kt kh
th
ks
-ḱ/ǵ/ǵh
-k/g/kh
kt kh
th
ks
-t/d/dh
-t/d/th
st sth
0s
-p/b/bh
-p/b/ph
pt ph
th
ps
-s -s/0 st sth
0s
7.3 The Mycenaean clusters and their development
Mycenaean development differs from the Classical Greek in preserving the old labiovelars;
otherwise it has, as far as limited data allow us to be sure, the same development.
For Mycenaean data, Aura Jorro/Adrados (1985; 1993) and Bartoněk 2003 will be used. With
Bartoněk, if quoting the grammatical data, the quotation on a given page will be used, but when
lemmata in the glossary are used, we will quote the given lemma (in the form Bartoněk 2007:
Lxxx). Beekes (2016; 2016) will be quoted for Mycenaean when quoting the same data as Aura
Jorro/Adrados and Bartoněk.
7.3.1 The clusters labial + t/dh
/s in Mycenaean
P + t = Myc. pt:
]e-na-ri-po-to (enaliptos) ‘greased, polished, paintedʼ (cf. Gr. ἐνάλειπτος, ἐναλείφω
‘anoint withʼ, ἀλείφω ‘anointʼ; < IE *√le p- (?); cf. OIA limpáti ‘smearʼ, OCS -lěpiti
‘glueʼ; cf. Pokorny IEW: 670–671; Frisk 1960: 67–68; Aura Jorro/Adrados 1985: 217;
Liddell/Scott 1996: 62, 553; LIV2
: 408; Bartoněk 2003: L205; Beekes 2016: 64);
ra-pte (rh
aptēr) (cf. Gr. ῥάπ-της ‘ one who stitches, clothes-menderʼ, ῥάπτω ‘sew, stitchʼ;
cf. OIA várpas- ‘artificeʼ; cf. Pokorny IEW: 1156; Frisk 1970: 643; Aura
Jorro/Adrados 1993: 221–223; Liddell/Scott 1996: 1565; LIV2
: 701; Bartoněk 2003:
L698; Beekes 2016: 1275–1276);
P + s = Myc. ps:
di-pi-si-jo (dipsios) ‘?ʼ (a dat. sg. ?) (Aura Jorro/Adrados 1993: 175–176; Bartoněk 2003:
L162);
e-we-pe-se-so-me-na (eu (h)epsēsomena) fut. part. ‘well cooked (?)ʼ (cf. Gr. εὐέψητος, ‘
readily cookedʼ?; < IE *√peku̯
-; cf. OIA pácati ‘cook, ripeʼ, L. coquō ‘cookʼ, OCS pekǫ
‘bakeʼ; cf. Pokorny IEW: 798; Frisk 1970: 519–520; Aura Jorro/Adrados 1993: 267
for other possible meanings; Liddell/Scott 1996: 712; LIV2
: 468; Bartoněk 2003:
L282; Beekes 2016: 1180–1181);
P + th
= Myc. ph
th
:
di-pte-ra (diphtherā) (cf. Gr. διφθέραʼprepared hide, piece of leatherʼ and probably Gr.
δέφω ‘softenʼ; < IE *√deph
- (?); cf. Arm. topʿem ‘beatʼ, Pol. deptać ‘step, treadʼ; cf.
166
Pokorny IEW: 203; Frisk 1960: 372–373, 400; Aura Jorro/Adrados 1993: 176–177;
Liddell/Scott 1996: 382, 438; Bartoněk 2003: L164; Beekes 2016: 320, 341);175
7.3.2 The clusters velar + t/dh
/s in Mycenaean
K + t = Myc. kt:
wa-na-ka-te (u̯ anaktei) (< PGr. u̯ anaks; cf. Gr. ἄναξ ‘rulerʼ; cf. Frisk 1960: 102–103;
Aura Jorro/Adrados 1993: 400–402; Liddell/Scott 1996: 114; Bartoněk 2003: L837;
NIL: 267–277; Beekes 2016: 98–99);176
K + s = Myc. ks:
ke-se-ne-wi-ja (ksenu̯ ia) (< PGr. ksenuu̯ i a; cf. ξένιος ‘friendlyʼ; cf. Pokorny IEW: 453;
Frisk 1970: 333–334; Aura Jorro/Adrados 1985: 353–354; Liddell/Scott 1996: 1188–
1189; Bartoněk 2003: L379; Beekes 2016: 1034);177
wa-na-ka (u̯ anaks) (< PGr. u̯ anaks; cf. Gr. ἄναξ ‘rulerʼ; Frisk 1960: 102–103; Aura
Jorro/Adrados 1993: 400–402; Liddell/Scott 1996: 114; Bartoněk 2003: L837; NIL:
267–277; Beekes 2016: 98–99);
7.3.3 The clusters palatovelar + t/dh
/s s in Mycenaean
Ḱ + t = Myc. kt:
mi-ka-ta (miktās) ‘mixerʼ (cf. Gr. μείγνυμι, μίγνυμι; < IE *√me ḱ-; cf. L. misceō, OCS
měšǫ ‘mixʼ; cf. Pokorny IEW: 714; Frisk 1970: 192–193; Aura Jorro/Adrados 1985:
452–453; Liddell/Scott 1996: 1092; LIV2
: 428–429; Bartoněk 2003: L477; Beekes
2016: 919–920);
pa-ke-te-re (pā̆ ktēres ?) ‘pin, plugʼ (cf. Gr. πήγνῡμι ‘stickʼ; < IE *√peH2ǵ-; cf. L. pangō
‘fasten, fixʼ; cf. Pokorny IEW: 787; Frisk 1970: 525–526; Aura Jorro/Adrados 1993:
71; Liddell/Scott 1996: 1399; LIV2
: 461; Bartoněk 2003: L560; Beekes 2016: 1184);
pe-ki-ti-ra2 (pektr ai) (cf. Gr. πεκτέω, πέκω ‘combʼ; < IE *√peḱ-; cf. L. pectō ‘comb,
shearʼ, Lith. pešù ‘pluckʼ; cf. Pokorny IEW: 797; Frisk 1970: 492–493; Aura
Jorro/Adrados 1993: 97; Liddell/Scott 1996: 1356, 1432; LIV2
: 467; Bartoněk 2003:
L589; Beekes 2016: 1164–1165);
Ḱ + s = Myc. ks:
a3-ka-sa-ma (aiksmans) ‘point of a spear/arrowʼ (cf. Gr. αἰχμή ‘point of a spearʼ; < IE
*√H2e ḱ-s(m)-; cf. L. īcō ‘hit, woundʼ, Pruss. aysmis ‘roasting spitʼ, OCS igla ‘needleʼ;
cf. Pokorny IEW: 15; Frisk 1960: 48; Aura Jorro/Adrados 1985: 132–133;
Liddell/Scott 1996: 45; LIV2
: 259; Bartoněk 2003: L121; Beekes 2016: 405–406);
de-ka-sa-to (deksato) (cf. Gr. δείκνυμι ‘showʼ; < IE *√de ḱ-; cf. OIA ádiṣṭa ‘showʼ, L.
dīcō ‘sayʼ; cf. Pokorny IEW: 188‒189; Frisk 1960: 355–356; Aura Jorro/Adrados
1985: 164; Liddell/Scott 1996: 373; LIV2
: 108‒109; Bartoněk 2003: L148; Beekes
2016: 309);
7.3.4 The clusters labiovelar + t/dh
/s in Mycenaean
The labiovelars are preserved in Mycenaean (or restored?):
175
Could be from *dipstéra, cf. Pokorny (l.c.)?
176
Assumed to be a substrate word.
177
Could be related to L. hostis, OCS gostь?
167
Ku̯
+ t = Myc. Ku̯
t:
ke-ni-qe-te-we (kh
erniku̯
tēu̯ es) ‘basinʼ (cf. Gr. χερνίπτομαι ‘wash (hands) (to purify)ʼ; cf.
OIA anijam, ninikta ‘washʼ, OIr. -nig ‘washʼ; cf. Pokorny IEW: 761; Frisk 1970: 319–
320; Aura Jorro/Adrados 1985: 342; Liddell/Scott 1996: 1175–1176, 1988; LIV2: 450;
Bartoněk 2003: L366; Beekes 2016: 1620–1621);
ra-qi-ti-ra2 (laku̯
tr ai ?) ‘takeʼ (cf. Gr. λαμβάνω ‘takeʼ; < IE *√sleH2gu̯
-; cf. Œ læccean
‘takeʼ; cf. Pokorny IEW: 958; Frisk 1970: 77–78; Aura Jorro/Adrados 1993: 223–224;
Liddell/Scott 1996: 1026–1027; LIV2
: 566; Bartoněk 2003: L700; Beekes 2016: 828–
826);
Ku̯
+ s = Myc. Ku̯
s:
a3-ti-jo-qo (Aithioku̯
s) PN ‘Aith
ioku̯
sʼ (cf. Gr. aἰθίοψ ‘burnt-face, negroʼ, ὤψ ‘face, eyeʼ,
ὄσσομαι ‘see, presageʼ; < IE *√H3eku̯
-; cf. OIA ī́kṣate ‘seeʼ; cf. Pokorny IEW: 775–
777; Frisk 1970: 407–408, 436, 1154; Aura Jorro/Adrados 1985: 140; Liddell/Scott
1996: 37, 1262, 1282–1283; LIV2
: 297; Bartoněk 2003: 138; Beekes 2016: 36–37,
1094, 1118, 1684–1685);
mo-qo-so (moku̯
sos) PN ‘Mopsosʼ (cf. Gr. PN Μόψος-; Aura Jorro/Adrados 1985: 457);
Ku̯
+ th
= Myc. ku̯ h
th
:
e-qi-ti-wo-e (eku̯
th
iu̯ o(h)e) ‘destroyedʼ (?) (cf. Gr. φθίνω ‘decay, wasteʼ; < IE *√dh
gu̯ h
e -;
(?)ʼ ; cf. OIA kṣiṇāti ‘destroyʼ, ON dvena ‘wane, destroyʼ; cf. Pokorny IEW: 487; Frisk
1970: 1014–1016; Liddell/Scott 1996: 1926; LIV2
: 150–152; Bartoněk 2003: L240;
Beekes 2016: 1570–1571);
7.3.5 The clusters dental + t/dh
/s s in Mycenaean
T + t = Myc. st:178
e-pi-da-to (epidastos = -d-t-) (cf. Gr. δᾰτέομαι ‘divideʼ; < IE *√dH1t-sa- (?); cf. OIA
dáyate ‘divideʼ; cf. Pokorny IEW: 175–176; Frisk 1960: 351–352; Aura Jorro/Adrados
1985: 223; Liddell/Scott 1996: 370; LIV2
: 103–104; Bartoněk 2003: L216; Beekes 2016:
305–306);
T + s = Myc. 0s:
pa-si (pansi = *pant-si) dat. pl. ‘allʼ (cf. πᾶς, πᾶσα, πᾶν; < IE *pH2-ent-; cf. ; cf. Pokorny
IEW: 593; Frisk 1970: x; Aura Jorro/Adrados 1985: 342; Liddell/Scott 1996: x; LIV2:
x; Bartoněk 2003: L365; Beekes 2016: 1154–1155);
pi-we-ri-si (Piu̯ erisi = *-d-si) dat. pl. ‘Pieridesʼ (cf. Gr. PN Πῑερίδες; Aura Jorro/Adrados
1993: 342; Liddell/Scott 1996: 1403; Bartoněk 2003: 145);
]o-da-sa-ṭọ (das(s)ato) ‘distribute, divideʼ (cf. δᾰτέομαι, fut. δάσομαι; < IE *√dH1t-sa(?);
cf. OIA dáyate ‘divideʼ; cf. Pokorny IEW: 175–176; Frisk 1960: 351–352; Frisk
1970: 126–127; Aura Jorro/Adrados 1993: 17, 252; Liddell/Scott 1996: 370, 1052;
LIV2
: 103–104; Bartoněk 2003: L140; Beekes 2016: 305–306, 864);179
7.3.6 The clusters sibilant + t/dh
/s in Mycenaean
s + t = Myc. st:180
178
Mycenaean script does not show more than t-, as in the case of the cluster of st-.
179
The relation between ἐναλείφω and ἀλείφω is doubted, cf. especially Beekes (l.c.).
180
A Mycenaean graphics does not show more than given t+vowel sign, as in the case of the cluster of Tt-.
168
ta-to-mo (stath
mos) ‘weight, stallʼ (cf. Gr. σταθμός ‘standing place, dwelling, balance,
weightʼ; Gr. ἵστημι ‘standʼ; < IE *√steH2-; cf. OIA ástāt, L. sistō ‘standʼ; cf. Pokorny
IEW: 1007; ; Frisk 1960: 739; Frisk 1970: 775; Aura Jorro/Adrados 1993: 321;
Liddell/Scott 1996: 841, 1632–1633; LIV2
: 590–592; Bartoněk 2003: L755; Beekes
2016: 601, 1388–1389);
s + s = Myc. 0s:
e-so-to (es(s)ontoi) 3rd
pl. fut. ‘beʼ (cf. Gr. ἐσσοῦνται; < IE *√H1es-; cf. Hitt. ēszi, OIA
ásti, L. est ‘beʼ; cf. Pokorny IEW: 340–341; Frisk 1960: 463; Aura Jorro/Adrados
1985: 203–204, 252; Liddell/Scott 1996: 487–489; LIV2
: 241–242; Bartoněk 2003:
L184; Beekes 2016: 389);
7.3.7 The overview of the Mycenaean development
The reconstructed (and not directly attested) outcomes are in brackets; note that Mycenaean
preserves labiovelars, usually with their labial value restored:
IE Mycenaean t- dh
- s-
-kṷ
/gṷ
/gṷh
-kṷ
/gṷ
/kṷh
kṷ
t kṷ
th
kṷ
s
-k/g/gh
-k/g/kh
kt (kh
th
) ks
-ḱ/ǵ/ǵh
-k/g/kh
kt (kh
th
) ks
-t/d/dh
-t/d/th
st (sth
) 0s
-p/b/bh
-p/b/ph
pt ph
th
ps
-s -s/h st (sth
) 0s
7.4 Trajectories of the Greek development
A typical feature of the Greek clusters of plosive + t/s is that only peripheral plosives are
preserved either with t- or with s- (Bubenik 2017: 646), while the central (i.e., dental) plosives
are sibilantized before t-/th
- or wholly lost before s- (probably after a sibilantization).
The status of two aspirates clusters of kh
th
, ph
th
<χϑ, φϑ> is questionable, the aspiration
of the first plosive is often disputed. On the following lines, we assume the full assimilation of
a left plosive both in voice and in aspiration, hence kh
th
, ph
th
in full accordance to the standard
orthography, though we cannot exclude the dialectical outcome without aspiration: kth
, pth181
,
since phonetically clusters of two aspirates in a single cluster are articulatory more complex
than those with an only left aspirate. For comparison, OIA does not have the clusters of kh
th
,
th
th
, ph
th
but only kth
, tth
, pth
(proposed by Brugmann 1885: 34–35; Brugmann 1890: 57;
Brugmann 1900: 96; expressed later by Jannaris 1897: 58; Hirt 1912: 189; Brugmann/Thumb
181
The graphic κϑ and πϑ are attested in archaic inscriptions dialectally (Lejeune 1972: 69), limited to the South
Aegean Islands and Crete (the “green variant” of Kirchhoff 1877). Note that the same dialects often use π and
κ both for p and ph
and k and kh
.
169
1913: 112; Meillet/Vendryes 1924: 64; Schwyzer 1939: 210;182
Lejeune 1952: 59; 1972: 68–
69). However, as Allen has noted (Allen 1968: 24–26), such clusters are not impossible at all,
cf. Georgian pʿkʿ, tʿkʿ, Arm. tʿkʿ, Abaza pʿqʿ (but note that all examples, on the other hand
could be simply a case of a special Sprachbund-feature of the Caucasus). Rix (1992: 95)
supports the two-aspirates solution, though he even accepts the variant kth
, pth
as well.
Note: Any reader who prefers T + th
= Tth
should accordingly modify the development of clusters formed by IE
*dh
-.
7.4.1 Development of the clusters labial + t/dh
/s
Similarly to the velar series, the clusters of labial plosive + t/dh
/s were affected only by the
devoicing of the IE *dh
, otherwise the clusters are preserved, the left plosive is assimilated both
in voice and aspiration to the right obstruent (cf. Meillet/Vendryes 1924: 53; Buck 1932: 144;
Lejeune 1947: 28–59, 62–63; Lejeune 1972: 68–69, 73; Sihler 1995: 203):
P + t > pt
P + dh
> bdh
→ p + th
> ph
th
P + s > ps
7.4.2 Development of the clusters (palato)velar + t/th
/s
Since there is no internal proof of the original distinction between reconstructed IE plain velars
and palatovelars, we will treat both series as a single one.
Old IE velars (both plain and palate-) are realized as the velar k before t/s- and as the
aspirate kh
before the aspirate th
- (< *dh
), i.e., assimilated both in voice and aspiration according
to the right obstruent (cf. Meillet/Vendryes 1924: 53, 63; Buck 1932: 145; Lejeune 1947: 28–
59, 62–63; Lejeune 1972: 68, 72; Rix 1992: 94–95; Sihler 1995: 203–204):
K + t > kt
K + dh
> gdh
→ Kth
> kh
th
K + s > ks
182
Schwyzer assumes the full aspiration of the first plosive was present in the earlier phase, replaced by a
dissimilated form later.
170
7.4.3 Development of the clusters labiovelar + t/th
/s
The Mycenaean in general preserves old labiovelars (though they were assimilated as plain
velars when in direct contact with u), as written above, even in clusters otherwise affected by
the assimilation of the voice and aspiration.
In Classical Greek, the old labiovelars are assimilated as other plosives, in a distinctive
way according to the given dialect. In contexts of t/th
/s, the prevailing realization is that of p
though k-variants are often to be met. The p-variant could be a result of a levelling (cf.
Meillet/Vendryes 1924: 59–61; Buck 1932: 129; Sihler 1995: 164). Classical Greek tends to
level one of the outputs of old labiovelars without regard to original contexts, though the
different outputs were originally context-conditioned. Examples are limited to p-output, since
k-output is limited to roots with preceding *u (cf. Hirt 1912: 201–205; Brugmann/Thumb 1913:
137; Meillet/Vendryes 1924: 57–61; Lejeune 1947: 36; Lejeune 1972: 44; Aura Jorro/Adrados
1985: 261–262; Rix 1992: 85–88; Sihler 1995: 156; Bartoněk 2003: 139):
i. Ku̯
+ t > ku̯
t (Mycenaean)
ii. Ku̯
+ t > Pt > pt (Classical Greek)
i. Ku̯
+ dh
> gu̯
dh
→ Ku̯
th
> ku̯ h
th
(Mycenaean)
ii. Ku̯
+ dh
> Pth
> ph
th
(Classical Greek)
i. Ku̯
+ s > ku̯
s (Mycenaean)
ii. Ku̯
+ s > Ps > ps (Classical Greek)
Note: νύξ, νυκτός ‘nightʼ could be an example of the old neutralization Ku̯
+t > kt (cf. Hit. nekuz), but this could
be as well an example of an old Indo-European process, since L. nox, noctis has k through the whole paradigm,
probably due to levelling (Lejeune 1947: 37; Lejeune 1975: 44; Sihler 1995: 230; Beekes 2016: 1027).
7.4.4 Development of the clusters dental + t/dh
/s
The classical trajectory is that of an affricate: the old *T is affricated; later the affricate is
sibilantized, and the cluster of *ss is then simplified. The affrication model was developed for
Indo-European languages by Kräuter (1877: 88)183
and popularized by Brugmann (firstly 1880:
140–142, used since), Brugmann successfully applied it to Greek development and was widely
accepted (cf. Brugmann 1885: 34;184
Brugmann 1890: 57; Brugmann 1900: 96, 99–100; Hirt
183
It is interesting that Kräuter speaks about affrication, but his description of the feature is that of a spirantization!
Verner (1878: 341–342) has a critical evaluation of the idea.
184
In this book, Brugmann uses the older version of the trajectory, assuming insertion of a spirant (tÞt), as he did
in Brugmann (1880: 140–142).
171
1912: 193, 209, 239; Brugmann/Thumb 1913: 112; Buck 1932: 143–145; Sihler 1995: 201–
202, 204).185
T + t > ts
t > tst > st
T + dh
> dz
dh
→ ts
th
> tsth
> sth
T + s > ts
s > tss > ss > 0s186
Note: Rix (1992: 96) simply assumes the (direct?) assibilation of the first plosive.
Another possible trajectory is that of spirantization, assuming the spirantization instead of
affrication, then sibilantization of the spirant (and its subsequent loss before s- due to
simplification). This trajectory is a better solution for the cluster of Ts since it does not require
a loss of t from the cluster tss, which would be easier to simplify to the original ts.
T + t > ϑt > st
T + dh
> δdh
→ tth
> ϑth
> sth
T + s > ϑs > ss > 0s
The status of clusters of *Tdh
in both trajectories is questionable: did the affrication or the
spirantization appear before the transition of *Dh
> Th
or not? Since the change of a dental
before an obstruent is attested in all IE languages (disputable in OIA), we prefer the older
existence of it before the devoicing of old IE voiced aspirate, but technically, we can even
construct trajectories with reversed ordering: the affricate trajectory: T + dh
> ddh
→ T + th
>
ts
th
> tsth
> sth
; the spirantization trajectory: T + dh
> ddh
→ T + th
> ϑth
> sth
.
7.4.5 Development of the clusters sibilant + t/dh
/s
Pre-Greek development is conservative. we can assume for the clusters *ss an (already IndoEuropean)
simplification of the cluster (Meillet/Vendryes 1924: 54). The cluster of sdh
was
revalued due to the loss of voice of the aspirate (the old cluster being zdh
) (cf. Schwyzer 1939:
328; Meillet/Vendryes 1924: 52; Lejeune 1947: 99; Lejeune 1972: 117–118; Sihler 1995: 220):
185
Interestingly, neither Brugmann nor anyone else tries to detail the trajectory for Ts (cf. Brugmann 1885: 39;
Brugmann 1890: 64; Brugmann 1900: 101).
186
The simplification ss > 0s is a standard in Attic, later Ionic, partially in Homeric; other dialects keep ss (cf.
Sihler 1995: 204).
172
s + t > st
s + dh
> zdh
→ s + th
> sth
s + s > 0s
7.5 Conclusion
The development of Greek is remarkably conservative.
The peripheral series are fully preserved in all three contexts (dh
-context being subjected
to the analogical remodelling due to the Pre-Greek loss of voice though). The old labiovelars
are restored before obstruents; only few traces of the old neutralization of the labial value of
the labiovelars can be found in Mycenaean).
The dental series underwent an old fricativization process, ending in the sibilantization
of a plosive in all three contexts. we prefer the spirantization trajectory of the two variants of
the development, since it better explains the loss of a sibilant in the cluster of *Ts: a
simplification of the assumed intermediate *ϑs to ss and later on 0s seems to be a more plausible
trajectory than the simplification of the *ts
s, which could probably only restore the old ts.
Similarly, IE clusters of *sT and ss are preserved (the second degeminated in some of
the dialects). The cluster of *sdh
was remodelled, according to the change IE *dh
> Gr. th
, as
were all other clusters within the dh
-context.
173
8 The development of the two-obstruent clusters from Indo-European into
Italic languages
8.0 Italic languages
The Italic languages include many languages, initially covering the area of Apennine peninsula except Etruria and
Apulia (and Greek settlements on Italian coast). The status of Venetian and Sicel is questionable and beyond the
scope of this study (for a short overview cf. Wallace 2017: 317–319).
There are at least two sub-branches inside the Italic family: Latin (–Faliscan) branch and Sabellic (=
Osco–Umbrian) branch, different in many aspects of our interest (for a complex overview of the differences
between both sub-branches, cf. especially Rix: 2004: 147–172). We will examine both branches independently
before trying to sketch the trajectories of the development of clusters of plosive + t/s.
8.1 Latin language
Latin is one of the most substantial languages for (not only) Comparative Indo-European linguistics, especially
since it is well attested both in its grammar and lexicon.
It represents, almost alone, the whole sub-branch of the Italic languages (but Faliscan, closely related is
poorly attested; cf. Baldi 1999: 123–125, on the Latin–Faliscan relationship cf. Baldi 1999: 170–171; an excellent
modern overview is that by Stuart-Smith 2004: 54–64).
8.1.1 Latin and Indo-European
The typical features separating Latin obstruent system from that of Indo-European are:
i. the preservation of the three modal classes, reconstructed for the Indo-European
protolanguage; the traditionally reconstructed voiced aspirates are realized either as
voiceless fricatives or voiced plosives (see below);
ii. the preservation of the labiovelars, at least voiceless and voiced;
iii. the development of the IE cluster of *Tt into a cluster of ss;
iv. the rhotacization of the intervocalic *s.
The first feature is shared with Indo-Aryan, Armenian, Greek and Germanic, the second with
Germanic, the third with other Italic languages and with Germanic and Celtic languages, the
fourth with other Italic languages and with Germanic.
8.1.2 Latin clusters and their IE origins
The clusters of plosive + t/s are not subjected, as in most Indo-European languages outside the
Indo-Iranian branch, to Bartholomae’s Law (cf. Meiser 1998: 124); hence all clusters, without
regard to the original voice or aspiration are realized as voiceless outcomes of their both
segments.187
Note: A remarkable feature of Latin historical phonology is the lengthening of the vowel preceding a -Ct- cluster
if the left plosive is a voiced unaspirated, known as Lachmann’s Law (first formulated by Lachmann 1850;
for the overviews of the literature, see Collinge 1985:105–114; Sukač 2013: 52–87). The examples on the law
187
On the possible origin of suffixes -dh
ro-/-dh
lo- (vs -tro-/-tlo-) due to Bartholomaeʼs Law, cf. Sihler (1995:
200–201).
174
are: legit vs lēctus ‘readʼ; edit vs ēsus ‘eatʼ; agit vs āctus ‘drive, actʼ; fragit vs frāctus ‘breakʼ. Lachmann’s
Law was linked to Winter’s Law and repeatedly interpreted and rejected (the existence of such a law was
rejected especially by Kent 1928, who considers such outcomes a result of analogy processes, in many aspects
similarly to de Saussure 1885). From other possible solutions we have to mention the glottalic explanation (cf.
Baldi 1991; Kortland 1989; 1999; Schrijver 1991: 134–138; but rejected by Meiser 1998: 79–80); there are
other different solutions based either on phonemic or morphemic analogy (most relevant are: Osthoff 1884:
112–113; Maniet 1956; Kuryłowicz 1968; Watkins 1968; Drinka 1991; Sihler 1995: 75–76; Jasanoff 2004).
Besides Latin, three modal classes are directly attested not only for Sabellic but also for Greek,
Armenian, and Germanic, while traces of the triad can be found in Balto-Slavic (Winter’s Law)
etc. Old Indo-Aryan preserved not only three old modal classes but also had a fourth. Such a
preservation is more the matter of the diachronic analysis than the synchronic reality of Latin
or in other words: not present but traceable. A characteristic feature of the Latin development
of the IE voiced aspirates is the often split of old voiced aspirates into context-given outcomes,
typically in the anlaut on the one hand and in the inlaut on the other (for development of IE
plosives into Latin cf. Meiser 1998: 97–104; Weiss 2009: 73–79, 149–150). However, in
contexts of our interest, the processes and their outcomes are regular, which enable us to treat
old IE voiced aspirates as a single modal class in our analysis.
Since Latin is a language with a large corpus of data and of the old grammarian tradition,
we can focus especially on the productive examples of ‘activeʼ clusters, preferring verbal
derivation and flexion (typically supines, passive particles perfecti, sigmatic perfects etc.), the
non-productive (‘etymologicalʼ) examples being used more to illustrate the common IE
examples of the clustering, notably numerals.
8.1.2.1 The clusters labial + t/s
The IE cluster of *Pt is realized without any changes as L. pt, but as 0t in the word-initial:
P + t = L. pt:
sup. aptum, pr. aptō, ppp. aptus (cf. pr. apō, -ere, apiō ‘fasten, bindʼ; < IE *√H1ep-; cf.
Hitt. ēpzi ‘take, grabʼ, YAv. āpa ‘reachʼ; cf. Lewis/Short 1879: 137–139; WH II: 57–
58; Pokorny IEW: 50–51; LIV2
: 237; de Vaan 2008: 47);
sup. cleptum (cf. pr. clepō, -ere ‘stealʼ; < IE *√klep-; cf. Gr. κλέπτω ‘stealʼ, Goth. hlifan
‘stealʼ; cf. Lewis/Short 1879: x; WH I: 232; Pokorny IEW: 604; LIV2
: 363–364; de
Vaan 2008: 120);
sup. serptum (cf. pr. serpō, -ere ‘crawlʼ; < IE *√serp-; cf. OIA sárpati ‘creep, crawlʼ, Gr.
ἕρπω ‘move slowlyʼ; cf. Lewis/Short 1879: 1680–1681; WH II: 524; Pokorny IEW:
912; LIV2
: 536; de Vaan 2008: 558);
sup. nūptum, ppp. nuptus (cf. pr. nūbō, -ere ‘marryʼ; < IE *√sneu̯ bh
-; cf. RuCS snubiti
‘make coupleʼ; cf. Lewis/Short 1879: 1222; WH I: 268; WH II: 183–184; Pokorny
IEW: 977–978; LIV2
: 574; NIL: 499–504; de Vaan 2008: 417–418);
sup. scrīptum, nom. scrīptor (cf. pr. scrībō, -ere ‘writeʼ; < IE *√skre bh
-; cf. ON hrífa
‘scratch, tearʼ, Latv. skrīpât ‘scratch, scribble, write downʼ; cf. Lewis/Short 1879:
175
1647–1648; WH II: 499; Pokorny IEW: 946–947; LIV2
: 562; de Vaan 2008: 546–
547);
num. septem ‘sevenʼ (cf. OIA saptá-, Gr. ἑπτά; cf. Lewis/Short 1879: 1675; WH II: 517–
518; Pokorny IEW: 909; Sihler 1995: 214–216; Coleman 1992: 248; Blažek 1999:
248; de Vaan 2008: 555);
sternuō ‘sneezeʼ (if related to Gr. πτάρνυμαι, πταίρω ‘sneezeʼ, Arm. pʾṙngam, pʾṙngem
‘squeezeʼ, W. ystrew ‘sneezingʼ; < IE *√pster-; cf. Lewis/Short 1879: 1758; WH I:
591; Pokorny IEW: 846–847; LIV2
: 494; de Vaan 2008: 587);
taceō, -ēre ‘be silentʼ (if from IE *√pteH2k- as states LIV2
495; then related to Gr.
πτώσσω ‘shrink fromʼ, Arm. tʿakʿeaw ‘hid himselfʼ, Goth. Þahan ‘to keep secretʼ,
OHG dagēn ‘be silentʼ; cf. Lewis/Short 1879: 1833; WH II: 641–642; Pokorny IEW:
1055; de Vaan 2008: 604–605);
Similarly, the IE cluster *Ps is realized without any changes internally, as 0s word-initially (cf.
Meiser (1998: 113); Weiss (2009: 170):188
P + s = L. ps:
s-pf. clepsī (cf. pr. clepō, -ere ‘stealʼ; < IE *√klep-; cf. Gr. κλέπτω ‘stealʼ, Goth. hlifan
‘stealʼ; cf. Lewis/Short 1879: x; WH I: 232; Pokorny IEW: 604; LIV2
: 363–364; de
Vaan 2008: 120);
s-pf. serpsī (cf. pr. serpō, -ere ‘crawlʼ; < IE *√serp-; cf. OIA sárpati ‘creep, crawlʼ, Gr.
ἕρπω ‘move slowlyʼ; cf. Lewis/Short 1879: 1680–1681; WH II: 524; Pokorny IEW:
912; LIV2
: 536; de Vaan 2008: 558);
s-pf. nūpsī (cf. pr. nūbō, -ere ‘marryʼ; < IE *√sneu̯ bh
-; cf. RuCS snubiti ‘make coupleʼ;
cf. Lewis/Short 1879: 1222; WH I: 268; WH II: 183–184; Pokorny IEW: 977–978;
LIV2
: 574; NIL: 499–504; de Vaan 2008: 417–418);
s-pf. scrīpsī (cf. pr. scrībō, -ere ‘writeʼ; < IE *√skre bh
-; cf. ON hrífa ‘scratch, tearʼ, Latv.
skrīpât ‘scratch, scribble, write downʼ; cf. Lewis/Short 1879: 1647–1648; WH II: 499;
Pokorny IEW: 946–947; LIV2
: 562; de Vaan 2008: 546–547);
nom. sabulum ‘sand, gravelʼ (cf. ON sandr, Gr. ψάμμος; cf. Lewis/Short 1879: 1609;
Stolz 1894: 297; Kent 1932b: 112; WH II: 458; Pokorny IEW: 145–146; Leumann
1977: 186; Meiser 1998: 113; NIL 45–46; de Vaan 2008: 531);
Note: In a limited number of cases, metathesis possibly appeared, cf. L. crispus ‘curlyʼ, MW. Crych, Gallo-L. PN
Crixsus (< IE *krip-so-), cf. Meiser (1998: 127); Weiss (2009: 170), but de Vaan (2008: 145) prefers *cris- (as
in L. crīnis ‘hair of the headʼ, crista ‘crest on the head of animal, plumeʼ).
8.1.2.2 The clusters velar + t/s
The velar cluster *Kt is fully preserved, similarly to the cluster *Ks, at least in the word-internal
position. However, the word-initial cluster *#Ks- results in #0s- only:
K + t = L. kt (<ct>):
sup. ductum, ppp. ductus (cf. pr. dūcō, -ere ‘leadʼ; < *√deu̯k-; cf. OW. -duch ‘lead, bringʼ,
Alb. n-duk ‘pullʼ; cf. Lewis/Short 1879: 615–616; WH I: 377–378; Pokorny IEW:
220–221; LIV2
: 124; de Vaan 2008: 181);
188
This process did not affect later borrowings from Greek.
176
sup. iūnctum, nom. iunctiō (cf. pr. iungō, -ere ‘jointʼ; < *√ eu̯g-; cf. OIA yunájmi ‘yokeʼ,
OCS igo ‘yokeʼ; cf. Lewis/Short 1879: 1017; WH I: 726–730; Pokorny IEW: 508–
510; LIV2
: 316; NIL: 397–404; de Vaan 2008: 314–315);
sup. pictum, ppp. pictum, nom. pictor (cf. pr. pingō, -ere ‘paintʼ; < *√pe g-; cf. OIA pěgъ
‘colourfulʼ; cf. Lewis/Short 1879: 1378; WH II: 305–306; Pokorny IEW: 794; LIV2
:
464; NIL: 546–548; de Vaan 2008: 465–466);189
sup. tectum, nom. tectio, tector (cf. pr. tegō, -ere ‘coverʼ; < IE *√(s)teg-; cf. Gr. στέγω
‘cover, fend, containʼ, Lith. stógas ‘roofʼ; cf. Lewis/Short 1879: 1845–1846; WH II:
654; Pokorny IEW: 1013–1014; LIV2
: 589; NIL: 634–636; de Vaan 2008: 608);
K + s = L. ks (<x>):
s-pf. dūxī, nom. dux (cf. pr. dūcō, -ere ‘leadʼ; < *√deu̯k-; cf. OW. -duch ‘lead, bringʼ, Alb.
n-duk ‘pullʼ; cf. Lewis/Short 1879: 615–616; WH I: 244–245; Pokorny IEW: 220–
221; LIV2: 124; de Vaan 2008: 181);
s-pf. iūnxī (cf. pr. iungō, -ere ‘jointʼ; < *√ eu̯g-; cf. OIA yunájmi ‘yokeʼ, OCS igo ‘yokeʼ;
cf. Lewis/Short 1879: 1017; WH I: 726–730; Pokorny IEW: 508–510; LIV2
: 316; NIL:
546–548; de Vaan 2008: 314–315);
s-pf. pīnxī pictor (cf. pr. pingō, -ere ‘paintʼ; < *√pe g-; cf. OIA pěgъ ‘colourfulʼ; cf.
Lewis/Short 1879: 1378; WH II: 305–306; Pokorny IEW: 794; LIV2
: 464; NIL: 546–
548; de Vaan 2008: 465–466);
pr. texō, s-pf. tēxī (cf. pr. tegō, -ere ‘coverʼ; < IE *√(s)teg-; cf. Gr. στέγω ‘cover, fend,
containʼ, Lith. stógas ‘roofʼ; cf. Lewis/Short 1879: 1845–1846; WH II: 654; Pokorny
IEW: 1013–1014; LIV2
: 589; NIL: 634–636; de Vaan 2008: 608);
sentis ‘thornʼ, sentus ‘shrubbyʼ (< *#√k(e)s-n̥ -ti/to-; cf. Hitt. kiszi ‘combʼ, Gr. ξαίνω
‘scratch, combʼ, MIr. eīr ‘combʼ, OCS češǫ ‘combʼ; cf. Lewis/Short 1879: 1673; Stolz
1894: 297; WH II: 516–517; Kent 1932b: 125; Pokorny IEW: 585; Meiser 1998: 113;
LIV2
: 357);190
8.1.2.3 The clusters palatovelar + t/s
The outcome of the development of the palatovelar + t is same as for *Kt(and *Ku̯
t):
Ḱ + t = L. kt (<ct>):
sup. dictum (cf. pr. dīcō, -ere ‘sayʼ; < IE *√de ḱ-; cf. OIA ádikṣi ‘pointʼ, Gr. δείκνυμι
‘showʼ; cf. Lewis/Short 1879: 570–571; WH I: 348–349; Pokorny IEW: 188–189;
LIV2
: 108–109; de Vaan 2008: 169–170);
sup. spectrum, nom. spectrum (cf. pr. speciō, -ere ‘observeʼ; < IE *√speḱ-; cf. OIA páśyati
‘seeʼ, Gr. σκέπτομαι ‘look about somethingʼ; cf. Lewis/Short 1879: 1737; WH II: 570–
571; Pokorny IEW: 984; LIV2
: 575–576; de Vaan 2008: 578–579);
sup. lēctum (cf. pr. legō, -ere, pf. lēgī ‘choose, readʼ; < IE *√leǵ-; cf. Gr. λέγω ‘pick upʼ;
cf. Lewis/Short 1879: 1047–1048; WH I: 351–352, 397, 780; Pokorny IEW: 658;
LIV2
: 397; de Vaan 2008: 332–333);
sup. rēctum, nom. rector ‘guideʼ (cf. pr. regō, -ere ‘ruleʼ; < IE *√H3reǵ-; cf. OIA rā́ jat
‘prevailʼ, MW. reag ‘stand upʼ; cf. Lewis/Short 1879: 1551–1553; WH I: 273, 415,
429–430; WH II: 426–427; Pokorny IEW: 854–857; LIV2
: 304–305; de Vaan 2008:
517–518);
189
If reconstructed as *√pe k- (cf. LIV2
l.c.), if related to OIA piṁśati “adornʼ, it should be reconstructed as *√pe ḱ(cf.
Pokorny IEW l.c., LIV2
: 464–465).
190
The cluster of #Ks is simplified on #0s-, cf. Meiser (1998: 113); Weiss (2009: 170).
177
sup. fictum, adj. fictus (cf. pr. fingō, -ere ‘moldʼ; < IE *√dh
e ǵh
-; cf. OIA díhanti ‘stack
upʼ, Goth. digan ‘form mouldʼ; cf. Lewis/Short 1879: 750–751; WH I: 501–502;
Pokorny IEW: 244–245; LIV2
: 140–141; NIL: 118–119; de Vaan 2008: 221–222);
sup. tractum, ppp. tractus (cf. pr. trahō, -ere ‘pullʼ; < IE *√dh
reǵh
-; cf. Gr. τρέχω ‘runʼ,
Goth. -dragan ‘carryʼ; cf. Lewis/Short 1879: 1885–1886; WH II: 697–699; Pokorny
IEW: 257, 273; LIV2
: 154; de Vaan 2008: 626–627);
sup. uectum, nom. vector (cf. pr. uehō, -ere ‘carryʼ; < IE *√u̯ eǵh
-; cf. OIA váhati ‘cart,
driveʼ, OCS vezǫ ‘driveʼ; cf. Lewis/Short 1879: 1962; WH II: 741–743; Pokorny IEW:
1118–1120; LIV2
: 661–662; de Vaan 2008: 658);
num. octō ‘eightʼ (< IE *oḱtō; cf. Goth. ahtau, Gr. ὀκτώ; cf. Lewis/Short 1879: 1254–
1255; WH II: 199–200; Pokorny IEW: 775; Coleman 1992: 266; Blažek 1999: 266;
de Vaan 2008: 424–425);
Similarly, the outcome of the development of the palatovelar + s is same as for *Ks (and *Ku̯
s):
Ḱ + s = L. ks (<x>):
s-pf. dīxī (cf. pr. dīcō, -ere ‘sayʼ; < IE *√de ḱ-; cf. OIA ádikṣi ‘pointʼ, Gr. δείκνυμι ‘showʼ;
cf. Lewis/Short 1879: 570–571; WH I: 348–349; Pokorny IEW: 188–189; LIV2
: 108–
109; de Vaan 2008: 169–170);
s-pf. spēxī (cf. pr. speciō, -ere ‘observeʼ; < IE *√speḱ-; cf. OIA páśyati ‘seeʼ, Gr.
σκέπτομαι ‘look about somethingʼ; cf. Lewis/Short 1879: 1737; WH II: 570–571;
Pokorny IEW: 984; LIV2
: 575–576; de Vaan 2008: 578–579);
s-pf. rēxī, nom. rēx ‘kingʼ (cf. pr. regō, -ere ‘ruleʼ; < IE *√H3reǵ-; cf. OIA rā́ jat ‘prevailʼ,
MW reag ‘stand upʼ; cf. Lewis/Short 1879: 1551–1553; WH I: 273, 415, 429–430;
WH II: 426–427; Pokorny IEW: 854–857; LIV2
: 304–305; de Vaan 2008: 517–518);
s-pf. finxī (cf. pr. fingō, -ere ‘moldʼ; < IE *√dh
e ǵh
-; cf. OIA díhanti ‘stack upʼ, Goth.
digan ‘form mouldʼ; cf. Lewis/Short 1879: 750–751; WH I: 501–502; Pokorny IEW:
244–245; LIV2
: 140–141; NIL: 118–119; de Vaan 2008: 221–222);
s-pf. trāxī (cf. pr. trahō, -ere ‘pullʼ; < IE *√dh
reǵh
-; cf. Gr. τρέχω ‘runʼ, Goth. -dragan
‘carryʼ; cf. Lewis/Short 1879: 1885–1886; WH II: 697–699; Pokorny IEW: 257, 273;
LIV2
: 154; de Vaan 2008: 626–627);
s-pf. uēxī (cf. pr. uehō, -ere ‘carryʼ; < IE *√u̯ eǵh
-; cf. OIA váhati, OCS vežǫ; cf.
Lewis/Short 1879: 1962; WH II: 741–743; Pokorny IEW: 1118–1120; LIV2
: 661–662;
de Vaan 2008: 658);
num. sex ‘sixʼ (< IE *seḱs; cf. Goth. saihs, Gr. ἕξ; cf. Lewis/Short 1879: 1687; WH II:
528–529; Pokorny IEW: 1044; Coleman 1992: 237; Blažek 1999: 237; de Vaan 2008:
560);
Note: L. uōx is an example of the neutralization of the labiovelar before -s, cf. OIA vā́k, Gr. ὄψ < IE *u̯ ōku̯
-s. This
neutralization was extended, by the analogy, to other cases of the paradigm (de Vaan 2008: 691–692); it even
became a base for new derivations: uocāre, uocālis, uocābulum, etc.
8.1.2.4 The clusters labiovelar + t/s
Labiovelars lose their labial component when clustering with t or s:
Ku̯
+ t = L. kt (<ct>):
sup. coctum, adj. coctus, nom. coctor (cf. pr. coquō, -ere ‘cookʼ; < *√peku̯
-; cf. OIA
pacati, OCS pečǫ ‘cookʼ; cf. Lewis/Short 1879: 468; WH I: 270–271; WH II: 338;
Pokorny IEW: 798; LIV2
: 468–469; NIL: 548–552; de Vaan 2008: 134);
178
sup. -lictum (cf. pr. -linquō, -ere ‘abandonʼ; < *√le ku̯
-; cf. OIA riṇákti ‘leaveʼ, Gr. λείπω
‘leave, quitʼ; cf. Lewis/Short 1879: 1557–1558; WH I: 808–809; Pokorny IEW: 669–
670; LIV2
: 406–408; de Vaan 2008: 344);
nom. sector (cf. pr. sequor ‘followʼ; < *√seku̯
-; cf. OIA sácate ‘followʼ, Gr. ἕπομαι
‘followʼ; cf. Lewis/Short 1879: 1677–1678; WH II: 519; Pokorny IEW: 896–897;
LIV2
: 526–527; de Vaan 2008: 555–556);
sup. -stīnctum (cf. pr. -stinguō, -ere ‘quenchʼ; < *√stengu̯
-; cf. Goth. stigqn ‘meet, adjoinʼ;
cf. Lewis/Short 1879: 1760; WH I: 706–707; Pokorny IEW: 1016–1017; LIV2
: 596–
597; de Vaan 2008: 588);
sup. ūnctum (cf. pr. unguō, -ere ‘anointʼ; < *√H2engu̯
-; cf. OIA anákti ‘anoint, smearʼ,
Arm. awcanem ‘anointʼ; cf. Lewis/Short 1879: 1931; WH II: 819–820; Pokorny IEW:
779; LIV2
: 267; de Vaan 2008: 641–642);
pr. nictō, nictor, nom. nictus (cf. pr. cōnīueō, -ere ‘close the eyesʼ; < *√kne gu̯h
-; cf. OHG
nīgan ‘bow, inclinedʼ; cf. Lewis/Short 1879: 1206; WH I: 260; Pokorny IEW: 608;
LIV2
: 410–411; Sihler 1995: 163; de Vaan 2008: 130), 410–411);
num. quīnctus, quīntus ‘fifthʼ (< Italic *ku̯
enku̯
to- < IE *pn̥ ku̯
to-; cf. OIA pakthá-, Gr.
πέμπτος; cf. Lewis/Short 1879: 1515; WH II: 407–408; Pokorny IEW: 808; Coleman
1992: 222; Blažek 1999: 221). De Vaan (2008: 509) assumes for quīntus the earlier
spirantization and later loss of the spirant (-nkt- > -nxt- > -nt-), similarly to the
spirantization of velars in Sabellic.
Ku̯
+ s = L. ks (<x>):
s-pf. cōxī (cf. pr. coquō, -ere ‘cookʼ; < *√peku̯
-; cf. OIA pacati, OCS pečǫ ‘cookʼ; cf.
Lewis/Short 1879: 468; WH I: 270–271; WH II: 338; Pokorny IEW: 798; LIV2
: 468–
469; NIL: 548–552; de Vaan 2008: 134);
s-pf. -stīnxī (cf. pr. -stinguō, -ere ‘quenchʼ; < *√stengu̯
-; cf. Goth. stigqn ‘meet, adjoinʼ;
cf. Lewis/Short 1879: 1760; WH I: 706–707; Pokorny IEW: 1016–1017; LIV2
: 596–
597; de Vaan 2008: 588);
s-pf. ūnxī (cf. pr. unguō, -ere ‘anointʼ; < *√H2engu̯
-; cf. OIA anákti ‘anoint, smearʼ, Arm.
awcanem ‘anointʼ; cf. Lewis/Short 1879: 1931; WH II: 819–820; Pokorny IEW: 779;
LIV2
: 267; de Vaan 2008: 641–642);
s-pf. cō-nīxī (cf. pr. cōnīueō, -ere ‘close the eyesʼ; < *√kne gu̯h
-; cf. OHG nīgan ‘bow,
inclinedʼ; cf. Lewis/Short 1879: 1206; WH I: 260; Pokorny IEW: 608; LIV2
: 410–411;
Sihler 1995: 163; de Vaan 2008: 130);
s-pf. ninxit, nom. nix (cf. pr. ninguō, ninguit, -ere, nom. g. sg. nivis ‘snowʼ; < *√sne gu̯h
;
cf. OIr. snigid ‘snowʼ, Goth. snaiws ‘snowʼ, OCS sněgъ ‘snowʼ; cf. Lewis/Short
1879: 1208; WH II: 169–170; Pokorny IEW: 974; LIV2
: 573; NIL 622–625; de Vaan
2008: 409–410);
Note: L. pr. uīuō, -ere, s-pf. uīxī, sup. uīcturus ‘liveʼ, though from IE *√gu̯
eH3- (cf. Gr. βίος, OIA jīva-, OCS. živъ
‘lifeʼ), seems to be translated from a u̯ -final to a labiovelar final and fits subsequently into the alternation
pattern (cf. de Vaan 2008: 686).
8.1.2.5 The clusters dental + t/s
The cluster *Tt is wholly sibilantized as ss:
T + t = L. (s)s:
sup. messum, nom. messis (cf. pr. metō, -ere ‘mow, reapʼ; < IE *√met-; cf. W. medi ‘mow,
harvestʼ, OCS metǫ, mesti ‘throw, sweepʼ; cf. Lewis/Short 1879: 1140; WH II: 82–83;
Pokorny IEW: 703–704; LIV2
: 442; de Vaan 2008: 377–378);
179
sup. sēnsum (cf. pr. sentiō, -īre ‘feelʼ; < IE *√sent-; cf. OCS sȩštъ ‘sensible, wiseʼ; cf.
Lewis/Short 1879: 1672–1678; WH II: 515–516; Pokorny IEW: 908; LIV2
: 533; de
Vaan 2008: 554);
sup. ēsum, inf. esse (besides edere), nom. ēsor, ppp. ēsus (cf. pr. ēdō, -ere ‘eatʼ; < IE
*√H1ed-; cf. OIA átti ‘eatʼ, OLith. edmi, esti ‘eatʼ; cf. Lewis/Short 1879: 626; WH I:
392–393; Pokorny IEW: 287–289; LIV2
: 230–231; NIL: 208–210; de Vaan 2008:
185–186);
sup. fūsum, nom. fūsiō (cf. pr. fundō, -ere ‘pourʼ; < IE *√ǵh
eu̯ d-; cf. OIA juhóti ‘pour,
sacrificeʼ, Goth. guitan ‘pourʼ; cf. Lewis/Short 1879: 792–793; WH I: 563–564;
Pokorny IEW: 447–448; LIV2
: 179–180; de Vaan 2008: 249–250);
sup. lūsum, ppp. lūsus (cf. pr. lūdō, -ere ‘playʼ; < IE *√le d-; cf. Gr. λίζει ‘playʼ, OIA
lédmi ‘playʼ; cf. Lewis/Short 1879: 1083; WH I: 829; Pokorny IEW: 666; LIV2
: 402–
403; de Vaan 2008: 350–351);
sup. suāsum, nom. suāsor (cf. pr. suādeō, -ēre ‘advice, recommendʼ; < IE *√su̯ eH2d-; cf.
OIA svádant ‘make savouryʼ, Gr. ἡδύς ‘pleasantʼ; cf. Lewis/Short 1879: 1770–1771;
WH II: 483, 611–612; Pokorny IEW: 1039–1040; LIV2
: 606–607; NIL: 670–672; de
Vaan 2008: 594);
sup. trūsum, ppp. trūsus (cf. pr. trūdō, -ere ‘thrust, pushʼ; < IE *√treu̯ d-; cf. OCS trudъ
‘labour, workʼ, OHG -driozan ‘cause sorrowʼ; cf. Lewis/Short 1879: 1905; WH II:
710; Pokorny IEW: 1095–1096; LIV2
: 651–652; de Vaan 2008: 630);
ppp. fīsus sum (cf. pr. fīdō, -ere ‘trustʼ; < IE *√bh
e dh
-; cf. Gr. πείθω ‘persuadeʼ, Alb. bē,
besë ‘faithʼ; cf. Lewis/Short 1879: 747–748; WH I: 493–494; Pokorny IEW: 117;
LIV2
: 71; NIL: 12–13; de Vaan 2008: 218–219);
sup. iussum (cf. pr. iubeō, -ēre ‘commandʼ; < IE *√H eu̯ dh
-; cf. yúdhyati ‘fightʼ; cf.
Lewis/Short 1879: 1014; WH I: 724–725; Pokorny IEW: 511–512; LIV2
: 225–226; de
Vaan 2008: 312–313);
In reconstructed clusters of *dh
t, the outcome is st attested in two examples: aestus ‘heat,
fervorʼ, aestās ‘summer (heat)ʼ (< *H2e dh
-) (Stolz 1894: 326; Sihler 1995: 203; Meiser 1998:
127; Hill 2003: 243–247; de Vaan 2008: 28). The outcome is irregular, hence minor, and
definitely not related to Bartholomae’s Law (which, if operative in Pre-Italic, would yield a
voiced outcome).
The same outcome is regular (since it is attested for all clusters of the same structure)
for clusters of *Ttr > str. Examples are: rōstrum ‘snout, beakʼ (cf. rōdō ‘grawʼ with s-pf rōsī
and ppp. rōsum, both regular; cf. WH II: 439–440; Pokorny IEW: 854; Leumann 1977: 190;
Meiser 1998: 124; de Vaan 2008: 526); similarly fūstis ‘stick, rodʼ (cf. -fūtāre ‘strikeʼ; cf. OE
bēatan, OHG bozan ‘strikeʼ; cf. WH I: 259–260, 573; Pokorny IEW: 112; Hill 2003: 229–238;
de Vaan 2008: 253) and cæstus ‘strip of leatherʼ (related to caedō ‘cut, hew, fellʼ; cf. WH I: 44,
129, 690; Pokorny IEW: 917; Untermann 2000: 364; Hill 2003: 229–238; de Vaan 2008: 79–
80).
Note: The old process of sibilantization does not affect the (later) clusters resulting from verbal prefix + verbal
root: attineō ‘holdʼ (= ad-teneō), cf. Sihler (1995: 203), Baldi (1999: 293).
180
T + s = L. (s)s:
s-pf. messuī (cf. pr. metō, -ere ‘mow, reapʼ; < IE *√met-; cf. W. medi ‘mow, harvestʼ,
OCS metǫ, mesti ‘throw, sweepʼ; cf. Lewis/Short 1879: 1140; WH II: 82–83; Pokorny
IEW: 703–704; LIV2
: 442; de Vaan 2008: 377–378);
s-pf. sēnsī (cf. pr. sentiō, -īre ‘feelʼ; < IE *√sent-; cf. OCS sȩštъ ‘sensible, wiseʼ; cf.
Lewis/Short 1879: 1672–1678; WH II: 515–516; Pokorny IEW: 908; LIV2
: 533; NIL:
208–220; de Vaan 2008: 554);
s-pf. lūsī (cf. pr. lūdō, -ere ‘playʼ; < IE *√le d-; cf. Gr. λίζει ‘playʼ, OIA lédmi ‘playʼ; cf.
Lewis/Short 1879: 1083; WH I: 829; Pokorny IEW: 666; LIV2
: 402–403; de Vaan
2008: 350–351);
s-pf. suāsī (cf. pr. suādeō, -ēre ‘advice, recommendʼ; < IE *√su̯ eH2d-; cf. OIA svádant
‘make savouryʼ, Gr. ἡδύς ‘pleasantʼ; cf. Lewis/Short 1879: 1770–1771; WH II: 483,
611–612; Pokorny IEW: 1039–1040; LIV2
: 606–607; NIL: 670–672; de Vaan 2008:
594);
s-pf. trūsī (cf. pr. trūdō, -ere ‘thrust, pushʼ; < IE *√treu̯ d-; cf. OCS trudъ ‘labour, workʼ,
OHG -driozan ‘cause sorrowʼ; cf. Lewis/Short 1879: 1905; WH II: 710; Pokorny IEW:
1095–1096; LIV2
: 651–652; de Vaan 2008: 630);
s-pf. iussī (cf. pr. iubeō, -ēre ‘commandʼ; < IE *√H eu̯ dh
-; cf. OIA yúdhyati ‘fightʼ; cf.
Lewis/Short 1879: 1014; WH I: 724–725; Pokorny IEW: 511–512; LIV2
: 225–226; de
Vaan 2008: 312–313);
s-pf. uāsī (cf. pr. uādō, -ere ‘goʼ; < IE *√u̯ eH2dh
-; cf. ON vaða ‘wadeʼ; cf. Lewis/Short
1879: 1951; WH II: 723–724; Pokorny IEW: 1109; LIV2
: 664; de Vaan 2008: 650);
Note: The process even afflicts secondary clusters like possum ‘canʼ (< *potis sum), cf. Meiser (1998: 116).
8.1.2.6 The clusters sibilant + t/s
Old Indo-European cluster *st is fully preserved; Indo-European cluster *ss is either preserved
or simplified (usually according to rhythmicity rule):
s + t = L. st:
nom. castus ‘pureʼ (cf. pr. careō, -ēre ‘lackʼ; < IE *√ḱes-; cf. OIA śāsti ‘orderʼ; cf.
Lewis/Short 1879: 291–292, 299; WH I: 167, 178; Pokorny IEW: 586; LIV2
: 329; de
Vaan 2008: 92–93);
pr. est (cf. pr. sum ‘be ’; < IE *√H1es-; cf. OIA ásti, OLith. esti, OCS jestъ ‘beʼ; cf.
Lewis/Short 1879: 1797–1800; WH I: 263, 420; WH II: 628–629; Pokorny IEW: 340–
341; LIV2
: 241–242; NIL: 235–238; de Vaan 2008: 599);
sup. pistum (beside pīnsum, pīnsitum), nom. pistor ‘bakerʼ (cf. pr. pīnsō, -ere ‘crushʼ; <
IE *√pe s-; cf. OIA pináṣṭi ‘crush, grindʼ, RuCS pъchati ‘thrust, sproutʼ; cf.
Lewis/Short 1879: 1878–1880; WH II: 302, 307–308; Pokorny IEW: 796; LIV2
: 466;
de Vaan 2008: 466–467);
sup. haustum (cf. pr. hauriō, -īre ‘drawʼ; < IE *√H2eu̯ s-; cf. Gr. αὔω ‘get a lightʼ, ON
ausa ‘scoopʼ; cf. Lewis/Short 1879: 842–843; WH I: 637; Pokorny IEW: 90; LIV2
:
275–276; de Vaan 2008: 281);191
pr. stō, sistō ( ‘stand ’; < IE *√steH2-; cf. OIA ásthāt ‘standʼ, Lith. stóti ‘standʼ; cf.
Lewis/Short 1879: 1711–1712, 1762–1763; WH II: 584, 587, 597–598, 632; Pokorny
IEW: 1004–1010; LIV2
: 590–592; NIL: 637–659; de Vaan 2008: 567, 589–590);
191
The initial h- in Latin is probably hypercorrect, l.c.
181
s + s = L. (0)s:
pr. esi (= *es-si; cf. pr. sum ‘be ’; < IE *√H1es-; cf. OIA ásti, OLith. esti, OCS jestъ ‘beʼ;
cf. Lewis/Short 1879: 1797–1800; WH I: 263, 420; WH II: 628–629; Pokorny IEW:
340–341; LIV2
: 241–242; NIL: 235–238; de Vaan 2008: 599);
s-pf. pīnsī (cf. pr. pīnsō, -ere ‘crushʼ; < IE *√pe s-; cf. OIA pináṣṭi ‘crush, grindʼ, RuCS
pъchati ‘thrust, sproutʼ; cf. Lewis/Short 1879: 1878–1880; WH II: 302, 307–308;
Pokorny IEW: 796; LIV2
: 466; de Vaan 2008: 466–467);
s-pf. hausī (cf. pr. hauriō, -īre ‘drawʼ; < IE *√H2eu̯ s-; cf. Gr. αὔω ‘get a lightʼ, ON ausa
‘scoopʼ; cf. Lewis/Short 1879: 842–843; WH I: 637; Pokorny IEW: 90; LIV2
: 275–
276; de Vaan 2008: 281);
8.1.2.7 Overview of Latin development192
Latin development has two innovations: the first is the Common Indo-European transformation
of the dentals in clusters of *Tt and *Ts; the second is old neutralization of labiovelars in both
contexts:
IE Latin t- s-
-kṷ
/gṷ
/gṷh
ku̯
/gu̯
(u̯ )/u̯ kt ks
-k/g/gh
k/g/h kt ks
-ḱ/ǵ/ǵh
k/g/h kt ks
-t/d/dh
t/d/d (b, f) ss ss
-p/b/bh
p/b/f (b) pt ps
-s s/r st ss
8.2 Sabellic languages
The Sabellic branch of Italic languages, though once widely used than the Latin-Faliscan branch, is worse attested
than its counterpart. In this paper, data will be used from Oscan and Umbrian. Texts are attested either in native
alphabets (adapted to a given language, principally similar to Etruscan and closely related to it) or in Greek
ἀλφάβητος (only for Oscan), later Latin script was used for both languages (Baldi 1999: 129–132; 136–140; StuartSmith
2004: 78–79, 100–101; Pocetti 2017: 739).
Since both Oscan and Umbrian are relict languages (Oscan is attested in about 200 documents, Umbrian
is attested in small glosses and especially in Tabulae Iguvinae), we have to use a minimal set of language data,
though we can cover the main tendencies in the development of clusters of plosive + t/s (and s + t/s) even with
attested data, albeit often very thinly (on the relationship between Oscan and Umbrian, cf. Baldi 1999: 174–176).
8.2.1 Sabellic and Indo-European
The typical features separating the Sabellic obstruent system from that of Indo-European are
(listed are only those relevant for our field of interest):
i. the loss of labiovelars (in contrast to Latin) in all positions including plosive + t/s (for this
more below), intervocalic in the way of merging with pure labials, cf. Os. pís vs L. quis
‘whoʼ, Os. nom. pl. bivus vs L. vīvus, Um. vufru ‘votiveʼ (< IE *u̯ogu̯
-ro-) (Meiser 2017:
749);
192
Valid only for the internal clusters.
182
ii. the transformation of old IE voiced aspirated plosives into voiceless spirants (Meiser 2017:
744);
iii. sibilantization of clusters of Tt as ss;
iv. spirantization/lenition of plosives before t/s;
v. degemination of ss (especially for metrical reasons).
Note: Though not the proper ‘consonantal developmentʼ, we should mention vowel syncope (Meiser 2017: 748),
producing secondary clusters of different origins and (often even outcomes) that result from ‘old clustersʼ.
Some examples we will mention below.
The first feature is shared with Ancient Greek (but not with Mycenaean!) and with P-Celtic
languages. The second feature is attested partially in Latin and Middle Greek. The third feature
is shared with Latin, Celtic, and Germanic languages, while the spirantization of plosives is
attested not only in Celtic and Germanic but even in Iranian, etc.
8.2.2 Sabellic clusters and their IE origins
There is no distinction between plain velars and palatovelars in all Italic languages (as there is
none in all centum-languages), but we will distinguish both series for better Indo-European
contextualisation. Beside old alternations, there are numerous secondary clusters, arising due
to the syncope of vowels, often with a different outcome than those of the primary clusters. This
distinction between primary and secondary clusters is very distinguishable.
8.2.2.1 The clusters labial + t/s
In Oscan, labiality is preserved. In Umbrian the labial fricative is delabialized:
P + t = (Sab. *φt), Os. ft, Um. (h)t):193
Os. scriftas, Um. screhto, screihtor ‘writtenʼ (cf. L. scrīptus; < IE *√skre bh
-; cf. ON hrífa
‘scratch, tearʼ, Latv. skrīpât ‘scratch, scribble, write downʼ; cf. von Planta 1892: 425;
Buck 1904: 78; Pokorny IEW: 946–947;Meiser 1986: 92; LIV2
: 562; Stuart-Smith
2004: 80, 113; de Vaan 2008: 546–547; Meiser 2017: 749);
Os. ufteis, uhftis ‘voluntatis’ (cf. L. optiō; < IE *√H3ep- (?); cf. Hitt. epp-zi
app- ‘take,
grabʼ; cf. Buck 1904: 78; Pokorny IEW: 781; LIV2
: 299; Stuart-Smith 2004: 95; de
Vaan 2008: 431–432);
Um. setums PN ‘Septimusʼ (cf. L. Septimus; < IE *septm-; cf. OIA saptá- ‘sevenʼ, Gr.
ἑπτά ‘sevenʼ; Pokorny IEW: 909; Coleman 1992: 248; Sihler 1995: 214–216; Blažek
1999: 248; de Vaan 2008: 555; Meiser 2017: 749);
Note: The secondary (having arisen due to syncopation) clusters in Umbrian seem to follow the same trajectory:
Um. hahtu, hatu, hatu ‘capitoʼ (cf. Os. hipid; Buck 1904: 78; Untermann 2000: 316).
P + s = Sab. (s)s:
193
This seems to be extended to secondary clusters, too: Um. hahtu, hatu “capitoʼ (cf. L. habeō) (Buck 1904: 78).
183
Os. osiin[ns-, Um. úpsim ‘(lit.) to be againstʼ (cf. L. ob-sint; < IE *√H1es-; cf. OIA ásti,
OLith. esti, OCS jestъ ‘beʼ; cf. von Planta 1892: 427; Buck 1904: 78; Pokorny IEW:
340–341; Untermann 2000: 248; LIV2
: 241–242; de Vaan 2008: 599);
Os. essuf, esuf, Um. esuf ‘himself; thereʼ (cf. L. ipse; PItal. < *eps(o)-ōn-s < IE *soso; cf.
von Planta 1892: 427; Buck 1904: 79; Pokorny IEW: 281–286; Untermann 2000: 235–
236; de Vaan 2008: 308);
Um. ostendu ‘should set upʼ (cf. L. ostendere ‘show, revealʼ; < IE *√(s)tend-; cf. OIA
tanóti ‘stretchʼ, Gr. τείνω ‘stretch, pull tightʼ, Goth. ufþanjan ‘extendʼ; cf. Pokorny
IEW: 1065–1066; Meiser 1986: 169; Untermann 2000: 812–814; LIV2
: 626–628; de
Vaan 2008: 612–613);
Note: The secondary (having arisen due to syncopation) clusters in Oscan are not subdued to this alternation, but
those in Umbrian are: Os. upsed, úpsannam (cf. L. fecit), Um. osatu, oseto (from *opesā-, cf. L. operor)
(Buck 1904: 79; Meiser 1986: 169, 173).
8.2.2.2 The clusters velar + t/s
The IE cluster *Kt is realized as ht (the h is often omitted). The outcomes are mixed with those
of IE *Ḱt (see below):
K + t = Sab. ht:
Os. saahtúm, Um. sahta, satam, sahatam ‘sanctified, holyʼ (cf. L. sanctum ‘holyʼ; < IE
*√sH2nk-i-; cf. Hitt. šāklāi ‘custom, ritesʼ, Celtiberian Sancilistara ‘money-fineʼ (?);
cf. Buck 1904: 89; Pokorny IEW: 878; Untermann 2000: 640–643; Stuart-Smith 2004:
95; de Vaan 2008: 532);
Um. uhtur ‘(a title of an official?)ʼ (cf. L. auctor ‘seller, authoritative personʼ; < IE
*√H2eu̯g-; cf. OIA ukṣáti ‘increaseʼ, Lith. áugu ‘growʼ; cf. Buck 1904: 89; Meiser
1986: 92; Pokorny IEW: 86–87; LIV2
: 84–85; Untermann 2000: 788–789; NIL: 328–
332; de Vaan 2008: 61–62);
Um. ahtisper, ahtimen ‘actʼ( cf. L. actiō ‘activityʼ; < *√H2eg-; cf. OIA ájati ‘driveʼ, Gr.
ἄγω ‘drive, lead, goʼ; cf. Pokorny IEW: 4–6; LIV2
: 255–256; Untermann 2000: 65–66;
NIL: 267–277; de Vaan 2008: 30-31).
Note: The secondary clusters arising due syncopation are not affected by this process, cf. Um. fiktu ‘formʼ (cf. L.
fingō; < IE *dh
e ǵh
-e/o-; cf. WH I: 501–502; Pokorny IEW: 244–245; LIV2
: 140–141; Stuart-Smith 2004: 112;
; de Vaan 2008: 221–222; Meiser 2017: 750). The secondary cluster from syncopated syllables could be
realized in Umbrian as it, but not in Oscan: cf. Um. aitu, aitu, Os. actud (cf. L. agit); Os. fac Um. feitu, fetu,
feetu (cf. L. factum). In such a case, the lenition process has outcome either a palatal approximant or zero due
to the elision (cf. Buck 1904: 89).
There are no solid examples for the development of the IE cluster of Ks, though we assume the
outcomes are same as for the IE cluster Ḱs (see below):
K + s = Sab. 0s:
not attested
Note: The k is sometimes restored due to analogy: Os. μεδδειξ, medixud = meddíss, meddis ‘(L.) meddix
(magistrate)ʼ (from *med-de ḱs; von Planta 1892: 376; Buck 1904: 91; Untermann 2000: 456–459; de Vaan
2008: 169–170).
184
8.2.2.3 The clusters palatovelar + t/s
Old IE palatovelars merged fully with their plain velar counterparts. The cluster of Ḱt is then
realized in the same way as Kt.
Ḱ + t = Sab. ht:
Os. ehtrad ‘outsideʼ (cf. L. extra; < IE *H1eǵh
-s-; cf. OIr. ess- ‘outʼ, Gr. ἐξ ‘fromʼ, OCS iz
‘outʼ; cf. Buck 1904: 89; Pokorny IEW: 292–293; Meiser 1986: 92; Untermann 2000:
202–203; Meiser 2017: 749; de Vaan 2008: 195–196);
Os. Úhtavis PN (cf. L. Octāvius; < IE *oḱto-; cf. Goth. ahtau, Gr. ὀκτώ; cf. von Planta
1892: 351; Buck 1904: 89; Pokorny IEW: 775; Meiser 1986: 92; Coleman 1992: 266;
Blažek 1999: 266; Stuart-Smith 2004: 95; de Vaan 2008: 424–425);
Um. speturie ‘spectoriae (augural)ʼ (cf. L speciō, -ere ‘observeʼ; < IE *√speḱ-; cf. OIA
páśyati ‘seeʼ, Gr. σκέπτομαι ‘look about somethingʼ; cf. Pokorny IEW: 984; LIV2
:
575–576; de Vaan 2008: 578–579; Meiser 2017: 749);
Um. rehte ‘rightʼ (cf. L. rēctā ‘directlyʼ; < IE *√H3reǵ-; cf. OIA rā́ jat ‘prevailʼ, MW reag
‘stand upʼ; cf. von Planta 1892: 352; Buck 1904: 89; Pokorny IEW: 854–857; Meiser
1986: 92; LIV2
: 304–305; Untermann 2000: 633; de Vaan 2008: 517–518);
The IE cluster Ḱs merged with the Ks cluster, as in all centum-languages, and as far as we can
judge from the lack of data for Ks (see above). The outcomes for Ḱs are:
Ḱ + s = Sab. (h)s:
Os. sehsimbrịịs ‘born in the sixth monthʼ, sehsík[ ‘?ʼ (cf. L. sextārius ‘measure of onesixthʼ;
< IE *√su̯eḱs-; cf. Gr. ἕξ, Goth. saihs ‘sixʼ; cf. Buck 1905: 91; Pokorny IEW:
1044; Coleman 1992: 237; Blažek 1999: 237; Untermann 2000: 91; de Vaan 2008:
560);
Os. destrst (⇔ L. ‘dextra estʼ), Um. desua, dersua ‘rightʼ, destrame ‘in the rightʼ (cf. L.
dexter ‘rightʼ; < IE *deḱs-; cf. OIA dákṣina- ‘rightʼ, Gr. δεξιά ‘right handʼ; cf. von
Planta 1892: 376; Buck 1904: 91; Pokorny 1959: 189–191; Untermann 2000: 169–
170; de Vaan 2008: 168);
Note: Os. sehsimbrịịs still preserves the older cluster hs, Os. destrst represents -*ḱs-tr-.
8.2.2.4 The clusters labiovelar + t/s
The development of old labiovelars before t is based on few etymologies. It seems that
labiovelars were delabialized before t as in Latin, as far as we can judge from very scarce
examples, but surprisingly, the outcoming velar is not spirantized as in case of Kt/Ḱt, but in all
examples quoted below, the clusters could be secondary, due to syncopation (cf. Buck 1904:
80; Meiser 1986: 179). Regardless, if secondary, the cluster-realization of the original
labiovelar (otherwise fully labialized) as a velar is remarkable, being a result of an independent
neutralization.
However, we lack reliable enough data to determine the outcome of IE Ku̯
t data.
185
Ku̯
+ t = Sab. kt (secondary?):
Um. fiktu ‘figitoʼ (< *fīku̯
tōd; cf. L. fīgō; < IE *dh
e Hgu̯
-; cf. Lith. díegti ‘stingʼ, Toch. B
tsäkaṃ ‘biteʼ (?); cf. Buck 1904: 95; Pokorny IEW: 243–244; Meiser 1986: 82;
Untermann 2000: 284; LIV2
: 142; de Vaan 2008: 219);
Um. ninctu ‘ninguitoʼ194
(< *ninku̯
tōd; cf. L. ninguit ‘snowʼ; < *√sne gu̯h
-; cf. OIr. snigid
‘snowʼ, Goth. snaiws ‘snowʼ, OCS sněgъ ‘snowʼ; cf. Buck 1904: 95; Pokorny IEW:
974; Meiser 1986: 84–86; Untermann 2000: 497–498; LIV2
: 573; de Vaan 2008: 409–
410);
Os. Púntiis PN ‘Quintiusʼ (but Os. Пομπτιες, pomtis), Um. nom. pl. puntes ‘quinionesʼ
(i.e., a group of five priests) (< PSab. *ponkto- ‘fiveʼ < Italic *ku̯
enku̯
to-; < IE *penku̯
te-;
cf. OIA pakthá-, Gr. πέμπτος; cf. Buck 1904: 95; Pokorny IEW: 808; Meiser 1986:
89; Coleman 1992: 222; Blažek 1999: 221–222; Untermann 2000: 608). De Vaan
(2008: 509) assumes for quīntus the earlier spirantization and later loss of the spirant
(-nkt- > -nxt- > -nt-), similarly to the spirantization of velars in Sabellic. Similar
development is valid in the case of Um. anstintu ‘distinguitoʼ (< *-stinku̯
tōd; Buck
1904: 95; Meiser 1986: 82; Untermann 2000: 106).
The velar is lost due to the position inside the cluster. Similarly, the intermediate labiovelar (or
more properly, its outcome) was lost in Um. umtu ‘anointʼ (< *H3engu̯
etōd; Meiser 1986: 80;
Untermann 2000: 797–798). Such examples could represent an old neutralization Ku̯
t >kt > ht
> 0t.
Note: An atypical outcome like Os. aftiím ‘unknown psyche/body part (?)ʼ (< *√H3egu̯
- ‘seeʼ ?; Meiser 1986: 90–
91; Untermann 2000: 60) could be a result of a levelling (and a regular spirantization); abovementioned Os.
Пομπτιες (to Os. Púntiis, pomtis) could be the same case (π could represent a spirant since in Greek φ had still
the value of /ph
/).
For the development of the cluster of Ku̯
s, we have at our disposal only Um. suboco ‘invocationʼ
(acc. sg.) and Um. subocau ‘callʼ (both from *u̯ok-s), analogically extended to other forms (as
in Latin), though the cluster is *ku̯
s in its origin (Meiser 1986: 90; Untermann 2000: 707–708).
As in the case of IE Ku̯
t, we lack data to demonstrate the Sabellic outcomes for IE Ku̯
s but we
can assume the delabialization of the cluster, later probably spirantized and lenited as the
clusters Ks and Ḱs.
8.2.2.5 The clusters dental + t/s
As in other IE languages (Latin, Germanic, Celtic), the IE cluster of dental plosive + t were
transformed into ss:
T + t = Sab. (s)s:
194
Surprisingly without a loss of the intermediate phoneme, cf. below on Um. anstintu, Os. Púntiis, pomtis, Um.
puntes, Um. umtu).
186
Os. Ϝερσορει ‛*Versori’ (epithet of Iuppiter = advertor; cf. Um. trahurfi ‘placed acrossʼ,
L. versus; < IE *√u̯ert-to-; cf. OIA vṛttá- ‘turnʼ, Pruss. wīrst ‘becomeʼ ; cf. von Planta
1892: 419; Buck 1904: 86; Pokorny IEW: 1156–1158; Untermann 2000: 844–845;
LIV2
: 691–692; de Vaan 2008: 666–667);
Um. sesust ‘sederitʼ (cf. L. sessus; < IE *√sed-t-; cf. OIA sattá- ‘sitʼ, OCS sěděti ‘sitʼ; cf.
von Planta 1896: 335; Buck 1904: 86; Pokorny IEW: 884–887; Untermann 2000: 680–
681; LIV2
: 513–515; NIL: 590–600; de Vaan 2008: 551–552)195
;
Um. frosetom ‘fraudatum’ (cf. OL. fraussus; < IE *√dh
reu̯-t/dh
- (?); cf. OIA dhrúti‘deception,
errorʼ; cf. Buck 1904: 86; Pokorny IEW: 277; LIV2
: 156; Untermann 2000:
300-301; de Vaan 2008: 240);
Os. castrous, Um. kastruvuf, castruo ‘(fenced) fieldʼ (?) (cf. L. castrum ‘fortʼ, castrō
‘castrateʼ; < IE *√ḱes-; cf. OIA śástra- ‘knifeʼ, Gr; cf. Buck 1904: 86; Pokorny IEW:
586; Untermann 2000: 374–375; LIV2
: 329–330; de Vaan 2008: 97–98);196
There are the secondary clusters, arising due to syncope, of two dentals, with the pattern Tt >
0t: Um. titu, tetu, ditu ‘should giveʼ (< PSab. *déδatōd < IE *di-dH3-tod); preuendu ‘advertitoʼ
(< PSab. *pra -u̯endetōd < IE *√u̯endh
-) (cf. Meiser 1986: 180).
Similarly, the IE clusters of Ts are realized as 0s:
T + s = Sab. (s)s:
Um. revestu ‘check’ (= *re-u̯e d-s-e-tōd; cf. L. re-vīsere ‘visitʼ; < IE *√u̯ei̯d-s-; cf. OIA
vittá- ‘findʼ, Lith. véizdi ‘look forʼ; cf. von Planta 1892: 390; Buck 1904: 85; Pokorny
IEW: 1125–1127; Untermann 2000: 634–635, 854–855; LIV2
: 665–667; de Vaan
2008: 676);
Um. Fise ‘deo Fidio’, Os. Fiísíais ‘*Fisiis’ (cf. L. fīsus ‘trustʼ; < IE *√bh
ei̯dh
-so-, but often
being considered from *√bh
ei̯dh
-t-; cf. von Planta 1892: 419; Buck 1904: 85; Pokorny
IEW:117; Untermann 2000: 286; LIV2
: 71–72; Stuart-Smith 2004: 104, 113; de Vaan
2008: 218–219);
Note: The secondary clusters of Ts, arising due to syncope, are realized as z in the native alphabet but as s in Latin
alphabet, cf. Os. húrz (cf. L. hortus), Um. taçez, tases (cf. L. tacitus), Os. puz, pous, Um. puze, puse ‘utʼ (<
*put-s) (von Planta 1892: 391; Buck 1904: 86; Meiser 1983: 173).
8.2.2.6 The clusters sibilant + t/s
The original cluster st is preserved in all cases:
s + t = Sab. st:
Os. púst, Um. post ‘after, behindʼ (cf. L. post; < IE *pos-ti; cf. Gr. Arc.-Cypr. πός ‘at, toʼ,
Cz. pozdě ‘lateʼ; cf. Buck 1904: 73; Pokorny IEW: 841; Untermann 2000: 618–624;
de Vaan 2008: 483);
Os. staít, Um. stahu ‘standʼ (cf. L. stō ‘stand ’; < IE *√steH2-; cf. OIA ásthāt ‘standʼ, Lith.
stóti ‘standʼ; cf. Buck 1904: 73; Pokorny IEW: 1004–1010; Untermann 2000: 697–
700; LIV2
: 590–592; NIL: 634–636; de Vaan 2008: 567, 589–590);
195
But von Planta (1892: 390) considers it to be from *set-s-!
196
This is an example for the development of the cluster *Ttr, resulting, as in Latin, in str.
187
Os. est, Um. est, est ‘beʼ (L. est; < IE *√H1es-; cf. OIA ásti, OLith. esti, OCS jestъ ‘beʼ;
cf. von Planta 1892: 473; Buck 1904: 73; Pokorny IEW: 340–341; Untermann 2000:
245–247; LIV2
: 241–242; NIL: 235–238; de Vaan 2008: 599);
The following single example of the development of the cluster of *ss could be interpreted as
a simplification of such a cluster as 0s in Sabellic, assuming that Um. nom.sg. meřs/mers, is a
s-stem followed by an -as of a nominative ending (i.e., -s-s-):
s + s = Sab. (s)s (?):197
Um. nom. sg. meřs, mers ‘lawʼ (s-stem, cf. d.-abl. pl. mersus; von Planta 1896: 71; Buck
1904: 130; Untermann 2000: 461–462).
8.2.2.7 Overview of Sabellic development
In the following table, only the primary clusters are listed. Note the spirantization of peripheral
series. The sibilantization of dental clusters is shared with Latin, Celtic and Germanic
languages:
IE Sabellic t- s-
-kṷ
/gṷ
/gṷh
-p/b/f [ht] [(s)s]
-k/g/gh
-k/g/h ht [(s)s]
-ḱ/ǵ/ǵh
-k/g/h ht (s)s
-t/d/dh
-t/d/f (s)s (s)s
-p/b/bh
-p/b/f ft/ht (s)s
-s -s st (s)s
8.3 Trajectories of the Italic development
The development in both branches remarkably differs in the development of both peripheral
series: Latin data show a remarkably conservative development of the peripheral series clusters,
where old clusters are preserved. On the other hand, Sabellic data show a progressive
development of spirantization/lenition, with this split between the closely related sub-branches
mirroring that of Indo-Iranian. The labiovelar series has special development, which is
neutralized as plain velars in Latin, while the Sabellic data are inconclusive.
Both sub-branches share the same development of the dental series, which has
undergone a typical Indo-European development, here sibilantization of both clusters of *Tt
and *Ts (as in Sabellic, Germanic, Celtic). Outcoming ss (of different origins) are often
simplified as 0s again in both sub-branches due to the moraic leveling.
197
However, the simplification of two sibilant clusters on 0s is known even from the secondary clusters resulting
from IE *Ts.
188
8.3.1 Development of clusters labial + t/s
For the development of the labial clusters with t-, Latin is once again a conservative language,
preserving both plosives. The Sabellic development is more complex, different even for Oscan
on the one side and Umbrian on the other, though we trace both developments to a common
Sabellic source. First, the Pt cluster was spirantized, and the spirant is preserved as the
labiodental spirant in Oscan (cf. Os. scriftas, but Um. screihtor, cf. L. scrīptus). In Umbrian,
the spirant was delabialized and later debuccalized (the outcoming h is sometimes elided):
i. P + t > pt (Latin)
ii. P + t > φt > ft (Oscan)
iii. P + t > φt > xt > ht (Umbrian)
Note: The Italic cluster of #pt- has another development in Latin: #pt- > st- (cf. L. sternuō ‘sneezeʼ, related to Gr.
πταρ-μός, πταίρω, Arm. pʾṙngam, pʾṙngem; Stolz 1894: 297; Meiser 1998: 113; de Vaan 2008: 587), but note that
in this case the cluster of*#pst- is usually reconstructed (Schrijver 1995: 454; LIV2
494–495). The L. taceō, -ēre
‘be silentʼ is reconstructed to be from IE *√pteH2k- (LIV2
495; related to Gr. πτώσσω ‘shrink fromʼ, Arm. tʿakʿeaw
‘hid himselfʼ, Goth. Þahan ‘to keep secretʼ, OHG dagēn ‘be silentʼ (but against this etymology cf. de Vaan 2008:
604–605; Untermann 2000: 731–332 reconstructs from *√tak(H1)-).
The IE cluster *Ps is fully preserved in Latin when word-internal, but spirantized, debuccalized
(and often simplified) in Sabellic:
i. P + s > ps (Latin)
ii. P + s > φs > hs > 0s (Sabellic)
Note: The cluster #ps- has another development: #ps- > 0s- (cf. L. sabulum ‘sand, gravelʼ), otherwise known from
Sabellic. For this analogy we presume that the initial labial was also spirantized, debuccalized and elided: #ps>
#φs- > #hs. > #0s- (cf. Weiss 2009: 170, who does not give details on the trajectory).
The development attested in L. crispus ‘curlyʼ, where *ps > L. sp is an example of an unproductive cluster,
where the metathesis of a sibilant (for the articulatory reasons) is attested, as in L. vespa ‘waspʼ (cf. Prus.
wobse, Œ wæfs, CS *(v)osь (Weiss 2009: 170).
Note: As with the cluster of Ks, there is an alternative trajectory for the Sabellic cluster, with a sibilantization
instead of debuccalization: Ps > φs > ss > 0s.
8.3.2 Development of the clusters velar + t/s
The developments of the cluster *Kt are remarkably different : Latin preserved an original
cluster, Sabellic clusters underwent a spirantization, followed by the debuccalization, the
outcoming h is sometimes elided):
i. K + t > kt (Latin)
ii. K + t > xt > ht (/> 0t) (Sabellic)
Note: The velar (of any origin) is lost in clusters of Rks (R = any liquid): L. fortis ‘strong, robustʼ (< OL. forctis);
L. ultus ‘revengeʼ (< *ulkto-, cf. ulcīscor) (Meiser 1998: 123; de Vaan 2008: 236–237, 363–637).
189
Similarly, the IE cluster of *Ks is realized in Latin as ks in the inlaut (cf. Meiser 1998: 116;
Weiss 2009: 170). For the Sabellic development of the cluster of Ks, we can model the trajectory
based on spirantization in the first phase, the spirant being later debuccalized and elided:
i. K + s > ks (Latin)
ii. K + s > xs > hs > 0s (Sabellic)
Note: But for the word-initial cluster Ks in Latin we have to state a more complex development, similar to that of
Sabellic, with a spirantization, debuccalization and elision: #Ks- > #xs- > #hs- > #0s-; cf. L. sentis ‘thornʼ,
sentus ‘shrubbyʼ (< *ksn̥ -ti/to-; cf. Gr. ξαίνω; Meiser 1998: 113).
Note: The alternative trajectory for the Sabellic languages differs in its middle phase, the spirant sibilantized and
later elided: Ks > xs > ss > 0s.
8.3.3 Development of the clusters palatovelar + t/s
The palatovelar IE clusters of *Ḱt and *Ḱs are realized in Latin in precisely the same way as IE
clusters of *Kt and *Ks (and *Ku̯
t, *Ku̯
s, cf. above), i.e, as kt and ks (cf. Meiser 1998: 116, 124–
125; Weiss 2009: 170–171), the trajectories are hence the same. The same is valid for Sabellic
languages.
8.3.4 Development of the clusters labiovelar + t/s
The development of the labiovelar series differs remarkably in both sub-branches: Latin
preserved them, while in Sabellic they were lost, merged with labials.
Note: The labiovelar is regularly neutralized before labial vowels o/u (Sommer 1914: 187; Meiser 1998: 124–125;
Baldi 1999: 278), cf. L. colō ‘inhabitʼ, OIA cárati ‘moveʼ, Gr. πέλομαι, Doric τέλομαι ‘becomeʼ < IE *√ku̯ elH1(de
Vaan 2008: 125); L. oculus ‘eyeʼ, Gr. ὤψ, OCS oko < IE *H3eKu̯ - (de Vaan 2009: 425).
The Latin labiovelars are neutralized before t-, which is probably an ancient Indo-European
feature (cf. Meiser 1998: 116, 124–125). The development is hence the same as with plain
velars (and assumed Indo-European palatovelars). The Sabellic development could not be
reconstructed: the attested outcomes kt are, with the highest probability, secondary, since they
are not spirantized. In clusters of -Nku̯
t- the labiovelar is lost regularly, hence the proposed
trajectory is more constructed than reconstructed:
i. Ku̯
+ t > kt (Latin)
ii. Ku̯
+ t > kt > xt > ht (Sabellic)
Similar ancient neutralization of a labiovelar is attested for the IE cluster of *Ku̯
s in Latin (cf.
Meiser 1998: 116, 124–125; Weiss 2009: 170). Again, we lack enough Sabellic data to
reconstruct the proper trajectory.
190
i. Ku̯
+ s > s (Latin)
ii. Ku̯
+ s > ks > xs > hs > (s)s (?) (Sabellic)
8.3.5 Development of clusters dental + t/s
The outcomes of the Italic development, in general, exclude any possibility of considering ss
an outcome of intermediate stage st (as in Balto-Slavic or Iranian), since the outcome of IE *st
is preserved as Latin st (see below). Hence, we have to reconstruct a trajectory not coming
though an st-stage at all. The outcome ss is the same as for the IE clusters *Ts and (as much as
we dare say from scarce data) for IE *ss.
The traditional trajectory assumed for the development of IE clusters of *Tt was
broadened especially under Brugmann’s influence (first Brugmann 1880: 140–142, used since),
though initially, the idea was by Kräuter (1877: 88).
This model assumes the affrication of the
left plosive, followed by dissimilation (the affrication trajectory), merging at the same point
with the development of clusters of ss/Ts. For Latin, it was proposed by Brugmann (1885: 183;
1890: 305); the affrication trajectory is already used by Stolz (1894: 315); later Leumann (1977:
197), Meiser (1998: 124), Görtzen (1998: 386–390), Baldi (1999: 287) or Weiss (2008: 173),
for the Indo-European context cf. especially Szemerényi (1996: 103).
i. T + t > ts
t > tss > (s)s (Italic)
Note: Meisser (1986: 36; 1998: 123–124) reconstructs: Tt > ts
t > ts > ss; Ts > ss (accepted by Kümmel 2007:
376).
Inside the affrication trajectory, the minor development Tt(r) to L. st(r) could be modelled as:
TT(r) > ts
t(r) > st(r), with t- restored, i.e. not within a sound law (von Planta 1892: 419–424;
Buck 1904: 86–87; Leumann 1977: 197–198; Sihler 1995: 201–203; Meisser 1998: 124–125;
Hill 2003: 226; Weiss 2009: 174), sinc ethe first element of the presumed affricate is lost.
Note: It is worth of notion that clusters of plosive + s + plosive are often simplified as s + plosive, as attested with:
suscipiō ‘take upʼ (subs-capiō), sustineō ‘supportʼ (subs-teneō), though otherwise the cluster is preserved:
subscrībō ‘write beneathʼ (Baldi 1999: 297). Examples of the loss of the first plosive in a cluster could be
counter-examples against the affrication trajectory. Clear examples of the loss of the internal syllable but the
sibilantization of the final dental are given by Stoltz (1894: 317), Leumann (1977: 197–198, 203), Sihler
(1995: 198) or Meiser (1998: 117): L. lustrō ‘make lightʼ, illūstris (= in-lou̯k-stri-) ‘illustriousʼ vs lūceō ‘shineʼ
(with examples for clusters with final p or k, Meiser (1998: 117) states that textus and extra are results of the
analogical restoration, not the old outcomes).198
198
As a counter-example could serve nom. dexter “rightʼ, PN Sextius, not speaking about prefixed words (cf.
Leumann 1977: 202–204; Sihler 1995: 198–199).
191
A different variant of the affrication model was proposed by Kent (1932a: 23; 1932b: 117–
118), who proposes that the second dental plosive (of the suffix) was lost and the cluster became
both merged with *Ts and subsequently sibilantized with it:
i. T + t > ts
t > tss > (s)s (Italic)
We propose an alternative trajectory (the spirantization trajectory), based on assumed
spirantization, instead of affrication of the left dental, and later spirantization of the whole
cluster and its sibilantization. The first to propose such a strategy for Latin was de Saussure
(1877: 375)199
, followed by Cocchia (1883: 16–58) and Bartholomae (1887: 83), critically to
this cf. Brugmann (1885: 183); Walde (1897: 487–492):
i. T + t > ϑt > ϑϑ > (s)s (Italic)
Within the spirantization trajectory, the aforementioned cases of the development*Tt(r) > st(r)
could be regularly modelled as: Ttr > ϑtr > str, with a spirant sibilantized regularly, analogically
for *Tt > ϑt > st as a minor development, which is phonetically entirely acceptable. It is a
process parallel to the above-mentioned regular major development in the development of the
first plosive, the second being preserved, since it is not spirantized before r (but cf. Hill 2003:
247; who attributes the different distribution of ss/st to the syllabic structure of a given word).
Similarly, the development of the cluster of *Ts can follow either the affrication trajectory, with
the affrication and simplification of the cluster, as with all such geminate clusters of various
origins, it could be degeminated, especially after a long vowel (cf. Meiser 1998: 116; Baldi
1999: 287):
i. T + s > ts
s > tss > ts > (s)s (Italic)
A variant model by Kent (1932a: 23; 1932b: 117–118) fits within the proposed general features.
Furthermore, the alternative spirantization trajectory differs in the first stages by a spirantization
first, followed by a sibilantization of the spirant, later often degeminated:
i. T + s > ϑs > (s)s (Italic)
199
De Saussure speaks of the development of -d+t- but there are no reasons do not apply it on all dental clusters.
192
8.3.6 Development of the clusters sibilant + t/s
The original IE clusters st are preserved both in Latin and in Sabellic; hence the trajectory is
straightforward:
i. S + t > st (Italic)
For the IE cluster *ss we also assume a simplification due to the degemination as 0s for all Italic
languages. In the case of the L. pr. es ‘thou artʼ the simplification could be considered to be
already Indo-European (cf. Kent 1932b: 127; Weiss 2009: 171). All attested clusters ss are of a
later development: L. esse, essem etc. are morphemic restorations given by analogy. The later
sequence ss of any origin is either preserved or degeminated according to different phonemic
conditions, especially if following a long vowel (cf. Kent 1932b: 131–132; Baldi 1999: 287):
i. S + s > (s)s (Italic)
8.4 Conclusion and final remarks
Confronting the Sabellic and Latin data we see that the development of the clusters with dentals
and clusters with sibilants are shared (not only within Italic but often outside, in the wider IE
family), while the development of the peripheral series remarkably differ inside the family,
Sabellic having the progressive, Latin the conservative attitude.
The dental clusters *Tt and *Ts are uniformly realized as Italic ss (as in Germanic and
Celtic). Of both possible fricativization strategies, we prefer that of spirantization (as proposed
by de Saussure and Cocchia) over affricativization (as developed by Kräuter and Brugmann)
for the following reason: the affricativization trajectory of the *Ts cluster assumes the
intermediate ts
s, which is very improbable to be simplified as ss, but the intermediate ϑs cluster
could be easily assimilated to ss. For the development of the *Tt clusters, both trajectories are
similarly probable. The spirantization trajectory can easily explain the development of the
clusters of *Ttr, resulting in Italic str: if the first plosive were affricativized, the structure of the
cluster would be even more complex (tstr!), but within the spirantization trajectory the
development is simpler: the left plosive is fricativized, the second is preserved as plosive before
r, and the first is later regularly sibilantized (ϑtr >str).
The Sabellic languages used the spirantization/lenition trajectory even for the
development of the peripheral series. That this gradual strategy is working is clear from
Umbrian, where the Oscan ft is realized as ht, i.e., after the debuccalization of the spirant as a
laryngeal approximant. Latin (besides the old Indo-European neutralization of the labial value
193
of the labiovelars in both examined contexts) is without any process of this type (Italian later
followed the trajectory of the gemination with the same clusters).
The cluster st is fully preserved; the cluster ss is usually degeminated, usually
contextually driven (after long vowels, due to consonantal context forming a moraic length,
etc.).
195
9 The development of the two-obstruent clusters from Indo-European into
Celtic
9.0 Celtic languages and their branches
Within the Celtic language group at least three different sub-branches are distinguished:
i. the Continental languages (Gaulish, Celtiberian, Galatian etc.), attested only as relict inscriptions, quotations
and proper names; all dead about 500 AD;
ii. Brythonic languages (Welsh, Cornish, Breton), with huge documentation since their Middle phase and (except
Cornish) living languages until our days; all descending from Brittonic language, diverging since 500 AD;
iii. Goidelic languages (Irish, Scottish Gaelic, Manx), also with numerous documentation since its Middle phase
and living languages (except Manx), all were descending from Irish Gaelic about 1000 AD.
Note: The second and third group are often considered sub-branches of Insular Celtic.
On the classification and mutual relationship between Celtic sub-branches, cf. Blažek 2009; Vath/Ziegler 2017:
1169–1170; MacAulay 1992: 1–8; McCone 1996: 67–104; Schijver 2007; Sims-Williams 2007; Sims-Williams
2017: 352.
Note: It was proposed many times (and also many times rejected) that Celtic languages form a branch inside the
broader Italo-Celtic branch. From the point of the examined material, such idea is acceptable, but this specific
and wider question (since affecting more fields of interests than just clusters of plosives and sibilants) is left
aside (for further reading, cf. Watkins 1966; Schmidt 1991 as examples of the ʽnegativeʼ approach, Kortlandt
1981; Eska 2010; Weiss 2012; as examples of the ʽpositiveʼ approach).
9.1 Celtic and Indo-European
The typical features separating the Celtic obstruent system from that of Indo-European are:
i. the typical Common Celtic loss of IE *p;
ii. the later loss of labiovelars, which are replaced either by labials (P-Celtic) or a velar (Q-
Celtic);
iii. the merging of IE voiced plosives with voiced aspirated plosives200
;
iv. the sibilantization of *Tt clusters;
v. spirantization of plosives in clusters of peripheral plosive + t/s-;
vi. the tendency of s either became h or disappear;
vii. the extensive processes of lenition and mutations (especially in Insular Celtic languages).
The first feature is a unique Celtic phenomenon (Armenian and Germanic p were subjected to
a shift, but we cannot speak about a shift in Celtic); labiovelars are otherwise preserved in Latin
(but not in other Italic languages) and Germanic, but the merging with labials is known from
Sabellic, and Greek (beside merging with dentals or velars). The Continental languages and
oldest attested forms of Goidelic have labiovelars preserved. The merging of the second and
the third modal class are known from Baltic, Slavic, Iranian languages, so the Celtic process is
an example of parallel development (a drift). The sibilantization of the IE *Tt is a common IE
200
Interesting feature of this development is IE *gu̯
> PCelt. b vs. IE *gu̯ h
> PCelt. gu̯
(Cowgill 1980; Sims-Williams
1981; Sims-Williams 2007: 128; Sims-Williams 2017: 363 but contrary view has McCone 1996: 38–42, who
argues for the merging of both IE phonemes before the delabialization of labiovelars).
196
process (Old Indo-Aryan being the particular case), and the outcome ss is attested even in both
sub-branches of Italic and in Germanic (another case of a drift or an area development?). The
spirantization of the peripheral plosives in clusters is known from Iranian, Sabellic and Slavic;
the Celtic development is specific in merging of all fricatives in a single one, either Brythonic
or Goidelic x: the Continental fricative is also x, which represents the original state. The loss
of s is known from Iranian and Greek, and again such processes are examples of independent
drifts, not of a dialect continuum.
Both living sub-branches underwent deep and numerous phonemic and morphemic
development; hence we will limit ourselves to etymological examples. Both living subbranches
are dealt with independently; the Continental languages will be represented by Gaulish
(examples taken mainly from epigraphic sources and literary quotations either in Latin or
Greek).
9.2 Brythonic clusters and their IE origins
A special feature of the Brythonic development, related to our objects of study, is the merge of
old labiovelars with labials (as in Sabellic). Common Celtic x (of different origins) > (cf. SimsWilliams
2017: 365).
9.2.1 The clusters labial + t/s
Similarly to all peripheral plosives in the contexts t/s-, the outcome of the labial plosive is a
palatal approximant; the outcome of *Ps is x:
P + t = Bryth. t:
MW. seith, W. saith; OCorn. syth, MCorn. seyth, OBr. seith, MBr. seiz (< PCelt. *sextam
< IE *septem; cf. OIA saptá-, L. septem ʽsevenʼ; cf. Pokorny IEW: 909; GPC III: 3170;
Greene 1992: 510, 515, 540; Blažek 1999: 248; Deshayes 2003: 648; Matasović 2009:
332);
MW. nith, Corn. noith, OBr. nith, MBr. nyz ʽnieceʼ (PCelt. *neftī; < IE *neptiH2; cf. L.
neptis, OHG nift ʽnieceʼ; cf. Pokorny IEW: 764; GPC III: 2584; Deshayes 2003: 538;
NIL 520–524; Matasović 2009: 286);
W. caeth ʽbond, captiveʼ, OCorn. caid ʽcaptivusʼ, MBr. quaez, NBr. keaz ʽunlucky, poorʼ
(< PCelt. *kaxto-; < IE *√keH2p-; cf. L. captus, ON haptr ʽcaptusʼ; cf. Pokorny IEW:
527; GPC I: 384–385; LIV2
: 344–345; Deshayes 2003: 583; Matasović 2009: 197);
P + s = Bryth. x:
MW. crych, MBr. crech ʽcurlyʼ (if < PCelt. *kaxto- with a metathesis < IE *kris-po-; cf.
L. crispus, crispō ʽcurlyʼ; cf. Pedersen 1913 I: 75; Lewis/Pedersen 1937: 19; Pokorny
IEW: 937–938; GPC I: 619; Deshayes 2003: 429; Matasović 2009: 226);
MW. uch, Corn. ugh, OBr. uh, Br. ucʼh ʽabove, overʼ (< PCelt. *ouxso < IE *H2eu̯ p-so;
cf. Gr. ὕψῐ), W. uchel, Corn. huhel, OBr. uchel, uhel ʽhighʼ (< PCelt.*ouxselo- < IE *
197
H2eu̯ pselo-; cf. Pedersen 1909: 75; Lewis/Pedersen 1937: 19; Pokorny IEW: 1107;
GPC IV: 3692–3693; Deshayes 2003: 755; Matasović 2009: 303);
9.2.2 The clusters velar+ t/s
Plain velar plosives are lenited as before t-, the cluster of *Ks is realized as x:
K + t = Bryth. t:
MW. mwyth ʽluxury, ease, pleasureʼ (< PCelt. *muxto-; < IE *√meu̯ k-t-; cf. L. mūcor
ʽmouldʼ, Gr. μύξᾰ ʽmucusʼ; cf. Pokorny IEW: 744–745; de Bernardo Stempel 1999:
438; GPC III: 2525; LIV2
: 443–444; Matasović 2009: 282);
K + s = Bryth. x:
MW. trech ʽstrongerʼ, Corn. tragh ʽvictoriousʼ, MBr. vrech ʽvictoryʼ (< PCelt. *trex-so;
< IE *treg-; cf. OE Þragjan ʽcourageʼ, ON Þrekr ʽstrengthʼ; cf. Pokorny IEW: 1090;
Schrijver 1995: 136; de Bernardo Stempel 1999: 252, 258, 389; GPC IV: 3571;
Deshayes 2003: 739; Matasović 2009: 389–390);
9. 2.3 The clusters palatovelar + t/s
Since there is no distinction between assumed IE plain and palatovelar plosives, Celtic being a
centum-language, the outcome of *Ḱt is equal to that of *Kt and *Ḱs to that of *Ks:
Ḱ + t = Bryth. t:
OW. oith, MW. wyth, Corn. eath, OBr. eith, MBr. eiz ʽeightʼ (< PCelt. *oxtū < IE *oḱtō;
cf. L. octō, Goth. ahtau; cf. Lewis/Pedersen 1937: 41; Pokorny IEW: 775; GPC III:
3746; Greene 1992: 510–511, 540; Blažek 1999: 266; Schrijver 1995: 350; Deshayes
2003: 212; Matasović 2009: 304);
MW. amaeth ʽploughman, tillerʼ, OBr. ambaith (< PCelt. *ambi-aktos; < IE *√H2eǵ-; cf.
Pokorny IEW: 4; LIV2
: 255–256; Schumacher 2007: 189–193; NIL 267–267;
Matasović 2009: 32);
OW. rhaith, MW. reyth ʽlaw, sermon, jury, verdictʼ, Br. reiz ʽorder, law, rightʼ (< PCelt.
*rextu- < IE *√H3reǵ-; cf. L. rēctus, Goth. raihts; cf. Lewis/Pedersen 1937: 42;
Pokorny IEW: 854–857; GPC III: 3033; Bernardo Stempel 1999: 291; LIV2
: 304;
Schumacher 2007: 530–534; Deshayes 2003: 619; Matasović 2009: 310–311);
MW. gweith ʽwork, act; time, -timesʼ, OCorn. gueid, MCorn. gweth, gwyth ʽ-timesʼ, OBr.
gueid ʽtimeʼ, MBr. gwez, gweach ʽ-timesʼ (< PCelt. *u̯ extā; < IE *√u̯ eǵh
-; cf. L. vector,
vectis ʽcarryʼ, OCS vezti ʽcarryʼ; cf. Pokorny IEW: 1118–1120; GPC II: 1563; LIV2
:
661–662; Deshayes 2003: 309; Matasović 2009: 419–420);
MW. lled-brith ʽcharmʼ, OBr. brith ʽincantationʼ (< PCelt. *brixtu/o-; < IE *√bh
erǵh
-; cf.
OIA brahmán-, ON bragr; cf. Pokorny IEW: 139; Delamarre 2003: 90; NIL 30–34;
Matasović 2009: 79–80);201
Ḱ + s = Bryth. x:
MW. echel, MBr. ahel ʽaxisʼ (< PCelt. *axsilā; < IE *H2eḱs-; cf. OIA ákṣa-, L. axis ʽaxleʼ;
cf. Lewis/Pedersen 1937: 20; Pokorny IEW: 6; GPC I: 1160; Deshayes 2003: 53; NIL
259–262; Matasović 2009: 50);
201
The connection to the reconstructed IE root is disputed (cf. l.c.).
198
MW. deheu, dehau ʽto the right, to the southʼ, OCorn. dehow, dyghow, OBret. dehou (<
PCelt. *dexs(o)u̯ o- < IE *deḱsu̯ o; cf. Gr. Gr. δεξῐτερός ʽon the right hand or sideʼ, L.
dexter; cf. Lewis/Pedersen 1937: 20; Pokorny IEW: 190; GPC I: 999; Deshayes 2003:
174; Matasović 2009: 97);
MW. chwe(ch), Corn. whe(gh), MBr. huech, Br. cʼhouecʼh ʽsixʼ (< PCelt. *su̯ exs < IE
*su̯ eḱs; cf. Gr. ἕξ, L. sex ʽsixʼ; cf. Lewis/Pedersen 1937: 20; Pokorny IEW: 1044; GPC
I: 864; Greene 1992: 510–511, 515, 539–540; Blažek 1999: 237; Deshayes 2003: 164;
Delamarre 2003: 285–286; Matasović 2009: 364);
MW. ech, eh, OBret. ech ʽout of, fromʼ (< PCelt. *exs- < IE *H1esǵs-; cf. L. ex, Lith. iš;
cf. Lewis/Pedersen 1937: 20; Pokorny IEW: 292; Pokorny 1969: 24; GPC I: 1160;
Delamarre 2003: 169; Matasović 2009: 119);
MW. nych ʽpainʽ (< PCelt. nexso- < IE *neḱ-; cf. OIA naṣṭá- ʽbe lostʼ, L. necō ʽkill, sayʼ;
cf. Pokorny IEW: 762; GPC I: 49; GPC III: 2602; de Bernardo Stempel 1999: 97;
Deshayes 2003: 65, 534; Matasović 2009: 37–38, 39, 290);
9.2.4 The clusters labiovelar + t/s
In Brythonic, the old IE labiovelars have merged with labials in all contexts; hence all outcomes
are equal to those of labiovelars in the same contexts. In fact, due to the development of plosives
before t-/s-, the outcomes are the same for all peripheral plosives:
Ku̯
+ t = Bryth. i̯t:
OW. he-noid ʽtonightʼ, MW. peu-noeth ʽevery nightʼ, Corn. haneth, MBr. hanoez
ʽtonightʼ, (PCelt. *noxt- < IE *noku̯
-t- ; cf. L. nox, noctis, Goth. nahts; cf.
Lewis/Pedersen 1937: 41; Pokorny IEW: 762–763; GPC III: 2790; LIV2
: 449; de
Bernardo Stempel 1999: 36; Delamarre 2003 : 237; NIL 504–513; Matasović 2009:
293–294);
MW. gwaethl ʽdisputeʼ (< PCelt. *u̯ oxtlo- ; < IE *u̯ oku̯
-; cf. OIA vā́ k-, L. uōx ʽvoiceʼ; cf.
Pokorny IEW: 1135–1136; GPC II: 1552; de Bernardo Stempel 1999: 229; LIV2
673–
674; Matasović 2009: 428–429);
OW. noid, MW. noeth, Corn. noyth, noeth, OBr. noit, MBr. noaz ʽnakedʼ (< PCelt.
*noxto- < IE *nogu̯
to-; cf. Goth. naqaÞs, OHG nackut, nachut ʽnakedʼ; cf. Pokorny
IEW: 769; GPC III: 2592; de Bernardo Stempel 1999: 440; Deshayes 2003: 539; NIL
513–515; Matasović 2009: 294) ;
Ku̯
+ s = Bryth. x:
W. ych ʽoxʼ (< PCelt. *uxso- < IE *√u̯ egu̯
-; cf. OIA ukṣan- ʽbullʼ, OHG ohso ʽoxʼ; cf.
Lewis/Pedersen 1937: 20; Pokorny IEW: 1118; GPC III: 3749; LIV2
602–603;
Deshayes 2003: 544; NIL 368–370; Matasović 2009: 400–401);
MW. techu, Corn. têgh, MBr. tec´hed ʽfleeʼ (< IE *√teku̯
-; cf. OIA takti ʽshoots awayʼ,
OCS těchъ ʽrunʼ; cf. Pedersen 1913: 639; Pokorny IEW: 1059–1160; LIV2
620–621;
Deshayes 2003 : 717; Schumacher 2007: 629–631; Matasović 2009: 377);
9.2.5 The clusters dental + t/s
In both sub-branches of Celtic languages, as in Italic and Germanic, the outcome of the IE
cluster *Tt is ss, often simplified to 0s, and the same outcome is valid for IE cluster *Ts:
199
T + t = Bryth. 0s:
W. ys ʽhe eatsʼ (< PCelt. *ed-ti < IE *√H1ed-; cf. OIA átti; cf. Pokorny IEW: 287; LIV2
:
230–321; GPC IV: 3821; Schumacher 2007: 377–380; Matasović 2009: 113);
W. gwys ʽknowledgeʼ, MBr. gous ʽwould be knownʼ (< IE *u̯ id-to-s; cf. L. uīsus ʽseenʼ,
OIA vitti- ʽknowʼ; cf. Jackson 1953: 531; Pokorny IEW: 1125–1127; GPC II: 1745,
1752; Schrijver 1995: 404; LIV2
: 665–667; Hill 2003: 257; Schumacher 2007: 664–
669, 690–701; NIL 717–722; Matasović 2009: 407–408; 422–423);
MW. moes ʽcustom, habitʼ, MBr. boas (< PCelt. *banssu-; < IE *√bh
endh
-; cf. OIA
bándhati ʽbindʼ, Goth. bindan ʽbindʼ; cf. Pokorny IEW: 127; GPC III: 2476; LIV2
: 75;
Hill 2003: 258; Deshayes 2003: 117; Matasović 2009: 55);
OW. guecrissou, MW. crys ʽgirdle, shirtʼ, OCorn. kreis ʽcamisiaʼ, MBr. cres ʽshirtʼ (<
PCelt. *kris-su- < IE *kr̥ dh
-tu.; cf. Ru. čérez ʽgirdleʼ; cf. Lewis/Pedersen 1937: 21;
Pokorny IEW: 279; de Bernardo Stempel 1999: 262, 574; Deshayes 2003: 431;
Matasović 2009: 225);
T + s = 0s:
W. is ʽlowerʼ (< PCelt. *fissu < IE *ped-su (loc. pl.); OIA pad-, L. pēs ʽfootʼ; cf.
Lewis/Pedersen 1937: 21; Pokorny IEW: 790–792; GPC II: 2031; NIL 520–540;
Matasović 2009: 131);
OW. nes, nesaf, Corn. nes, nessa, MBr. nes, nessaff ʽcloseʼ (< PCelt. *nesso- < IE *Hnedskō;
cf. Os. nessimas ʽproximaeʼ, OHG nezzi ʽnetʼ; cf. Lewis/Pedersen 1937: 21;
Pokorny IEW: 758; GPC III: 2573; McCone 1996: 48; Deshayes 2003: 535; Matasović
2009: 289–290);202
9.2.6 The clusters sibilant + t/s
A general feature of the Celtic development, present also in Brythonic, is the loss of the plosive
in *sC clusters (cf. Stifter 2017: 68). The outcome of IE cluster *ss is not securely identifiable
(cf. Pedersen 1909: 75; Lewis/Pedersen 1937: 18):
S + t = ss:203
OW. serenn, MW. ser, syr, OCorn. steren, MBr. sterenn ʽstarʼ (< PCelt. *sterā- < IE
*H2stēr-; cf. Gr. ἀστήρ ʽstarʼ, L. stella ʽstarʼ; cf. Lewis/Pedersen 1937: 20; Pokorny
IEW: 1128; GPC III: 3226; de Bernardo Stempel 1999: 47; Delamarre 2003: 282; NIL
348–354; Matasović 2009: 355);
MW. assen, Corn. asen ʽribʼ (< PCelt. *astn(i )o- < IE *H2ostH1-; cf. Hitt. hastāi-, OIA
ásthi-, L. os, ossis; cf. Pokorny IEW: 783; GPC I: 219, 1198; de Bernardo Stempel
1999: 368; Schrijver 1995: 53–54; Matasović 2009: 44–45);
OW. PN Con-bresel, Corn. bresel, MBr. brezel, bresel ʽwarʼ (< PCelt. *brestā < IE
*bh
res-t-; cf. OHG brestan; cf. Pokorny IEW: 166–167; Deshayes 2003: 135;
Matasović 2009: 76–77);204
202
Matasović (l.c.) assume either *√Hned-t- or *√Hned-s-, NIL (590–600) relates to IE *√sed-.
203
Schrijver (1995: 406) argues that there are instances when -st- is preserved in Brythonic (cf. for list of items,
cf. Schrijver 1995: 410-413). Since some examples are dubious and other could be results of an analogy, but
definitely not the results of a sound law, we do not repeat his list here.
204
According to Pokorny IEW and LIV2
, (l.c.) the root is *√bh
re -, enlarged by -s. The cognates then are: L. friō
“break, crumble”, RuCS briti “shave, cut”.
200
MW. bys, OCorn. bis, bes, Bret. biz, bis ʽfingerʼ (< PCelt. *bisti- ʽfingerʼ < IE *gu
is-ti-;
cf. ON. kvistr- ʽbranchʼ, Alb. gisht ʽfingerʼ; cf. Pokorny IEW: 481; GPC I: 367;
Deshayes 2003: 111; Matasović 2009: 66–67);
9.2.7 Overview of Brythonic development
In the overview are the regular outcomes of IE clusters in reconstructed Brythonic. The outcome
of the IE cluster *ss, since not directly attested, is reconstructed by analogy and in brackets.
IE Brythonic t- s-
-kṷ
/gṷ
/gṷh
-p/b t x
-k/g/gh
-k/g t x
-ḱ/ǵ/ǵh
-k/g t x
-t/d/dh
-t/d 0s 0s
-p/b/bh
-0/b t 0x
-s -s 0s (0s)
9.3 Goidelic clusters and their IE origins
In contrast to Brythonic, Goidelic preserved old labiovelars for a longer period but finally
merged them with plain velars (this development is typical otherwise for satəm-languages, but
it is also partially attested in Greek). Again, in contrast to Brythonic, old x (a result of a
spirantization before a plosive) is preserved (cf. Sims-Williams 2017: 366), Goidelic hence
being generally more conservative in the development of the consonantal clusters than
Brythonic languages.
9.3.1 The clusters labial + t/s
IE labials are realized as x before t- and as s before s-, similarly to all velar plosives:
P + t = Goid. xt:
Ir. secht N
ʽsevenʼ seiz (< PCelt. *sextam < IE *septem; cf. OIA saptá-, L. septem ʽsevenʼ;
cf. Pokorny IEW: 909; LEIA S-66; Greene 1992: 510, 515, 540; Blažek 1999: 248;
Matasović 2009: 332);
OIr. necht ʽnieceʼ (PCelt. *neftī < IE *neptiH2; cf. L. neptis, OHG nift ʽnieceʼ; cf. Pokorny
IEW: 764; LEIA N-15; NIL: 520–524; Matasović 2009: 286);
OIr. cacht ʽfemale slaveʼ, Ir. cath ʽservantʼ (< PCelt. *kaxto- < IE *√keH2p-; cf. L. captus,
ON haptr ʽcaptusʼ; cf. Pokorny IEW: 527; LEIA C-261; LIV2
: 344–345; Matasović
2009: 197);
P + s = Goid. 0s:
OIr. úas, ōs ʽabove, overʼ (< PCelt. *ouxso < IE *H2eu̯ p-so; cf. Gr. ὕψῐ), OIr. ūasal ʽhighʼ
(< PCelt.*ouxselo- < IE * H2eu̯ pselo-; cf. Pedersen 1909: 75; Lewis/Pedersen 1937:
19; Pokorny IEW: 1107; Matasović 2009: 303);
201
9.3.2 The clusters velar + t/s
Plain velar plosives are lenited as x before t-; the cluster of *Ks is realized as 0s:
K + t = Goid. xt:
MIr. rucht ʽtunic, mantleʼ (< PCelt. *rouk-tu; < IE *√Hreu̯ k-;205
cf. W. rhuch ʽfilm,
pellicle, jerkin, coatʼ; cf. Pokorny IEW: 867; LEIA R-50; Matasović 2009: 317–318);
OIr. ucht ʽbreastʼ (< PCelt. *fextu < IE *pektu-; cf. OIA pákṣa- ʽshoulderʼ, L. pectus
ʽbreastʼ; cf. Pokorny IEW: 792; LEIA U-14–15; de Vaan 2008: 453; Matasović 2009:
130);
MIr. mocht ʽsoft, tenderʼ (< PCelt. *muxto- < IE *meu̯ kt-t cf. L. mūcor ʽmouldʼ, Gr. μύξᾰ
ʽmucusʼ; cf. Pokorny IEW: 744–745; LEIA M-58;de Bernardo Stempel 1999: 438;
LIV2
: 443–444; Matasović 2009: 282);
K + s = Goid. ss:
OIr. tress ʽfightʼ (< PCelt. *tregso; < IE *√treg-; cf. OE Þragjan ʽcourageʼ, ON Þrekr
ʽstrengthʼ; cf. Pokorny IEW: 1090; LEIA T-136; Schrijver 1995: 136; de Bernardo
Stempel 1999: 252, 258, 389; Matasović 2009: 389–390);
9.3.3 The clusters palatovelar + t/s
Since the palatavelar plosives clusters of Ḱt have merged with plain velars, they are lenized as
x before t; the cluster of *Ks is realized as 0s, as in the case with plain velars, labiovelars and
labials:
Ḱ + t = Goid. xt:
OIr. écht ʽslaughterʽ (< PCelt. *anxtu- < IE *neḱ-tu-; cf. OIA naśyati ʽbe lostʼ, L. necō
ʽkill, sayʼ; cf. Pokorny IEW: 762; LEIA N-11; de Bernardo Stempel 1999: 97;
Matasović 2009: 37–38, 39, 290–291);
Ir. ocht N
ʽeightʼ (< PCelt. *oxtū < IE *oḱtō; cf. L. octō, Goth. ahtau; cf. Lewis/Pedersen
1937: 41; Pokorny IEW: 775; LEIA O-7; Greene 1992: 510–511, 540; Blažek 1999:
266; Matasović 2009: 304);
OIr. recht ʽlawʼ (< PCelt. *rextu- < IE *√H3reǵ-; cf. L. rēctus, Goth. raihts; cf.
Lewis/Pedersen 1937: 42; Pokorny IEW: 854–857; LEIA R-12; Bernardo Stempel
1999: 291; LIV2
: 304; Schumacher 2007: 530–534; Matasović 2009: 310–311);
OIr. bricht ʽmagical formulaʼ (< PCelt. *brixtu/o- < IE *√bh
erǵh
-; cf. OIA brahmán-, ON
bragr; cf. Pokorny IEW: 139; LEIA B-89; Delamarre 2003: 90; NIL 30–34; Matasović
2009: 79–80);
OIr. fecht ʽtravel, time, -timesʼ (< PCelt. *u̯ extā; < IE *√u̯ eǵh
-; cf. L. vector, vectis ʽcarryʼ,
OCS vezti ʽcarryʼ; cf. Pokorny IEW: 1118–1120; LIV2
: 661–662; Matasović 2009:
419–420);
Ḱ + s = Goid. ss:
MIr. aiss ʽbackʼ206
(< PCelt. *axsilā; < IE *H2eḱs-; cf. OIA ákṣa-, L. axis ʽaxleʼ; cf.
Lewis/Pedersen 1937: 20; Pokorny IEW: 6; LEIA A-50; Greene 1992: 510–511, 515,
539–540; NIL 259–262; Matasović 2009: 50);
205
Pokorny (l.c.) reconstructs *reu̯ ǵ-.
206
Cf. Matasović (2009: 50) on the semantic motivation.
202
OIr. dess ʽto the right, to the southʼ (< PCelt. *dexs(o)u̯ o- < IE *deḱsu̯ o; cf. Gr. δεξῐτερός
ʽright handʼ; L. dexter; cf. Lewis/Pedersen 1937: 20; Pokorny IEW: 190; LEIA D-61–
62; Matasović 2009: 97);
OIr. sesser ʽsix menʼ207
(< PCelt. *su̯ exs < IE *su̯ eḱs; cf. Gr. ἕξ ʽsixʼ, L. sex ʽsixʼ; cf.
Lewis/Pedersen 1937: 20; Pokorny IEW: 1044; LEIA S-59; Blažek 1999: 237;
Delamarre 2003: 285–286; Matasović 2009: 364);
OIr. ess-, ass- ʽout of, fromʼ (< PCelt. *exs- < IE *H1esǵs-; cf. L. ex, Lith. iš; cf.
Lewis/Pedersen 1937: 20; Pokorny IEW: 292; Pokorny 1969: 24; Delamarre 2003:
169; Matasović 2009: 119);
OIr. ness ʽwoundʽ (< PCelt. *nexso- < IE *neḱ-; cf. OIA naṣṭá- ʽbe lostʼ, L. necō ʽkill,
sayʼ; cf. Pokorny IEW: 762; LEIA N-11; de Bernardo Stempel 1999: 97; Matasović
2009: 37–38, 39, 290);
9.3.4 The clusters labiovelar + t/s
As all peripheral plosives are, labiovelars are realized as x in the context of t- and as a sibilant
before s-:
Ku̯
+ t = Goid. xt:
Ir. in-nocht ʽtonightʼ (< PCelt. *noxt- < IE *noku̯
t- ; cf. L. nox, noctis, Goth. nahts;
Lewis/Pedersen 1937: 41; Pokorny IEW: 762–763; LEIA N-19; LIV2
: 449; NIL: 504–
513; Matasović 2009: 293–294) ;
Ir. nocht ʽnakedʼ < PCelt. *noxto- < IE *nogu̯
to-; cf. Goth. naqaÞs, OHG nackut,
nachut; Pokorny IEW: 769; LEIA N-19; de Bernardo Stempel 1999: 440; NIL: 513–
515; Matasović 2009: 294) ;
Ir. snecht(a)e ʽrainsʼ (with t-suffix, cf. Gr. νῐφετός ʽfalling snow, snowstormʼ, L. nix < IE
*sne gu̯h
tos; cf. Pokorny IEW: 974; LEIA S-153; LIV2
: 573; Schumacher 2007: 597–
598; NIL: 622–623; Matasović 2009: 349)
Ku̯
+ s = Goid. ss:
Ir. oss ʽstag, cowʼ (< *uxso- < *uku̯
sō < *√u̯ egu̯
-; cf. OIA ukṣan- ʽbullʼ, OHG ohso ʽoxʼ;
cf. Lewis/Pedersen 1937: 20; Pokorny IEW: 1118; LEIA O-34; LIV2
602–603; NIL:
368–370; Matasović 2009: 400–401);
OIr. no-tes ʽfleeʼ (< IE *√teku̯
-; cf. OIA takti ʽshoots awayʼ, OCS těchъ ʽrunʼ; cf. Pedersen
1913: 639; Pokorny IEW: 1059–1160; LEIA T-40; LIV2
620–621; Schumacher 2007:
629–631; Matasović 2009: 377);
9.3.5 The clusters dental + t/s
Both IE clusters of *Tt and *Ts are realized as Goidelic ss:
T + t = Goid. ss:
Ir. fiuss ʽknowledgeʼ (< IE *u̯ id-to-s; cf. L. uīsus ʽseenʼ, OIA vitti-; cf. Pokorny IEW:
1126; Schrijver 1995: 404; LIV2
: 665–667; Hill 2003: 258; Schumacher 2007: 664–
669, 690–701; NIL 717–722; Matasović 2009: 407–408; 422–423);
OIr. bés (< PCelt. *banssu-; < IE *√bh
endh
-; cf. OIA bándhati ʽbindʼ, Goth. bindan ʽbindʼ;
cf. Pokorny IEW: 127; LEIA B-43; LIV2
: 75; Hill 2003: 257; Matasović 2009: 55);
207
Note that -er < fer < *u̯ īros; Blažek (1999: 237).
203
Ir. cri(u)ss ʽgirdleʼ (< PCelt. *kris-su- < IE *kr̥ dh
-tu.; cf. Ru. čérez ʽgirdleʼ; cf.
Lewis/Pedersen 1937: 21; Pokorny IEW: 279; LEIA C-238–239; de Bernardo Stempel
1999: 262, 574; Matasović 2009: 225);
OIr, mess ʽjudgementʼ (< IE *med-tu-; cf. MW. meddu ʽpossess, ruleʼ, L. modus
ʽmeasureʼ, Goth. mitan ʽmeasureʼ; cf. Pokorny IEW: 705–706; LEIA M-48–49;
Lambert 1994: 44; LIV2
: 423; Hill 2003: 257; Delamarre 2003: 223; Schumacher
2007: 478–483; NIL: 463–465; Matasović 2009: 261);
T + s = Goid. ss:
OIr. ís ʽbelow, underʼ (< PCelt. *fissu < IE *ped-su (loc. pl.); OIA pad-, L. pēs ʽfootʼ; cf.
Lewis/Pedersen 1937: 21; Pokorny IEW: 790–792; de Bernardo Stempel 1999: 31;
NIL 520–540; Matasović 2009: 131);
Ir. nessa, nessam ʽnearer, nearestʼ (< PCelt. *nesso- < IE *Hned-skō; cf. Os. nessimas
ʽproximaeʼ, OHG nezzi ʽnetʼ; cf. Lewis/Pedersen 1937: 21; Pokorny IEW: 758; LEIA
N-12; McCone 1996: 48; Matasović 2009: 289–290);208
Ir. īss ʽwill eatʼ (< *i-ed-s-; < IE *√H1ed-; cf. OIA atsyáti; cf. Pokorny IEW: 287; LIV2
:
230; Schumacher 2007: 377–380; Matasović 2009: 113);
9.3.6 The clusters sibilant+ t/s
The IE cluster of *sT is realized as ss (or simplified as 0s) (as generally clusters of *sC are; cf.
Stifter 2017: 68), there is probably only one “secure” outcome of the IE cluster *ss again (cf.
Pedersen 1909: 75; Lewis/Pedersen 1937: 18):
S + t = Goid. (s)s:
OIr. ser ʽstarʼ (< PCelt. *sterā- < IE *H2stēr; cf. L. stella; cf. Lewis/Pedersen 1937: 20;
Pokorny IEW: 1128; LEIA S-90; de Bernardo Stempel 1999: 47; Delamarre 2003:
282; NIL 348–354; Matasović 2009: 355);
OIr. bres ʽfight, combatʼ, PN Bres-(u)al (< PCelt. *brestā < IE *bh
res-t-; cf. OHG
brestan; cf. Pokorny IEW: 166–167; LEIA B-86; Matasović 2009: 76–77);209
OIr. asna, esna ʽribʼ (< PCelt. *astn(i )o- < IE *H2ostH1-; cf. Hitt. hastāi-, OIA ásthi-, L.
os, ossis ʽboneʼ; cf. Pokorny IEW: 783; de Bernardo Stempel 1999: 368; Schrijver
1995: 53–54; Matasović 2009: 44–45);
MIr. bissig ega (dat.pl.) ʽicicleʼ (< PCelt. *bisti- ʽfingerʼ < IE *gu
is-ti-; cf. ON. kvistrʽbranchʼ,
Alb. gisht ʽfingerʼ; cf. Pokorny IEW: 481; LEIA B-53; Delamarre 2003: 76;
Matasović 2009: 66–67);
S + s = Goid. ss:
OIr. céiss ʽlike to createʼ (< *keis-; < IE *√kē (s)-; cf. Gr. κίω ʽgo awayʼ L. cieō ʽput in
motionʼ; cf. Pedersen 1913: 490–491; Pokorny IEW: 538–539; Schumacher 2007:
391–393);
208
Matasović (l.c.) assume either *√Hned-t- or *√Hned-s-, NIL (590–600) relates to IE *√sed-.
209
According to Pokorny IEW and LIV2
, (l.c.) the root is *√bh
re -, enlarged by -s, the cognates then are: L. friō
“break, crumble”, RuCS briti “shave, cut”.
204
9.3.7 The overview of the Goidelic development
The IE peripheral clusters tend to form a cluster of xt in the t-contexts. The clusters of *Tt and
all clusters formed in the sibilant context are realized as ss:
IE Goidelic t- s-
-kṷ
/gṷ
/gṷh
k/b/g xt ss
-k/g/gh
k/g xt ss
-ḱ/ǵ/ǵh
k/g xt ss
-t/d/dh
t/d ss ss
-p/b/bh
0/b xt ss
-s s ss ss
9.4 Gaulish clusters and their IE origins
Gaulish development has more archaic features that than Insular languages, preserving spirants before *s. Since
we have only a limited data pool to work with, our examples are usually onomastic.
Only attested cluster are listed, sources are quoted continuously:
9.4.1 The clusters labial + t/s
The labial clusters develop in exactly the same way as velar clusters in the same context:
P + t = Gal. xt:
Gal.(–L.) Pagus Sextan-mandu[us] (CIL XIII 3149) (< PCelt. *sextam < IE *septem; cf. L. septem ʽsevenʼ;
cf. Pokorny IEW: 909; Evans 1967: 223; Lambert 1994: 44; Greene 1992: 510, 515, 540; Blažek 1999:
248; Matasović 2009: 332);
Gal. PN Caxtos (< PCelt. *kaxto- < IE *√keH2p-; cf. L. captus, ON haptr ʽcaptusʼ; cf. Pokorny IEW: 527;
Delamarre 2003: 112; LIV2
: 344–345; Matasović 2009: 197);
P + s = Gal. xs:
Gal. PN Crixus, Crixsus (< PCelt. *kaxto- after a metathesis < IE *kris-po-; cf. L. crispus, crispō; cf.
Pedersen 1909: 75; Lewis/Pedersen 1937: 19; Evans 1967: 398, 445; Delamarre 2003: 130; Matasović
2009: 226);
Gal. TN Οὔξελον, Uxello-dūnum, Uxama (< PCelt.*ouxselo < IE *oupselo-; cf. Pedersen 1909: 75;
Lewis/Pedersen 1937: 19; Pokorny IEW: 1107; Lambert 1994: 44; Delamarre 2003: 330);
9.4.2 The clusters velar + t/s
The IE cluster of *Ks is realized as Gaulish x, the cluster of *Kt is not attested, and outcomes are the same for IE
clusters of *Ḱt and *Ḱs:
K + s = Gal. x:
Gal. PN Trexius, Trexa, Trenus (< PCelt. *trexso-; < IE *treg-; cf. OE Þraka ʽcourageʼ, ON Þrekr
ʽstrengthʼ; cf. Pokorny IEW: 1090; de Bernardo Stempel 1999: 252, 258, 389; Delamarre 2003: 301;
Matasović 2009: 389–390);
9.4.3 The clusters palatovelar + t/s
The old IE palatovelars merged with old plain velars. The outcome of a *Ḱt cluster is xt, and of *Ḱs is xs:
205
Ḱ + t = Gal. xt:
Gal. oxtumetos ʽeightʼ (< PCelt. *oxtū-m- < IE *oḱto-; cf. L. octō, Goth. ahtau ʽeightʼ; cf. Lewis/Pedersen
1937: 41; Pokorny IEW: 775; Greene 1992: 510–511, 540; Blažek 1999: 266; Delamarre 2003: 304;
Matasović 2009: 304);
Gal.(–L.) ambactus ʽservantʼ (< PCelt. *ambi-aktos < IE *√H2eǵ-; cf. Pokorny IEW: 4; Evans 1967: 128,
135–136; LIV2
: 255–256; Delamarre 2003: 40; Schumacher 2007: 189–193; NIL 267–267; Matasović
2009: 32);
Gal. PN Rextu-genus (< PCelt. *rextu- < IE *√H3reǵ-; cf. L. rēctum, Goth. raihts ʽrightʼ; cf. Lewis/Pedersen
1937: 42; Pokorny IEW: 854–857; Evans 1967: 109; Bernardo Stempel 1999: 291; Delamarre 2003:
255; LIV2
: 304; Schumacher 2007: 530–534; Matasović 2009: 310–311);
Gal. brixtia (Chamalières) (< PCelt. *brixtu/o- < IE *√bh
erǵh
-; cf. OIA brahmán-, ON bragr; cf. Delamarre
2003: 90; Matasović 2009: 79–80);
Gal. PN Vectirix, Vecturius (< PCelt. *u̯ extā; < IE *√u̯ eǵh
-; cf. L. vector, vectis ʽcarryʼ, OCS vezti ʽcarryʼ;
cf. Pokorny IEW: 1118–1120; Evans 1967: 281; Delamarre 2003: 309; LIV2
: 661–662; Matasović 2009:
419);
Ḱ + s = Gal. xs:
Ga. (La Graufesenque) suexos ʽsixthʼ (< *su̯ eḱsos < IE *su̯ eḱs; cf. Gr. ἕξ, L. sex ʽsixʼ; cf. Lewis/Pedersen
1937: 20; Pokorny IEW: 1044; Greene 1992: 510–511, 515, 539–540; Blažek 1999: 237; Delamarre
2003: 285–286; Matasović 2009: 364);
Gall. PN Dexsiua (< PCelt. *dexs(o)u̯ o- < IE *deḱsu̯ o-; cf. Gr. δεξῐτερός ʽright handʼ; L. dexter;
Lewis/Pedersen 1937: 20; Pokorny IEW: 190; de Bernardo Stempel 1999: 212, Delamarre 2003: 143;
Matasović 2009: 97);
Gal. ex- (< PCelt. *exs- < IE *H1esǵs-; cf. L. ex, Lith. iš; cf. Lewis/Pedersen 1937: 20; Pokorny IEW: 292;
Evans 1967: 398; Pokorny 1969: 24; Delamarre 2003: 169; Matasović 2009: 119);
9.4.4 The clusters labiovelar + t/s
There are no secure data on the development of IE labiovelars in contexts of t/s-.
9.4.5 The clusters dental + t/s
Both clusters *Tt and *Ts are often realized in ʽtau gallicumʼ, usually considered to be a dental affricate (but also
a dental fricative, cf. Evans 1967: 398), here marked always as đ. Note that the same symbol could express *st.
Note: The phonemic nature of ʽtau gallicumʼ is assumed either as a dental affricate, fricative or sibilant (cf. Dottin
1920: 48; Weisgerber 1935: 317; Schmidt 1957: 101; Evans 1967: 419) or a lenited t (Eska 1998: 120–125).
T + t = Gal. đđ/ss:
Gal. PN Međđu-gnatus, Međđi-gnatius, Μεϑϑιλος, Meϑilos, Messillus, Medsillus (< IE *med-tu-; cf. MW.
meddu ʽpossess, ruleʼ, L. modus ʽmeasureʼ; cf. Pokorny IEW: 705–706; Evans 1967: 293, 367–368,
411, Lambert 1994: 44; LIV2
: 423; Delamarre 2003: 223; Schumacher 2007: 478–483; Matasović 2009:
261);
Gal. PN Meliđđus, Meliđđius, Melissus etc. (< IE *mélit-t-; cf. G. μέλῐ, Goth. miliÞ , W. melys ʽsoftʼ; cf.
Pokorny IEW: 723–724; Evans 1967: 115, 293, 297; Delamarre 2003: 224; Matasović 2009: 263);
T + s = Gal. đđ/ss:
Gal. neđđamon ʽnextʼ (gen. pl., Banassac) ʽproximorumʼ (< PCelt. *nesso- < IE *Hned-skō; cf. Os. nessimas
ʽproximaeʼ, OHG nezzi ʽnetʼ; cf. Lewis/Pedersen 1937: 21; Pokorny IEW: 758; McCone 1996: 48;
Delamarre 2003: 233; Hill 2003 : 254–256; Matasović 2009: 289);210
9.4.6 The clusters sibilant + t/s
210
Matasović (l.c.) assume either *√Hned-t- or *√Hned-s-, NIL (590– 600) relates to IE *√sed-.
206
The IE cluster *st is realized either as a sibilant s, or alternately with đ, i.e. ʽtau gallicumʼ. There are no secure
examples of the IE cluster *ss, just as there are none in both Insular branches (cf. Pedersen 1909: 75;
Lewis/Pedersen 1937: 18):
S + t = Gal. ss/đđ:
Gal. TN Sirona, Đirona, Thirona, Dirona (< PCelt. *sterā- < IE *H2stēr; cf. Gr. ἀστήρ ʽstarʼ, L. stella; cf.
Lewis/Pedersen 1937: 20; Pokorny IEW: 1128; Evans 1967: 412–414, 419; de Bernardo Stempel 1999:
47; Delamarre 2003: 282; NIL 348–354; Matasović 2009: 355);
Gal.(-Germ). PN Bissula Bissus, Bisso, Bisius, Bessius (< PCelt. *bisti- ʽfingerʼ < IE *gu
is-ti-; cf. ON.
kvistr- ʽbranchʼ, Alb. gisht ʽfingerʼ; cf. Pokorny IEW: 481; Delamarre 2003: 76; Matasović 2009: 66);
9.4.7 Overview of the Gaulish development
The outcome in the following table is, due to the lack of data, only partially attested, outcomes based on the analogy
are in brackets:
IE Gaulish t- s-
-kṷ
/gṷ
/gṷh
p/b (xt) (xs)
-k/g/gh
k/g xt xs
-ḱ/ǵ/ǵh
k/g xt xs
-t/d/dh
t/d đđ/ss đđ/ss
-p/b/bh
0/b xt xs
-s s đđ/ss ?
9.5 Trajectories of the development
Celtic development can be classified, according to its inputs and the context, into three (sub)blocks.
The first is that of *Tt and *Ts: both blocks were subjected to the old Common IndoEuropean
process, here leading, as in Italic and Germanic, towards ss in Goidelic and Brythonic
(but with much interesting outcome in Continental Celtic, see below). Another block is that of
all peripheral plosives, which underwent a wide spirantization and lenition of the left plosive to
a fricative or approximant before t-, but which has led towards the simplification of clusters
with s-. The last block contains IE cluster *st (since the output of the IE cluster *ss is not
securely attested in Celtic), resulting in Insular languages in a sibilant, but with the same
interesting outcome in Continental Celtic.
9.5.1 Development of clusters labial + t/s
The development of the labial series mirrors that of all velar series; we assume a spirantization
(as a bilabial spirant, later probably shifted to labiodental spirant), and the loss of labiality. In
Brythonic, the spirant became a palatal approximant (as with all peripheral series), while in the
other two sub-branches the outcome is a velar spirant.
The cluster of *Pt underwent the same spirantization (attested in Gaulish), with both
spirants later assimilated either as x (in Brythonic) or ss (in Goidelic), perfectly mirroring the
207
development known from velar clusters (cf. Thurneysen 1909: 138; Pedersen 1909: 93, 117;
Lewis/Pedersen 1937: 43–44; Jackson 1953: 529, 535-540; McCone 1996: 43–44; Schumacher
2004: 129; Kümmel 2007: 387–389; Stifter 2017: 1191):
i. P + t > φt > (ft > ht >) t (Brythonic)
ii. P + t > φt > (ft >) xt (Goidelic)
iii. P + t > φt > (ft >) xt (Continental)
i. P + s > φs > xx > 0x (Brythonic)
ii. P + s > φs > ss (Goidelic)
iii. P + s > φs > xs (Continental)
9.5.2 Development of cluster velar (and palatovelar) + t/s
The development of the plain velars and palatovelars can be reconstructed as follows: with a
spirantization of a velar in its first phase, both before t and s, this state is preserved in
Continental Celtic in its fullness. Both Insular branches have more advanced development of
the clusters with s-, where both fricatives merged either as x (in Brythonic) or as ss (in Goidelic).
The Brythonic spirant was lenited to before t- (with an intermediate stage ht?; cf. Thurneysen
1909: 133–134; Pedersen 1909: 77–78; Lewis/Pedersen 1937: 43–44; Jackson 1953: 404-411,
535-540; McCone 1996: 43–44; Kümmel 2007: 387–389):
i. K + t > xt > (ht >) t (Brythonic)
ii. K + t > xt (Goidelic)
iii. K + t > xt (Continental)
i. K + s > xs > xx > 0x (Brythonic)
ii. K + s > xs > ss (Goidelic)
iii. K + s > xs (Continental)
9.5.3 Development of clusters labiovelar + t/s
Since the lack of Continental Celtic data and both outputs of the development of clusters with
a left-standing labiovelar in Insular Celtic211
give us no hint to presume that Proto-Celtic
clusters had already lost their labiality in the neutralization position before t/s (as they did in
Latin)212
or that labiality was preserved, we will, for simplicity omit the question of labiality
present here, using a model essentially the same as for plain velars (Thurneysen 1909: 134–
135; Pedersen 1909: 77-78, 129–130; Jackson 1953: 535-540; McCone 1996: 43–44):
211
Since in both sub-branches of the Insular Celtic the outcome is the same for velars, labiovelars or labials.
212
For the neutralization of labiality before t/s-; cf. Lewis/Pedersen 1937: 43–44; Kümmel 2007: 387.
208
i. Ku̯
+ t > xt > (ht >) t (Brythonic)
ii. Ku̯
+ t > xt (Goidelic)
(iii. Ku̯
+ t > xt (Continental) )
i. Ku̯
+ s > xs > xx > 0x (Brythonic)
ii. Ku̯
+ s > xs > ss (Goidelic)
(iii. Ku̯
+ s > xs (Continental) )
9.5.4 Development of clusters dental + t/s
The processes affecting the dental series clusters are the oldest layer of the development of
clusters, since we meet the same processes in Germanic and Italic and very similar in all other
Indo-European languages (beside Indic).
An interesting development is that attested in Gaulish, where *Tt > đđ/ss. (cf. Vendryès
1908: 53; Evans 1967: 410–420; Lambert 1994: 44; Görtzen 1997: 394–399; Delamarre 1999:
223; McCone 1996: 48; Sims-Williams 2017: 364), the precise phonetic value of đđ is either a
dental affricate or a spirant (cf. Evans 1967: 398). There is no reason to doubt its geminate
nature, and its alternation with ss, numerously attested, is either a sign of the problem of
expressing a given sound in the alphabet used or its unstable and dialectally different phonetic
nature. However, both possibilities could be valid at the same time.
Both Insular languages have the same outcome ss, and it is worthy of notice that either
in Continental or Insular languages clusters of *Tt/Ts/st always have the same singular
outcome.213
For the cluster *Tt, its trajectory is usually modelled using the affricate model,
popularized by Brugmann (first 1880: 140–142, used since), though initially the idea was by
Kräuter (1877: 88), who assumed the affrication of the first plosive, followed by the
assimilation of the second plosive and later simplification of both affricates as two sibilants.
For the clusters of *Ts the process could be modelled as affricativizaon and the assimilation of
the affricate to a sibilant. The Continental Celtic forms with ʽtau gallicumʼ could be assumed
to be examples of preserved affricates:
i. T + t > ts
t > tsts > ss (Insular)
ii. T + t > ts
t > tsts > đđ/ss214
(Continental)
i. T + s > ts
s > tsts > ss (Insular)
ii. T + t > ts
s > tsts > đđ/ss15
(Continental)
213
It should be noted that Jackson (1953: 709) and Sims-Williams (2007: 338–339) argue that Ts was preserved
as an affricate in Old welsh and Ogamic Irish.
214
Here đđ marks two dental affricates.
209
Note: Pedersen (1909: 136 and later) assumes that *Tt > ts
t > st > ss, this trajectory seems to be overcomplicated;
however, the very same idea was supported by Jackson (1953: 529), similarly Hill (2003: 310) assumes that *Tt
realized first as Proto-Celtic *st, i.e. similarly to the development in Iranian, Baltic, Slavic, Greek.
The affricate model hardly explains the sibilantization of both clusters, especially that of *Ts:
if such a cluster were affricated, the simple phonetic solution would be the merging of both
sibilants (Tss > ts), i.e., resulting in the original cluster again. The phonetic value of đđ is
unknown, and its affricate value is derived from the Brugmann’s model, not otherwise, hence
as a self-evident proof it is useless).
We propose an alternative model, which was first proposed for Italic by de Saussure
(1877), Cocchia (1883: 16–58) and Bartholomae (1887), who assumed that IE *Tt > Italic *ϑϑ
(similarly Walde 1897: 503 for IE *dh
t215
). Since Celtic (and Germanic and Italic) share the
same features of the process, we dare to apply it in the Celtic spirantization model, assuming
the spirantization of the left dental plosive and the assimilation of the right plosive to it,
similarly the spirantization of the (left) dental plosive before s. All dental plosives merged later
with s; hence both clusters have a singular input ss in Insular Celtic. The Continental geminate
đđ is an old cluster of two dental spirants ϑϑ, its alternation with ss preserved in Gaulish data
(it is possible that ϑs also gave ϑϑ):
i. T + t > ϑt > ϑϑ > ss (Insular)
ii. T + t > ϑt > ϑϑ > đđ/ss216
(Continental)
i. T + s > ϑs > ϑϑ (?) > ss (Insular)
ii. T + t > ϑs > ϑϑ > đđ/ss16
(Continental)
9.5.5 Development of clusters sibilant + t/s
We have mentioned above that even *st clusters are realized in the same way as clusters of *Tt
and *Ts (cf. Vendryès 1908: 53; Jackson 1953: 527, 530; Hill 2003: 291–306), i.e. as sibilant
clusters in Insular Celtic and often as a ʽtau gallicumʼ cluster of đđ/ss in Continental Celtic (in
the word-initial, only đ/s are present). The assimilation of Vst as Vss appeared in Insular Celtic
en bloc (McCone 1997: 33).
Lewis / Pedersen (1937: 20) assume the affrication of *st to *ts and relate this directly
to the ʽtau gallicumʼ, Celt. *ts being realized as s in Insular (cf. Sims-Williams 2007: 339). The
trajectory could be modelled as:
i. s + t > ts > ss (Insular)
ii. s + t > ts > đđ/ss (Continental)
215
Note that Walde assumes that IE voiced aspirates were in fact voiced spirants.
216
Here đđ marks two dental spirants ϑϑ.
210
Again, we assume that the spirantization trajectory is more appropriate, assuming that the right
dental plosive was spirantized and later assimilated to s; in Gallic, both obstruents are
assimilated as đđ/ss:
i. s + t > sϑ > ϑϑ > ss (Insular)
ii. s + t > ϑs > ϑϑ > đđ/ss (Continental)
For this development, we see a parallel in the development of all clusters of s + plosive in
Celtic. IE *#sp- gives Goidelic #s- (lenited to #f) and Brythonic *#f- (Lewis/Pedersen 1937: 18;
Stifter 2017: 1190), cf. Ir. seir ʽheelʼ (acc. du. di pherid), MW. ffer ʽancleʼ, OCorn. fer from IE
*sperH1-o- (cf. OIA sphuráti ʽkickʼ Gr. σφῠρόν ʽankleʼ; Lewis/Pedersen 1937: 18; Matasović
2009: 333); OIr sine ʽteatʼ (cf. ON speni ʽnippleʼ; < PCelt. *sfen o-; Lewis/Pedersen 1937: 18;
Matasović 2009: 333); MIr. sond ʽstake, beamʼ, OW. finn, MW. fonn ʽstick, staffʼ (cf. L. sponda
ʽbedstead, bedʼ, OE spōn ʽsliver, shavingʼ Gr. σφήν ʽwedgeʼ; < IE *spH2en-; Lewis/Pedersen
1937: 18; Matasović 2009: 334). Though in the internal position we can assume the metathesis
*-sp- > -ps- (cf. MW. crych, MBr. crech ʽcurlyʼ, Gal. PN Crixus, Crixsus (< PCelt. *kaxtoafter
a metathesis < IE *kris-po-; cf. L. crispus, crispō; Pedersen 1909: 75; Lewis/Pedersen
1937: 19; Evans 1967: 398, 445; Delamarre 2003: 130; Matasović 2009: 226). Since the
reflexes of *#sp- and *-ps- remarkably differ, they could hardly be outputs of a single input,
resulting from the metathesis of sp > ps. This is, however, improbable in an anlaut, since it is
harder to pronounce, though such a metathesis is proposed by Lewis/Pedersen (1937: 18). Since
the position of IE *p was remarkably weak in Proto-Celtic, we assume a spirantization of its
clusters of *#sp- as *#sφ- or (later on) *#sf- (Thurneysen 1909: 140; Jackson 1953: 529-531;
cf. also reconstructed forms in Matasović 2009: 332–335) – we assume a similar lenition for
the IE clusters of *#st- as well: *#st- > *#sϑ-.
However, the clusters of *#sK- (where K marks any voiceless velar plosive217
) were
subjected only partially (in Welsh, there is often a prosthetic vowel; cf. Jackson 1953: 527):
preserved in: OIr. scáth ʽshadowʼ, MW. isgaud, cy-sgawd, OCorn. scod, Br. skeud (< *Celt.
skāto- < IE *skeH3t-, cf. Gr. σκότος ʽdarkness, gloomʼ, Goth. skadus ʽshadowʼ; cf. Pedersen
1909: 75; Lewis/Pedersen 1937: 19; de Bernardo Stempel 1999: 324, 528; Kümmel 2007: 388–
389; Matasović 2009: 340); OIr. scian ʽknifeʼ, MBr. squieaff, squeigaff ʽcutʼ (< *Celt. ski -o-
217
List of clusters *#sku̯
- as given by Pedersen (1909: 75); Lewis/Pedersen (1937: 20) is not persuasive, as it not
the list of possible etymologies in Matasović (2009: 338–339), hence we have to limit ourselves on clusters
*#sk-/*#sḱ- (on Insular development cf. Jackson 1953: 534–535).
211
< IE *sḱe -, cf. L. sacéna, scēna ʽdolabra pontificalisʼ, OIc. skeggia ʽaxeʼ; cf. Lewis/Pedersen
1937: 15, 19; Schrijver 1995: 107; de Bernardo Stempel 1999: 254; Matasović 2009: 343) but
lost in Brythonic: Ir. scend- ʽspringʼ, W. cy-chwynn-af ʽI startʼ (but also MW. ysgeinnyaw
ʽdisperseʼ with *sk- preserved!) (< *Celt. skando- < IE *skend, cf. OIA skándati, L. scando;
cf. Lewis/Pedersen 1937: 19; Matasović 2009: 339); Ir. scéith ʽact of vomitingʼ, MW. chwydu,
Corn. huedzha, MBr. huedaff (< *Celt. ski -o- < IE *sḱe -d, cf. Lith. skíesti ʽto have diarrhoeaʼ,
OIc. skeggia ʽaxeʼ; cf. Lewis/Pedersen 1937: 19; Matasović 2009: 343). The spirantization of
plosives after *s appeared then first for *st-, later it was applied to *sp-, and the clusters of *sKwere
affected last and only partially in Brythonic, but not in Goidelic.
9.6 Conclusion and final remarks
The oldest development affected the dentals in both contexts, as in all Indo-European languages
(OIA being a false exception). Usually, this change is explained using the affricate trajectory;
however we prefer, for the reasons listed above, the spirantization trajectory, assuming *Tt >
ϑt > ϑϑ > ss, which is supported by the existence of the ʽtau gallicumʼ, still documenting the
intermediate state. Similarly, for *Ts we model the trajectory: *Ts > ϑs > ss, and a very similar
trajectory is possible even for *st- (> sϑ > ss), with the spirantization of plosives after *s is well
attested for *sp- as well.
The later development of peripheral plosives before *t/s fits into a range of
spirantization/lenition: peripheral plosives were spirantized, later lenited as h (attested in
Goidelic) or as a palatal approximant (in Brythonic).
The development of the cluster of *st has been already described few lines above, the
cluster of *ss is not securely attested, hence its possible development is left aside.
213
10 The development of the two-obstruent clusters from Indo-European into
Germanic
10.0 Germanic languages and their branches
Germanic is first attested, beside glosses and short runic inscriptions, since the Christianization of given Germanic
nation, by the Gothic language (late 4th
century), later by Old English (since mid-late 7th
century), by Old High
German (mid 8th
century), Old Saxon (first half of 9th
century), Old Norse and Old Frisian (since 11th
century)
(Bousquette/Salmos 2017: 387).
Gothic represents the East Germanic branch, Old English, Old High German, Old Saxon and Old Frisian
represent West Germanic branch, Old Norse the North Germanic. Details of the relationship between branches are
beyond the scope of this study (cf. Prokosch 1939: 21–34; Krahe/Meid 1969: 10–40; Nedoma 2017:875–888).
10.1 Germanic and Indo-European
The typical features separating the Germanic obstruent system from that of Indo-European,
relevant for this study, are:
i. the Germanic shift of plosives (erste Lautverschiebung; Rask/Grimm’s Law218
, with Verner’s
Law modifying its outcomes) (followed by a later similar shift within Old High German),
with a spirantization of the IE voiceless non-aspirate; devoicing of the IE voiced plosives,
and deaspirantion/spirantization of the IE voiced aspirated plosives;
ii. preservation of the Indo-European modal opposition of three classes within the limits of the
consonantal shift;
iii. preservation of old labiovelars in its older stages219
.
A similar shift is well known from Armenian (it but hardly could be anything more than a
common drift, not the sign of a single phenomenon), and the loss of *p in Celtic could be a
result of a similar process. Labiovelars are attested in Latin inside the Italic languages and from
archaic Celtic languages – a remarkable feature is a later split of old labiovelars into velars and
labials; the process is known best from Greek (cf. Prokosch 1939: 71–74; Stiles 2017: 895).
10.2 Germanic clusters and their IE origins
(Rask–)Grimm´s Law did not affect voiceless stops following another obstruent (cf.
Braune/Heidermanns 2004: 73; Ringe 2006: 97), hence IE *t- is fully preserved in clusters of
plosive or s + t, i.e., in the clusters of our interest.
The primary language of our analysis will be Gothic; other language data will be used
when necessary, otherwise as supporting material. Beside etymological examples, we will use
218
Since the first scholar describing the principle was Rask, we will use his name to mark the law itself. However,
since popular knowledge connects this rule with Grimm, who used it in the second edition of his Deutsche
Grammatik and since there is no chance to disconnect this false authorship, we will use Grimm’s name as a
second part of the compound name.
219
Labiovelars are however later lost in the development of later phases of Germanic languages (cf. Markey1980
for a short overview of the Germanic development of labiovelars).
214
given productive clusters from the Gothic derivation/flexion to demonstrate the synchronic
form and value of described processes.
10.2.1 Clusters plain velar + t/s in Germanic
The plain velars are hard to reconstruct from the Germanic data, being merged with the
reconstructed IE palatovelars in all their outcomes:
K + t = Germ. ht:
Goth. prt. mahta ‘have power, be ableʼ, nom. mahts, OHG, OS math, OE meaht, might
‘powerʼ (< PGerm. *mahta; < IE *√magh
-; cf. OIA maghá- ‘powerʼ, OCS mogǫ, mošti;
cf. Pokorny IEW: 695; Streitberg 1974: 114; Lehmann 1986: 239–240; LIV2
: 422;
Kroonen 2013: 347; D. G. Miller 2019: 30);
Goth. gasahts ‘rebukeʼ (< PGerm. *ga-sahti-; < IE *√seH2g-; cf. Gr. ἡγέομαι, L. sāgus
‘prophetic, prescientʼ, OIr. saigid ‘try to reachʼ; cf. Pokorny IEW: 876–877; Lehmann
1986: 292–293; LIV2
: 520; Kroonen 2013: 423)
K + s = Germ. hs:
OHG, OS sahs, OE seax ‘knifeʼ, ON sax ‘short sword; scissorsʼ (< PGerm. *sahsa-; < IE
*√sekH-; cf. L. secō, OCS sěkǫ ‘cutʼ; cf. Pokorny IEW: 895–896; LIV2
: 524–525;
Kroonen 2013: 421);
Goth. maihstus, OHG mist ‘dung, manureʼ (< PGerm. *mihstu- < IE *√H3migh
-so-220
;cf.
OIA méhati, mīḍhá- ‘urinateʼ, Gr. ὀμείχω ‘urinateʼ, L. meiō, mixi, mictum ‘peeʼ; cf.
Pokorny IEW: 713; Lehmann 1986: 241; LIV2
: 301–302; NIL: 384–385; Kroonen
2013: 369);
10.2.2 Clusters palatovelar + t/s in Germanic
The reconstructed palatovelars have the same outcomes as plain velars:
Ḱ + t = Germ. ht:
Goth. ahtau ‘eightʼ (< PGerm. *ohtōu < IE *oḱtō; cf. OIA aṣṭáu, L. octō ‘eightʼ; cf.
Pokorny IEW: 775; Lehmann 1986: 12; Ross/Berns 1992: 602–609, 618; Blažek 1999:
266; Ringe 2006: 96; Kroonen 2013: 6-7);
Goth. raíhts, OE, OS, OHG reht ‘rightʼ (cf. < IE *√H3reǵ-; OIA rā́ ṣṭi ‘ruleʼ, L. pr. regō,
-ere ‘ruleʼ; sup. rēctum, nom. rector ‘guideʼ; cf. Pokorny IEW: 856; Krahe/Meid 1969:
109; Lehmann 1986: 281; LIV2
: 304–305; Kroonen 2013: 408);
Goth. waihts, ON véttr ‘thingʼ, OS with ‘somethingʼ (< IE *√u̯ eǵh
-ti-; cf. OCS veštь
‘thing; cf. Pokorny IEW: 1136; Lehmann 1986: 388–389; Kroonen 2013: 578);221
Ḱ + s = Germ. hs:
Goth. saíhs, OE siex ʽsixʼ (< PGerm. *sehs < IE *su̯ eḱs; cf. Gr. ἕξ, L. sex; cf. Pokorny
IEW: 1044; Lehmann 1986: 290–291; Ross/Berns 1992: 585, 617, 628–629; Blažek
1999: 237; Ringe 2006: 96; Kroonen 2013: 431);
Goth. taíhswa, OHG zeso, zesawēr ‘right (-hand)ʼ (< PGerm. *tehswaz < IE *deḱsu̯ o; cf.
Gr. δεξιός ‘on the right hand or sideʼ, L. dexter ‘rightʼ; cf. Pokorny IEW: 190;
Lehmann 1986: 338–339; Ringe 2006: 97; Kroonen 2013: 512);
220
Pokorny (1966: 713) reconstructs me ǵh
-.
221
Pokorny (l.c.) reconstructs IE *√u̯ eku̯
-ti-; i.e., the deverbative noun from √u̯ eku̯
- (cf. L. vōx etc.).
215
OHG ahsa ‘axleʼ (< PGerm. *ahsō < IE *H2eḱs-; cf. OIA ákṣa- L. axis; cf. Pokorny IEW:
6; NIL: 259–262; Kroonen 2013: 6);
10.2.3 Clusters labiovelar + t/s in Germanic
The labiovelars were neutralized before t/s, as is attested in Latin, but this development was
often corrupted by analogy (cf. Ringe 2006: 95). The plosive was spirantized before t:
Ku̯
+ t = Germ. ht:
Goth. nahts, OHG naht ‘nightʼ (< IE *nóku̯
t- ~ néku̯
t-; cf. Hitt. nekuz ‘evening timeʼ, Gr.
νύξ, νυκτός, L. nox, noctis; cf. Pokorny IEW: 762–763; Streitberg 1974: 114; Lehmann
1986: 262; LIV2
: 449; Ringe 2006: 93, 97; NIL: 504–513; Kroonen 2013: 381)
Goth. leiht, OE lēoht, OS liht- ‘lightʼ (< Germ. liht( )a- < IE *legu̯ h
-t-; cf. L. levis ‘lightʼ,
MW. líei ‘lessʼ, OCS lьgъkъ ‘lightʼ; cf. Pokorny IEW: 660–661; Streitberg 1974: 111;
Lehmann 1986: 229–230; Kroonen 2013: 338–339);
Note: Labiovelar is sometimes restored before t- due to analogy, cf. Goth. pt. saƕt (analogy to saiƕan ‘seeʼ)
(Streitberg 1974: 111; Lehmann 1986: 291; Braune/Heidermanns 2004: 70), but cf. OS siht, OHG siht with the
old neutralization.
Ku̯
+ t = Germ. ft:
Goth. fimfta-táihunda ‘fifteenthʼ, OHG fimfto, fingto, OS fifto ‘fifthʼ (< IE *penku̯
t- cf. OIA paṅktí ‘number
fiveʼ, Gr. πέμπτος ‘fifthʼ; cf. Prokosch 1939: 287; Pokorny IEW: 808; Lehmann 1986: 117; Ross/Berns
1992: 584–585, 600, 616–617, 628; Blažek 1999: 223–224; Kroonen 2013: 140–141);222
Ku̯
+ s = Germ. hs:
Goth. gen. pl. aúhsne, OE oxa ʽoxʼ (< PGerm. *uhso- < IE *uku̯
-sō < *√u̯ egu̯
-so-; cf. OIA
ukṣan-, OHG ohso; cf. Pokorny IEW: 1118; Lehmann 86 49; LIV2
: 602–603; Ringe
2006: 97; NIL: 368–370; Kroonen 2013: 558);
Note: Labiovelars were also neutralized in auslaut, cf. Goth. nih ‘und nichtʼ (cf. L. neque), but the labiovelar is
restored in Goth. nêƕ ‘naheʼ (cf. Lehmann 1986: 265–266; Braune/Heidermanns 2004: 70) as it is in .Goth.
saƕ/saíƕ analogy to saiƕan ‘sehenʼ.
10.2.4 Clusters dental + t/s in Germanic
The IE clusters of Tt are realized, similarly to Italic or Celtic, as ss, as are clusters of Ts:
T + t = Germ. ss:
Goth. unwiss ‘uncertainʼ (cf. Goth. unwita ‘foolʼ), OE gewiss ʽcertainʼ (< PGerm.
*(ga)waissaz; < IE *√u̯ id-to-; cf. OIA pf. véda, Gr. pf. οἶδα ‘knowʼ; cf. Prokosch 1939:
85; Pokorny IEW: 1125–1127; Lehmann 1986: 406–407; LIV2
: 665–667; Ringe 2006:
87; Kroonen 2013: 588);
Goth. gawiss ‘jointʼ (< PGerm. *(ga)wissiz; < IE *√u̯edh
-ti-; cf. OIA vádhram ‘leather
strapʼ, W. gwedd ‘yokeʼ; cf. Pokorny IEW: 1116–1117; Krahe/Meid 1969: 110;
Lehmann 1986: 153–154; LIV2
: 660; Ringe 2006: 87; Kroonen 2013: 577);
OE, ON sess ‘seatʼ (< PGerm. *sessaz; < IE *√sed-to-; cf. L. sedēre; cf. Prokosch 1939:
85; Pokorny IEW: 884–887; LIV2
: 513; Ringe 2006: 87; NIL: 560–600; Kroonen
2013: 433);
222
The numeral “five” is irregular even in its cardinal form: Goth. fimf, OE, OS fif, OHG fimf/finf, though the last
plosive is reconstructed as *-ku̯
(cf. L. quinque, Gr. πέντε, OIA paňca-, Lith. penkì), hence the ft of the ordinals
could be a result of a transition due to the analogy with an ordinal form.
216
Note: In some cases, the cluster of *Tt is realized as st, due to analogy, cf. Goth. preterits, waist (wait), haihaist
(haitan), qast (qiÞan), warst (waírÞan), ufsnaist (*sneiÞan), cf. Bammesberger 1986: 95;
Braune/Heidermanns 2004: 73, 75; Hill 2003: 78––217; D. G. Miller 2019: 30), though Kögel (1879; similarly
Brugmann 1880: 132–133 and later editions) assumes the phonetic explanation due to the original position of
accent.
T + s = Germ. ss:
Goth. missō ‘mutuallyʼ, ON ý-miss ‘alternatelyʼ (< IE *√mith
-sto- ?; cf. OIA adv. mithas
‘mutuallyʼ; cf. Prokosch 1939: 85; Pokorny IEW: 715; Lehmann 1986: 257);
ON eisa ‘embersʼ (< IE *√Həts/Hədh
s ?-; cf. Gr. αἶθος ‘burning heat, fireʼ; cf. Prokosch
1939: 85; Pokorny IEW: 11–12; Kroonen 2013: 14);
Goth. us-stagg imp. (hapax legomenon) ‘ἐξελε (pluck out)ʼ (von Grienberger 1900: 233–
234; Brugmann 1913: 180; Brugmann 1913/1914: 284–285 reconstruct *uz-staggan
< IE *√stegh
-, cf. Pokorny IEW: 1014–1015; Lehmann 1986: 380, 383; LIV2
: 589;
Kroonen 2013: 480);
10.2.5 Clusters labial + t/s in Germanic
In Gothic, in productive clusters f regularly replaces any IE labial plosive before t, a similar
process is attested in other Germanic languages in the same contexts.
P + t = Germ. ft:
Goth. gaskafts ‘createʼ, ufar-skafts ‘first fruitsʼ (cf. pt. part. gaskapans; ON scapa; < IE
*√(s)keH-p; cf. Gr. σκῆπτρον ‘stickʼ, L. scāpus ‘shaftʼ; cf. Braune/Heidermanns 2004:
63; Lehmann 1986:223
148–149; Kroonen 2013: 440);
Goth. haftam ‘the married, laden with, subject toʼ, OE hæft, ON haptr ʽcaptive, prisonerʼ
(< IE *√keH2p-; cf. Gr. κάπτω ‘gulp downʼ; L. captus ‘takenʼ; cf. Pokorny IEW: 527;
Streitberg 1974: 114; LIV2
: 344–345; Lehmann 1986: 168; Ringe 2006: 96; Kroonen
2013: 198);
OE nift ‘niece, granddaughterʼ, ON nift ‘female relative, sisterʼ, OHG nift ‘nieceʼ (<
PGerm. neftī- < IE *nept-iH2; cf. OIA naptī́, YAv. napti ‘nieceʼ; cf. Pokorny IEW:
764; Streitberg 1974: 114; NIL: 520–524; Kroonen 2013: 558);
Note: The Germanic cluster of *ft is often lenited as ht in Old Saxon: kraht vs kraft ‘kraftʼ, eht vs eft ‘wiederʼ,
ohto vs ofto ‘oftʼ (cf. Gallée 1993: 164). Similar lenition is known from Celtic and Sabellic.
P + s = Germ. fs:
OE wæfs, wæps, wæsp, OHG wefsa, wafsa, waspa ʽwaspʼ (< IE *u̯ obh
seH2; concurrent
clusters either without spirantization or metathesis; cf, Lith. vapsvà, OCS osa ‘waspʼ;
cf. Pokorny IEW: 1179);
Goth. wulfs, ON ulfr ʽwolfʼ (< IE *u̯ l̥ ku̯
o-; secondary cluster; labial irregular replacement
of IE labiovelar; cf. Pokorny IEW: 1178–1179; Kroonen 2013: 598);
Goth. af-stassais (bokos) ʽnotice of divorceʼ (< IE H2epo- + *√steH2-; hapax legomenon,
an ad hoc calque; cf. Lehmann 1986: 7);224
10.2.6 Clusters sibilant + t/s in Germanic
The IE cluster of *st is preserved in Germanic:
223
Cf. l.c. for the very complicated semantic explanation!
224
Lehmann (1969: l.c.) lists five more similar forms with the prefix af-.
217
S + t = Germ. st:
Goth. staírno, OE steorra ʽstarʼ (< PGerm. *sternan- < IE *H2stēr-; cf. Gr. ἀστήρ ‘starʼ,
L. stella ‘starʼ; cf. Pokorny IEW: 1128; Lehmann 1986: 322; Ringe 2006: 97; NIL:
348–354; Kroonen 2013: 495);
Goth., OHG pr. ist 3rd
sg. ‘beʼ (< PGerm. *isti < IE *H1es-ti; cf. OIA ásti, Gr. ἐστί, L.
est, OLith. esti, OCS jestъ ‘beʼ; cf. Pokorny IEW: 340–341; Lehmann 1986: 205;
LIV2
: 241–242; Braune/Heidermanns 2004: 170; Ringe 2006: 97; Kroonen 2013:
170);
Goth. gast, OE giest ʽguestʼ (< PGerm. *gastiz < IE *H2ostH1-; cf. L. hostis ‘enemyʼ,
OCS gostь ‘guestʼ; cf. Pokorny IEW: 783; Lehmann 1986: 151; Ringe 2006: 97; NIL:
173; Kroonen 2013: 170);
The following example of IE *ss > 0s is probably already Indo-European. There are no
persuasive examples for Germanic clusters of ss arising from IE *ss, so due to analogy we can
expect ss to be preserved (or compensatory shorted after a long vowel):
S + s = Germ. ss:
Goth. pr. 2nd
sg. is ʽbeʼ ( IE *H2es-si; cf. OIA ási, Gr. Aeol. ἔσσι,225
L. es, OLith. esi,
OCS jesi ‘beʼ; cf. Pokorny IEW: 340–341; Lehmann 1986: 205; LIV2
: 241–242;
Braune/Heidermanns 2004: 170; Ringe 2006: 97; Kroonen 2013: 170);
10.2.7 Overview of Germanic development
Germanic development (represented here by Gothic) has a typical sibilantization of both dental
clusters into ss; the velar and labial clusters are replaced by spirant clusters; sibilant clusters are
preserved:
IE Gothic t- s-
-ḱ/ǵ/ǵh
-x/k/g ht hs
-k/g/gh
-x/k/g ht hs
-kṷ
/gṷ
/gṷh
-xṷ
/kṷ
/gṷ
ht hs
-t/d/dh
-Þ/t/d ss ss
-p/b/bh
-f/p/b ft fs
-s s/z st (ss)
10.3 Trajectories of the Germanic development
The development of the Indo-European clusters of plosive + t/s and s + t/s into Germanic can
be split into several sub-blocks, based either on the contexts (t-context, s-context), or on the
series involved.
The oldest layer is the development of the dental series, resulting in both contexts in s,
similarly to the development in Italic and Celtic (more appropriately: in Insular Celtic). Such
225
This cluster is more probably a result of the re-archaization than a relict.
218
development is connected to the analogous process *Tt > st, known from Iranian (but not from
Indic), Slavic, Baltic and Greek.
A younger development is that of all peripheral (grave) series, i.e., of plain velars
(including palatovelars), labiovelars and labials, leading towards spirantization of the plosive
before t/s, again, such development has its counterpart in the development of Sabellic, Celtic,
Iranian and Slavic. The outcomes of this spirantization were same as those reconstructed as
outcomes of Rask-Grimm’s Law, i.e. x, (xu̯
), φ (cf. Ringe 2006: 93–94).
The block of sibilants is relatively stable, as it is in all IE languages outside Celtic.
10.3.1 Development of clusters labial + t/s
Labials, as all peripheral series in Germanic languages, were spirantized in both contexts, and
later debuccalized. As the first step, we assume spirantization of the bilabial spirant, later
shifted to a labiodental spirant (cf. Streitberg 1974: 114; Görtzen 1998: 441; Ringe 2006: 219):
P + t > pt > φt > ft (Germanic)
P + s > ps > φs> fs (Germanic)
Note: The above mentioned Old Saxon development of IE *Pt > OS ht (parallel to the Standard Germanic *ft,
also known from Old Saxon) is further lenition, parallel to that known from Celtic and Sabellic (cf. Gallée
1993: 164), its trajectory is:
P + t > pt > φt > ft (/> ht) (Old Saxon)
10.3.2 Development of clusters velar and palatovelar + t/s
Since there is no distinction between reconstructed plain- and palatovelar series in all centumlanguages,
the development of both series will be dealt with as a single one.
All peripheral series in Germanic languages were spirantized in both contexts, in the
case of the (palato)velars we assume spirantization and later debuccalization (cf. Streitberg
1974: 114; Görtzen 1998: 441; Ringe 2006: 219):
K + t > kt > xt > ht (Germanic)
K + s > ks > xs > hs (Germanic)
10.3.3 Development of clusters labiovelar + t/s
The labiovelar series was regularly neutralized before t/s, as it is in Latin, but the analogy
process corrupted the system. Since the labial marker is regularly neutralized, the output is a
219
plain velar and follows the same trajectory as all velars, with spirantization and debuccalization
(cf. Streitberg 1974: 114; Voyles 1992: 44–45; Ringe 2006: 219; Kümmel 2007: 391):
Ku̯
+ t > kt > xt > ht (Germanic)
Ku̯
+ s > ks > xs > hs (Germanic)
10.3.4 Development of clusters dental + t/s
The traditional trajectory used to describe the development of IE clusters of dental plosive + t
is that of affrication, used since Brugmann (first used 1880: 140–142 and continuously used
since, but the idea was initially stated by Kräuter 1877: 88), assuming the affrication of the first
plosive, later expanded on the right dental plosive, followed by a sibilantization of the whole
cluster.
This model is used in all standard overviews (cf. Prokosch 1939: 85; Görtzen 1998: 441;
Ringe 2006: 18, 87–88; Kroonen 2013: xxxiii; D. G. Miller 2019: 30) and it is, at least its older
stage, a common Indo-European phenomenon, though the output ss is known only from
Germanic, Italic and Celtic, forming either a dialectal continuum or parallel drifts (cf. Ringe
2006: 88). The development of the cluster of *Ts is infrequently modelled, but we can assume
a similar trajectory (inside the affricatization model) with affrication and later simplification:
Note: Miller (2019: 30) brings the analogy to the insertion of s (i.e. for the affrication of Tt) in the speech manner
of Bernese Swiss German, when speaking High German.
T + t > ts
t > tsts > ss (Germanic)
T + s > ts
s > tss > ss (Germanic)
Note: Voyles (1992: 18) reconstructs: Tt > ts
t > ts > ss, Ts > ss (accepted by Kümmel 2007: 391).
Our objections to this model are:
i. the too complex articulation of tsts;
ii. the unknown reason why tss should change to more complex articulation tsts, instead of
using (the phonetically already present!) form ts (however, tss could be easily simplified as ts).
From parallel Continental Celtic we know many examples of tau gallicum (from IE *Tt),
alternating with ss,226
of insecure phonetic value, definitely representing the older stage of the
development, which in its essence had to be the same either in Celtic, Italic or Germanic.
226
Cf. Gal. PN Meliđđus, Meliđđius, Melissus etc. (< *mélit-t-; cf. G. μέλῐ, Goth. miliÞ , W. melys ʽsoftʼ; cf. Pokorny
IEW: 723–724; Evans 1967: 115, 293, 297; Delamarre 2003: 224; Matasović 2009: 263);
220
For this reason, we dare to propose the alternative model, first stated for the Italic development
by de Saussure (1877), Cocchia (1883: 16–58) and Bartholomae (1887) (but cf. also Walde
1897: 498), who assumed that IE *Tt > Italic *ϑϑ. For explicit parallelism between Italic,
Germanic and Celtic, we apply it to Germanic as well227
, assuming the spirantization of the left
dental plosive and the later assimilation of the right plosive to it, both dental spirants being later
sibilantized. Similarly, the IE cluster of *Ts was spirantized first in its left dental plosive; later
this secondary plosive was sibilantized:
T + t > ϑt > ϑϑ > ss (Germanic)
T + s > ϑs > ss (Germanic)
Note: The Germanic cluster of ss (of any origin) could be simplified as 0s due to rhythmical rules (cf. Kroonen
2013: xxxiii).
Note: The different outcomes of IE *Tt either as major ss or minor st were explained as a phonetic regular process
by Kögel (1879), who relates the different outcomes to different accent positions in given words. This solution
was accepted even by Brugmann (cf. Brugmann 1880–132–133 and later). However, Görtzen (1998: 146, 234–
236) assumes that st was an older stage and its preserved examples are archaisms. In contrast, Bammesberger
(1986: 96) argues that the only regular outcome is ss, and that st is a morphological restoration based on the
analogy (on the detailed debate on this subject, cf. Hill 2003: 79–84).
10.3.5 Development of clusters sibilant + t/s
The IE cluster of *st is fully preserved. The IE cluster of *ss is securely attested only for 2nd
sg.
*√es-si, which had been already simplified to *esi in Indo-European (OIA asi, Av. ahi, Gr. εἶ
all prove that the input was -s-, not -ss-). We have to assume the very same situation for
Germanic, otherwise we have not enough data to model any trajectory for IE *ss cluster:
s + t > st (Germanic)
s + s > ss (?) (Germanic)
10.4 Conclusion and final remarks
The oldest development affected, as in other IE languages, is the dental series in both contexts.
It could be summed under the term of the fricativization of the left dental, which followed either
Kräuter/Brugmannian affricativization or de Saussure/Cocchia/Bartholomae’s spirantization
trajectory. Since the development of the *Ts cluster is more probably via spirantization (the *ts
s
cluster is highly improbable to give an ss-output), we prefer the spirantization trajectory.
227
Spirantization is assumed (after an affricatization) by Ringe (2006: 97) at least for original IE *tst/ts: *Tt > ts
t
> ϑt > ss, though he assumes the traditional affricate trajectory otherwise, cf. Ringe 2006: 88.
221
The spirantization later affected even the peripheral series; similar processes are known
from Iranian, Celtic and other IE language branches.
A remarkable feature, probably related to the spirantization of plosives before a t- is that
this t was not affected by Rask/Grimm’s Law, as it was not original *st-cluster. We can deduce
that any fricative (either sibilant or a spirant) in a given cluster blocked the first leg of
Rask/Grimm’s Law (cf. Ringe 2006: 94). However, in a later development, the spirant segment
was debuccalized (except for ft and fs clusters) and the resulting laryngeal approximant was
later fully elided in later Germanic languages. Note that Ringe (2006: 97) noticed that if the
affrication of the ts
t cluster (from IE *Tt) survived until the occurrence of Rask/Grimm’s Law,
the initial t- would be rightly expected to change into ϑ- (the ϑϑ cluster according to the
spirantization trajectory is, due its nature, also immune to Rask/Grimm’s Law). However, we
dare to speculate this proves that the spirantization of the peripheral plosives also preceded
Rask/Grimm’s Law. Such a sequence of the developments could be logical: first, the voiceless
plosives were spirantized in contexts of t/s; later, this spirantization dispersed to other positions
since the position before t/s is a typical neutralization position in the best tradition of
Trubetzkoy way (plosives are neutralized in their voice and aspiration). With the voiceless
plosive, replaced in the neutralization position by a voiceless spirant, the spirant could be
considered a new neutralized allophone and hence a neutral, unmarked form of the phoneme,
which could cause the whole process of the first phase of the Germanic consonantal shift. The
other phases are the consequences of this first phase: with originally voiceless plosives
transformed into voiceless spirants, the original voiced plosives can lose their no longer
necessary voicedness and become newly voiceless and similarly, original voiced aspirated
plosives could be marked now only by the voicedness, aspiration being a redundant value.
We can model transformations of this kind as a sequence three of stages (T = any nonaspirated
voiceless plosive; D = any voiced non-aspirated plosive; Dh
= any aspirated voiced
plosive; Θ = any voiceless spirant):
i. in the first stage (the input = Late Indo-European) the neutralization before *t/s- is regularly
realized; since this position is the neutralization, the original basic value is preserved in the
non-alternating position (here the antevocalic position);
ii. after spirantization in the old neutralization position before *t/s-, the old basic value is
preserved in the antevocalic position;
iii. the spirant in the neutralization position is re-evaluated as an unmarked member, and
spirantization is extended on the old voiceless non-aspirates through all positions. With the
transition IE T > Germ. Θ, the old voiced non-aspirates do not have to be marked by the
voicedness and are devoiced IE D > Germ. T; similarly old voiced non-aspirates could be
revaluated as simple voiced plosives/spirants, IE Dh
> Germ. D/Ð :
222
input spirantization output/revaluation
antevocalic neutralization antevocalic neutralization antevocalic neutralization
position position position position position position
TV Tt TV Θt ΘV Θt
Ts Θs Θs
DV TT > DV ΘT > TV ΘT
Ts Θs Θs
Dh
V TT Dh
V ΘT DV ΘT
Ts Θs Θs
The development of the clusters of sibilant + t/s is conservative (sibilants block Rask/Grimms’s
Law on t-). We assume similar process for the IE cluster *ss.
223
11 The development of the two-obstruent clusters from Indo-European into
Anatolian
11.0 Anatolian languages
The Anatolian subgroup of Indo-European languages is attested from the middle of the second millennium BC to
last centuries BC, with evidence of various quality and quantity, including either administrative archives and long
texts of Hittite on the one hand and but few inscriptions for Pisidian and other later languages (cf. Melchert 2017:
171–172) on the other hand.
In the following lines, we will focus exclusively on Hittite data, representing the oldest and best-attested
segment of Anatolian; other languages (very poorly attested, beside Luwian) are omitted. Hittite was preserved in
the Assyrian cuneiform script, containing not only syllabograms sensu stricto but also Akkadograms or
sumerograms (the reason why, for example, our knowledge of Hittite numerals is sketchy). A remarkable feature
is that Hittite script does not use the Assyrian contrast of voiceless/voiced obstruents – such features of the script
obscure our understanding of the Hittite data (for such reasons was Hittite long considered to do now have any
modal opposition), which affects our understanding of the phonemic data more than anything else, as we will see
below (cf. especially Hoffner/Melchert 2008: 9–24).
The particular problem is the writing of the consonantal clusters, since the cuneiform script has graphemes
for vowels, consonant + vowel, vowel + consonant and consonant + vowel + consonant and the cluster had to be
expressed by a combination only within of this limited set (cf. Friedrich 1974: 29; Hoffner/Melchert 2008: 11).
11.1 Anatolian and Indo-European
The typical features separating the Anatolian obstruent system from that of Indo-European are:
i. the merging of the reconstructed voiced non-aspirated and voiced aspirated plosives in a
single modal series (the original voiceless non-aspirated plosives are preserved);
ii. the merging of the assumed palatovelars and plain velars in a single series: the often claimed
preservation of palatovelars in Luwo-Lycian has to be abandoned (cf. Melchert 2012;
Melchert 2017: 176);
iii. the preservation of the labiovelar series228
(cf. Kronnasser 1956: 65–68; Lindemann 1965;
Melchert 1994: 92, 120);
Note: The transliteration we will use is that of Friedrich 1990, including writing z to mark an affricate (ts). The
precise phonetic value of a single Hittite sibilant is unknown and a matter of debate, but since it represents a
single IE sibilant and stays a single Hittite sibilant, the phonemic status is the same, regardless of its phonetic
value. Since for a graphic realization the ideogram used is that of Akkadian š-graphemes, we will use the š
sign, according to the prevailing tradition, especially since Ugaritic writing of the name of Hittite king
Šuppiluliuma is with š-, not with s-sign (Ugaritic has this option) (cf. Hoffner/Melchert 2008: 38). However,
for its phonemic value will use s-signs when describing trajectories, i.e., the probable palatalization is entirely
omitted.
11.2 Hittite clusters and their IE origins
The merging of the old voiced non-aspirates and the voiced aspirates is a common process in
various Indo-European languages: similar processes are known from Iranian, Baltic, Slavic,
Albanian, and Celtic languages. In some cases these could be related (in the Balto-Slavic),
although they are often not (Celtic and other languages).
228
Note that Sturtevant (1933: ) and Sturtevant/Hahn (1951: 55) reject the phonemic status of labiovelars in Hittite.
224
In Anatolian two modal classes are distinguished; the phonetic value of this opposition
is still a matter of debate. For simplicity, we will use symbols and terms for the voiceless and
voiced modal classes (cf. Melchert 1994: 13–21; Hoffner/Melchert 2008: 25–26; Melchert
2017: 176–177), though we have to mention the different view of Kloekhorst, working with the
opposition of geminated and non-geminated stops (Kloekhorst 2008: 21–25). The re-evaluation
of the opposition in terms of fortis/lenis is another option (Melcher 1994: 18–20). However, we
are focused on the position in neutralization contexts, not on the phonetic values in other
contexts.
Note: However, voiceless stops often became voiced, either between unaccented vowels or after an accented long
vowel or diphthong (Eichner 1973: 79–80, 10086
; Eichner 1980: 14669
, Melchert 1994: 60–61)
The labiovelars are otherwise preserved in Mycenaean, Latin, the older phase of Goidelic
Celtic, Tocharian and Germanic, and often delabialized in later stages of the development of
the given branches.
11.2.1 The clusters labial + t/s
The Indo-European clusters of the labial plosive + t/s are fully preserved:
P + t = Hitt. pt/pts
:229
Hitt. šiptamiya- ‘a liquid consisting of seven ingredientsʼ; Luw. sap(pa)tammimma/i‘sevenfoldʼ
(< PAnat. *šiptama- ‘sevenʼ < IE *septmó-; cf. OIA saptá-, L. septem
‘sevenʼ; cf. Hrozný 1917: 96; Sturtevant 1933: 117; Sturtevant/Hahn 1951: 60;
Kronasser 1956: 61; Pokorny IEW: 909; Eichner 1992: 84–85; Friedrich 1990: 194;
Melchert 1994: 156; Blažek 1999: 247; Kloekhorst 2008: 775–756; CHD Š: 400);
Hitt. epzi ‘take grab, captureʼ (< IE *√H1ep-; cf. OIA ī́psati ‘reachʼ, OL. apiō ‘fastenʼ; cf.
Pokorny IEW: 50–51; Friedrich 1974: 79; Friedrich 1990: 41–42; LIV2
: 237; HED 1–
2: 273–282; Kloekhorst 2008: 242–243);
Hitt. pr. 3rd
sg. teripzi ‘ploughʼ (< IE *√trep-; cf. Gr. τρέπω, L. trepō ‘turnʼ; cf. Pokorny
IEW: 1094; Sturtevant 1933: 125; Sturtevant/Hahn 1951: 60; Friedrich 1990: 221;
LIV2
: 650; Kloekhorst 2008: 871–872);
Hitt. 3rd
sg. pret. lapta ‘glowʼ (< IE *√leH2p-; cf. Gr. λάμπω ‘give light, shineʼ, Lith. lópė
‘lightʼ; cf. Kronasser 1956: 60; Pokorny IEW: 652–653; CHD L–N: 39–40; Friedrich
1990: 127; LIV2
: 402; Kloekhorst 2008: 519–520);
P + s = Hitt. ps:
Hitt. tepšu- ‘dry (?), reduced (?)ʼ (< IE *√dh
ebh
-; cf. OIA ī́dabhnóti ‘deceiveʼ, Lith. dóbiu
‘invalidateʼ; cf. Pokorny IEW: 540; Friedrich 1990: 221; Melchert 1994: 156; LIV2
:
132–133; NIL ; HED 1–2: 273–282; Kloekhorst 2008: 866–868);
Hitt. epši ‘take grab, captureʼ (< IE *√H1ep-; cf. cf. OIA ī́psati ‘reachʼ, OL. apiō ‘fastenʼ;
Pokorny IEW: 50–51; Friedrich 1974: 79; Friedrich 1990: 41–42; LIV2
: 237; NIL 85–
86; HED 1–2: 273–282; Kloekhorst 2008: 242–243);
229
The dental could be affricated due to the palatal context.
225
11.2.2 The clusters plain velar + t/s
The IE cluster *Kt is preserved as such. There is no sure example for *Ks (but since merging
of IE clusters of *Ks and *Ḱs, we can be sure that the outcome was in Hittite also ks).
K + t = Hitt. kt/kts
:230
Hitt. 3rd
sg. pr. lukzi, 3rd
sg. pret. lukta ‘shineʼ (< IE *√leu̯ k-; cf. Gr. λευκός ‘light, bright,
clear, whiteʼ, L. lūx ‘lightʼ, Lith. laũkas ‘white-spottedʼ; cf. Sturtevant 1933: 116;
Sturtevant/Hahn 1951: 57; Kronasser 1956: 64, 73; Pokorny IEW: 687–690; CHD L–
N: 76; Friedrich 1990: 130; Melchert 1994: 156; HED 1–2: 103–108; LIV2
: 418–419;
Kloekhorst 2008: 530–531);
Hitt. pr. 3rd
sg. harakzi, 2nd
pl. pr. ḫarakteni ‘perishʼ (< IE *√H3erg-ti; cf. Arm. harkanem
‘split, fell, smiteʼ; cf. Sturtevant/Hahn 1951: 57; Kronasser 1956: 65, 166; Friedrich
1990: 57; HED 3: 135–137; LIV2
: 301; Kloekhorst 2008: 304);
Hitt. šaktā(i) ‘sick–maintainʼ (< IE *√sok-to- e-; OIr socht ‘stuporʼ; cf. Friedrich 1990:
177; Melchert 1994: 93, 156; HED 10: 47–48; Kloekhorst 2008: 701; CHD 10: 51–
52)
Hitt. ikt-/ekt-ikdu-/ekdu- ‘sole, a part of the foot (?)ʼ/ʼhunting net (?)ʼ (< IE *√ ek-t; cf.
Friedrich 1990: 81; Melchert 1994: 156; HED 1–2: 259–260; Kloekhorst 2008: 235)
11.2.3 The clusters palatovelar + t/s
Since there is no distinction between the IE plain velars and the palatovelars, even outcomes of
IE clusters of *Ḱt are the same as *Kt. The outcome of IE *Ḱs is regularly ks in Hittite:
Ḱ + t = Hitt. kt/kts
:231
Hitt. 3rd
sg. pr. u̯ ekzi, 3rd
sg. pret. u̯ ekta ‘askʼ (< IE *√u̯ eḱ-; cf. OIA vaṣṭi ‘wishʼ, Gr. ἑκών
‘willinglyʼ; cf. Sturtevant 1933: 117; Sturtevant/Hahn 1951: 57; Pokorny IEW: 672;
Friedrich 1974: 80; Friedrich 1990: 251; LIV2
: 672–673; Kloekhorst 2008: 996–997);
Hitt. 3rd
sg. pr. šalikzi, 3rd
sg. pret. šalikta ‘fallʼ, 2nd
pl. šalikteni ‘prostrateʼ (< IE *√sle ǵh
-
/legh
- (?); cf. Gr. λέχος ‘bedʼ, L. lectus ‘bedʼ, Goth. ligan ‘lieʼ; cf. Sturtevant 1933:
117; Sturtevant/Hahn 1951: 57; Pokorny IEW: 658–659; Friedrich 1990: 179–180;
LIV2
: 398; CHD 10: 100–104; Kloekhorst 2008: 711);
Ḱ + s = Hitt. kš:
Hitt. nom. takšan- ‘centre, joint, combinationʼ, adv. takšan ‘togetherʼ, nom. takšeššar
‘combinationʼ, (verbal forms with an anaptyctic vowel: 3rd
sg. pr. takkešzi, 2nd
pl. pr.
takkešteni ‘put together, undertakeʼ) (< IE *√tek-s-; Gr. τέχνη ‘skillʼ, OHG dehsala‘axeʼ;
cf. Pokorny IEW: 1058; Melchert 1994: 156; Friedrich 1990: 204–205; LIV2
:
619–620; Kloekhorst 2008: 813–814; CHD Š: 100–104);
11.2.3 The clusters labiovelar + t/s
230
The dental could be affricated due to the palatal context.
231
The dental could be affricated due to the palatal context.
226
The neutralization of the labial value of the IE labiovelar *Ku̯
t was originaly probably regular
(as in Latin) but is often replaced by analogy. The cluster of *Ku̯
s is preserved only without
neutralization:
Ku̯
+ t = Hitt. ku̯
t/ku̯
ts
:232
Hitt. 3rd
sg. pret. ekuta, 3rd
sg. pr. ekuzi (beside e-uk-zi with neutralization?) ‘drinkʼ (< IE
*√H1egu̯ h
-; cf. Toch. A yok- ‘drinkʼ, L. ēbrius ‘drunkʼ; cf. Pokorny IEW: 23; Friedrich
1990: 40; Melchert 1994: 92; LIV2
: 231; HED 1–2: 261–268; Kloekhorst 2008: 236–
237);233
Hitt. nekut- ‘twilight, eveningʼ (< IE *√negu̯
-; cf. Gr. νύξ, L. nōx, L. naktìs; cf. Pokorny
IEW: 762–763; Puhvel 1972: 112; CHD L–N: 434–437; Friedrich 1990: 150; Melchert
1994: 61–62, 156; LIV2
: 449; NIL: 504–513; HED 7: 79–83; Kloekhorst 2008: 602);
Note: But note that there is an etymological cluster of *Ku̯
t > Hitt.: Hitt. 3rd
sg. pr. ḫuekzi, 3rd
sg. pret. ḫuekta
‘conjureʼ (< IE *√u̯ eku̯
-; cf. OIA vakti, Gr. (ϝ)έπος, L. vōx; cf. Sturtevant 1933: 121; Kronasser 1956: 66;
Pokorny IEW: 1135–1136; Friedrich 1990: 70; LIV2
: 673; Kloekhorst 2008: 347–348), similarly probably euk-zi,
noted above as a variant of 3rd
sg. pr. ekuzi.
Ku̯
+ s = Hitt. ku̯
s:
Hitt. tekkuššāi-, tekkuššanu- ‘showʼ (< IE *√deku̯
-s-; Av. daxš- ‘to teachʼ; cf. Pokorny
IEW: 189; Friedrich 1990: 220; Melchert 1994: 61–62, 113, 156; LIV2
: 112;
Kloekhorst 2008: 865; Kimball 2017: 253);
Hitt. 2nd
sg. pr. ekušši ‘drinkʼ (< IE *√H1egu̯ h
-s-; cf. Toch. A yok- ‘drinkʼ, L. ēbrius
‘drunkʼ; cf. Pokorny IEW: 23; Friedrich 1990: 40; Melchert 1994: 92; LIV2
: 231; HED
1–2: 261–268; Kloekhorst 2008: 236–237);
Hitt. nana(n)kušši(iye) ‘grow darkʼ (< PAnat. *no-noku̯
-s- < IE *√ne-negu̯
-s.; cf. Pokorny
IEW: 762–763; CHD L–N: 394–395; Melchert 1994: 61–62; LIV2
: 449; NIL: 504–
513; HED 7: 60–62, 79–83; Kloekhorst 2008: 595);
11.2.4 The clusters dental + t/s s
The first dental in the cluster of *Tt is affricated; the cluster *Ts is preserved:
T + t = Hitt. ts
t (<zt>):
Hitt. 3rd
sg. pret. ezta, 2nd
pl. pr. ezzteni ‘eatʼ (< IE *√H1ed-; cf. OL. est ‘he eatsʼ, Gr. Att.
pr. ἐσθίω; cf. Hrozný 1917: 89; Sturtevant 1933: 127, 129; Sturtevant/Hahn 1951: 61,
63; Kronasser 1956: 61; Pokorny IEW: 287–289; Friedrich 1990: 44; Melchert 1994:
97, 109; Kimball 1999: 285–286; LIV2
: 230; NIL: 208–220; HED 1–2: 315–321;
Kloekhorst 2008: 261–263);
Hitt. 3rd
sg. pret. ḫazta ‘dry upʼ (< IE *√H2ed-; cf. Gr. pr. ἄζω; cf. Pokorny IEW: 68–69;
Friedrich 1990: 64; Kimball 1999: 286; LIV2
: 255; HED 3: 247–248; Kloekhorst 2008:
328–329);
Hitt. 3rd
sg. pr. ḫuezta ‘pullʼ (< IE *√H2eu̯ t-(?); cf. Friedrich 1990: 72–74; Kimball 1999:
286; LIV2
: 294; HED 3: 343–352; Kloekhorst 2008: 349–351);234
Note: Some of the old tt geminates are preserved without any change: Hitt. atta- ‘fatherʼ (cf. Friedrich 1990: 38;
Melchert 1994: 150; Kloekhorst 2008: 225).
232
The dental could be affricated due to the palatal context.
233
But note that e-uk-zi is delabialized, cf. Friedrich 1990: 40).
234
Not attested outside Anatolian.
227
T + s = Hitt. ts
(<z>):
Hitt. 2nd
sg. pr. ezši ‘eatʼ (< IE *√H1ed-; cf. OL. es ‘he eatsʼ, Gr. Att. pr. ἐσθίω; cf. Hrozný
1917: 89; Pokorny IEW: 287–289; Friedrich 1990: 44; Melchert 1994: 97, 109; LIV2
:
230; NIL: 208–220; HED 1–2: 315–321; Kloekhorst 2008: 26, 261–263);
Hitt. ḫazziya- ‘strikeʼ (< PAnat. Hats e/o- < IE *H2et-?); cf. Friedrich 1990: 67; Melchert
1994: 96; LIV2
: 274; HED 3: 248–255; Kloekhorst 2008: 330–332);235
Hitt. loc. suffix -zzi(ya)- (< PAnat. -ts o-; cf. Melchert 1994: 96);
Note: There are examples of the development of *Ts > šš, attested only over the clitic boundary (cf.
Hoffner/Melchert 2008: 41).
11.2.6 The clusters sibilant + t/s
The cluster of sibilant + t is fully preserved, as is the ss cluster, as far as we can depend on the
scarce data:
S + t = Hitt. st/sts
:236
Hitt. talukašti, dalugašti ‘lengthʼ (< IE *√dolgh
-; cf. OCS dlъgostь; cf. Hrozný 1917: 23;
Kronasser 1956: 65; Pokorny IEW: 197; Friedrich 1990: 206; Kloekhorst 2008: 819–
820);
Hitt. kašt ‘hungerʼ (< IE *√gh
os-; cf. Toch. A kaṣt, Toch. B. kest, OIA ghasati, kṣut
‘hungerʼ; cf. Sturtevant 1933: 118; Sturtevant/Hahn 1951: 58; Kronasser 1956: 65;
Pokorny IEW: 452; Friedrich 1990: 104; LIV2
: 198–199; HED 4: 121–123; Kloekhorst
2008: 461–463);
Hitt. 3rd
sg. ešzi, 2nd
pl. pret. ešten ‘beʼ (< IE *√H1es-; cf. OIA ásti, Gr. ἐστί, L. est, OLith.
esti; cf. Kronasser 1956: 169; Pokorny IEW: 340–342; Friedrich 1990: 42; LIV2
: 241–
242; NIL 235–238; HED 1–2: 285–291; Kloekhorst 2008: 250–252);
Hitt. ḫaštāi ‘boneʼ (< IE *√H3esth
-; cf. OIA ásthi, Gr. ὀστέον; cf. Sturtevant 1933: 124,
139; Sturtevant/Hahn 1951: 59, 65; Pokorny IEW: 783; Friedrich 1990: 63; HED 3:
233–237; Kloekhorst 2008: 325)
Hitt. ḫašduir ‘brushwoodʼ (< IE *√H2stH1-gu̯
er-; cf. Gr. ὄζος, Aeol. ὔσδος ‘bough,
branch, twigʼ; cf. Sturtevant 1933: 139; Sturtevant/Hahn 1951: 65; Kronasser 1956:
68; Pokorny IEW: 786; Friedrich 1990: 64; HED 3: 239–240; Kloekhorst 2008: 326)
Hitt. ḫaštēr ‘starʼ (< IE *√H2ster-; cf. Gr. ἀστήρ, L. stella; cf. Sturtevant 1933: 77, 124;
Kronasser 1956: 16, 204; Pokorny IEW: 1027–1028; NIL 348–354; HED 3: 238–239;
Kloekhorst 2008: 326);
S + s = Hitt. ss:
Hitt. pr. 2nd
sg. ešši ‘beʼ (< IE *√H1es-; cf. OIA ási, L. es, OLith. esi; cf. Kronasser 1956:
169; Pokorny IEW: 340–342; Friedrich 1990: 42; LIV2
: 241–242; NIL 235–238; HED
1–2: 285–291; Kloekhorst 2008: 250–252, especially see the commentary here);
Note: Though examples of the development of the original IE *ss clusters are scarce, Hittite sibilants were often
geminated, cf. Melchert 1994: 120–123).
235
Not attested outside Anatolian.
236
The dental could be affricated due to the palatal context.
228
11.2.7 Overview of Hittite development
The Hittite development of our clusters is highly conservative, only IE *Tt clusters are
alternating (as in other IE languages). The assumed neutralization of the IE old labiovelars in
the t- and s-contexts is usually replaced by analogous forms, the old neutralization preserved
exceptionally:
IE Hittite t- s-
-kṷ
/gṷ
/gṷh
-kṷ
/gṷ
k(u̯ )
t ku̯
s
-k/g/gh
-k/g kt ks
-ḱ/ǵ/ǵh
-k/g kt (ks)
-t/d/dh
-t/d tst ts
-p/b/bh
-p/b pt ps
-s -s st ss
11.3 Trajectories of the Hittite development
Though some examples are scarce, we still can postulate that the development of clusters of
plosive/s + t/s in Hittite is very conservative, as the single series affected by any changes is the
dental one. We assume the devoicing before both t/s (cf. Kimball 1999: 300–301).
11.3.1 Development of clusters labial + t/s
Similarly to all peripheral series, labial clusters with t/s are fully preserved in Hittite:
P + t > pt
P + s > ps
11.3.2 Development of clusters plain velar and palatovelar + t/s
Since there is no distinction in Hittite between the Indo-European reconstructed plain velars
and palatovelars, the trajectory is the same (cf. Kloekhorst 2008: 72); the velar plosives are
fully preserved as such:
K + t > kt
K + s > ks
11.3.3 Development of clusters labiovelar + t/s
For the development of the cluster of labiovelar + t, we assume the original neutralization of
the labial marker, attested otherwise fully in Latin and in diffused examples in other centumlanguages.
Later, this old neutralization is replaced by analogical clusters with reintroduced
229
labiovelars. The very same process is attested even for clusters with s (cf. Melchert 1994: 61–
62, 113; Kloekhorst 2008: 72):
Ku̯
+ t > kt/→ ku̯
t
Ku̯
+ s > (ks ?)/→ ku̯
s
11.3.4 Development of clusters dental + t/s
The IE cluster *Tt developed in Hittite into a cluster of ts
t (cf. Melchert 1994: 22, 62, 97, 109,
117120; Hoffner/Melchert 2008: 37, 44). The first segment is a dental (voiceless) affricate, and
the same affricate is a result of the development of the dental plosive before a palatal *i (cf.
Melchert 1994: 54, 116–117; Kümmel 2007: 350).
This development supports the traditional theory assuming that IE *Tt regularly
developed into *ts
t (before a further development either in st or ss, according to the given branch
of Indo-European languages). This model originates with Kräuter (1877: 88), but became a popular
model due to Brugmann’s influen
ce
(since Brugmann 1880: 140–142); Anatolian then would be a single
IE branch preserving the older state, lost in other branches due to further development.
Note: Kloekhorst (2008: 26) assumes that the outcome of the affrication in Tt and Ts clusters is different from that
of T , but this distinction is not fully phonetically explained (cf. l.c.). A simple solution could be that proposed by
Hoffner/Melchert (2008: 37): one of the outcomes was palatalized, probably that originating from T . His statement
that even the outcomes of clusters of dental plosive + t/s are different according to the original voice(less)ness of
the input cluster remains more questionable, especially since there is no similar distinction from other clusters with
plosives other that dental ones.
T + t > ts
t
Note: Görtzen (1998: 424) lists išduu̯ a- ‘announceʼ as an example of the Bartholomae’s Law operatibility in Hittite
but this seems hardly probably (cf. HED 1–2: 483–485; Kloekhorst 2008: 419–420 for other etymologies).
The Indo-European cluster *Ts is either preserved as such (i.e., as a biphonematic sequence ts)
or as an affricate (ts
) (cf. Melchert 1994: 96; Kloekhorst 2008: 72). Another possibility is that
the plosive was subjected to affricatization, as in the case before t and this cluster was simplified
as ts. We prefer this solution because of the analogy:
T + s > (ts
s >) ts
Note: Kloekhorst (2019) argues that Ts clusters are realized in Hittite regularly as a ts cluster, never as an affricate,
at least in the antevocalic positon or before a diareme.
11.3.5 Development of clusters sibilant + t/s
The IE clusters of sibilant + t/s are fully preserved in Hittite (cf. Kloekhorst 2008: 70–71),
though for the cluster *ss we have a minimal set of examples:
230
s + t > st
s + s > ss
11.4 Conclusion and final remarks
The Hittite developments of the Indo-European clusters of our interest is an example of a very
conservative development.
The dental series was subjected, as in all other IE languages, to the alternation of the left
dental, which was affricated and preserved as such. A remarkable feature is that the IE *Ts
cluster is preserved as such (in other languages the cluster is realized often as a double or single
sibilant – cf. the Iranian, Slavic, Baltic, Greek, Italic, Celtic and Germanic) – in this case either
the left dental was not affected by any process, or the probable affricativization (parallel to that
of the *Tt-cluster) was later simplified/re-archaized.
The clusters of the peripheral series, on the contrary, are fully preserved in both contexts.
The labiovelar series, originally with the neutralization of the labial value (as in Latin) is often
restored in this feature due to analogy processes.
231
12 The development of the two-obstruent clusters from Indo-European into
Tocharian
12.0 Tocharian languages
There are two Tocharian languages, termed Tocharian A and Tocharian B,237
both derived from a single centumlanguage,
attested from findings in the Tarim Basin.
Texts are written in the local version of the brāhmī script (a recent overview by Malzahn 2007b) and are
either translations and adaptations of Indian Buddhist texts or non-literal texts from the Buddhist milieu usually
written on paper, but also on wood and especially graffiti (Penney 2017; for an overview of manuscripts and their
location, cf. especially Malzahn 2007a).238
Since the Tocharian B is better attested than Tocharian A, the following analysis will be based on this
language. Tocharian A serves, if possible, as a complement (though listed, for alphabetic reasons, before B
examples). However, both languages do not usually differ in outcomes of the analysed clusters, as we will see
below.
12.1 Tocharian and Indo-European
Both Tocharian languages have been subjects of linguistic examination for just one century,
and many details of their phonemic development are not fully understood yet.
The typical features separating Tocharian obstruent systems from that of Indo-European are:
i. the merging of the three modal classes, reconstructed for the Indo-European protolanguage;
ii. the preservation of old labiovelars (though in minimal scale due to later processes);
iii. the palatalization of plosives (p vs ṕ; k vs ś; t vs c; old labiovelars are palatalized in the same
way as plain velars) and of s (as ṣ).
Note: Winter (1962b: 24–25) brings three examples of the alleged validity of Grassmann’s Law in Tocharian, all
of the deaspiration of the root-initial *dh
- (not any other voiced aspirate), cf. Ringe 1996: 47. However,
Grassmann’s Law is outside of the scope of our interest. Similarly, Winter (2011) tries to demonstrate that
Proto-Tocharian had a similar lengthening of vowels before IE voiced non-aspirated plosives to that of
Balto-Slavic (i.e., to Winter’s Law in sensu stricto and to Lachmann’s Law).
The first feature is unknown in any other IE language (though older opinions wrongly assumed
the same development also for Hittite); the second feature is a parallel process common in all
centum-languages; the third feature appears independently in various IE languages.
12.2 Tocharian clusters and their IE origins
Both Tocharian languages preserve old clusters with peripheral series. The dental series
underwent a similar development as in all Indo-European languages, in this feature it does not
differ from Hittite, Latin and Greek, which, as centum-languages, similarly keep the clusters of
237
The name is a misnomer, resulting from a false connection between Bactrian Τόχαροι and the Tarim Basin
inhabitants, but is preserved due to a long tradition.
238
Though there are at least two known fragments written in the Manichean script, one a hymn to Mani and the
other a hymn to Jesus.
232
peripheral series relatively intact (satəm Old Indo-Aryan or Lithuanian have similar
development).
12.2.1 The clusters labial + t/s
As far as we can judge from the single attested data, the clusters of *Pt are preserved as such,
at least in Tocharian A. The *Ps cluster is reduced to 0s in the word-initial (but we have a single
example only of such a development and it is doubtful), or often an epenthetic vowel splits the
cluster:
P + t = Toch. pt/pts
:
Toch. A num. pl. ṣäptäntu, ṣäpta- (in compounds),239
Toch. B num. suk, ṣukt ‘sevenʼ (<
Toch.*ṣäp(ä)t, ṣäktu < IE *septm̥ ;240
cf. L. septem, OIA saptá-, Gr. ἑπτά; cf.
Sieg/Siegling 1908: 927; Pokorny IEW: 909; Winter 1992b: 109, 137–138; Ringe
1996: 67; Blažek 1999: 250; Adams 2013: 720);
Toch. A pr. kroptär, inf. kroptsi, Toch. B pr. krauptär ‘gather, assembleʼ (< Toch.
*√kreup- < IE *kreu̯ bh
-; cf. Gr. κρύπτω ‘hideʼ, without p-suffix cf. OCS kryjǫ ‘coverʼ;
cf. van Windekens 1976: 235; Carling 2009: 173–174; Malzahn 2010: 614–616;
Adams 2013: 236–238);
Toch. B pr. kleptär ‘touch, investigateʼ (< IE *√klep-; cf. LIV2
: 246; Malzahn 2010: 618–
619; Adams 2013: 246);
Toch. B inf. yaptsi ‘enter, set (sun)ʼ (< Toch. *√ äp- < IE *√ ebh
-; cf. OIA yábhati, OCS
jebǫ ‘fuckʼ; cf. Pokorny IEW: 298; van Windekens 1976: 605241
; LIV2
: 309; Malzahn
2010: 796–798; Adams 2013: 537–538);
P + s = Toch. ps:
Toch. B pr. lupṣtär, grd. lupṣalle ‘smear, rubʼ (< Toch.*√läup- < IE *√sleu̯ bh
-; cf. L.
lūbricus ‘slipperyʼ, Goth. sliupan ‘slink, crawlʼ; cf. Pokorny IEW: 963–964; van
Windekens 1976: 269; Malzahn 2010: 858–859; Adams 2013: 606);
Toch. B pt. yopsa ‘enter, set (sun)ʼ (< Toch. *√ äp- < IE *√ ebh
-; cf. OIA yábhati, OCS
jebǫ ‘fuckʼ; cf. Pokorny IEW: 298; van Windekens 1976: 605; LIV2
: 309; Malzahn
2010: 796–798; Adams 2013: 537–538);
Note: P + s = Toch. pVs (with an epethetic vowel) is attested in Toch. A päśśäṃ, Toch. B päścane ‘breast; [in
plural] seat of wisdomʼ (< IE *pstḗm; cf. Hitt. istanza ‘soulʼ, Av. fštāna-; cf. van Windekens 1976: 103; Ringe
1996: 71; Adams 2013: 386);
#Ps > Toch. #0s: Toch. A sāt ‘hotʼ, Toch. B satāsk ‘exhaleʼ (< IE *ps-ōd-, derived from IE *√bh
es-; cf. OIA
√bhas- ‘breathʼ; cf. van Windekens 1976: 103, 419–420; Adams 2013: 736–737 doubt this etymology);
12.2.2 The clusters velar + t/s
239
Tocharian A ṣpät “seven” is a result of a metathesis.
240
The Tocharian B form is affected by the numeral eight. This particular trajectory was given by van Windekens
(1976: 461) as: *septm̥ > *śäptäṃ > ṣäptu → ṣäktu > *ṣukt; in contrast, by Winter (1992b: 109) as: *ṣäwät >
*swät > *ṣut → ṣukt.
241
Van Windekens gives a different etymology, c. l.!
233
Though secure examples of the development of IE plain velars + t/s are relatively scarce, we
have to assume there the outcomes are not different from the outcomes of IE palatovelars + t/s
(cf. below), since Tocharian is a centum-language:
K + t = Toch. kt/kts
:
Toch.B pr. klyeñktär, inf, kläṅktsi ‘doubtʼ (Toch. *√kläṅk- < IE *√kleng-; cf. L. clingō,
OE hlinc ‘ringeʼ; Pokorny IEW: 603; Carling 2009: 177; Malzahn 2010: 623–624;
Adams 2013: 240);
Toch. B prtcl. täṇktsi ‘up to, until; including, evenʼ (historical infinitive of Tocharian
*√täṅk- < IE *√tengh
-; cf. OIA tȩgnǫti ‘pullʼ, ON Þungr ‘heavyʼ; Pokorny IEW: 1167;
van Windekens 1976: 502; LIV2
: 567; Malzahn 2010: 648; Adams 2013: 307);
Toch. B pr. wākṣtär-ś ‘split apartʼ (< Toch. *√u̯ āk- < IE *√u̯ ag-; cf. Gr. ἄγνῡμι ‘break
apartʼ, L. uāgīna ‘sheathʼ (?); cf. Pokorny IEW: 1110; van Windekens 1976: 550–551;
LIV2
: 664–665; Malzahn 2010: 862–863; Adams 2013: 635–636);
K + s = Toch. ks/kṣ:
Toch. A grd. taṅkṣäl, Toch. B pr. taṇksäṁ ‘check, stop, hinderʼ (Toch. *√täṅk- < IE
*√tengh
-; cf. OIA tȩgnǫti ‘pullʼ, ON Þungr ‘heavyʼ; Pokorny IEW: 1167; van
Windekens 1976: 502; LIV2
: 567; Malzahn 2010: 648; Adams 2013: 306);
Toch. B pr. trikṣäṃ ‘go astray, be confusedʼ, nom. trikṣo-’error, mistakeʼ (Toch. *√trik<
IE *√tre k- (?); cf. L. trīcae ‘trifles, nonsense; vexations, troublesʼ, trīcārī ‘make
difficulties; shuffle; trifleʼ; Pokorny IEW: 1167; van Windekens 1976: 514–515; LIV2
:
514–515; Malzahn 2010: 668–670; Adams 2013: 334–335);
Toch. A pr. okṣoññäṃ, Toch. B pr. auksäṣṣäṁ ‘grow, increaseʼ (< Toch. *√ṣäk(ä)s- < IE
*√H2eu̯ ks; cf. OIA vakṣáyati ‘growʼ, Gr. ἀέξω ‘increase, fosterʼ, Goth. wahsjan
‘growʼ; cf. Pokorny IEW: 85; van Windekens 1976: 329; LIV2
: 288–289; NIL: 368–
370; Malzahn 2010: 549–550; Adams 2013: 1387);
Toch. A nt-part. wākṣantāṃ ‘split apartʼ (< Toch. *√u̯ āk- < IE *√wag-; cf. Gr. ἄγνῡμι
‘break apartʼ, L. uāgīna ‘sheathʼ (?); cf. Pokorny IEW: 1110; van Windekens 1976:
550–551; LIV2
: 664–665; Malzahn 2010: 862–863; Adams 2013: 635–636);
12.2.3 The clusters palatovelar + t/s
The clusters of IE palatovelar + t/s are preserved as kt/ks:
Ḱ + t = Toch. kt/kts
:
Toch. A okät,242
B ok(t) ‘eightʼ, Toch. A oktuk ‘eightyʼ, Toch. A oktänt, Toch. B oktante,
oktunte ‘eighthʼ (Toch. *oktu < IE *oḱtō; cf. L. octō, Goth. ahtau, Gr. ὀκτώ; cf.
Pokorny IEW: 775; van Windekens 1976: 330–332; Winter 1992b: 110–112; Blažek
1999: 268; Carling 2009: 83; Adams 2013: 115);
Adams (2013: 202–203) relates Toch. B ketseñe ‘bodyʼ to a verbal abstract *ku̯
oḱ-ti- (<
IE *√ku̯
eḱ- ‘see, appearʼ; cf. Pokorny IEW: 638–639; LIV2
: 664–665; van Windekens
(1976:103, 187–188) relates it to OIA cakṣus- ‘aspect, formʼ (< IE √ku̯
oḱ-s-, i.e., to the
same root but with a different suffix, Adams considers this form irregular), Pinault
1999 derives from a collective *koḱse-den-);
Toch.B inf. praktsi ‘askʼ (< Toch. *pärk-sa- < IE *√preḱ-; cf. OIA pṛcchati ‘askʼ, praśná‘questionʼ,
cf. Pokorny IEW: 821–822; van Windekens 1976: 386; Ringe 1996: 68;
LIV2
: 490–491; Malzahn 2010: 707–708; Adams 2013: 398)
242
The cluster is split here, but cf. the form of the ordinal numeral further with the cluster preserved.
234
Ḱ + s = Toch. ks/kṣ:
Toch. B laks ‘fishʼ (< IE *loḱsi; cf. OHG lahs, Lith. la͂ šiš ‘salmonʼ; cf. Pokorny IEW:
653; van Windekens 1976: 254–255; Adams 2013: 590);
Toch. B kakse ‘[a body part?]ʼ (probably < IE *kuḱsi or IE *koḱso; could be related to
OIA kukṣí- ‘bellyʼ, kákṣa- ‘armpitʼ, L. coxa ‘hipʼ, OHG hahsa ‘back of kneeʼ; cf.
Pokorny IEW: 611; cf. especially Adams 2013: 143–144 for the detailed analysis);
Toch. A pr. praksam, grdv. prakṣäl, Toch. B pr. preksau, pr. part. preksemane, imp.
parksat ‘askʼ (< Toch. *pärk-sa- < IE *√preḱ-; cf. OIA pṛcchati ‘askʼ, praśná‘questionʼ,
cf. Pokorny IEW: 821–822; van Windekens 1976: 386; Ringe 1996: 68;
LIV2
: 490–491; Malzahn 2010: 707–708; Adams 2013: 398);
12.2.4 The clusters labiovelar + t/s
The phonemic status of labiovelars in Tocharian languages is a matter of debate (cf. especially
Hilmarsson 1993a; Kim 1999 and Fellner 2006). The author of the present article presumes that
there was a synchronic phoneme ku̯
in both Tocharian languages (as does Kim explicitly 1999:
177). However, the old IE labiovelars were later subjected to delabialization in numerous other
contexts (as they were in Romance vs Latin or later Germanic languages vs their older stages).
The clusters of IE labiovelar + t/s are preserved as kt/kts (i.e. the same as the palatovelars
in the same context), with a loss of the labialization:
Ku̯
+ t = Toch. kt/kts
:
Toch. A pänt, Toch. B piṅkte ‘fifthʼ (< Toch. *p änkte < IE *penku̯
-to-; cf. L. quīnctus,
quīntus ‘fifthʼ; OIA pakthá-, Gr. πέμπτος; cf. Pokorny IEW: 808; van Windekens
1976: 360–361; Winter 1992b: 107–108, 119–120, 136–137; Blažek 1999: 224;
Adams 2013: 411, 415–416).
Toch. A naktsu, Toch. B nektsīye adv. ‘last night, at nightʼ, Toch. A nokte ‘at nightʼ,
noktiṃ ‘last nightʼ (< Toch. *neku̯
t- < IE *neku̯
t-; cf. L. Gr. νύξ, νυκτός, L. nox, noctis,
Goth. nahts; Pokorny IEW: 762–763; van Windekens 1976: 319–320; Pinault 1990:
181–190; LIV2
: 449; NIL: 504–513; Adams 2013: 363);
Toch. B laṅktse ‘easy, lightʼ (< IE *H1ln̥ gu̯ h
-t o-; cf. Gr ἐλαφρός ‘light in weightʼ, Goth.
leihts, Lith. len͂ gvas ‘lightʼ; cf. Pokorny IEW: 660–661; van Windekens 1976: 255–
256; Adams 1988: 37; NIL: 243–245; Adams 2013: 590–591).
Toch. B mäkte ‘asʼ (< IE *mén-ku̯
-tō-; van Windekens 1976: 286; Adams 2013: 484–
485).
Toch. A pt. pakt-äṁ (< Toch. *√päku̯
-s- < IE *√peku̯ s-;
cf. non-extended and s-extended
forms: OIA pr. pácati, ao. pákṣat, L. pr. coquō, pf. cōxī ‘cookʼ; cf. Pokorny IEW: 798;
van Windekens 1976: 355; LIV2
: 468; NIL: 548–552; Malzahn 2010: 700–701; Adams
2013: 393–394);
Ku̯
+ s = Toch. ks:
Toch. [A ops,] Toch. B okso ‘ox, cowʼ (< Toch. *oku̯
so- < IE * H2uku̯
son-; cf. OIA ukṣan,
Av. uxšan- ‘bull, oxʼ, W. ych, OHG ohso ‘oxʼ; cf. Sieg/Siegling 1908: 927; van
Windekens 1976: 333; NIL: 368–370; Adams 2013: 117);
Toch. A pr. imp. pakṣānt, Toch. B pr. päksäṁ, part. pr. päksemane ‘become ready for
eating, cook, ripenʼ (< Toch. *√päku̯
-s- < IE *√peku̯ s-;
cf. unextended and s-extended
235
forms: OIA pr. pácati, ao. pákṣat, L. pr. coquō, pf. cōxī ‘cookʼ; cf. Pokorny IEW: 798;
van Windekens 1976: 355; LIV2
: 468; NIL: 548–552; Malzahn 2010: 700–701; Adams
2013: 393–394);
Toch. B mäksu ‘which, who, whatʼ (< IE *mén-ku̯
-sō-; van Windekens 1976: 285; Adams
2013: 485).
12.2.5 The clusters dental + t/s
The attested outcomes of the IE cluster *Tt are usually ts or ts
ts
, often attested from the same
root (van Windekens 1976: 105).
Note: The outcome tts, reconstructed in the example of TB wrattsai could be a reading mistake, according to
Adams 2018: 670, who reconstructs wrantsai, and gives a different etymology (cf. l.c.).
T + t = Toch. ts
t/ts
ts
:
Toch. A wärts, Toch. B aurtse, wartstse ‘broad, wideʼ (< Toch. *wärtse < IE *u̯ r̥ dh
-to-;
cf. OIA vṛddhá- ‘enlarged, bigʼ; cf. Pokorny IEW: 1167; van Windekens 1976: 105,
562–563; Adams 1988: 39; Görtzen 1998: 412; Adams 2018: 139–140);
Toch. B orotstse, wrotstse ‘great, big, large; adultʼ (< Toch. *nätsw < IE *ā-√u̯ r̥ H2dh
-to;
cf. OIA vrādhant- ‘being bigʼ; cf. Pokorny IEW: 1167; van Windekens 1976: 105,
341; Görtzen 1998: 412; Adams 2018: 127–128, note here on detailed discussion);
Toch. A nätsw-, Toch. B mätsts-243
‘starveʼ (< Toch. *nätsw < IE *n-H1d-tu̯ - e/o-; cf. Gr.
νηστεύω ‘fastʼ, νῆστις ‘fastingʼ; cf. Pokorny IEW: 287–288; van Windekens 1976:
105, 316–317; Adams 2018: 139–140);
Toch. B wrattsai244
‘againstʼ (< IE *u̯ r̥ t-to-; cf. L. versus; cf. Pokorny IEW: 1156–1157;
van Windekens 1976: 105, 583; Adams 1988: 39);
Toch. B inf. lyutsi ‘cross, go out, leaveʼ (< Toch. *√läut- < IE *√H1leu̯ dh
-; cf. Gr.
ἐλεύσεται ‘goʼ, OIr luid ‘wentʼ; cf. Pokorny IEW: 306–307; van Windekens 1976:
269–270; LIV2
: 248–249; Malzahn 2010: 856–858; Adams 2013: 605–606);
Toch. B wästarye ‘liverʼ (< Toch. *wästär ä- < IE *ud-tr̥ o -; cf. OIA udára- ‘bellyʼ, Gr.
ὑστέρα, L. uterus ‘wombʼ; cf. Pokorny IEW: 1104; van Windekens 1976: 565; Adams
1988: 39; Hilmarsson 1993b: 216–217; Ringe 1996: 71; Adams 2018: 651);
T + s = Toch. ts:
Toch. A pr. lutseñc, grd. lutṣäl, Toch. B pt. lyutsāmai ‘cross, go out, leaveʼ (< Toch.
*√läut- < IE *√H1leu̯ dh
-; cf. Gr. ἐλεύσεται ‘goʼ, OIr luid ‘wentʼ; cf. Pokorny IEW:
306–307; van Windekens 1976: 269–270; LIV2
: 248–249; Malzahn 2010: 856–858;
Adams 2013: 605–606);
Toch. A pr. yātṣānt, Toch. B pr. yātṣäṃ’be capable of; succeed; tameʼ (< Toch. *√yēt- <
IE *√yet-; cf. Av. yateiti ‘place in order, strive afterʼ; cf. Pokorny IEW: 506–507;
LIV2
: 313–314; Malzahn 2010: 785–787; Adams 2013: 527–528, but van Windekens
1976: 645 assume a borrowing!);
secondary: Toch. A, tsepant ‘danceurʼ, B verb. tsip- nom. tsaipe ‘danceʼ (cf. OIA túṣyati ‘be satisfiedʼ; cf. van
Windekens 1976: 110; but sceptical Adams 2018: 808, 812);
243
The Tocharian B form is with a distant assimilation on following *u̯ , cf. Adams 2013: 493.
244
But Adams 2003: 670 reconstructs wrantsai, based on the graphical evidence, this form absolutely excludes
the etymology given in the entry above (cf. l.c.).
236
12.2.6 The clusters sibilant + t/s
The IE cluster *st is preserved (often cerebralized, at least in writing); there are no examples
for the development of the IE cluster *ss:
S + t = Toch. st:
Toch. A, kaṣt, Toch. B kest ‘hungerʼ (< Toch. *kestā < IE *√kost-; cf. Hitt. kašt, OIA
ghasati, kṣut ‘hungerʼ ; cf. Pokorny IEW: 841–842; van Windekens 1976: 189; LIV2
:
198–199; Carling 2009: 107; Adams 2013: 213);
Toch. B pest, päst ‘[some particle]ʼ, postäṃ ‘later, afterwardsʼ (< Toch. *scar-i ē < IE
*post; cf. L. post, Gr. ἀστήρ; cf. Pokorny IEW: 841–842; van Windekens 1976: 367,
383–384; Adams 2013: 408–409, 430–431, 436–437);
Toch. A ṣtare ‘effortʼ, Toch. B ścīre ‘hard, harsh, rough, crudeʼ (< Toch. *scärēn < IE
*sterH1-eH1-en; cf. Gr. στερεός, στερρός ‘firm, solidʼ, OHG starēn ‘stare atʼ; cf.
Pokorny IEW: 1022, 1029–1030; van Windekens 1976: 482; LIV: 597–598245
; Adams
2013: 700–701);
Toch. [A śreñ],246
Toch. B n. [f. pl.] ścirye ‘starʼ (< Toch. *scar-i ē < IE *H2stēr; cf. L.
stella, Gr. ἀστήρ; cf. Pokorny IEW: 1027–1028; van Windekens 1976: 486; NIL: 348–
354; Adams 2013: 701);
Toch. B pr. yaṣtär ‘excite, touchʼ (< Toch. *√ as- < IE *√ es-; cf. OIA yásati ‘froths up;
strives afterʼ, Gr. ζέω ‘boil, seetheʼ; cf. Pokorny IEW: 506; van Windekens 1976: 595;
LIV2
: 312–313; Malzahn 2010: 533; Adams 2013: 802–803);
Toch. A inf. wassi (!), B inf. wastsi/wassi (!) ‘dress, wear clothesʼ, nom. wastsi ‘clothingʼ
(< Toch. *u̯ äs-t/dh
- < IE *√u̯ es-; cf. OIA váste ‘be dressedʼ, Alb. vesh ‘wearʼ, Goth.
wasjan ‘wearʼ; cf. Pokorny IEW: 1172–1173; van Windekens 1976: 564; LIV: 692–
693; Malzahn 2010: 896; Adams 2013: 635, 649);
S + s = Toch. ss:
Toch. B pr. āṣṣäṁ ‘bring, fetchʼ (< Toch. *ās-sk-; etymology unclear, could be a
borrowing; cf. van Windekens 1976: 624; Malzahn 2010: 533; Adams 2013: 63);
Toch. B pr. keṣäṁ ‘quench, extinguishʼ (< Toch. *käs-s- < IE *√gu̯
es-; cf. OIA jásate ‘be
exhaustedʼ, OCS -gasiti ‘extinguishʼ; cf. Pokorny IEW: 479–480; van Windekens
1976: 210; LIV: 541–542; Carling 2009: 160; Malzahn 2010: 594; Adams 2013: 188);
12.2.7 The overview of the Tocharian development
In the following table, the palatalized outcomes are omitted, since specifically Tocharian and
their outcomes could be easily deduced from the non-palatalized outcomes.
IE Tocharian t- s-
-kṷ
/gṷ
/gṷh
ku̯
kt ks
-k/g/gh
k kt ks
-ḱ/ǵ/ǵh
k kt ks
-t/d/dh
t ts
ts
(?) ts
-p/b/bh
p pt ps
-s s st ss
245
LIV does not list Tocharian with the root!
246
The Tocharian A form is from the syncopated form, with t/c lost in a newly created cluster.
237
12.3 Trajectories of the development
The development of IE clusters in Tocharian is, besides the dental series, very conservative.
The labiovelars are neutralized on plain velars (as they are in Latin, for comparison).
Surprisingly the clusters ss are fully preserved (or restored). The clusters of dental plosive + t
are usually realized either as simple affricates or as clusters of two affricates.
12.3.1 Development of clusters labial + t/s
Labials, like all other peripheral series in Tocharian, did not undergo any particular
development in clusters formed by t/s: the plosives are fully preserved (at least in the internal
clusters):
P + t > pt
P + s > ps
Note: The attested development of the IE *#Ps > Toch. *#0s: Toch. A sāt ‘hotʼ, B satāsk ‘exhaleʼ (< IE *ps-ōd-,
derived from IE *√bh
es-; van Windekens 1976: 103, 419–420; doubted as genuine by Adams 2013: 736–737)
is a regular process not a chance one, especially since Peyrot (2008: 72) lists many examples of variation of
the Tocharian B clusters #PC- (where C is any obstruent).
We meet another example in IE *Ps > Toch. *#pVs attested in Toch. A päśśäṃ, B päścane ‘breast; [in plural]
seat of wisdomʼ (< IE *pstḗm; cf. Hitt. istanza ‘soulʼ-; cf. van Windekens 1976: 103; Ringe 1996: 71; Adams
2013: 386) is simply a case of an anaptyctic vowel. To judge if this process of anaptyxis is regular for wordinitials
is also hard due to lack of other examples.
12.3.2 Development of clusters plain velar and palatovelar + t/s
The plain velar and palatovelar plosives are preserved as plain velar plosives before t- and s-,
since, as Tocharian languages are both centum-languages, there is no trace of any phonemic
distinction preserved:
K/Ḱ + t > kt
K/Ḱ + s > ks
Note: Tocharian A alternation between tā ‘(vers) oùʼ vs Toch. A te ‘(L.) -neʼ (both from *ku̯
u- + *to-, van
Windekens 1976: 105) could be a sign of an alternation #Kt ~ #0t-, as it is with Toch. B kuse, kuce vs se, ce of
the same compound pronoun (Peyrot 2008: 71–72).
Peyrot (2008: 72) lists examples of a variation of Toch. B kṣ ~ k in borrowings from Sanskrit, but not in
autochthonous words; Peyrot connects it to a received pronunciation of this cluster in Middle Indo-Aryan and
a stylistic marker sui generis.
12.3.3 Development of clusters labiovelar + t/s
For the development of IE labiovelars in clusters with t/s we assume the neutralization of the
labial marker before any obstruent, similarly as attested fully in Latin, partially in other
languages. This neutralization of labiality was later extended to other contexts (cf. Adams
1988:37; Kim 1999, especially pp. 177–182):
238
Ku̯
+ t > kt
Ku̯
+ s > ks
12.3.4 Development of clusters dental + t/s
The traditional trajectory for IE clusters of *Tt is assumed as that for affrication, either of the
first member or both dental plosives, cf. Krause (1952: 18) or van Windekens (1976: 105).
Adams (1988: 39–40) assumes that IE *Tt > st, *Tt > tsts, i.e. that the outcomes differentiate
due to the palatal context; Ringe (1996: 71) doubts the fact that there is a secure example of the
development of IE *Tt at all (otherwise he states *Tt > st, based on the singular example of
Toch. B wästarye ‘liverʼ), and Kümmel (2007: 352) lists the same outcome. However, the
above-noted examples are of the wide range: tsts, tst, ts, even st. It would be easiest to assume
affrication since the early phase, often doubled (either as progressive assimilation or due to a
secondary palatalization), partially simplified as tts.
Note: The single example of st of Toch. B wästarye ‘liverʼ; cf. van Windekens 1976: 565; Adams 1988: 39;
Hilmarsson 1993b: 216–217; Ringe 1996: 71; Adams 2018: 651, though etymologically entirely acceptable as
reflecting the same formation as OIA udára-, remains a mystery, could be a result of a unique process, or of
an unknown context feature.
The development of IE cluster *Ts could be modelled either as a preservation of the old cluster
(but note that other IE languages often have a specific simplification as *0s) or as restoration,
the process we prefer (cf. Görtzen 1998: 415):
T + t > ts
t > ts
ts
/ts
t
T + s > ts
s > ts
Note: We have to reject that there is a Bartholomae-like distinction in Pre-Tocharian, as Görtzen (1998: 415), who
models *Tt > ts
t/ts
ts
/ts and *dh
t > dz
dh
> zdh
> st, proposes.
12.3.5 Development of the clusters sibilant + t/s
The IE clusters of a sibilant + t/s are preserved as such, surprisingly in the case of ss clusters,
since we have reasons to assume that such clusters were often simplified as 0s in other IE
languages, probably even in the proto-language itself, which is usually typical and well-attested
for clusters with the IE root *H1es-, cf. OIA ási, Gr. εἶ, Lith. esì, OCS jesi. For Tocharian, we
can assume either preservation, or restitution (cf. Gr. Aeol. ἔσσι, Ep. and Dor. ἐσσί, OL. essis,
ēs vs L. es):
s + t > st
s + s > ss
239
12.4 Conclusion and final remarks
The development of the IE obstruent clusters in Tocharian is very conservative. The single
progressive feature is the fricativization of the IE cluster *Tt, shared with other IE languages.
In Tocharian this process is affricativization, attested otherwise only in Hittite, though usually
generalized for the development of all IE languages as a first stage of a more complex
development, usually ending in sibilant outputs. The cluster *Ts has a natural affricate outcome
either due to the preservation of the original state or through the re-archaization of the
intermediate *ts
s.
The peripheral series fully preserves the original plosives in all clusters, as do clusters
with the sibilant. What is remarkable is the development of labiovelars, for which we assume
the neutralization of a labial marker in the contexts of t/s, the same is known either fully
preserved (Latin) or at least partially (especially Germanic languages), but we have to keep in
mind that delabialization of labiovelars is often present in other contexts in Tocharian, being
almost universal.
Note: Beyond the scope of this work, there are interesting developments of three-obstruent clusters, which could
be worthy of a reader´s interest: Adams (1988: 38–39) notes that IE clusters of *Ksḱ, *Tsḱ (sḱ is here a part of
an iterative/intensive suffix) are realized as Tocharian sk and tk respectively. He assumes that *Psḱ would
realize as sk, if we accept the analogical process with all peripheral series, but knows no examples. The example
for *Ksḱ > sk is: IE *u̯oku̯
-sḱe/o- > Toch. AB wesk ‘speakʼ (cf. Pokorny IEW: 1135–1136; Adams 2013: 658–
660), for *Tsḱ > tk then: IE *snud-sḱe/o- > Toch. B snätk- ‘permeateʼ (cf. Pokorny IEW: 972; Adams 2013:
779). A remarkable feature of the development is different outcomes: in the case of *Ksḱ > Toch. sk the left
velar is lost and a sibilant preserved, in the case of *Tsḱ > tk the sibilant is lost. Neither development mirrors
the other.
241
13 The development of the two-obstruent clusters in the Indo-European
languages: the summary and conclusions
The processes affecting the development of clusters of obstruent + t/dh
/s in various IndoEuropean
sub-branches could be classified as:
i. the shared-processes, i.e. the processes in similar contexts, of the same origin (e.g., the
Common Indo-European first phase of the development of the dental + t/dh
/s clusters;
another example is the earliest phase of the development of the palatovelar + t/dh
/s clusters
in the satəm-languages; the third example is the old neutralization of labiovelars in the
context of + t/dh
/s, securely attested at least in some of the centum-languages);
ii. the drift-processes, i.e. parallel processes in similar contexts, independently caused (e.g., the
spirantization of the peripheral series in Iranian, Celtic, Sabellian, Slavic);
iii. the zero-processes; i.e. the retention of the original state (e.g., the preservation of the
peripheral series in Latin, Greek, Vedic, Baltic). Zero-processe seems to be trivial, but we
have to remember that the preservation of a state is as important as a change of it, especially
in comparison. The zero-processes has to be distinguished from re-archaization processes
(as is the restitution of a plosive in clusters of dental + t/s/dh
- in Indic).
In the following lines, we will list the known outcomes of the IE clusters in given daughter
languages and branches and then sketch up possible trajectories. The boldly marked outcomes
are attested; the lightly marked outcomes are constructs. Analogous forms are listed, but not
included in the trajectories graphs.
13.1 The development of the central series
There are two developments of the central series: the shared development of the dental series,
present in some form in all Indo-European languages, with the seemingly (and false) exception
of Indo-Aryan; and the shared development, limited to the satəm-languages, of the palatovelar
series.
A similarity in the development of both series is striking since both series, however,
with original plosive inputs, usually have sibilant outcomes (Hittite and Tocharian being
remarkable exceptions with affricate outcomes of the IE cluster *Tt, Nūristānī and Armenian
with a zero outcome for the same cluster, Albanian with a zero outcome both for the IE *Tt, but
also for IE *Ḱt).
For the development of both central series we can draw two possible trajectories through
the ‘black box’, which could be summed under terms of the affricativization trajectory and the
spirantization trajectory.
242
13.1.1 The development of central series I: the dental series
The singular development of the IE cluster *Tt, common to all sub-branches of Indo-European
languages, is the development of the dental series.
Note: Surprisingly, Kümmel (2007: 349–350) gives a shortlist of shared IE consonantal developments without at
least the first phase of the development of the dental series, but deals with the whole process with as a series
of later developments in each of the IE sub-branches (Kümmel 2007: 350–411).
This development has the following outcomes:
i. IE Tt* > ts
t(s)
, attested in Anatolian and Tocharian;
ii. IE Tt* > tt, attested in Old Indo-Aryan (and probably attested in Nūristānī);
iii. IE Tt* > st, attested in Iranian, Greek, Balto-Slavic;
iv. IE Tt* > ss, attested in Italic, Celtic247
and Germanic;
v. IE Tt* > u̯ t, attested in Armenian;
vi. IE Tt* > c, attested Albanian;
The Albanian outcome could not be the result either of an older st-outcome (BaltoSlavic/Iranian/Greek
style) since IE *st is realized as Albanian št nor could it be a result of the
Italic/Celtic/Germanic *ss-outcome, since the outcome of the IE *ss in Albanian is 0š (as it is
of the IE *Ts). The most probable predecessor of IE *Tt in the earlier stages of the Albanian
development was probably an affricate (Anatolian/Tocharian style) or a fricative (assumed as
on older stage for all developments outside the Anatolian/Tocharian model).
The Armenian outcome could not be attributed directly to one of the four types
mentioned above either, since the stage *st is impossible, since the IE *st cluster is fully
preserved and if this cluster merged with that of IE *Tt, the outcome would be the very same.
Moreover, the *ss outcome for *Tt is also impossible since IE *ss is realized as Arm. 0s.
The development of the cluster of dental plosive + s in various IE branches can be listed as:
i. IE *Ts > ts, attested Old Indo-Iranian, Hittite and Tocharian;
ii. IE *Ts > 0s, attested in Avestan, Baltic, Slavic, Greek and Albanian;248
iii. IE *Ts > ss, attested in Italic, Celtic (the Gaulish ðð being its variant) and Germanic;
iv. IE *Ts > tš
, attested in Nūristānī;
v. IE *Ts > cʿ, attested in Armenian;
The Hittite and Tocharian developments are securely archaic, preserving the older state. The
Indo-Aryan development can not be detached from the Iranian – we have all the reasons to
assume that both branches had a shared development, from which we have to model both
247
The Gallic outcome đđ (of the insecure phonetic value) alternates with ss and is considered both a variant and
a predecessor of the ss outcome.
248
In Albanian, the outcome is 0š, due to later palatalization of the sibilant.
243
trajectories; therefore the preservation of the older state (as in Hittite and Tocharian) is then
impossible for Indo-Aryan. We assume the spirantization of T before s not only for Indo-Iranian
but for all IE languages (outside Hittite and Tocharian); the Indic outcome is the later rearchaization,
while the Iranian state is a progressive outcome, resulting from the merging of the
dental spirant with a sibilant. All other languages with 0s outcomes followed the same
trajectory.
The development of the clusters of dental plosive + dh
can be securely reconstructed in a
few languages, and we have to highlight that the outcome is always given by the voicedness of
these right contexts, even in Greek, where the IE *dh
was devoiced, and subsequently, the
clusters were analogically remodelled. The outcomes in Indo-Iranian, Greek and Baltic are:
i. IE *Tdh
> ddh
, attested in Old Indo-Aryan as the major outcome;
ii. IE *Tdh
> zd, attested in Iranian and Baltic as a regular outcome, in OIA in the form of 0dh
as the minor outcome;
iii. IE *Tdh
> sth
, attested in Greek;
The traditional affricativization trajectory, first formulated by Kräuter (1977: 88)249
,
evaluated by Verner (1878: 341–342) and popularized by Brugmann (1880 and passim) is
usually assumed for the whole development, which can be modelled for first four outcomes as
follows: with a sibilantization of the left plosive (= the loss of the plosive segment of the
affricate), with a loss of the sibilant segment (= re-archaization) or with a further affrication of
the whole cluster and its simplification of sibilants:
Tt > tt > ts
t > st
> tt
> ts
ts
> ss
Note: Both the Armenian and Albanian outcomes could hardly be put within the affricativization trajectory, hence
we do not dare to propose any trajectories of their developments.
The great advantage of the affricativization trajectory is the attested preservation of affricates
in Hittite and Tocharian (exceptions in Tocharian could be explained as a further development
along the very same trajectory). What is problematic is application for the development of
Armenian, which could not be attached to the affricate stage, to a sibilant or to the double
sibilant stage. Similarly we might be tempted to attach the Albanian outcome to the doublesibilant
outcome, but the Albanian outcome of Tt is different from that of Ts and ss, hence
impossible. That the Indo-Aryan outcome is re-archaized is supported by the fact that the IE
cluster *ss is realized in OIA usually as ts, which demonstrated that outcomes of IE *Tt and *ss
249
Also Verner (1878: 341–342).
244
were merged in some point of development. Italo-Celtic-Germanic development is the result of
a similar trajectory.
Note: A specific development of the final -dh
, affected by Bartholomae’s Law, is known from Indo-Iranian. The
trajectory would be modelled, according to the affricativization strategy, as:
dh
t > dz
dh
> ddh
> zd(h)
The affricativization trajectory for the development of IE clusters of *Ts was reconstructed fully
only by Lipp (Lipp 2009a: 169) for the Indo-Iranian languages. It is remarkable that both
languages with affricate outcomes of the *Tt cluster (Hittite and Tocharian) do not show any
trace of the more complex developments attested in languages without the attested
affricativization of *Tt clusters. The Hittite/Tocharian development could be either an archaic
feature or a simplification of the affricate back to ts by the loss of one of the sibilants. We can
also consider Indic, Nūristānī and Armenian outcomes as archaic (or re-archaized) (here with
later aspiration).
Ts > ts
> ts
s > ss > 0s
> ts (?)
The affricativization trajectory assumes affricativization, followed in OIA by re-plosivation,
and by a sibilantization in other languages (and the same process is the minor process in OIA,
the loss of a voiced sibilant allophone is known from proper sibilants in the same context, cf.
below). The Greek trajectory is affected by analogy remodelling, hence omitted:
Tdh
> dz
dh
> ddh
> zd(h)
> 0dh
The spirantization trajectory was independently brought forward for Italic by de Saussure
(1877), independently by Cocchia (1883: 16–58)250
and Bartholomae (1887: 83; Bartholomae
1895: 16), followed by Leumann (1942: 13) and Morgenstierne (1942: 80; for Iranian only).
Applying this trajectory to the development of the IE cluster of plosive + t, the trajectories for
the st-group (Iranian, Greek, Balto-Slavic), Indic and the ss-group (Celto-Italic-Germanic)
could be modelled as:
250
And we have to remark that his idea was dismissed by Brugmann (1885: 183).
245
Tt > tt > ϑt > st
> tt
> ϑϑ > ss
That the Indo-Aryan outcome is re-archaized is supported by the fact that the IE cluster *ss is
usually realized in OIA as ts, which demonstrated that outcomes of IE *Tt and *ss were merged
in some point of development. Italo-Celtic-Germanic development is the result of a similar
trajectory: IE *Tt was spirantized as ϑt, assimilated first to ϑϑ and later to ss, merging both with
the original IE *Tt and *st clusters.
Albanian development, according to the spirantization trajectory is a series of
spirantization and later affricativization and sibilantization: Tt > ϑt > ϑϑ > ts
(= c) > 0s.
Armenian development has (after spirantization) the later debuccalization of the fricative: Tt
> ϑt > ht > u̯ t.
Note: A specific development of -dh
, affected by Bartholomae’s Law, is known from Indo-Iranian. For the
trajectory of Bartholomae’s Law, we assume spirantization for all series , followed by a re-plosivation in OIA:
dh
t > ðð > ddh
> zd(h)
The development of the cluster of dental plosive + s according to the spirantization trajectory
is similar to that of *Tt: the spirant was often sibilantized (and the geminate simplified). In Indic
the spirant was fortified as a plosive, and a similar process, followed by affricativization and
later aspiration, is attested in Armenian and probably in Nūristānī (here the re-plosivation is
probably related to that of OIA). The Gallic state shows the free variantion of the dental spirants
with dental sibilants.
Ts > ϑs > ss > 0s
> ts > cʿ
Note: Bartholomae’s Law applies even on clusters of dh
s in Indo-Iranian, but Indic outcomes are levelled:
dh
s > ðz > zz > 0z
The proposed trajectory of the development of the Tdh
-clusters within the spirantization/lenition
trajectory assumes first spirantization, followed in OIA by re-buccalization as a plosive, or by
a sibilantization in other languages (and the same process is the minor process in OIA, as the
loss of a voiced sibilant allophone is known from proper sibilants in the same context, cf.
below). The Greek trajectory is affected by analogy remodelling, hence omitted:
246
Tdh
> ðdh
> ddh
> zd(h)
> 0dh
Note: Principally the same development is valid for the development of IE clusters of *dh
dh
into Indo-Iranian.
To sum up: there are two languages with affricate outcomes both for the *Tt and *Ts clusters,
namely Hittite and Tocharian, both being peripheral languages, and there is no other possible
trajectory for both languages than the affricativization trajectory. On the other hand, for all other
languages the spirantization trajectory is more probable, since especially the development of
the *Ts > (s)s excludes the possibility of affricativization (the intermediate *ts
s could be easily
simplified on ts
but we have full sibilant outcomes). Affricativization development is wholly
impossible for Indic: if we accepted affricativization development both for the dental and the
palatovelar series, it would be impossible for the assumed (*Tt >) ts
t (> Indic tt) to lose the
plosive segment and the parallel and contemporary (since both processes are operating after the
split of the Indo-Iranian languages) (*Ḱt >) tš
t (> Indic ṣṭ) would lose the fricative segment of
the affricate – this paradox is not present within the spirantization trajectory.
The spirantization trajectory also makes it easier to explain Italic-Celtic-Germanic ssoutcome;
the merging of the IE *Ts and *ss in a single ϑs output also explains why the Indic
outcome of the *ss is surprisingly ts (and of *šs is kṣ), which is otherwise a solitary and isolated
process.
However, the distinction between the affricativization and the spirantization trajectory
is not as wide as it could see: both fell within the same frame of fricativization, and while the
affricativization presumes the insertion of the fricative segment into a given cluster, the
spirantization presumes the fricativization of the already existing segment. Both peripheral
languages (Tocharian and Hittite) used the affricativization variant of the fricativization
trajectory, but all other languages used the spirantization variant of the same trajectory.
13.1.2 The development of the central series II: the palatovelar series
The development of the IE clusters of *Ḱt into given satəm-languages usually has an outcome
in the form of a sibilant (either palatal or non-palatal) + t. The single exception is Albanian,
where the outcome is 0t:
i. IE Ḱt* > št, attested in Iranian, Lithuanian;
ii. IE Ḱt* > ṣṭ, attested in Indo-Aryan and Nūristānī; the geographical variant of the preceding
development;
iii. IE Ḱt* > st, attested in Slavic and Armenian;
iv. IE Ḱt* > 0t, attested in Albanian;
(v. IE Ḱt* > kt, attested in the centum-languages).
247
The development of the cluster of palatovelar + s in various IE branches can be listed as:
i. IE *Ḱs > kṣ, attested Old Indo-Iranian;
ii. IE *Ḱs > 0š, attested in Avestan, Lithuanian and Albanian;251
iii. IE *Ḱs > 0s, attested in Slavic;
iv. IE *Ḱs > ts
, attested in Nūristānī; the outcome cʿ, attested in Armenian, is a variant;
(v. IE *Ks > 0š, attested in the centum-languages)
The affricativization strategy assumes the affricativization of the original palatovelar in the
satəm-languages, and its later sibilantization (in Indo-Iranian, Armenian and Balto-Slavic).252
Ḱt > tš
t > št > st
> ṣṭ
Note: Bartholomae’s cluster of *ǵh
t, according to the affricativization trajectory, can be modelled as follows both
for Indic ( ḍh
) and Iranian (žd):
ǵh
t > ʝdh
> ždh
> ẓḍh
> ḍh
> žd
The development for the cluster of *Ḱs in the satəm-languages within the affricativization
trajectory can be modelled as follows (with affricativization, sibilantization, simplification for
Iranian, Lithuanian and Albanian with depalatalization for Slavic; with de-affricativization and
location shift for Indic; with the later aspiration of the affricate in Armenian; a simplified
affricate is attested in Nūristānī):
Ḱs > tš
s > šš > 0š > 0s
> ṭṣ > kš
> ts > c(ʿ)
Note: Bartholomae’s cluster of *ǵh
s, the affricate model is (the Indic outcome is due to the analogy):
ǵh
s > ʝžh
> džh
> ḍẓ (→ kṣ)
> žžh
> žž > 0ž
Within the spirantization strategy, we assume that the clusters were spirantized, either as a
palatal spirant or as a velar one (either directly from the ‘neutralization’ form kt or due to the
depalatalization of *çt). The palatal spirant was later sibilantized (eventually depalatalized
later), the velar spirant was debuccalized as simple ht or even fully elided (we model this
development for Albanian, since in Albanian even IE *Kt realizes as 0t).
251
The Albanian outcome should be listed with Slavic, since old sibilants merged into š in Albanian.
252
The Albanian development (based on Schumacher 2013: 243) assumes deaffricativization (technically
gemination) and simplification of the cluster (Ḱt > tš
t > tt > 0t).This development definitely must have been
later than the development of the IE cluster *Tt, since it had not merged with it.
248
Ḱt > çt > št > st
(> kt) > xt > ht > 0t
Note: As with all series, there is a specific development of -ǵh
+t in Indo-Iranian, affected by Bartholomae’s Law.
We model the following spirantization trajectory, assuming the spirantization of the palatovelar:
ǵh
t > ʝð > žð > ẓḍh
> 0ḍh
> žd
Within the spirantization strategy, for the development of the cluster of a palatovelar plosive
+ s-, we assume first the neutralization of a palatalization (and palatalization of a sibilant due
to the ruki-rule), followed in many languages by spirantization, sibilantization and
simplification:
Ḱs > kš > kṣ
> çš > šš > 0š
> ϑš > ts > 0s
Note: We can model the ‘Bartholomaen’ development of the cluster of *ǵh
s according to the spirantization
trajectory as (valid for Iranian; Indic development was replaced by the analogy):
ǵh
s > ʝz > ʝž > žž > 0ž
The development of the clusters of palatovelar plosive + dh
can be securely reconstructed in a
few satəm-languages (technically: Indo-Iranian and Baltic), and the outcomes are always
voiced. The outcomes in Indo-Iranian and Baltic are:
i. IE *Ḱdh
> ḍḍh
, attested in Old Indo-Aryan;
ii. IE *Ḱdh
> žd, attested in Iranian and Baltic;
The proposed trajectory assumes the spirantization of the palatovelar and later sibilantization
of the spirant, this sibilant is lost in Indic:
Ḱdh
> ʝdh
> žḍh
> ḍdh
> ždh
> žd253
Note: Essentially the same development is valid for the development of IE clusters of *ǵh
dh
into Indo-Iranian,
since they share the same outcome.
13.2 The development of the peripheral series
The set of the input peripheral series differ according to the centum/satəm languages dichotomy,
the first having labiovelars preserved (at least on a re-constructible level) beside plain velars
253
The outcome is zd after depalatalization in Prussian and Latvian.
249
and labials, the second having the old IE labiovelars merged fully with the plain velars
(however, such merging is known from the centum-languages as well).
Note: The logical consequence of this feature is clear: the positive marker we can use to distinguish the
centum/satəm languages is not the presence of the labiovelars series, but the presence of the palatovelar series.
In other words: the centum-languages are all languages without the presence of the original palatovelar series,
so the centum-languages are hence ‘negatively’ defined and the satəm-languages are defined ‘positively’.
Regarding the data of the attested Indo-European languages, we can state that there are two
strategies in general: the conservative strategy (with the ‘zero’ trajectory) and the progressive
strategy (with the spirantization/lenition trajectory). We can express the distribution of both
strategies in the following table:
CONSERVATIVE STRATEGY PROGRESSIVE STRATEGY
Old Indo-Aryan Old Iranian
Baltic Slavic
Armenian
Albanian
Greek (Middle Greek)
Latin Sabellian
Celtic
Germanic
Hittite
Tocharian
A remarkable feature is that the split between both strategies could run through a given subbranch,
as we can see in the examples of the Indo-Iranian, Balto-Slavic or Italic language
families.
Note: As we can see in the example of Middle Greek, spirantization could affect archaic clusters later. Similarly
process we can see in the development of French from (Vulgar) Latin.
Note: Beside spirantization, there is another progressive strategy, far less attested: gemination. The process of
gemination of the consonantal clusters is known from Middle Indo-Aryan and Italian. It is also usually
attributed to the Pre-Slavic development.
13.2.1 The development of the peripheral series I: the plain velar series
A development of the plain velar series can be listed as:
i. IE *Kt > kt, attested in Old Indo-Aryan, Baltic, Greek, Latin, Hittite and Tocharian;
ii. IE *Kt > xṭ, attested in Iranian, Gallic, Irish;
iii. IE *Kt > ht, attested in Sabellic and Gothic;
iv. IE *Kt > t, attested in Brythonic;
v. IE *Kt > 0t, attested Nūristānī, Armenian, Albanian and as a minor outcome in Slavic;
vi. IE *Kt > št, attested in OCS as the major outcome.
The first group is within the zero-process, the old plain velars are preserved as plain velar stops.
250
The outcomes from ii. to v. are products of various lenitions, which we can order according to
decreasing consonantal strength, being the palatal counterpart of h:
Kt > kt > xt > ht > 0t
> t
The št- outcome of OCS (c/ć-outcome in other Slavic languages) is an original prepalatal
variant, extended, due to analogy, to all productive clusters.
Note: The Armenian outcome displays aspiration of the t-context, known from the development of the all
peripheral series (but not the central series!) in Armenian, cf. the development of labials in Armenian below.
The older form was highly probably a spirant: the trajectory is: Kt >xt > xϑ >hϑ > 0tʿ.
Note: We model the ‘Bartholomaen’ development of the cluster of -gh
+s as:
gh
t > γð > gd(h)254
The development of the cluster of plain velar + s in various IE branches can be listed as:
i. IE *Ks> ks, attested in Hittite, Tocharian, Greek, Latin and in Baltic255
;
ii. IE *Ks> kṣ, attested Old Indo-Aryan;
iii. IE *Ks > xš, attested in Avestan;
iv. IE *Ks > xs, attested in Gaulish;
v. IE *Ks > hs, attested in Gothic;
vi. IE *Ks > 0x, attested in Brythonic and Slavic (beside its palatalized variant 0š);
vii. IE *Ks > 0š, attested in Armenian, Albanian and Slavic (beside its non-palatalized variant
0x);
viii. IE *Ks > ss, attested in Goidelic, with 0s attested in Sabellian;
The ks-outcome is the conservative one, with OIA kṣ as its ruki-variant attested in OIA. The
progressive outcomes follow the spirantization/lenition trajectory:
Ks > ks > xs > hs
> xx > 0x
> ss > 0s
> kš256
> xš > šš > 0š
Note: Similarly, we model the ‘Bartholomaen’ development of the cluster of -gh
+s as follows (the outcome is
attested in Iranian, Indic has the analogous levelling again):
gh
s > γz > gž257
254
This reconstruction is also valid for the development of *Ku̯
t clusters, since there is no distinction between
plain velars and labiovelars in Indo-Iranian.
255
The Baltic outcome is surprising, since it is the position where the ruki-rule is supposed to be operating.
256
Also represents OIA kṣ here.
257
Similarly to the development of the cluster of *Ku̯
t, this reconstruction is also valid for the development of
*Ku̯
s clusters, since there is no distinction between plain velars and labiovelars in Indo-Iranian.
251
The development of the plain velar plosive + dh
clusters can be reconstructed securely in IndoIranian,
Baltic and Greek, and the outcomes are always voiced. The outcomes in Indo-Iranian,
Greek and Baltic are:
i. IE *Kdh
> gḍh
, attested in Old Indo-Aryan;
ii. IE *Kdh
> gd, attested in Iranian and Baltic;
iii. IE *Kdh
> kh
th
, attested in Greek;
The trajectory is simple for OIA and Baltic, but in Iranian we meet a typical spirantization and
Greek has an analogy-based outcome, omitted below:
Kdh
> gdh
> γḍh
> γð/gd
Note: The clusters of *gh
dh
are similarly developed.
13.2.2 The development of the peripheral series II: the labiovelar series
The development of the cluster of labiovelar + t in various IE branches can be listed as:
i. IE *Ku̯
t > ku̯
t, attested in Hittite and Mycenaean;
ii. IE *Ku̯
t > kṭ, attested in Latin, Tocharian, partially in Greek;
iii. IE *Ku̯
t > pt, attested in partially in Greek;
iv. IE *Ku̯
t > xt, attested in Goidelic (and probably in Gaulish);
v. IE *Ku̯
t > ht, attested in Sabellic and Gothic;
vi. IE *Ku̯
t > t, attested in Brythonic;
The first group could be suspected to represent the zero-process, but it is more probable that
this development is the result of a secondary analogous levelling, the older state being preserved
in the second outcome, with the neutralization of the related series on the plain velars.
Note: The outcome in the conservative satəm-languages (OIA, Baltic) is also kt, the outcomes in the progressive
satəm-languages follows the development of the kt-clusters (see above). We dare to propose that it is was the
neutralization of labiality in the t/s/dh
-contexts (and in some other context too, especially before labial vowels,
the process well known from Italic) which caused the final loss of labiality of the old labiovelars in later satəmlanguages
(the process with its parallel in Tocharian).
The third version of the development is limited to Greek and it is a secondary outcome of the
development of labiovelars in Greek; here a labial is a direct heir of Mycenaean ku̯
.
The last three outcomes are all results of the spirantization (attested directly in Goidelic)
or of a further lenition (attested in Sabellic, Gothic and Brythonic), all within the progressive
strategy, following the development of plain velars, as described above258
:
Ku̯
t > kt > xt > ht
> t
258
A remarkable difference is the non-existence of the elided form 0t, since it is attested for velars only in the
satəm-languages.
252
The development of the cluster labiovelar + s in various IE branches could be listed as:
i. IE *Ku̯
s> ku̯
s, attested in Hittite and Mycenaean;
ii. IE *Ku̯
s> ks, attested in Latin, Tocharian, partially in Greek;
iii. IE *Ku̯
s > ps, attested in partially in Greek;
iv. IE *Ku̯
s > hs, attested in Gothic;
v. IE *Ku̯
s > ss, attested in Goidelic and 0s attested in Sabellic;
vi. IE *Ku̯
s > 0x, attested in Brythonic;
The ku̯
s-outcome is in our opinion a result of analogical levelling (as ku̯
t is), while the psoutcome
is a result of the specific Greek development of the levelled clusters of ku̯
s, hence the
conservative outcome follows the spirantization/lenition trajectory:
Ku̯
s > ks > xs > hs
> xx > 0x
> ss > 0s
Note: The satəm-languages follow the same trajectory as the Ks clusters (see above), with the exception of Baltic
languages.
Note: Since there are no wide and secure examples of the development of the IE clusters of *Ku̯
+ dh
in the centumlanguages
(Greek clearly restoring both the labialization of the labiovelar and remodelling the cluster due to
the loss of voicedness of the dh
), we willingly omit to reconstruct the trajectory.
Note: The development of the clusters of *Ku̯
dh
and *gu̯ h
dh
is essentially the same as the developments of the
clusters of *Kdh
and *gh
dh
, respectively (cf. above).
13.2.3 The development of the peripheral series III: the labial series
The development of the cluster of labial + t can be summed up as:
i. IE *Pt > pt, attested in Old Indo-Aryan, Avestan, Baltic, Greek, Latin, Hittite and Tocharian;
ii. IE *Pt > fṭ, attested in Oscan, Gothic, reconstructed for Old Persian;
iii. IE *Pt > ṭ, attested in Brythonic;
iv. IE *Pt > 0t, attested in Nūristānī, Slavic (as a major outcome), Armenian, Albanian;
v. IE *Pt > st, attested as a minor outcome in Slavic;
The first outcome is a conservative one, other clusters falling within the spirantization/lenition
trajectory, the oldest stage of which is roughly represented in the second outcome (for the first
stage *φt seems to be a more probable variant). Brythonic attests the further weakened
approximant, the outcome 0t the final, elided form. The minor outcome st known from Slavic
is a result of a parallel process: sibilantization instead of lenition.
Pt > pt > φt > ft
> ht > 0t
> t
> st
253
Note: The Armenian outcome displays the aspiration of the t-context, known from the development of all the
peripheral series (but not the central series!) in Armenian. The older form was highly probably a spirant; the
trajectory is: Pt >φt > φϑ >hϑ > 0tʿ.
Note: We model the ‘Bartholomaen’ development of the cluster of -bh
+t again with a spirantization in the first
phase:
bh
s > βð > bd(h)
The development of the cluster of labial plosive + s in various IE branches can be listed as:
i. IE *Ps> ps, attested in OIA, Lithuanian, Greek, Latin, Hittite and Tocharian;
ii. IE *Ps> fs, attested in Avestan (beside fš)259
and Gothic;
iii. IE *Ps > 0s, attested in Slavic, Armenian, Albanian260
and Sabellic;
iv. IE *Ps > xs, attested in Gaulish;
v. IE *Ps > ss, attested in Goidelic;
vi. IE *Ps > 0x, attested in Brythonic;
Ps > ps > φs > fs > 0s
> xs > 0x
> ss > 0s
Note: Similarly, we can model the ‘Bartholomaen’ development of the cluster of -bh
+s with an early spirantization
as follows (again, Indic development is based on the analogy):
bh
s > βz > bz
The development of the clusters of labial plosive + dh
can be reconstructed again securely in
Indo-Iranian, Baltic and Greek, and the outcomes are always voiced. The outcomes in IndoIranian,
Greek and Baltic are:
i. IE *Pdh
> bḍh
, attested in Old Indo-Aryan;
ii. IE *Pdh
> bd, attested in Iranian (the two-spirant cluster βð being its variant) and Baltic;
iii. IE *Pdh
> ph
th
, attested in Greek;
The trajectory is simple for OIA and Baltic; in Iranian we meet a typical spirantization; Greek
has an analogy-based outcome, omitted below:
Pdh
> bdh
> βḍh
> bd
Note: Clusters of *bh
dh
are similarly developed.
259
This outcome is an extension of the ruki-rule in Iranian (actually attested in Avestan only, not in Old Persian).
260
The variant Albanian outcome f is a result of a metathesis of IE *ps on *sp. The trajectory is: ps > sp > hf >
0f, see above. The Albanian outcome of a cluster without this metathesis is 0š, i.e., with a typical Albanian
palatalization of a sibilant.
254
13.3 The development of the sibilant clusters
The set of phonemes of the reconstructed IE obstruent system has a single sibilant phoneme *s
(with a positional allophone *z before voiced plosives). The satəm-languages had another
phoneme *š, resulting from the split of the old single sibilant due to Pedersen’s law (the rukirule),
securely attested for four of the six satəm-branches, but insecure for Albanian and
Armenian.
The development of the sibilant clusters are remarkably stable, the Celtic development
being an exception.
The developments of the IE cluster of *st could be summed as:
i. IE *st > st, attested in all branches except Celtic and Albanian;
ii. IE *st > šṭ, attested in Albanian;
iii. IE *st > ss, attested in Celtic languages (in Gaulish beside the variant ðð of the insecure
phonemic value);
The conservative st-outcome is a regular one, the Albanian št-outcome is a later result of the
independent Albanian development.
The Celtic development, tied with the development of the IE clusters *Tt, *Ts (see
above), can be explained only within this wider frame. If the Gaulish ðð had a value of an
affricate, the trajectory would be, if we accept the proposal of Lewis/Pedersen (1937: 20): st >
ts > ss, or, as we dare to propose: st > sts
> ts
ts
> ss.
st > st > sϑ > ϑϑ > ss
The developments of the satəm-cluster of *št can be summed up as:
i. IE *št > št, attested in Iranian and Lithuanian;261
ii. IE *st > ṣṭ, attested in OIA and Nūristānī;
Note: The OIA–Nūristānī outcome is just an area variation of the preceding regular outcome, as it is in the case
of IE clusters of *Ḱt > OIA ṣṭ.
The development of the clusters of sibilant + s usually leads towards the degeminated form of
a single sibilant or to the preservation of the geminate. The bisibilant cluster could be considered
not as a preserved one, but as an analogical restoration, since we have all the reason to consider
the simplification *ss > 0s as already being Indo-European at least in the 2nd
sg. pr. of the root
*√H1es- ‘be’ (cf. OIA asi, L. es, OCS jesi, Lith esi). A remarkable development is attested in
OIA, where the verbal forms have the outcome ts. This form can be traced back to the older
261
The Albanian outcome can not be distinguished from the outcome of the IE cluster *st, since the old IE
sibilant was regularly palatalized in Albanian.
255
spirant stage, resulting from the levelling of this cluster with a cluster of ϑs, which we
reconstruct as a stage of the development of the IE cluster Tt, later re-buccalized in OIA.
ss > ss > 0s
> ϑs > ts
The development of the clusters formed by a ruki-sibilant + s in the satəm-languages262
usually
also leads towards the degeminated form of a single sibilant or to the preservation of the
geminate. Again, OIA is an exception, where the right sibilant is replaced by a plosive, this time
velar (and the sibilant is cerebral). Even in this case we assume the shift of the original sibilant
towards a spirant and later re-buccalization of the spirant (and an area cerebralization of a right
sibilant):
šs > šš > 0š
> xš > kṣ
The development of the clusters of sibilant + dh
is simple outside Indic (where any voiced
sibilant of any origin is regularly lost), with the voicing of a sibilant. The Greek development
is remodelled due to devoicing of the original IE context *dh.
For the IE *s the development can
be modelled as:
i. IE *sdh
> 0dh
, attested in Old Indo-Aryan;
ii. IE *sdh
> zd, attested in Iranian and Baltic;
iii. IE *sdh
> sth
, attested in Greek;
And similarly, for the ruki-sibilant *ž the outcomes are in the satəm-languages, with a loss of a
sibilant in OIA, preserved as voiced in Iranian and Baltic:
i. IE *šdh
> 0ḍh
, attested in Old Indo-Aryan;
ii. IE *šdh
> žd, attested in Iranian and Baltic;
The trajectories for both sibilants we can model as:
sdh
> zd(h)
> zð > ðð > 0dh
šdh
> žd(h)
> žð > ð ̣ð ̣> 0ḍh
262
Again, the Albanian and Armenian data do not distinguish this outcome from that of *ss, since the validity of
Pedersen’s Law in these languages is questionable.
257
Appendix I: The comparative table of IE clusters plosive + t/s- in the given Indo-European languages263
i. the clusters dental plosive + t/s-:
IE OIA Av. N. Lith. OCS Arm. Alb. Gr. L. Os. Gal. Ir. W. Goth. Hitt. TB
Tt tt st 0t st st u̯t 0s st ss ss ðð
ss
ss ss ss ts
t ts
ts
Ts ts 0s ć 0s 0s cʿ 0š 0s ss ss ðð
ss
ss ss ss ts ts
tV tV tV tV tV tV tʿV tV tV tV tV tV tV tV þV tV tV
ii. the clusters palatovelar plosive + t/s-:
IE OIA Av. N. Lith. OCS Arm. Alb. Gr. L. Os. Gal. Ir. W. Goth. Hitt. TB
Ḱt ṣṭ št ṣṭ št st st 0t kt kt ht xt xt t ht kt kt
Ḱs kṣ 0š c 0š 0s cʿ 0š ks ks 0s xs ss 0x hs ks ks
ḱV śV šV cV šV sV sV ϑV kV kV kV kV kV kV hV kV kV
iii. the clusters velar plosive + t/s-:
IE OIA Av. N. Lith. OCS Arm. Alb. Gr. L. Os. Gal. Ir. W. Goth. Hitt. TB
Kt kt xt 0t kt št
0t
0tʿ 0t kt kt ht xt xt t ht kt kt
Ks kṣ xš ? ks 0x
0š
0š 0š ks ks 0s xs ss 0x hs ks ks
kV kV
cV
kV
cV
kV
cV
kV
cV
kV
čV
kʿV kV kV kV kV kV kV kV hV kV kV
iv. the clusters labiovelar plosive + t/s-:
IE OIA Av. N. Lith. OCS Arm. Alb. Gr. L. Os. Gal. Ir. W. Goth. Hitt. TB
Ku̯
t kt xt kt kt št 0tʿ 0t pt264
kt ht (xt) xt t ht ku̯
t kt
263
The examples are limited on the non-Bartholomaean clusters. The dh
-clusters are fully omitted since not securely attested in many daughter languages. The cluster
obstruent + vowel is added to demonstrate the unmarked form of the given obstruent (secondarily palatalized velars from Indo-Iranian and Slavic are omitted, as similar
secondary forms).
264
ku̯
t, ku̯
s in Mycenaean.
258
0t kt
Ku̯
s kṣ xš ? ks 0x
0š
0š 0š ps
ks
ks 0s (xs) ss 0x hs k u̯
s ks
ku̯
V kV
cV
kV
cV
kV
cV
kV kV
čV
kʿV kV pV ku̯
V pV pV kV pV hu̯
V ku̯
V kV
v. the clusters labial plosive + t/s-:
IE OIA Av. N. Lith. OCS Arm. Alb. Gr. L. Os. Gal. Ir. W. Goth. Hitt. TB
Pt pt pt 0t pt 0t
st
0tʿ 0t pt pt ft265
xt xt t ft pt pt
Ps ps fs
fš
ps 0s 0s 0š
(0f)266
ps ps 0s xs ss 0x fs ps ps
pV pV pV pV pV pV hV pV pV pV pV 0V 0V 0V fV pV pV
vi. the clusters sibilant + t/s-:
IE OIA Av. N. Lith. OCS Arm. Alb. Gr. L. Os. Gal. Ir. W. Goth. Hitt. TB
st st st st st st st št st st st ðð
ss
ss ss st st st
ss ts 0s 0s 0s 0s 0š 0s 0s 0s 0s 0s 0s (ss) ss ss
sV sV hV sV sV sV hV gjV
šV
hV 0s 0s 0s 0s 0s sV sV sV
(št) ṣṭ št ṣṭ št st st št st st st ðð
ss
ss ss st st st
(šs) kṣ 0š 0š 0š 0s 0š 0s 0s 0s 0s 0s 0s (ss) ss st
(šV) ṣV šV ṣV šV šV hV šV hV sV sV sV sV sV sV sV sV
265
Umbr. ht
266
Due to metathesis of *ps on *sp, later fricativized, debuccalized and elided.
259
Appendix II: The pan-chronic overviews of the given developments
i. The pan-chronic overview of the Indic development:
Kt > kt
Ḱt > çt > št> ṣṭ (SPT) vel Ḱt > tš
t > št > ṣṭ (AFT)
Tt > ϑt > tt (SPT) vel Tt > ts
t > tt (AFT)
Pt > pt
gh
t > γδ > gdh
ǵh
t > ʝδ > žδ > ẓḍh
> i̯ḍh
(SPT) vel ǵh
t > dž
dh
> ždh
> ẓḍh
> i̯ḍh
(AFT)
dh
t > δδ > ddh
(SPT) vel dh
t > dz
dh
> ddh
(AFT)
bh
t > βδ > bdh
st > st
št > št > ṣṭ
Kdh
> γδ > gdh
Ḱdh
> ʝδ > ʝʝ > ḍḍh
(SPT) vel Ḱdh
> dž
dh
> ždh
> ḍḍh
(AFT)
Tdh
> δδ > ddh
/0dh
(SPT) vel Tdh
> dz
d(h)
> ddh
/0dh
(AFT)
Pdh
> βδ > bdh
(?)
gh
dh
> γδ > gdh
(?)
ǵh
dh
> ʝδ > žḍh
> i̯dh
> 0ḍh
(SPT) vel ǵh
dh
> dž
dh
> ždh
> i̯dh
> 0ḍh
(AFT)
dh
dh
> δδ > i̯dh
(SPT) vel dh
dh
> dz
dh
> i̯dh
(AFT)
bh
dh
> βδ > bdh
(?)
sdh
> zδ > hδ > i̯dh
šdh
> žδ > ʝδ > ḍḍh
Ks > kš > kṣ
Ḱs > çš > xš/ϑš > kṣ (SPT) vel Ḱs > tš
š > kš/ṭṣ > kṣ (AFT)
Ts > ϑs > ts (SPT) vel Ts > ts
s > ts (AFT)
Ps > ps
gh
s > γž → kš > kṣ
ǵh
s > ʝž → kš > kṣ (SPT) vel ǵh
s > dž
ž → kš > kṣ (AFT)
dh
s > δz → ts (SPT) vel dh
s > dz
z → ts (AFT)
bh
s > βz → ps
ss > ϑs > ts
šs > çš > kṣ
ii. Pan-chronic overview of the Iranian267
development:
Kt > kt > xt
Ḱt > çt > št (SPT) vel Ḱt > tš
t > št (AFT)
Tt > ϑt > st (SPT) vel Tt > ts
t > st (AFT)
Pt > pt > φt  pt/*ft (?)
267
There is no difference in the general features of the development between Avestan and Old Persian.
260
gh
t > γδ > gd
ǵh
t > ʝδ > žd (SPT) vel ǵh
t > dž
d(h)
> žd (AFT)
dh
t > δδ > zd (SPT) vel dh
t > dz
d(h)
> zd (AFT)
bh
t > βδ > bd
st > st
št > št
Kdh
> γδ > gd
Ḱdh
> ʝδ > žd (SPT) vel Ḱdh
> dž
d(h)
> žd (AFT)
Tdh
> δδ > zd (SPT) vel Tdh
> dz
d(h)
> zd (AFT)
Pdh
> βδ > bd
gh
dh
> γδ > gd (?)
ǵh
dh
> ʝδ > žd (SPT) vel ǵh
t > dž
d(h)
> žd (?) (AFT)
dh
dh
> δδ > zd (SPT) vel dh
t > dz
d(h)
> zd (AFT)
bh
dh
> βδ > bd (?)
sdh
> zd
šdh
> žd
Ks > kš > xš
Ḱs > çš > šš > 0š (SPT) vel Ḱs > tš
š > šš > 0š (AFT)
Ts > ϑs > ss > 0s (SPT) vel Ts > ts
s > ss > 0s (AFT)
Ps > ps > φs > fs/fš
gh
s > γž
ǵh
s > ʝž > žž > 0ž (SPT) vel ǵh
s > dž
ž > žž > 0ž (AFT)
dh
s > δz > zz > 0z (SPT) vel dh
s > dz
z > zz > 0z (AFT)
bh
s > βz > βž
ss > 0s
šs > šš > 0š
iii. Pan-chronic overview of the Nūristānī development:
Kt > kt > tt > 0t
Ḱt > çt > št (SPT) vel Ḱt > tš
t > št (AFT)
Tt > ϑt > tt > 0t (SPT) vel Tt > ts
t > tt > 0t (AFT)
Pt > pt > tt > 0t
st > st > st/št
št > št > ṣṭ (/> ṭṭ > 0ṭ ?)
šdh
> žd(h)
> žd (/> ḍḍ > 0ḍ ?)
Ḱs > çš > ϑš > ts (SPT) vel Ḱs > tš
š > tš > ts (AFT)
Ts > ϑs > ϑs > tš (?) (SPT) vel Ts > ts
s > tš (?) (AFT)
261
iv. Pan-chronic overview of the Baltic268
development:
K(u̯ )
t > kt
Ḱt > çt > št (SPT) vel Ḱt > > tš
t > št (AFT)
Tt > ϑt > st (SPT) vel Tt > ts
t > st (AFT)
Pt > pt
K(u̯ )
s > ks
Ḱs > çš > šš > 0š (SPT) vel Ḱs > tš
š > šš > 0š (AFT)
Ts > ϑs > ss > 0s (SPT) vel Ts > ts
s > ss > ts (AFT)
Ps > ps
st > st
št > št
ss > 0s
šs > 0š
v.: Pan-chronic overview of the Old Church Slavonic development:269
K(u̯ )
t > xt > çt > št (SPT) vel K(u̯ )
t > tt > št (GET)
K(u̯ )
t > xt > ht > 0t (SPT) vel K(u̯ )
t > tt > 0t (GET)
Ḱt > çt > ϑt > st (SPT) vel Ḱt > tš
t > št > st (AFT)
Tt > ϑt > st (SPT) vel Tt > ts
t > st (AFT)
Pt > φt > ht > 0t (SPT) vel Pt > tt > 0t (GET)
Pt > φt > st (SPT) vel Pt > ? > st (GET)
K(u̯ )
s > xx/šš > 0x/0š (SPT) vel K(u̯ )
s > kš > šš > 0x/0š (GET)
Ḱs > çs > ϑs > ss > 0s (SPT) vel Ḱs > tš
s > šs > ss > 0s (AFT)
Ts > ϑs > ss > 0s (SPT) vel Ts > ts
s > ss > 0s (AFT)
Ps > φs > ss > 0s (SPT) vel Ps > ss > 0s (GET)
st > st
št > st
ss > 0s
šs > > šš > 0š
vi. Pan-chronic overview of the Armenian development:
K(u̯ )
t > xϑ > hϑ > 0tʿ
Ḱt > çt > śt > st
Tt > ϑt > ht > u̯ t
Pt > φϑ > hϑ > 0tʿ
K(u̯ )
s > xš > šš > 0š270
Ḱs > çs > ϑs > cʿ
Ts > ϑs (?) > cʿ
Ps > φs > hs > 0s271
268
Here demonstrated on the Lithuanian data.
269
The “strategy of simplification”, since trivial, is omitted in this overview.
270
Alternativelly with the same frame: K(u̯ )
s > xš > hš > 0š.
271
Alternatively within the same frame: Ps > φs > ss > 0s.
262
St > st
Ss > 0s
vii. Pan-chronic overview of the Albanian development:
K(u̯ )
t > kt > xt > ht > 0t
Ḱt > çt > xt > ht > 0t
Tt > ϑt > ϑϑ > c > 0s (SPT) vel Tt > ts
t > cc > 0c > 0s (AFT)
Pt > pt > φt > ht > 0t
K(u̯ )
s > ks > xs > hs > 0s > 0š272
Ḱs > çs > xs > hs > 0š273
Ts > ϑs > ss > 0s > 0š (SPT) vel Ts > ts
s > ss > 0s > 0š (AFT)
Ps > ps  sp > hf > 0f
St > st > št
Ss > ss > 0s > 0š
viii. Pan-chronic overview of the Ancient Greek development:
K(´)
t > kt
Ku̯
t > pt
Tt > ϑt > ϑϑ > ss (SPT) vel Tt > ts
t > tst > st (AFT)
Pt > pt
Kdh
> gdh
 Kth
> kh
th
Ku̯
dh
> gu̯
dh
 Ku̯
th
> ph
th
Tdh
> > δdh
 tth
> ϑth
> sth
(SPT) vel Tdh
> dz
dh
 ts
th
> tsth
> sth
(AFT)
Pdh
> bdh
 pth
> ph
th
K(´)
s > ks
Ku̯
s > ps
Ts > ϑs > ss > 0s (SPT) vel Ts > ts
s > tss > ss > 0s (AFT)
Ps > ps
st > st
sdh
> zdh
 sth
> sth
ss > 0s
ix. Pan-chronic overview of the Mycenaean development:
K(´)
t > kt
Ku̯
t > ku̯
t
Tt > ϑt > ϑϑ > ss (SPT) vel Tt > ts
t > tst > st (AFT)
Pt > pt
272
In the case of the validity of the ruki-rule for Proto-Albanian: Ku̯
s > kš > hš > 0š.
273
If we assume that Albanian was affected by ruki-rule, the development will be: Ḱs > kš > xš >hš > 0š.
263
Kdh
> gdh
 Kth
> kh
th
Ku̯
dh
> gu̯
dh
 Ku̯
th
> ku̯ h
th
Tdh
> > δdh
 tth
> ϑth
> sth
(SPT) vel Tdh
> dz
dh
 ts
th
> tsth
> sth
(AFT)
Pdh
> bdh
 pth
> ph
th
K(´)
s > ks
Ku̯
s > ku̯
s
Ts > ϑs > ss > 0s (SPT) vel Ts > ts
s > tss > ss > 0s (AFT)
Ps > ps
st > st
sdh
> zdh
 sth
> sth
ss > 0s
x. Pan-chronic overview of the Latin development:
K(´)
t > kt
Ku̯
t > kt
Tt > ϑt > ϑϑ > (s)s (SPT) vel Tt > ts
t > tss > (s)s (AFT)
Pt > pt
K(´)
s > ks
Ku̯
s > ks
Ts > ϑs > (s)s (SPT) vel Ts > ts
t > tss > (s)s (AFT)
Ps > ps
st > st
ss > (s)s
xi. Pan-chronic overview of the Sabellic development:
K(´)
t > xt > ht (/> 0t)
Ku̯
t > kt > xt > ht
Tt > ϑt > ϑϑ > (s)s (SPT) vel Tt > ts
t > tss > (s)s (AFT)
Pt > > φt > ft274
K(´)
s > xs > hs > 0s
Ku̯
s > ks > xs > hs > (s)s
Ts > ϑs > (s)s (SPT) vel Ts > ts
t > tss > (s)s (AFT)
Ps > > φs > hs > 0s
st > st
ss > (s)s
274
Umbrian: Pt > φt > xt > ht
264
xii.: Pan-chronic overview of the Brythonic development:
K(´)
t > xt (> ht) > t
Ku̯
t > xt (> ht) > t
Tt > ϑt > ϑϑ > ss (SPT) vel Tt > ts
t > tsts > ss (AFT)
Pt > φt > (ft > ht >) t
K(´)
s > xs > xx > 0x
Ku̯
s > xs > xx > 0x
Ts > ϑs > ss (SPT) vel Ts > ts
s > tsts > ss (AFT)
Ps > φs > xx > 0x
st > sϑ > ϑϑ > ss (SPT) vel st > ts > ss (AFT)
xiii. Pan-chronic overview of the Goidelic development:
K(´)
t > xt
Ku̯
t > xt
Tt > ϑt > ϑϑ > ss (SPT) vel Tt > ts
t > tsts > ss (AFT)
Pt > φt > (ft >) xt
K(´)
s > xs > ss
Ku̯
s > xs > ss
Ts > ϑs > ss (SPT) vel Ts > ts
s > tsts > ss (AFT)
Ps > φs > ss
st > sϑ > ϑϑ > ss (SPT) vel st > ts > ss (AFT)
xiv. Pan-chronic overview of the Gallic development:
K(´)
t > xt
(Ku̯
t > xt)
Tt > ϑt > ϑϑ > đđ/ss (SPT) vel Tt > ts
t > tsts > đđ/ss275
(AFT)
Pt > φt > (ft >) xt
K(´)
s > xs
(Ku̯
s > xs)
Ts > ϑs > ϑϑ > đđ/ss (SPT) vel ts
s > ts
t > tsts > đđ/ss (AFT)
Ps > φs > xs
st > ϑs > ϑϑ > đđ/ss (SPT) vel st > ts > đđ/ss (AFT)
xv.: Pan-chronic overview of the Gothic development:
K(´)
t > kt > xt > ht
Ku̯
t > kt > xt > ht
Tt > ϑt > ϑϑ > ss (SPT) vel Tt > ts
t > tsts > ss (AFT)
Pt > pt > φt > ft
K(´)
s > ks > xs > hs
Ku̯
s > ks > xs > hs
Ts > ϑs > ss (SPT) vel Ts > ts
s > tss > ss (AFT)
275
Here đđ marks two dental affricates, probably voiceless.
265
Ps > ps > φs> fs
st > st
ss > ss (?)
xvi. Pan-chronic overview of the Hittite development:
K(´)
t > kt
Ku̯
t > kt /  ku̯
t
Tt > ts
t > ts͜ t
Pt > pt
K(´)
s > ks
Ku̯
s > (ks ?) /  ku̯
s
Ts > ts
s > ts
Ps > ps
st > st
ss > ss
xvii. Pan-chronic overview of the Tocharian development:
K(´)
t > kt
Ku̯
t > kt
Tt > ts
t > ts
ts
/tts
Pt > pt
K(´)
s > ks
Ku̯
s > (ks ?)
Ts > ts
s > ts
Ps > ps
st > st
ss > ss
267
Abbreviations of languages
Aeol. – Aeolic
Alb. – Albanian
Arc. – Arcadian
Arm. – Armenian
Att. – Attic
Av. – Avestan
B. – Bulgarian
Br. – Breton
Bryth. – Brythonic
Celt. – Celtic
Corn. – Cornish
CS – Common Slavic
Cypr. – Cypriotic
Cz. – Czech
Dor. – Dorian
Gal. – Gaulish
Germ. – Germanic
Goid. – Goidelic
Goth. – Gothic
Gr. – Greek
Hitt. – Hittite
Hom. – Homeric
IE – Indo-European
Ir. – Irish
L. – Latin
Latv. – Latvian
Lith. – Lithuanian
Luw. – Luwian
MBr. – Middle Breton
MCorn. – Midle Cornish
MIr. – Middle Irish
MHG – Middle High German
MW. – Middle Welsh
N. – Nūristānī
NP – New Persian
OAlb. – Old Albanian
OAv. – Old Avestan
OBr. – Old Breton
OCorn. – Old Cornish
OCS– Old Church Slavonic
OE – Old English (Anglo-Saxon)
OGeg. – Old Gegh
OHG – Old High German
OIA – Old Indo-Aryan
OIr. – Old Irish
OL. – Old Latin
OLith. – Old Lithuanian
ON – Old Norse
OP – Old Persian
OPol. – Old Polish
OS – Old Saxon
Os. – Oscan
PAlb. – Proto-Albanian
PAnat. – Proto-Anatolian
PArm. – Proto-Armenian
PCelt. – Proto-Celtic
PGerm. – Proto-Germanic
PGr. – Proto-Greek
Phl. – Pahlavi
PItal. – Proto-Italic
Pol. – Polish
PPAlb. – Pre-Proto-Albanian
Pruss. – (Old) Prussian
PSab. – Proto-Sabellian
Ru. – Russian
RuCS – Russian Church Slavonic
Sab. – Sabellian
SCr. – Serbo-Croatian
Slk. – Slovakian
Sln. – Slovenian
Toch. – Tocharian
Uk. – Ukrainian
Um. – Umbrian
W. – Welsh
YAv. – Young Avesta
269
Literature
Adams, Douglas Q. 1988. Tocharian Historical Phonology and Morphology. New Haven:
American Oriental Society.
Adams, Douglas Q. 20132
. A Dictionary of Tocharian B. Revised and Greatly Enlarged.
Volume 1. Volume 2. Amsterdam – New York: Rodopi.
Allen, W. Sidney. 1968. Vox Graeca. A Guide to the Pronunciation of Classical Greek.
Cambridge: Cambridge University Press.
Allen, W. Sidney. 1978. The PIE velar series: Neogrammarian and other solutions in the light
of attested parallels. Transactions of the philological society 76. 87–110.
Andersen, Hennig. 1968. IE *s after i, u, r, k in Baltic and Slavic. Acta Linguistica 11, 171–
190.
Anderson, Stephen R. 1970. On Grassmann’s Law in Sanskrit. Linguistic Inquiry 1(4). 387–
396.
Andreev, Nikolai D. 1957. Periodizacija istorii indoevropejskogo prajazyka. Voprosy
jazykoznanija 2. 3–18.
Arumaa, Peeter. 1976. Urslavische Grammatik. II. Band. Konsonantismus. Heidelberg: Carl
Winter-Universitätsverlag.
Aura Jorro, Francisco / Adrados, Francisco R. 1985. Diccionario griego-español. Anejo 1.
Diccionario micénico. Volumen 1. Madrid: Consejo Superior de Investigaciones
Cientifícas.
Aura Jorro, Francisco / Adrados, Francisco R. 1993. Diccionario griego-español. Anejo 2.
Diccionario micénico. Volumen 2. Madrid: Consejo Superior de Investigaciones
Cientifícas.
Baldi, Philip. 1991. Lachmannʼs Law in the Light of the Glottalic Theory of PIE Consonantism.
In: Coleman, Robert (ed.). New Studies in Latin Linguistics: Selected Papers from the 4th
International Colloquium on Latin Linguistics, Cambridge, April 1987. Amsterdam /
Philadelphia: John Benjamins. 3–21.
Baldi, Philip. 1999. The Foundations of Latin. Berlin / New York: Mouton de Gruyter.
Bammesberger, Alfred. Der Aufbau des germanischen Verbalsystems. Heidelberg: Carl Winter
Universitätsverlag.
Bartholomae, Christian. 1882. Die arische Vertretung von med. Asp. + t und med. Asp. +s.
Arische Forschungen I. 3–24. Halle: Niemeyer.
Bartholomae, Christian. 1883. Handbuch der altiranischen Dialekte. Leipzing: Breitkopf &
Härtel.
Bartholomae, Christian. 1885. Θυγάτερ. KZ 27. 206–207.
Bartholomae, Christian, 1887. Die Vertretung des altitalischen ss im oskischen (etc.).
Bezzenbergers Beiträge 12. 80–92.
Bartholomae, Christian. 1895–1901. I. Vorgeschichte der iranischen Sprachen. II. Awestasprache
und Altpersisch. In: Geiger, Wilhelm / Kuhn, Ernst (eds.). Grundriß der
iranischen Philologie. Erster Band. I. Abteilung. Strassburg: Karl J. Trübner: 1–248.
Bartoněk, Antonín 2003. Handbuch des mykenischen Griechisch. Heidelberg:
Universitätsverlag C. Winter.
Beekes, Robert S. P. 1988. A Grammar of Gatha-Avestan. Leiden: Brill.
Beekes, Robert S. P. 2003. Historical phonology of Classical Armenian. In: Kortlandt, Fredrik.
Armeniaca. Comparative Notes. Ann Arbor: Caravan Books, 133–211.
Beekes, Robert S. P. 2011. Comparative Indo-European Linguistics. Armsterdam/Philadelphia:
John Benjamins Publishing Company.
Beekes, Robert, S. P. 2016a. Etymological dictionary of Greek. Volume one. Leiden: Brill.
270
Beekes, Robert, S. P. 2016b. Etymological dictionary of Greek. Volume two. Leiden: Brill.
Bernabé Pajares, Alberto. 1971. Aportaciones al estudio fonológico de las guturales
indoeuropeas, Emérita, Revista de Lingüística y Filología Clásica, XXXIX, 63–107.
Bezzenberger, Adalber. 1890. Die indogermanischen Gutturalreihen. Beiträge zur Kunde
der indogermanischen Sprachen. 16, 234–260.
Belić, Aleksandar. 1921. Srpskohrvatske glasovne grupe -jt-, -jd- i
praslovansko -kti-, -gdi-, -hti- [The Serbo-Croatian phonemic clusters -jt-, -jd- and PreSlavic
-kti-, -gdi-, -hti-]. Južnoslovenski Filolog 5. 217–226.
de Bernardo Stempel, Patrizia. 1999. Nominale Wortbildung der älteren Irischen.
Stammbildung und Derivation. Tübingen: Niemayer.
Blažek, Václav. 1999. Numerals. Brno: Masarykova univerzita.
Blažek, Václav 2009. On the position of Gaulish within Celtic from the point of view of
glottochronology. Indogermanische Forschungen 114: 257−299.
Blažek, Václav / Hegedüs, Irén. 2012: On the position of Nuristani within Indo-Iranian. In:
Sukač, Roman / Šefčík, Ondřej (eds.). The Sound of Indo-European 2. München: Lincom
Europa, 40–66.
Bloch, Jules. 1965. Indo-Aryan from Vedas to Modern times. (translated by Master, Alfred).
Paris: Librairie d’Amérique et d’Orient. Adrien – Maisonneuve.
Bloomfield, Leonard. 1911. The Indo-European Palatals in Sanskrit. The American Journal of
Philology 32/1. 36–57.
Blust. Robert. 2009. The Austronesian Languages. Canberra: Pacific Linguistics. Research
School of Pacific and Asian Studies. The Australian National University.
Bousquette, Joshua / Salmons, Joseph. 2017. Germanic. In: Kapović, Mate (ed.). The IndoEuropean
Languages. New York: Routledge, 387–420.
von Bradke, Peter. 1890. Über Methode und Ergebnisse der arischen Alterthumswissenshaft.
Giessen: J. Ricker'che Buchhandlung.
Brandenstein, Wilhelm / Mayrhofer, Manfred. 1964. Handbuch des Altpersichen. Wiesbaden:
Harrassowitz.
Braune, Wilhelm / Ebbinghaus, Ernst A. 1981. Gotische Grammatik mit Lesestücken und
Wörterverzeichnis. Tübingen: Max Niemeyer.
Braune, Wilhelm / Heidermanns, Frank. 2004. Gotische Grammatik mit Lesestücken und
Wörterverzeichnis. Tübingen: Max Niemeyer.
(von) Brücke, Ernst. 1856. Grundzüge der Physiologie und Systematik der Sprachlaute für
Linguisten und Taubstummenlehrer. Wien: Carl Gerold´s Sohn.
Brugmann, Karl. 1880. Die lautgruppe dentale explosive + t im indogermanischen. In: Osthoff,
Hermann / Brugmann, Karl. Morphologische Untersuchungen auf dem Gebiete der
indogermanischen Sprachen. Teil. 3. Leipzig: S. Hirzel, 131–147.
Brugmann, Karl. 1885. Handbuch der klassischen Altertums-Wissenschaft in systematischer
Darstellung mit besonderer Rücksicht auf Geschichte und Methodik der einzelnen
Disziplinen. Nördlingen: C.H. Beck’schen Buchhandlung.
Brugmann, Karl. 1886. Grundriss der vergleichenden Grammatik der indogermanischen
Sprachen. Erster Band: Einleitung und Lautlehre. Strassburg: Karl. J. Trübner.
Brugmann, Karl. 1890. Handbuch der klassischen Altertums-Wissenschaft in systematischer
Darstellung mit besonderer Rücksicht auf Geschichte und Methodik der einzelnen
Disziplinen. Nördlingen : C.H. Beck’schen Buchhandlung.
Brugmann, Karl. 1897. Vergleichende Laut-, Stammbildungs- und Flexionslehre der
indogermanischen Sprachen. Zweite Bearbeitung. Erster Band: Einleitung und
Lautlehre. Zweite Hälfte. Strassburg: Karl. J. Trübner.
271
Brugmann, Karl. 1900. Handbuch der klassischen Altertums-Wissenschaft in systematischer
Darstellung mit besonderer Rücksicht auf Geschichte und Methodik der einzelnen
Disziplinen. Nördlingen: C.H. Beck’schen Buchhandlung.
Brugmann, Karl. 1913. Zu den Ablautverhältnissen der sogennannten starken Verba des
Germanischen. Indogermanische Forschungen 32: 179–195.
Brugmann, Karl. 1913/1914. Gotisch usstagg ʽstich ausʼ. Indogermanische Forschungen 33:
284–285.
Brugmann, Karl/Thumb, Albert. 1913. Griechische Grammatik. Lautlehre, Stammbildungsund
Flexionslehre, Syntax. München: C.H. Beck’schen Buchhandlung.
Brust, Manfred. 2018. Historische Laut- und Formenlehre des Altpersischen : mit einem
etymologischen Glossar. Innsbruck: Institut für Sprachen und Literaturen der Universität
Innsbruck, Bereich Sprachwissenschaft.
Bubenik, Vit. 1996. The Structure and Development of Middle Indo-Aryan Dialects. Delhi:
Motilal Banarsidass.
Bubenik, Vit. 2017. The Phonology of Greek. In: Klein, Jared / Joseph, Brian / Fritz / Matthias
(eds). Handbook of Comparative and Historical Indo-European Linguistics (HSK 41.1).
Berlin / Boston: Walter de Gruyter, 637–653.
Buck, Carl D. 1904. A Grammar of Oscan and Umbrian with a Collection of Inscriptions and
a Glossary. Boston: The Athenæum Press, Ginn & Company.
Buck, Carl D. 1933. Comparative Grammar of Greek and Latin. Chicago: The University of
Chicago Press.
Buddruss, Georg. 1977. Nochmals zur Stellung der Nuristan-Sprachen des Afghanistanischen
Hindukusch. Münchener Studien zur Sprachwissenschaft 36. 19–38.
Bugge, Sophus. 1889. Beiträge zur etymologischen Erläuterung der armenischen Sprache. .
Leipzig: Breitkopf und Härtel.
Bugge, Sophus. 1893. Beiträge zur etymologischen Erläuterung der armenischen Sprache.
Zeitschrift für vergleichende Sprachforschung 32, 1–87.
Burrow, Thomas. 1955. The Sanskrit Language. London: Faber and Faber.
Çabej, Eqrem. 1972. Über einige Lautrefeln des Albanischen. Die Sprache 18, 132–154.
Camaj, Martin. 1966. Albanische Wortbildung. Wiesbaden: Otto Harrassowitz.
Cantera, Alberto. 2017. The phonology of Iranian. In: Klein, Jared / Joseph, Brian / Fritz
Matthias (eds). Handbook of Comparative and Historical Indo-European Linguistics.
Berlin/Boston: De Gruyter Mouton. 481–503.
Cardona, George / Jain, Dhanes. 2003. General introduction. In: Cardona, George/Jain,
Dhanesh (eds). The Indo-Aryan Languages. 1–45. London: Routledge.
Cardona, George 2017. The documentation of Indic. In: Klein, Jared / Joseph, Brian / Fritz
Matthias (eds). Handbook of Comparative and Historical Indo-European Linguistics.
Berlin/Boston: De Gruyter Mouton. 309–325.
Carling, Gerd [et al.]. 2009. Dictionary and Thesaurus of Tocharian A. Volume 1: A–J.
Wiesbaden: Harrassowitz Verlag.
Carlton, Terence. R. 1991. Introduction to the Phonological History of the Slavic Languages.
Columbus: Slavica Publishers.
Cavoto, Fabrice. 2001. Did PIE have Three Velar series? MSS 61, 29–56.
Čekman, Valerij M. (1974): O reflexach indoevropejskix *ḱ, *ǵ v balto-slavjanskom
jazykovom areale. Balto-slavjanskie issledovanija. 116–135
CHD L–N = Günterbock, Hans G. / Hoffner, Harry. A. et al. 1989. The Hittite Dictionary of
the Oriental Institute of the University of Chicago. L – N. Chicago: The University of
Chicago.
CHD P = Günterbock, Hans G. / Hoffner, Harry. A. et al. 1997. The Hittite Dictionary of the
Oriental Institute of the University of Chicago. P. Chicago: The University of Chicago.
272
CHD Š = Günterbock, Hans G. / Hoffner, Harry. A. et al. 2019. The Hittite Dictionary of the
Oriental Institute of the University of Chicago. Š. Chicago: The University of Chicago.
Cheung, Johny. 2007. Etymological Dictionary of the Iranian Verb. Leiden: Brill.
Clackson, James. 1994. The Linguistic Relationship between Armenian and Greek. Oxford:
Blackwell.
Cocchia, Enrico. 1883: Questioni di fonologia Latina. Rivista di filologia e d´instrutione
classica XI. 16–101.
Coleman, Robert. 1992. Italic [numerals]. In: Gvozdanović, Jadranka (ed.). Indo-European
Numerals. Berlin/New York: Mouton de Gruyter, 389–445.
Collinge, Neville E. 1985. The Laws of Indo-European. Amsterdam/Philadelphia: John
Benjamins.
Collins, Daniel. 2017. The Phonology of Slavic. In: Klein, Jared / Joseph, Brian / Fritz /
Matthias (eds). Handbook of Comparative and Historical Indo-European Linguistics
(HSK 41.3). Berlin / Boston: Walter de Gruyter, 1414–1538.
Comrie, Bernard. 1992. Balto-Slavonic [Numerals]. In: Gvozdanović, Jadranka. IndoEuropean
Numerals. Berlin / New York. 717–834.
Cowgill, Warren. 1980 : The Etymology of Irish guidid and the Outcome of *gu̯ h
in Celtic. In:
Marhofer, Manfred / Peters, Martin / Pfeiffer, Oscar E. (eds.) Lautgeschichte und
Etymologie. Wiesbaden: Reichert. 49–78.
Curtius, Georg. 1853. Die aspîraten der indogermanischen sprachen. KZ 2. 321–336.
Darden, Bill. J. 1978. On the Slavic reflexes of Indo-European *pt. GL 18, 10–13.
Degener, Almuth 2002. The Nuristani Languages. In: Sims-Williams, Nicolas (eds.). IndoIranian
Languages and Peoples. 103–117. Oxford: Oxford University Press.
Delamarre, Xavier. 2003. Dictionaire de la langue gauloise. Une approche linguistique du
vieux-celtique continental. Paris: Editions Errance.
Demiraj, Bardhyl. 1997: Albanische Etymologie (Untersuchungen zum albanischen
Erbwortschatz). Amsterdam/Atlanta: Rodopi.
Demiraj, Shaban. 1993: Historische Grammatik der albanischen Sprache. Wien: Verlag der
Österreichischen Akademie der Wissenschaften.
Derksen, Rick. 2008. Etymological dictionary of the Slavic Inherited Lexicon. Leiden: Brill.
Derksen, Rick. 2015. Etymological Dictionary of the Baltic Inherited Lexicon. Leiden/Boston:
Brill.
Deshayes, Albert. 2003. Dictionnaire étymologique du breton. Douarnenez: Le Chasse-Marée.
de Vaan, Michiel. 2008. Etymological Dictionary of Latin and the other Italic Languages.
Leiden / Boston: Brill.
de Vaan, Michiel. 2018. The phonology of Albanian. In: Klein, Jared / Joseph, Brian / Fritz,
Matthias (eds.) Handbook of Comparative and Historical Indo-European Linguistics
Volume 3 (HSK 41.3). Berlin/Boston: De Gruyter Mouton, 1732–1748.
Dini, Pietro U. 2014. Foundations of Baltic Languages. Vilnius: Eugrimas.
Dottin, Georges. 1920. La langue gauloise. Grammaire, textes et glossaire. Paris: Librairie C.
Klincksieck.
Dressler, Wolfgang. 1976. Das Altarmenische und die Phonologietheorie. Handes Amsorya 90.
901–976.
Drinka, Bridget. 1991. Lachmannʼs Law: A Phonological Solution. Indogermanische
Forschungen 91. 52–74.
Džaukjan276
, Gevork B. 1967. Očerki po istorii dopisʹmennogo perioda armjanskogo jazyka.
Erevan: Izdatel´stvo Akademii nauk armjanskoj SSR.
276
Name is also transliterated as: J̌ahukyan, Djahukian. Since the quoted work is in Russian, we use the
transliteration of the Russian variant of transliteration..
273
Ėdeľman, Džoj I. 1973. K tipologii indoevropejskich guttural´nyx. Izvestija Akademii Nauk
SSSR. Serija Literatury i Jazyka 32. 540–546.
Ėdeľman, Džoj I. 1994. K predposylkam v etapam perexoda i.-e. *s v š v jazykax grupy satem.
Baltistica Priedas 4, 56–66.
Eichner, Heiner. 1973. Die Etymologie von heth. mēhur. Münchener Studien zur
Sprachwissenschaft 31, 53–107.
Eichner, Heiner. 1980. Phonetik und Lautgesetze des Hethitischen – ein Weg zur ihrer
Entschlüsselung. In: Mayrhofer, Manfred / Peters, Martin, Pfeiffer, Oskar E. (eds.),
Lautgeschichte und Etymologie. Akten d. VI. Fachtagung d. Indogermanischen
Gesellschaft, Wien, 24.-29. September 1978. Wiesbaden: Reichert. 120-165.
Eichner, Heiner. 1992. Anatolian [numerals]. In: Gvozdanović, Jadranka (ed.). Indo-European
Numerals. Berlin/New York: Mouton de Gruyter, 29–96.
Ejerhed, Eva I. 1981. The Analysis of Aspiration in Sanskrit Phonology. Nordic Journal of
Linguistics 4(2). 139–159.
Elbourne, Paul. 1998. Proto-Indo-European Voiceless Aspirates. HF 111. 1-30.
Elbourne, Paul. 2000. Plain voiceless stop plus laryngeal in Indo-European. HF 113. 2-28.
Emmerick, Ronald E. 1992a. Old Indian [Numerals]. In: Gvozdanović, Jadranka (ed.). IndoEuropean
Numerals. Berlin/New York: Mouton de Gruyter, 163–198.
Emmerick, Ronald E. 1992b. Iranian [Numerals]. In: Gvozdanović, Jadranka (ed.). IndoEuropean
Numerals. Berlin/New York: Mouton de Gruyter, 289–345.
Endzelin(s), J(ānis). 1923. Lettische Grammatik. Heidelberg: Carl Winter’s
Universitätsbuchhandlung.
Endzelin(s), Jānis. 1939. Über den slavisch-baltischen Reflexe vom idg. sḱ. ZfslPh 16, 107–
115.
Endzelin(s), J(ānis). 1944. Altpreussische Grammatik. Rīga: Verlag Latvju Grāmata.
Erhart, Adolf. 1956. Zum ie. Wechsel Media : Media aspirata. Sborník prací Filozofické fakulty
brněnské univerzity. A 5. 5– 18.
Erhart, Adolf. 1980. Struktura indoíránských jazyků [The Structure of the Indo-Iranian
Languages]. Brno: Univerzita J. E. Purkyně.
Erhart, Adolf. 1982. Indoevropské jazyky – srovnávací fonologie a morfologie [Indo-European
Languages – the Comparative Phonology and Morphology]. Praha: Academia.
Erhart, Adolf. 1984: Baltské jazyky [Baltic Languages]. Brno: Univerzita J. E. Purkyně.
ESJS = Erhart, Adolf / Eva Havlová / Ilona Janyšková (eds.), 1989–2018. Etymologický slovník
jazyka staroslověnského, 1–19 [Etymological dictionary of Old Church Slavonic
Language, 1–19]. Praha: Academia/ Brno: Tribun.
Eska, Joseph F. 1998. Tau Gallicum. Studia Celtica 32. 115–127.
Eska, Joseph F. 2010. The Emergence of the Celtic Languages. In: Ball, Martin J. / Müller,
Nicole (eds.). The Celtic Languages. London: Routledge. 22–27.
Evans, D. Ellis. 1967. Gaulish Personal Names. Oxford: Clarendon Press.
EWAi = Mayrhofer, Manfred. 1992. Etymologisches Wörterbuch des Altindoarischen I. +
Mayrhofer, Manfred. 1996. Etymologisches Wörterbuch des Altindoarischen II. +
Mayrhofer, Manfred. 2001. Etymologisches Wörterbuch des Altindoarischen III.
Heidelberg: Carl Winter.
Heidelberg: Carl Winter.
Fahs, Achim. 1989. Grammatik des Pali. Leipzig: VEB Verlag Enzyklopädie.
Fallon, Paul. D. 2002. The synchronic and diachronic Phonology of Ejectives. New York &
London: Routledge.
Fellner, Hannes A. 2006. On the Development of Labiovelars in Tocharian. In: Jones-Bley,
Karlene / Huld, Martin E. / Della Volpe, Angela (eds). Proceedings of the seventeenth
annual UCLA Indo-European conference, Los Angeles, October 27-28, 2005. 51–65.
274
Forssman, Berthold. 2001. Lettische Grammatik. München: J. H. Röll Verlag.
Fortunatov, Filipp. F. 1888. Phonetische Bemerkungen veranlasst durch Miklosich’s
Etymologisches Wörterbuch der slavischen Sprachen. Archiv für Slavische Philologie
11. 561–575.
Fraenkel, Ernst. 1950a. Die baltischen Sprachen. Heidelberg: Carl Winter Universitätsverlag.
Fraenkel, Ernst. 1950b. Slavisch gospodъ, lit. viẽšpats, preuss. Waispattin und Zubehör.
Zeitschrift für slavische Philologie 20. 51–89.
Fraenkel LEW = Fraenkel, Ernst. 1962. Litauisches etymologisches Wörterbuch I.
Heidelberg/Göttingen: Carl Winter Universitätsverlag/Vandenhoeck & Ruprecht +
Fraenkel, Ernst. 1965. Litauisches etymologisches Wörterbuch II. Heidelberg/Göttingen:
Carl Winter Universitätsverlag/Vandenhoeck & Ruprecht.
Friedrich, Johannes. 1974. Hethitisches Elementarbuch I. Heidelberg: Carl Winter
Universitätsverlag.
Friedrich, Johannes. 1990. Kurzgefaßtes Hethitisches Wörterbuch. Heidelberg: Carl Winter
Universitätsverlag.
Frisk, Hjalmar. 1960. Griechisches etymologisches Wörterbuch. Band I, A-Ko. Heidelberg:
Carl Winter Universitätsverlag.
Frisk, Hjalmar. 1970. Griechisches etymologisches Wörterbuch. Band II, Kr-Ō. Heidelberg:
Carl Winter Universitätsverlag.
Frisk, Hjalmar. 1972. Griechisches etymologisches Wörterbuch. Band III, Nachträge,
Wortregister, Corrigenda, Nachwort.. Heidelberg: Carl Winter Universitätsverlag.
Fussman, Gérard. 1972. Atlas linguistique des parlers dardes et kafirs. II. Commentaire. Paris:
École française dʼExtrême-Orient.
Gallée, Johan H. Altsächsische Grammatik. Dritte Auflage mit Berichtigungen und
Literaturnachträgen von Heinrich Tiefenbach. Tübingen: Max Niemeyer.
Gamkrelidze, T. V. / Ivanov Vyacheslav V. 1972. Lingvističeskaja tipologija i rekonstrukcija
sistemy indoevropejskix smyčnyx. In: Bernštein, Samuil B. et al. (eds). Konferencija po
sravnitelno-istoričeskoj grammatike indoevropejskix jazykov. Moskva: Institut slavjanovedenija
i baltistiki AN SSSR. 15–18.
Gerullis, Georg. 1922. Die altpreußischen Ortsnamen: gesammelt und sprachlich behandelt.
Berlin: Vereinigung wissenschaftlicher Verleger Walter de Gruyter & Co.
Gippert, Jost. 2002. Neues zu “Slavisch st aus älterem pt”?. In: Anreiter, Peter / Ernst, Peter/
Hausner, Isolde (eds.): Namen Sprachen und Kulturen. Imena, Jeziki i Kulture. Festschrift
für Heinz Dieter Pohl zum 60. Geburtstag. Wiedn: Edition Praesens, 239–256.
Godel, Robert. 1975. An Introduction to the Study of Classical Armenian. Wiesbaden: Dr.
Ludwig Reichert Verlag.
Görtzen, Jens. 1998. Die Entwicklung der indogermanischen Verbindungen von dentalen
Okklusiven mit besonderer Berücksichtigung des Germanischen. Innsbruck: Institut für
Sprachwissenschaft der Universität Innsbruck.
Götze, Albert. 1928. Madduwattaš. Leipzig: Hinrichs.
GPC = [eds]. 1997–2002. Geriadur Prifysgol Cymru. Aberystwyth: Gwasg Prifysgol Cymru.
Greene, David. 1992. Celtic [numerals]. In: Gvozdanović, Jadranka (ed.). Indo-European
Numerals. Berlin/New York: Mouton de Gruyter, 498–554.
Greenberg, Marc. L. 2017. Slavic. In: Kapović, Mate (ed.). The Indo-European Languages.
New York: Routledge, 519–5151.
von Grienberger, Theodor. 1900. Untersuchungen zur gotischen Wortkunde. Wien: Gerold.
Griffen, Toby D. 1989. Nostratic and Germano-European. General Linguistics 29/3. 139–149.
Gunnarsson, Jón. 1971. On the Indo-European “dental spirants”. Norsk Tidskrift for
Sprogvidenskap 24. 21–82.
275
Hackstein, Olav. 2017. The Phonology of Tocharian. In: Klein, Jared / Joseph, Brian / Fritz /
Matthias (eds). Handbook of Comparative and Historical Indo-European Linguistics
(HSK 41.2). Berlin / Boston: Walter de Gruyter, 1304–1335.
Haider, Hubert 1985. The Fallacy of Typology. Remarks on the PIE Stop-System. Lingua 65.
1–27.
Hammerich, Louis L. 1955. Die germanische und die hochdeutsche Lautverschiebung. Beiträge
zur Geschichte der deutschen Sprache und Literatur (PBB(T)) 77. 1–29, 165–203.
Hammerich, Louis L. 1967. Ketzereien eines alten Indogermanisten. In: To Honor of Roman
Jakobson II. The Hague: Mouton. 839–849.
Hamp, Eric P. 1961a. Albanian be, besë “oath”. Zeitschrift für vergleichende Sprachforschung
auf dem Gebiete der indogermanischen Sprachen 77/3–4. 252–253.
Hamp, Eric P. 1961b. IE *bhend- in Albanian. Zeitschrift für vergleichende Sprachforschung
auf dem Gebiete der indogermanischen Sprachen 77/3–4. 253–254.
Hamp, Eric P. 1961c. Albanien natë “night”. Zeitschrift für vergleichende Sprachforschung auf
dem Gebiete der indogermanischen Sprachen 77/3–4. 254–256.
Hamp, Eric P. 1967. On IE *s after i, u in Baltic. Baltistica 3. 7–11.
Hamp, Eric. 1976. *gweiH- ʽliveʼ. In: Davies, Anna Morpurgo / Meid, Wolfgang (eds.). Studies
in Greek, Italic and Indo-European Linguistics offered to Leonhard R. Palmer.
Innsbruck: University of Innsbruck, 87–91.
Hamp, Eric P. 1992. Albanian [numerals]. In: Gvozdanović, Jadranka (ed.). Indo-European
Numerals. Berlin/New York: Mouton de Gruyter, 835–921.
HED 1–2 = Puhvel, Jaan. 1984. Hittite Etymological Dictionary. Vol. 1, Words Beginning with
A. Hittite Etymological Dictionary. Vol. 2, Words Beginning with E and I. Berlin / New
York / Amsterdam: Mouton.
HED 3 = Puhvel, Jaan. 1991. Hittite Etymological Dictionary. Vol. 3, Words Beginning with H.
Berlin / New York: Mouton.
HED 4 = Puhvel, Jaan. 1997. Hittite Etymological Dictionary. Vol. 4, Words Beginning with K.
Berlin / New York: Mouton.
HED 5 = Puhvel, Jaan. 2001. Hittite Etymological Dictionary. Vol. 5, Words Beginning with L.
Indices to volumes 1–5. Berlin / New York: Mouton.
HED 6 = Puhvel, Jaan. 2004. Hittite Etymological Dictionary. Vol. 6, Words Beginning with
M. Berlin / New York: Mouton.
HED 7 = Puhvel, Jaan. 2007. Hittite Etymological Dictionary. Vol. 7, Words Beginning with N.
Berlin / New York: Mouton.
HED 8 = Puhvel, Jaan. 2011. Hittite Etymological Dictionary. Vol. 8, Words Beginning with
PE, PI, PU. Berlin / New York: Mouton.
HED 9 = Puhvel, Jaan. 2013. Hittite Etymological Dictionary. Vol. 9, Words Beginning with
PA. Berlin / New York: Mouton.
HED 10 = Puhvel, Jaan. 2017. Hittite Etymological Dictionary. Vol. 10, Words Beginning with
SA. Berlin / New York: Mouton.
Hegedüs, Irén. 2012. The RUKI-rule in Nuristani. In: Nielsen Whitehead, Benedicte et al. (eds).
The Sound of Indo-European. Phonetics, Phonemics, and Morphophonemics. 145–167.
Copenhagen: Museum Tusculanum Press.
Hiersche, Rolf. 1964. Untersuchungen zur Frage der Tenues Aspiratae im Indogermanischen.
Wiesbaden: Otto Harrassowitz.
Hill, Eugen. 2003. Untersuchungen zum inneren Sandhi des Indogermanischen. Der Zusammenstoß
von Dentalplosiven im Indoiranischen, Germanischen, Italischen und
Keltischen. Bremen: Hempen Verlag.
Hillmarsson, Jörundur. 1993a. Development of labiovelars (and tectals plus u̯) in initial position
in Tocharian (an overview). Die Sprache 35/2. 176–186.
276
Hillmarsson, Jörundur. 1993b. The fate of TTR/RTT in Indo-European. In: Meiser, Gerhard
(ed.) Indogermanica et italic. Festschrift für Helmut Rix zum 65. Geburtstag. Innsbruck:
Innsbrucker Beiträge zur Sprachwissenschaft 72. 209–222.
Hirt, Hermann. 1899. Zur Lösung der Gutturalfrage im Indogermanischen. BB 24. 218–291.
Hirt, Hermann. 1912. Handbuch der griechischen Laut- und Formenlehre. Heidelberg: Carl
Winter´s Universitätsbuchhandlung.
Hirt, Hermann. 1927. Indogermanische Grammatik. Teil 1, Einleitung. 1. Etymologie. 2.
Konsonantismus. Heidelberg: Carl Winter´s Universitätsbuchhandlung.
Hirt, Hermann. 1939. Die Hauptprobleme der indogermanischen Sprachwissenschaft. Halle:
Max Niemeyer Verlag.
Hock, Wolfgang. 2004. Forschungsbericht: Baltoslavisch. I. Teil: Phonologie. Kratylos 49. 1-
32.
Hock, Wolfgang. 2005. Zur Vorgeschichte des albanischen Lautsystems. In: Meiser,
Gerhard/Hackstein, Olav (eds.). Sprachkontakt und Sprachwandel. Akten der XI.
Fachtagung der Indogermanischen Gesellschaft, 17.—23. September 2000, Halle an der
Saale. Wiesbaden: Dr. Ludwig Reichert Verlag, 261–274.
Hoffmann, Karl. 1956. Notizen zu Wackernagel–Delbrunner, Altindische Grammatik II.2. MSS
8. 5–24.
Hoffmann, Karl / Forssman, Bernard. 1996. Avestische Laut- und Flexionslehre. Innsbruck:
Institut für Sprachwissenschaft der Universität Innsbruck.
Hoffner, Harry A. / Melchert, H. Craig. 2008. A Grammar of the Hittite Language. Part 1.
Reference Grammar. Winona Lake: Eisenbrauns.
Holst, Jan Henrik. 2001. Lettische Grammatik. Hamburg: Helmut Buske Verlag.
Hopper, Paul. 1973. Glottalized and murmured Occlusives in Indo-European. Glossa 7. 141–
166.
Horrock, Geoffrey. 2017. The Evolution of Greek. In: Klein, Jared / Joseph, Brian / Fritz /
Matthias (eds). Handbook of Comparative and Historical Indo-European Linguistics
(HSK 41.1). Berlin / Boston: Walter de Gruyter, 717–745.
Hrozný, Bedřich. 1917. Die Sprache der Hethiter. Leipzig: Hinrichs.
Hübschmann, Heinrich. 1883. Armenische Studien I. Grundzüge der armenischen Etymologie.
Leipzig: Breitkopf und Härtel.
Hübschmann, Heinrich. 1897. Armenische Grammatik I. Armenische Etymologie. Leipzig:
Breitkopf und Härtel.
Hujer, Oldřich. 1913. Skupina kt ve slovanštině [The Cluster kt in Slavic]. Časopis moderních
filologů 3. 356–360.
Huld, Martin E. 1984. Basic Albanian Etymologies. Columbus, Ohio: Slavica Publishers.
Huld, Martin E. 1986. On the unacceptability of the Indo-European Voiced stops as Ejectives.
IF 91, 67–78.
Huld, Martin E. 1997. Satəm, centum and Hokum. In: Adams, Douglas Q. (ed.), Festschrift for
Eric P. Hamp I. Washington: Institute for the study of Man. 115–138.
Insler, Stanley. 1972. Some irregular Vedic imperatives. Language 48/3. 551–565.
Insler, Stanley. 1975. The Vedic type dheyām. Sprache 21/1. 1–22.
Jackson, Kenneth. 1953. Language and History in Early Britain. Edinburgh: Edinburgh
University Press.
Jakobson, Roman / Morris Halle. 1956. Fundamentals of Language. ‘S-Gravenhage: Mouton
& Co.
Jakobson, Roman. 1958. Typological Studies and their Contribution to Historical comparative
linguistics. In: Sievertsen (ed.): Proceedings of the Eighth International Congress of
Linguists. Oslo: Oslo University Press. 17–35.
277
Jamison, Stephanie W. 1997. Syntactic Constraints on Morphological Change: the Vedic
Imperatives bodhí, dehí and dhehí. In: Pirart, Eric (ed.). Syntaxe des langues indoiraniennes
anciennes: colloque inernational. 63–80. Barcelona: Universidad de
Barcelona.
Jannaris, Antonius N. 1897. An Historical Greek Grammar, Chiefly of the Attic Dialect as
Written and Spoken from Classical Antiquity Down to the Present Time: Founded upon
the Ancient Texts, Inscriptions, Papyri and Present Popular Greek. London: Macmillan.
Jasanoff, Jay. 2002. The Vedic Imperatives yódhi “fight” and bódhi “heed”. Journal of the
American Oriental Society 122/2. 290–295.
Jasanoff, Jay. 2004. Plus ça change... Lachmann's Law in Latin. In: Penney, J. H. W. (ed.).
Indo-European Perspectives: Studies in Honour of Anna Morpurgo Davies. Oxford:
Oxford University Press. 405–416.
Job, Michael. 1995. Zum Lautwandel im Armenischen: Probleme einer relativen Chronologie.
In: Smoczyński, Wojciech (ed.). Kuryłowicz Memorial Volume. Part One. Cracow:
Universitas, 291–311.
Johansson, Karl F. 1903. Arische Beiträge. Indogermanische Forschungen 14: 265–339.
Johansson, Karl F. 1906. Arische Beiträge. Indogermanische Forschungen 19: 112–139.
Jokl, Norbert. 1923. Linguistisch-kulturhistorische Untersuchungen aus dem Bereiche des
Albanischen. Berlin/Leipzig: Walter de Gruyter & Co.
Jokl, Norbert. 1937. Ein Beitrag zur Lehre von der alb. Vertretung der idg. Labiovelare. In:
Hjelmslev, Louis (ed.). Mélanges linguistiques offerts à M. Holger Pedersen à lʼoccasion
de son soixante-dixième anniversaire. København : Universitetsforlaget i Aarhus Levin
& Munksgaard, 127–161.
Jokl, Norbert. 1963. Die Verwandtschaftsverhältnisse des Albanischen zu den übrigen
indogermanischen Sprachen. Die Sprache 9. 113–156.
Kapović, Mate. 2017. Proto-Indo-European Phonology. In: Kapović, Maté. (ed.). 2017: The
Indo-European Languages. London – New York: Routledge. 13–60.
Karaliūnas, Simas. 1966. K voprosu ob i.-e. *s posle i, u v litovskom jazyke. Baltistika 1, 113–
126.
Karulis, Konstantīns. 1992a. Latviešu etimoloğijas vārdnīca I. A – O. Rīga: Avots.
Karulis, Konstantīns. 1992b. Latviešu etimoloğijas vārdnīca II. P – Ž. Rīga: Avots.
Kausen, Ernst. 2012. Die indogermanischen Sprachen. Hamburg: Helmut Buske Verlag.
Kellens, Jean. 1976. Un prétendu present radical. Münchener Studien zur Sprachwissenschaft
34. 59–71.
Kellens, Jean. 1989. Avestique. In: Schmitt, Rüdiger (ed.). Compendium linguarum
Iranicarum. 32–55.Wiesbaden: Dr. Ludwig Reichert Verlag.
Kellens, Jean. 1984. Le verbe avestique. Wiesbaden: Dr. Ludwig Reichert Verlag.
Kellens, Jean. 1995. Liste du verbe avestique. Wiesbaden: Dr. Ludwig Reichert Verlag.
Kent, Roland G. 1928. Lachmannʼs Law of Vowel Lengthening. Language 4. 181–190.
Kent, Roland G. 1932a. The Development of the Indo-European Dental Groups. Language 8/1.
18–26.
Kent, Roland G. 1932b. The Sounds of Latin. A Descriptive and Historical Phonology.
Language 8/3. (Language Monograph 12). 11–13, 15-216.
Kent, Roland G. 1936. Assimilation and dissimilation. Language 12/4: 245–258.
Kent, Roland G. 1950. Old Persian. Grammar. Texts. Lexicon. New Haven: American Oriental
Society.
Kim, Ronald. 1999. The development of labiovelars in Tocharian: a closer look. Tocharian and
Indo-European Studies 8. 139–187.
Kim, Ronald I. 2016. Studies in Armenian Historical Phonology I: Aspiration and
Spirantization of PIE Voiceless Stops. In: Byrd, Andrew M. / DeLiss, Jessica / Wenthe,
278
Mark (eds.). Tavet Tat Satyam. Studies in Honor of Jared S. Klein on the Occasion of his
Seventieth Birthday. Ann Arbor/New York: Beech Stave Press, 151–167.
Kim, Ronald. 2018. Greco-Armenian: The persistence of a myth. Indogermanische
Forschungen 123/1, 247–271.
Kimball, Sara. 1999. Hittite Historical Phonology. Innsbruck: Innsbrucker Beiträge zur
Sprachwissenschaft.
Kimball, Sara. 2017. The Phonology of Anatolian. In: Klein, Jared / Joseph, Brian / Fritz /
Matthias (eds). Handbook of Comparative and Historical Indo-European Linguistics
(HSK 41.1). Berlin / Boston: Walter de Gruyter, 249–256.
Kirchhoff, Adolf. 1877. Studien zur Geschichte des griechischen Alphabets. Berlin: Ferd.
Dümmler´s Verlags-Buchshandlung.
Klingenschmitt, Gert. 1982. Das altarmenische Verbum. Wiesbaden: Dr. Ludwig Reichert
Verlag.
Kloekhorst, Alwin. 2008. Etymological Dictionary of the Hittite Inherited Lexicon. Leiden /
Boston: Brill.
Kloekhorst, Alwin. 2019. The spelling of clusters of dental stop + sibilant in Hittite. Münchener
Studien zur Sprachwissenschaft. 73/1. 55–72.
Kobayashi. Masato. 2004. Historical Phonology of Old Indo-Aryan Consonants. Tokyo:
Research Institute for Languages and Cultures of Asia and Africa. Tokyo University of
Foreign Studies.
Kobayashi. Masato. 2017. The phonology of Indic (Old Indo-Aryan). In: Klein, Jared / Joseph,
Brian / Fritz Matthias (eds). Handbook of Comparative and Historical Indo-European
Linguistics 4.1. Berlin/Boston: De Gruyter Mouton. 325–344.
Ködderitzsch, Rolf. 1991. Historische Phonologie des Albanischen: Probleme und Aufgaben.
In: Breu, Walter / Ködderitzsch, Rolf / Sasse, Hans-Jürgen (eds). Aspekte der
Albanologie: Akten des Kongresses „Stand und Aufgaben der Albanologie heute“. 3.–5.
Oktober 1988, Uniersität Zu Köln. Wiesbaden: Otto Harrassowitz. 121–130.
Kögel, Rudolf. 1879. Ueber einige germanische dentalverbindungen. Beiträge zur Geschichte
der deutschen Sprache und Literatur 7. 171–201.
Kortlandt, Frederik. 1975. A Note on the Armenian Palatalization. Zeitschrift für vergleichende
Sprachforschung 89. 43–45.
Kortlandt, Frederik. 1976. Notes on Armenian Historical phonology I. Studia Caucasica 3, 91–
100 (also in: Kortlandt/Beekes 2003: 1–9).
Kortlandt, Frederik. 1978a. Notes on Armenian Historical phonology II. Studia Caucasica 4,
9–16 (also in: Kortlandt/Beekes 2003: 20–25).
Kortlandt, Frederik. 1978b. I.-E. palatovelars before Resonants in Balto-Slavic. In: Fisiak, Jack
(ed): Recent developments in historical phonology. The Hague: Mouton. 237–243.
Kortlandt, Frederik. 1978c. PIE obstruents. Indogermanische Forschungen 78. 107–120.
Kortlandt, Frederik. 1978d. Comment on W. Winter’s paper. In: Fisiak, Jack (ed.). Recent
developments in historical phonology. The Hague: Mouton., 447.
Kortlandt, Frederik. 1980a. On the Relative Chronology of Armenian Sound Changes. Studia
Caucasica 4, 9–16 (also in: Kortlandt/Beekes 2003: 26–33).
Kortlandt, Frederik. 1980b. Albanian and Armenian. Zeitschrift für vergleichende
Sprachforschung 94/1–2, 243–251 (also in: Kortlandt/Beekes 2003: 13–19).
Kortlandt, Frederik. 1981. More Evidence for Italo-Celtic. Ériu 32.1–22.
Kortlandt, Frederik 1982. IE. *pt in Slavic. Folia Linguistica Historica 3 (1). 25–28.
Kortlandt, Frederik. 1985. Proto-Indo-European glottalic stops: the comparative evidence.
Folia Linguistica Historica VI/2. 183–201.
279
Kortlandt, Frederik. 1986. Armenian and Albanian. In: Leroy, Maurice / Mawet, Francine
(eds.). La place de l'arménien dans les langues indo-européennes. Leuven: Peeters, 38–
47 (also in: Kortlandt/Beekes 2003: 68–74).
Kortlandt, Frederik. 1987. PIE *s in Albanian. Studies in Slavic and General Linguistics 10.
219–226.
Kortlandt, Frederik. 1988. Reflexes of Indo-European Consonants in Albanian. Orpheus.
Journal of Indo-European and Thracian Studies 8 (Georgiev Memorial Volume). 35–37.
Kortlandt, Frederik. 1989. Lachmannʼs Law. In: Vennemann, Theo (ed.). The New Sound of
Indo-European (Essays in Phonological Reconstruction). Berlin: Mounton de Gruyter.
103–105.
Kortlandt, Frederik H. H. 1994a. From Proto-Indo-European to Slavic. The Journal of IndoEuropean
Studies 22. 91–112.
Kortlandt, Frederik. 1994b. Proto-Armenian Numerals. In: Rasmussen, Jens E. (ed.) In
honorem H. Pedersen. Kolloquium der Indogermanischen Gesellschaft vom 25. bis 28.
März 1993 in Kopenhagen. Wiesbaden: Reichert, 253–257. (also in: Kortlandt/Beekes
2003: 98–101).
Kortlandt, Frederik. 1995. The Sigmatic Forms of the Armenian Verb. Annual of Armenian
LInguistics 16, 13–17. (also in: Kortlandt/Beekes 2003: 107–109).
Kortlandt, Frederik. 1996. Proto-Armenian Verbal System. In: Sakayan, Dora (ed.) Proceeding
of the 5th International Conference on Armenian Linguistics. Delmar, N.Y.: Caravan
Books, 35–43. (also in: Kortlandt/Beekes 2003: 110–116).
Kortlandt, Frederik. 1999. Lachmannʼs Law Again. In: Polomé, Edgar C.; Justus, Carol F.
(eds.). Language Change and Typological Variation: In Honor of Winfred P. Lehmann
on the Occasion of his 83rd Birthday. Vol. 1 : Language change and phonology.
Washington, D.C.: Institute for the Study of Man. 246–248.
Kortlandt, Frederik / Beekes, Robert S. P. 2003. Armeniaca. Comparative Notes. Ann Arbor:
Caravan Books
Krahe, Hans / Meid, Wolfgang. 1969. Germanische Sprachwissenschaft. I. Eingleitung und
Lautlehre. (Sammlung Göschen Band 238). Berlin: Walter de Gruyter & Co.
Kräuter, Johann F. 1877. Zur Lautverschiebung. Strassburg: Karl J. Trübner.
Krause, Wofgang 1952. Westtocharische Grammatik. Band I. Das Verbum. Heidelberg: Carl
Winter Universitätsverlag.
Krause, Wofgang / Thomas, Werner. 1960. Tocharisches Elementarbuch. Band I. Grammatik.
Heidelberg: Carl Winter Universitätsverlag.
Kronasser, Heinz. 1956. Vergleichende Laut- und Formenlehre des Hethitischen. Heidelberg:
Carl Winter Universitätsverlag.
Kroonen, Gus. 2013. Etymological Dictionary of Proto-Germanic. Leiden/Boston: Brill.
Kümmel, Martin J. 2007. Konsonantenwandel: Bausteine zu einer Typologie des Lautwandels
und ihre Konsequenzen für die vergleichende Rekonstruktion. Wiesbaden: Reichert.
Kümmel, Martin J. 2012. Typology and reconstruction. In: Nielsen Whitehead, Benedicte /
Olander, Thomas / Olsen, Birgit A. / Rasmussen, Jens E. (eds.): The Sound of IndoEuropean.
Copenhagen: Museum Tusculanum Press. 291–329.
Kurschat LDW = Kurschat, Alexander et al. 1968. Litauisch-deutsches Wörterbuch. Thesaurus
Linguae Lituanicae I. Göttingen: Vandenhoeck & Ruprecht. + Kurschat, Alexander et al.
1970. Litauisch-deutsches Wörterbuch. Thesaurus Linguae Lituanicae II. Göttingen:
Vandenhoeck & Ruprecht.
Kuryłowicz, Jerzy. 1927. ə indo-européen et ¯ hittite. In: Symbolae grammaticae in honorem
Ioannis Rozwadowski I. Kraków. 95–104.
Kuryłowicz, Jerzy. 1935. Études indoeuropéennes I. Kraków: Skład główny w ksiȩgarni
Gebethnera i Wolffa.
280
Kuryłowicz, Jerzy. 1951. Le VIIe
aoriste indien. In: Kuryłowicz, Jerzy 1973. 126–130.
Kuryłowicz, Jerzy. 1956. L´apophonie en indo-européen. Prace jȩzykoznawcze 9. Wrocław:
Polska Akademia Nauk.
Kuryłowicz, Jerzy. 1964. The inflectional Categories of Indo-European. Heidelberg: Carl
Winter Universitätsverlag.
Kuryłowicz, Jerzy. 1968. A Remark on Lachmann's Law. Harvard Studies in Classical
Philology 72. 295–299.
Kuryłowicz, Jerzy. 1973. Esquisses linguistiques I. München: Fink.
Kuryłowicz, Jerzy. 1977. Problèmes de linguistique indo-européenne. Prace jȩzykoznawcze 90.
Wrocław/Warszawa/Kraków/Gdańsk: Polska Akademia Nauk.
Kuiper, Franciscus B. J. 1967. The Sanskrit Nom. Sg. víṭ. Indo-Iranian Journal 10. 103–125.
Lachmann, Karl. 1850. Lucretius: De rerum natura (reprinted 1979. New York/ London:
Garland Publ.).
Lambert, Pierre-Yves. 1994. La langue gauloise. Paris: Editions Errance.
Lambert, Pierre-Yves / Pinault, Georges-Jean (eds). 2007. Gaulois et Celtique continental.
Paris: Droz.
Lamprecht, Arnošt. 1987. Praslovanština [The Pre-Slavic language]. Brno: Univerzita J. E.
Purkyně.
Langston, Keith. 2017. The Documentation of Slavic. In: Klein, Jared / Joseph, Brian / Fritz /
Matthias (eds). Handbook of Comparative and Historical Indo-European Linguistics
(HSK 41.3). Berlin / Boston: Walter de Gruyter, 1397–1413.
Larson, Jenny H. / Bukelskytė-Čepelė, Kristina. 2018. The documentation of Baltic. In: Klein,
Jared / Joseph, Brian / Fritz Matthias (eds). Handbook of Comparative and Historical
Indo-European Linguistics 4.3. Berlin/Boston: De Gruyter Mouton. 1621–1639.
Lehmann, Winfred P. 1952. Proto-Indo-European Phonology. Austin: The University of Texas
Press/Linguistic Society of America.
Lehmann, Winfred P. 1986. A Gothic Etymological Dictionary, based on the third edition of
Vergleichendes Wörterbuch der Gotischen Sprache by Sigismund Feist. Leiden: E. J.
Brill.
LEIA = Vendryès, Joseph. 1959–. Lexique étymologique de l'irlandais ancien. Paris: Centre
national de la recherche scientifique.
Lejeune, Michel. 1947. Traité de phonétique grecque. Paris: Librairie C. Klincksieck.
Lejeune, Michel. 1972. Phonétique historique du mycénien et du grec ancien. Paris: Éditions
Klincksieck.
Leumann, Manu. 1942. Idg. sḱ im Altindischen und Litauischen. Indogermanische
Forschungen 58. 1–26, 113-130.
Leumann, Manu. 1977. Lateinische Laut- und Formenlehre. München: C. H. Beck´sche
Verlagsbuchandlung.
Lewis, Charlton Th. / Short, Charles / Freund, Wilhelm / Andrews, Ethan Allen. 1879. A Latin
dictionary : founded on Andrews' edition of Freund's Latin Dictionary, rev., enl. and in
great part rewritten by Charlton T. Lewis and Charles Short. Oxford: Clarendon, 1879.
Lewis, Henry / Pedersen, Holger. 1937. A Concise Comparative Celtic Grammar. Göttingen:
Vandenhoeck & Ruprecht.
Liddell, Henry G. / Scott, R. (/ Jones, Henry S.), 1996. A Greek-English Lexicon. A new edition.
Oxford: Clarendon Press.
Lindemann, Fredrik O. 1965. Note phonologique sur hittite eku- 'boire'. Revue hittite et
asianique 23. 29–32.
Lipp, Reiner. 2009a. Die indogermanischen und einzelsprachlichen Palatale im Indoiranischen.
Band I: Neurekonstruktion, Nuristan-Sprachen, Genese der indoarischen
Retroflexe, Indoarisch von Mitanni. Heidelberg: Universitätsverlag Winter.
281
Lipp, Reiner. 2009b. Die indogermanischen und einzelsprachlichen Palatale im Indoiranischen.
Band II: Thorn-Problem, indoiranische Laryngalvokalisation. Heidelberg:
Universitätsverlag Winter.
LIV2
= Rix, Helmut / Kümmel, Martin J. / Zehnder, Thomas / Lipp, Reiner / Schirmer, Brigitte.
2001. Lexikon der indogermanischen Verben. Wiesbaden: Dr. Ludwig Reichert Verlag.
Lombardi, Linda. 1991. Laryngeal Features and Laryngeal Neutralization. Dissertation,
University of Massachusetts.
Lubotsky, Alexander. 1995. Sanskrit h < *dh, bh. In: Gurov, Nikita V. / Vasiľkov, Yaroslav V.
(eds). Sthāpakašrāddham. Professor G. A. Zograf Commemorative Volume. 124–144. St.
Petersburg: Peterburgskoe Vostokovedenie.
Lubotsky, Alexander. 2018. The phonology of Proto-Indo-Iranian. In: Klein, Jared / Joseph,
Brian / Fritz Matthias (eds). Handbook of Comparative and Historical Indo-European
Linguistics. Berlin/Boston: De Gruyter Mouton. 1876–1888.
Macak, Martin. 2018. The Phonology of Classical Armenian. In: Klein, Jared / Joseph, Brian /
Fritz, Matthias (eds.) Handbook of Comparative and Historical Indo-European
Linguistics Volume 2 (HSK 41.2). Berlin/Boston: De Gruyter Mouton, 1037–1079.
MacAulay, Donald (ed). 1992. The Celtic Languages: an Overview. In: MacAulay, Donald
(ed). The Celtic Languages. Cambridge: Cambridge University Press. 1–8.
MacDonell, Arthur A. 1910: Vedic Grammar. Strassburg: Karl J. Trübner.
MacDonell, Arthur A. 1916: A Vedic Grammar for Students. Oxford: Clarendon Press.
Malzahn, Melanie. 2007a. Tocharian Texts and Where to Find Them. In: Malzahn, Melanie
(ed.). Instrumenta Tocharica. Heidelberg: Universitätsverlag Winter. 79–112.
Malzahn, Melanie. 2007b. A Tocharian Brahmi Chart. In: Malzahn, Melanie (ed.). Instrumenta
Tocharica. Heidelberg: Universitätsverlag Winter. 223–354.
Malzahn, Melanie. 2010. The Tocharian Verbal System. Leiden / Boston: Brill.
Maniet, Albert. 1956. La “loi de Lachmann” et les antinomies de l´allongement compensatoire.
In : Hommages à Max Niedermann. Bruxelles: Latomus. 230–237.
Mann, Stuart E. 1952. The Indo-European Consonants in Albanian. Language 28/1. 31–40.
Mann, Stuart E. 1963. Armenian and Indo-European (Historical Phonology). London: Lusac
& Co. Ltd.
Mareš, František V. 19691
. Diachronische Phonologie des Ur- und Frühslavischen. München:
Verlag Otto Sagner.
Mareš, František V. 19992
. Diachronische Phonologie des Ur- und Frühslavischen. Frankfurt
am Main: Peter Lang.
Markey, Thomas L. 1980. Delabialization in Germanic. Folia Linguistica Historica 1/2. 235–
293.
Marsh, Gordon H. 1941. The Voiced Sibilants in Sanskrit. JAOS 61. 45–50.
Martinet, André. 1951. Concerning some Slavic and Aryan reflexes of IE *s. Word 7. 91–95.
Martinet, André. 1952. Langues à syllabes ouvertes: le cas du slave commun. Zeitschrift für
Phonetik 6. 145–163.
Martinet, André. 1953. Remarques sur le consonantisme sémitique. Bulletin de la Société de
Linguistique de Paris 49, 67–78.
Martinet, André. 1955. Économie des changements phonètiques. Traité de phonologie
diachronique. Berne: A. Francke.
Martirosyan, Hrach, K. 2010. Etymological Dictionary of the Armenian Inherited Lexicon.
Leiden: Brill.
Martirosyan, Hrach, K. 2013. The Place of Armenian in the Indo-European language family:
the relationship with Greek and Indo-Iranian. Journal of Language Relationship 10, 85–
137.
Masica, Colin P. 1991. The Indo-Aryan Languages. Cambridge: Cambridge University Press.
282
Matasović, Ranko. 2009. Etymological Dictionary of Proto-Celtic. Leiden/Boston: Brill.
Matzinger, Joachim. 2006. Der altalbanische Text Mbsuamae e krështerë (dottrina cristiana)
des Lekë Matrënga von 1572: eine Einführung in die albanische Sprachwissenschaft.
Dettelbach: Röll.
Matzinger, Joachim. 2009. Die Albaner als Nachkommen der Illyrier aus der Sicht der
historischen Sprachwissenschaft. In: Schmitt, Oliver J. / Frantz, Eva (eds.). Albanische
Geschichte. Stand und Perspektiven der Forschung. München: R. Oldenburg Verlag, 13–
35.
Matzinger, Joachim. 2012. Zwischensprachen – areallinguistische Bemerkungen ausdem
Bereich des Balkanindogermanischen. In: Sadovski, Velizar / Stiffler, David (eds.).
Iranistische und indogermanistische Beiträge in memoriam Jochem Schindler (1944–
1994). Wien: Verlag der Österreichischen Akademie der Wissenschaften, 137–159.
Mayrhofer, Manfred 1951. Handbuch des Pali. Heidelberg: Winter.
Mayrhofer, Manfred. 1984. Lassen sich Vorstufen des Uriranischen nachweisen? Anzeiger der
Österreichischen Akademie der Wissensschaften, Phil.-hist. Klasse, 120: 249–255.
Mayrhofer, Manfred. 1986. Indogermanische Grammatik I.2. Lautlehre. Heidelberg: Winter.
Mayrhofer, Manfred. 1989. Vorgeschichte der iranischen Sprachen; Uriranisch. In: Schmitt,
Rüdiger (ed.). Compendium linguarum Iranicarum. 4–24,Wiesbaden: Dr. Ludwig
Reichert Verlag.
Mayrhofer, Manfred. 1997. L’Indo-iranien. In: Bader, Françoise (ed.). Langues indoeuropéennes.
Paris: CNRS Edition. 101–122.
Mayrhofer, Manfred. 2004. Die Hauptprobleme der indogermanischen Lautlehre seit Bechtel.
Wien: Verlag der österreichischen Akademie dr Wissenschaften.
Mažiulis, Vytautas. 2013: Prūsų kalbos etimologijos žodynas. Vilnius: Mokslo ir enciklopeidjų
ledidybos centras.
McCone, Kim. Towards a relative chronology of ancient and medieval Celtic sound change.
Maynooth: Department of Old Irish, St. Patrick’s College.
Meillet, Antoine. 1893. De quelques difficultés de la théorie des gutturales indo-européeennes.
Memoires de la Société de Linguistique de Paris 8. 8. 277–304.
Meillet, Antoine. 1902. Études sur l’etymologie et le vocabulaire du vieux slave I. Paris:
Bouillon.
Meillet, Antoine. 1903. Esquisse d´une grammarie comparée de l´Arménien classique. Vienne:
Imprimerie des PP. Mékhitharistes.
Meillet, Antoine. 1913. Altarmenisches Elementarbuch. Heidelberg: Carl Winters Universitäts-
Buchhandlung.
Meillet, Antoine. 1922. Les dialectes indo-européens. Paris: Librairie ancienne Édouard
Champion.
Meillet, Antoine. 1924. Le slave commun. Paris: Librairie ancienne Honoré Champion.
Meillet, Antoine. 1934. Introduction a l’étude comparative des langues indo-européennes.
Paris: Librairie Hachette
Meillet, Antoine. 1936. Esquisse d´une grammarie comparée de l´Arménien classique.2
Vienne: Imprimerie des PP. Mékhitharistes.
Meillet, Antoine / Vendryes, Joseph. 1924. Traité de grammaire comparée des langues
classiques. Paris: Librairie Ancienne Édouard Champion.
Meiser, Gerhard. 1986. Lautgeschichte der Umbrischen Sprache. Innsbruck: Institut für
Sprachwissenscahft der Universität Innsbruck.
Meiser, Gerhard. 1998. Historische Laut- und Formenlehre der lateinischen Sprache.
Darmstadt: Wissenschaftliche Buchgesellschaft.
283
Meiser, Gerhard. 2017. The Phonology of Italic. In: Klein, Jared / Joseph, Brian / Fritz /
Matthias (eds). Handbook of Comparative and Historical Indo-European Linguistics
(HSK 41.2). Berlin / Boston: Walter de Gruyter, 743–751.
Melchert, H. Craig. 1987. PIE Velars in Luvian. In: Watkins, Calvert (ed.) Studies in Memory
of Warren Cowgill, 182-204. Berlin  New York: de Gruyter.
Melchert, H. Craig. 1994. Anatolian Historical Phonology. Amsterdam / Atlanta: Rodopi.
Melchert, H. Craig. 2012. Luvo-Lycian Dorsal Stops Revisited. In: Sukač, Roman / Šefčík,
Ondřej (eds). The Sound of Indo-European 2. Papers on Indo-European Phonetics,
Phonemics and Morphophonemics. Munich: Lincom Europa, 206–218.
Melchert, H. Craig. 2017. Anatolian. In: Kapović, Mate (ed.). The Indo-European Languages.
New York: Routledge, 171–201.
Merlingen, Weriand. 1958. Idg. x. Die Sprache 4. 39–73.
Mey, Jacob L. 1972. Was Bartholomae really a Grassmann? Norwegian journal of linguistics
1. 81-89
Meyer, Gustav. 1891. Etymologisches Wörterbuch der albanesischen Sprache. Strassburg:
Verlag von Karl. J. Trübner.
Michelson, Truman. 1911/1912. The alleged change of Indo-European tst(h) to st(h).
Indogermanische Forschungen 29: 221–226.
Mikkola, Jooseppi J. 1942. Urslavische Grammatik. II. Teil. Konsonantismus. Heidelberg: Carl
Winter-Universitätsverlag.
Miller, Brett. 2012. Sonority and the Larynx. In: Parker, Steve (ed.). The Sonority Controversy.
Berlin: De Gruyter. 257–288.
Miller, D. Garry. 1977. Bartholomaeʼs Law and an IE Root Structure Constraint. In: Hopper,
Paul J. (ed.): Studies in Descriptive and Historical Linguistics. Festschrift for W. P.
Lehmann. Amsterdam: John Benjamins. 365–392.
Miller, D. Garry. 2019. The Oxford Gothic Grammar. Oxford: Oxford University Press.
Morgenstierne, Georg. 1926. Report on a Linguistic Mission to Afghanistan. Oslo: H. Aschehou
& Co.
Morgenstierne, Georg. 1942. Orthography and sound-system of the Avesta. Norsk Tidsskrift for
Sprogvidenskap 12: 30–82.
Morgenstierne, Georg. 1945. Indo-European ḱ in Kafiri. Norsk Tidsskrift for Sprogvidenskap
13: 225–238.
Morgenstierne, Georg. 1965. Dardic and Kāfir Languages. In: Lewis, Bernard/ Pellat,
Charles/Schacht Joseph (eds.) Encyclopaedia of Islam II. 138–139. Leiden: Brill.
Morpurgo Davies, Anna / Hawkins, J. David. 1988. A Luwian Heart. In: Imparati, Fiorella
(ed.). Studi di storia e di filologia anatolica dedicate a Giovanni Publiese Carratelli.
Firenze: Elite. 169–192.
Nedoma, Robert. 2017. The Documentation of Germanic. In: Klein, Jared / Joseph, Brian / Fritz
/ Matthias (eds). Handbook of Comparative and Historical Indo-European Linguistics
(HSK 41.2). Berlin / Boston: Walter de Gruyter, 875–888.
Nelson, David N. 1986. The Historical Development of the Nuristani Languagues. (Ph.D.
thesis, University of Minnesota).
NIL = Wodtko, Dagmar S. / Irslinger, Britta / Schneider, Carolin. 2008. Nomina im
indogermanischen Lexikon. Heidelberg: Winter Universitätsverlag.
Oettinger, Norbert. 1979. Die Stammbildung des hethitischen Verbums. Nürnberg: Hans Carl
Verlag.
Ölberg, Hermann M. 1976. Zwei oder drei Gutturalreihen? Vom Albanischen aus gesehen. In:
Scritti in onore di Giuliano Bonfante II. Brescia: Paideia Editrice. 561–570.
Olsen, Birgit A. 1999. The Noun in Biblical Armenian: Origin and Word-formation: with
special emphasis on the Indo-European heritage. Berlin / New York: Mouton de Gruyter.
284
Olsen, Birgit A. 2018. The Morphology of Armenian. In: Klein, Jared / Joseph, Brian / Fritz,
Matthias (eds.) Handbook of Comparative and Historical Indo-European Linguistics
Volume 2 (HSK 41.2). Berlin/Boston: De Gruyter Mouton, 1080–1099.
Olsen, Birgit A. 2017: Armenian. In: Kapović, Mate (ed.). The Indo-European Languages. New
York: Routledge.
Orël, Vladimir. 1998. Albanian Etymological Dictionary. Leiden: Brill.
Orël, Vladimir. 2000. A Concise Historical Grammar of the Albanian Language. Leiden: Brill.
Osthoff, Hermann. 1884. Zur Geschichte des Perfects im Indogermanischen: Mit besonderer
Rücksicht auf Griechisch und Lateinisch. Berlin: Karl I. Trübner.
Otkupščikov, Jurij V. 1984. Zakon Laxmana v svete indoevropeijskix dannyx. Gipotezy i fakty.
Voprosy Jazykoznanija 2. 83–90.
Otrȩbski, Jan. 1954. Slavjano-baltiskoe jazykovoe jedinstvo. Voprosy jayzkoznanija 5, 27–42,
6, 28–46.
Otrȩbski, Jan. 1958. Gramatyka jȩzyka litewskiego I. Warszawa: Państwowe wydawnictwo
naukowe.
Otrębski, Jan. 1963. Über die Vervollkommnung der Forschungsmethoden. Lingua
Posnanensis 9. 7–28.
Patri, Sylvain. 2003. Les traitements contradictoires de */pt/ en slave commun (avec une note
sur «þ» indo-européen). Münchener Studien zur Sprachwissenschaft 63. 121–137.
Pedersen, Holger. 1895. Das indogermanische s im Slavischen. Indogermanische Forschungen
5. 33–87.
Pedersen, Holger. 1896. Das indogermanische s im Slavischen. Indogermanischen
Forschungen 5. 33–87.
Pedersen, Holger. 1897. Aspirationen i Irsk. Ph.D. Dissertation, Copenhagen. Leipzig:
Spirgatis.
Pedersen, Holger. 1900a. Die Gutturale im Albanesischen. Zeitschrift für vergleichende
Sprachforschung auf dem Gebiete der indogermanischen Sprachen 36/3, 277–340.
Pedersen, Holger. 1900b. Albanesisch und Armenisch. Zeitschrift für vergleichende
Sprachforschung auf dem Gebiete der indogermanischen Sprachen 36/3, 340–341.
Pedersen, Holger. 1904. Armenisch und die nachbarsprachen. Zeitschrift für vergleichende
Sprachforschung 39, 334–485.
Pedersen, Holger. 1908. Die idg.-semitische Hypothese. Indogermanische Forschungen 22.
341–365.
Pedersen, Holger. 1909. Vergleichende Grammatik der keltischen Sprachen. Erster Band.
Einleitung und Lautlehre. Göttingen: Vanderhoeck und Ruprecht.
Pedersen, Holger. 1913. Vergleichende Grammatik der keltischen Sprachen. Zweiter Band.
Bedeutungslehre (Wortlehre). Göttingen: Vanderhoeck und Ruprecht.
Pedersen, Holger. 1926. La cinquième declinaison latine. KDVS. Hist.-fil. Medd. XI. 5.
Pedersen, Holger. 1951. Die gemein-i.e. und die vorie. Verschusslaute. Copenhagen: Acad.
Peeters, Christian 1971. A phonemic Definition of I-E bh dh gh. KZ 85/1. 1–4.
Penney, J. W. H. 2017. The Documentation of Tocharian. In: Klein, Jared / Joseph, Brian / Fritz
/ Matthias (eds). Handbook of Comparative and Historical Indo-European Linguistics
(HSK 41.2). Berlin / Boston: Walter de Gruyter, 1298–1303.
Peters, Martin. 1997. Der armenische Flexionstyp gitem, gitacʿ und das ion.-att.
Plusquamperfekt. In: Lubotsky, Alexander (ed). Sound Law and Analogy: Papers in
honor of Robert S. P. Beekes on the occasion of his 60th
Birthday. Amsterdam/Atlanta:
Rodopi, 209–217.
Petit, Daniel. 2018. The Phonology of Baltic. In: Klein, Jared / Joseph, Brian / Fritz Matthias
(eds). Handbook of Comparative and Historical Indo-European Linguistics 4.3.
Berlin/Boston: De Gruyter Mouton. 1640–1651.
285
Peyrot, Michaël. 2008. Variation and change in Tocharian B. Amsterdam / New York: Rodopi.
Pinault, Georges-Jean. 1990. Notes sur le manuscripts de Maitreyasamit. Tocharian and IndoEuropean
Studies 4. 119–202.
Pinault, Georges-Jean. 1999. Tokharien A kapśañi, B. kektseñe. In: Eichner, Heiner /
Luschützky, Hans Chr. Compositiones Indogermanicae in memoriam Jochem Schindler.
Praha: enigma corporation. 457–478.
Pisani, Vittore. 1947. Balto e Slavo. In: Pisani, Vittore. Linguistica generale e indeuropea,
saggi e discorsci I. Milano: Libreria editrice scientifico universitaria. 65–82.
Pisani, Vittore. 1951. Studi sulla fonetica dell’armeno (II). Ricerche linguistiche 2, 47–74.
Pisani, Vittore. 1961. La ricostruzione dell´indoeropeo e del suo sistema fonetico. Archivio
glottologico italiano XLVI. 1-31.
Pisani, Vittore. 1976. Sanskrit dehí und dhehí. Sprache 22/2. 166.
von Planta, Robert. 1892. Grammatik der Oskisch-Umbrischen Dialekte. Erster Band.
Einleitung und Lautlehre. Strassburg: Verlag von Karl J. Trübner.
von Planta, Robert. 1897. Grammatik der Oskisch-Umbrischen Dialekte. Zweiter Band.
Formenlehre, Syntax, Sammlung der Inschriften und Glossen, Anhang, Glossar.
Strassburg: Verlag von Karl J. Trübner.
Pocetti, Paolo. 2017. The Documentation of Italic. In: Klein, Jared / Joseph, Brian / Fritz /
Matthias (eds). Handbook of Comparative and Historical Indo-European Linguistics
(HSK 41.2). Berlin / Boston: Walter de Gruyter, 733–743.
Pohl, Hans D. 1980. Slavisch st aus älterem *pt? In Die Sprache 26. 62–63.
Pokorny IEW = 1959. Indogermanisches etymologisches Wörterbuch. I. Band, A–KN. Bern:
Francke Verlag; 1966. Indogermanisches etymologisches Wörterbuch. II. Band, KO–
WR. Bern: Francke Verlag. 1969. Indogermanisches etymologisches Wörterbuch. III.
Band. Bern: Francke Verlag.
Pokorny, Julius. 1969. Altirische Grammatik. Berlin: Walter de Gruyter & Co.
Porzig, Walter. 1984. Die Gliederung des indogermanischen Sprachgebiets. Heidelberg: Carl
Winter.
Prokosch, Eduard. 1918–1919. Media Aspirata (I–IV). Modern Philology 15/10. 621–628,
16/2. 99–115, 16/6. 325–336, 16/10. 543–552.
Prokosch, Eduard. 1939. A Comparative Germanic Grammar. Philadelphia: Linguistic Society
of America/University of Pennsylvania.
Rasmussen, Jens E. 1987. On the status of the aspirated tenues and the Indo-European
Phonation Series. Acta Linguistica Hafniensia 20. 81-109.
Reichelt, Hans. 1909. Awestisches Elementarbuch. Heidelberg: Carl Winter’s
Universitätsbuchhandlung.
Reichelt, Hans. 1922. Die Labiovelare. IF 40. 40–81.
Rejzek, Jiří. 2008. K vývoji psl. kt [On the development of Pre-Slavic kt]. Slavia 77. 165–170.
Renou, Louis. 19963
. Grammaire Sanscrite. Paris: Librairie d’Amérique et d’Orient. Adrien
Maisonneuve.
Ribezzo, Francesco. 1903: Il problema capitale delle gutturali ie. Rendiconti dell´Accademia
di Napoli 17. Napoli : Stab. Tipografico della R. Universita A. Tessitore e figlio.
Ribezzo, Francesco. 1922–1923. Per la genesi delle tre serie gutturali ie. RIGI 6. 225–241, RIGI
7. 41–62.
Ribezzo, Francesco. 1929. La teoria Ascoliana delle gutturali ie. Allo stato presente della
glottologia. AGI 22-23. 131–151.
Ringe, Don. 1996. On the Chronology of Sound Changes in Tocharian. Volume 1. From ProtoIndo-European
to Proto-Tocharian. New Haven: American Oriental Society.
Ringe, Don. 2006. From Proto-Indo-European to Proto-Germanic. Oxford: Oxford University
Press.
286
Rix, Helmut. 1992. Historische Grammatik des Griechischen. Laut- und Formenlehre.
Darmstadt: Wissenschaftliche Buchgesellschaft.
Rix, Helmut. 2004. Ausgliederung und Aufgliederung der italischen Sprachen. In:
Bammesberger, Alfred / Vennemann, Theo et al. (eds). Languages in Prehistoric Europe.
Heidelberg : Universitätsverlag Winter, 147–172.
Ross, Alan S. C. / Berns, Jan. 1992. Germanic [numerals]. In: Gvozdanović, Jadranka (ed.).
Indo-European Numerals. Berlin/New York: Mouton de Gruyter, 555–715.
Rozwadowski, Jan Michał. 1961. O pierwotnym stosunku wzajemnym jȩzyków bałtyckich i
słowiańskich. In: Rozwadowski, J. M. Wybór pism II. Jȩzykoznawstwo indoeuropejskie.
Warszawa: Państwowe wydawnictwo naukowe, 96–113.
Rusakov, Alexander. 2017. Albanian. In: Kapović, Mate (ed.). The Indo-European Languages.
New York: Routledge, 552–608.
Sag, Ivan A. 1974: The Grassmann’s Law Ordering Pseudoparadox. Linguistic Inquiry 5(4).
591–607.
Sag, Ivan A. 1976: Pseudosolutions to the Pseudoparadox: Sanskrtit Diaspirates Revisited.
Linguistic Inquiry 7(4). 609–622.
Safarewicz, Jan. 1945. Pochodzenie trzech szeregów spólgłosek tylnojęzykowych w prajęzyku
indoeuropejskim. Sprawozdania z czynności i posiedzeń Polskiej akademii umiejętności
XLVI. Kraków. 37.
Salmons, Joseph C. 1993. The Glottalic Theory. McLean, VA: Institute for the Study of Man.
de Saussure, Ferdinand. 1877. La transformation latine de *tt en ss suppose-t-elle un
intermédiaire *st ?. Mémoires de la Société de linguistique de Paris 3. 293–301 (quoted
the version published in: de Saussure, Ferdinand. 1922. Recueil des publications
scientifiques de Ferdinand de Saussure. Heidelberg: Carl Winters
Universitätsbuchhandlung. 370–375).
de Saussure, Ferdinand. 1885. Sur un point de la phonétique des consonnes en i.-e. Mémoires
de la Société de linguistique de Paris 6. 246–257.
de Saussure, Ferdinand. 1892. Contribution à lʼhistoire des aspirées sourdes. BSL 7. 118.
Schenker, Alexander M. 2002. Proto-Slavonic. In: Bernard Comrie / Greville G. Corbett (eds),
The Slavonic languages, 60–121. London: Routledge.
Schindler, Jochem. 1967: Das idg. Wort für “Erde” und die dentalen Spiranten. Die Sprache
13. 191–205.
Schindler, Jochem. 1976: Diachronic and Synchronic Remarks on Bartholomae’s Law and
Grassmann’s Law. Linguistic Inquiry 7(4). 622–637.
Schleicher, August. 18611
. Compendium der vergleichenden Grammatik der indogermanischen
Sprachen. Weimar: Herman Böhlau.
Schleicher, August. 18662
. Compendium der vergleichenden Grammatik der indogermanischen
Sprachen. Weimar: Herman Böhlau.
Schmidt, Gernot. 1973. Die iranischen Wörter für „Tochter“ und „Vater“ und die Reflexe des
interkonsonantischen H (ǝ) in den idg. Sprachen. Zeitschrift für Vergleichende
Sprachforschung 87. 36-83.
Schmidt, Johannes. 1881. Zwei arische a-Laute undi die Palatalen. KZ 25. 1–179.
Schmidt, József: 1912: Kisérlet az indogermán gutturális probléma medoldására [Essay at a
resoluton to the IE guttural Problem]. Budapest: Acad.
Schmidt, Karl H. 1957. Die Komposition in gallischen Personennamen. Zeitschrift für celtische
Philologie. 26/1-4. 33–301.
Schmidt, Karl H. 1980. Armenian and Indo-European. In: Greppin, John A.C. (ed.). First
International Conference on Armenian Linguistics. New York: Caravan Books. 35–58.
Schmidt, Karl H. 1991. Latin and Celtic: Genetic Relationship and Areal Contacts. Bulletin of
the Board of Celtic Studies 38. 1–19.
287
Schmidt, Wolfgang P. 1966. Alteuropa und der Osten im Spiegel der Sprachgeschichte.
(Innsbrucker Beiträge zur Kulturwissenschaft, Sonderheft 22). Innsbruck: Institut für
Sprachwissenschaft der Universität Innsbruck.
Schmitt, Rüdiger. 1989. Altpersisch. In: Schmitt, Rüdiger (ed.). Compendium linguarum
Iranicarum. 56-85,Wiesbaden: Dr. Ludwig Reichert Verlag.
Schmitt, Rüdiger. 1989. Die Iranischen Sprachen im Überblick. In: Schmitt, Rüdiger (ed.).
Compendium linguarum Iranicarum. 25-31,Wiesbaden: Dr. Ludwig Reichert Verlag.
Schmitt, Rüdiger. 2007. Grammatik des Klassisch-Armenischen mit sprachlichen
Erläuterungen. Innsbruck: Institut für Sprachen und Literaturen der Universität
Innsbruck.
Schmitt-Brandt, Robert. 1998. Einführung in die Indogermanistik. Tübingen: Francke.
Schrijver, Peter. 1995. Studies in British Celtic Historical Phonology. Amsterdam / Atlanta:
Rodopi.
Schrijver, Peter. 2007. Some common developments of Continental and Insular Celtic. In:
Lambert, Pierre-Yves / Pinault, Georges-Jean (eds). 2007. Gaulois et Celtique
continental. Paris: Droz. 355–371.
Schumacher, Stefan. 2004. Die Keltischen Primärverben. Ein vergleichendes, etymologisches
und morphologisches Lexikon. Innsbruck: Institut für Sprachen und Literaturen der
Universität Innsbruck. Abteilung Sprachwissenschaft.
Schumacher, Stefan / Matzinger, Joachim. 2013. Die Verben des Altalbanischen.
Belegwörterbuch, Vorgeschichte und Etymologie. Unter Mitarbeit von Anna-Maria
Adaktylos (= Albanische Forschungen 33). Wiesbaden: Harrassowitz.
Schumacher, Stefan. 2013. Historische Phonologie. In: Schumacher, Stefan / Matzinger,
Joachim. Die Verben des Altalbanischen. Belegwörterbuch, Vorgeschichte und
Etymologie. Unter Mitarbeit von Anna-Maria Adaktylos (= Albanische Forschungen 33).
Wiesbanden: Harrassowitz, 205–276.
Schwyzer, Eduard. 1934. Dissimilatorische Geminatenauflösung als Folge yon Übersteigerung,
zunächst im Neugriechischen und im Spätaltgriechischen. KZ 61/3-4. 222–252.
Schwyzer, Eduard. 1939. Griechische Grammatik auf der Grundlage von Karl Brugmanns
griechischer Grammatik. Erster Band. Allgemeiner Teil, Lautlehre, Wortbildung,
Flexion. München: C. H. Beck’sche Verglasbuchhandlung.
Šefčík, Ondřej. 2012. On the Integration of Old Indo-Aryan Voiceless Aspirates into the
system. In: Sukač, Roman  Šefčík, Ondřej (eds.): Sounds of Indo-European 2. München:
Lincom Europa.
Šefčík, Ondřej. 2016. The problem of Proto-Balto-Slavic ‘aspirates’ revisited. 77/4. 300–315.
Šefčík, Ondřej 2017. Notes on the Function of the Indo-European Velar Series. In: Hansen,
Bjarne Simmelksjær et al. (eds). Etymology and the European Lexicon. 401–414.
Wiesbaden: Reichert Verlag.
Senn, Alfred. 1966. Handbuch der litauischen Sprache I. Grammatik. Heidelberg: Carl Winter.
Shevelov, George Y. 1964. A Prehistory of Slavic: the Historical Phonology of Common Slavic.
Heidelberg: Carl Winter-Universitätsverlag.
Shields, Kenneth. 1981. A New Look at the Centum/Satem Isogloss, KZ 95, 203-213
Shields, Kenneth. 2002. On the distribution of velar consonants in the suffixes and endings of
Indo-European. Indogermanische Forschungen 107. 96–105.
Sihler, Andrew L. 1995. New Comparative Grammar of Greek and Latin. Oxford: Oxford
University Press.
Sihler, Andrew L. 1997. The Myth of Direct Reflexes of the PIE Palatal Series in Kati. In:
Diesterherft, Dorothy / Huld, Martin / Greppin, John (eds.). Studies in Honor of Jaan
Puhvel. Part One. Ancient Languages and Philology. Washington D. C.: Institute for the
Study of Man. 187–194.
288
Sims-Williams, Patrick. 1981. The Development of the Indo-European Voiced Labiovelars in
Celtic. Bulletin of the Board of Celtic Studies 29. 201–229.
Sims-Williams, Patrick. 2017. Iranian. In: Kapović, Mate (ed.). The Indo-European Languages.
New York: Routledge, 352–386.
Sims-Wiliams Patrick. 2007. Common Celtic, Gallo-Brittonic and Insular Celtic. In: Lambert,
Pierre-Yves / Pinault, Georges-Jean (eds). 2007. Gaulois et Celtique continental. Paris:
Droz, 309– 353.
Sims-Williams, Patrick. 2017. Celtic. In: Kapović, Mate (ed.). The Indo-European Languages.
New York: Routledge, 352–386.
Skjærvo, Prods Oktor. 2017. The documentation of Iranian. In: Klein, Jared / Joseph, Brian /
Fritz Matthias (eds). Handbook of Comparative and Historical Indo-European
Linguistics 4.1. Berlin/Boston: De Gruyter Mouton. 472–481.
Sköld, Hannes 1924: Die indogermanischen Labiovelare. KZ 52, 147-151.
Smoczyński, Wojciech. 2001. Jȩzyk litewski w perspektywie porównawczej. Kraków:
Wydawnictwo uniwersytetu Jagiellońskiego.
Smoczyński, Wojciech. 2005. Lexicon der altpreussischen Verben. Innsbruck: Institut für
Sprachen und Literaturen der Universität Inssbruck.
Smoczyński, Wojciech. 2007. Słownik etymologiczny jȩzyka litewskiego. Wilno: Uniwersytet
Wileński, Wydział Filologiczny.
Solta, Georg R. 1960. Die Stellung des Armenischen im Kreise der indogermanischen
Sprachen: eine Untersuchung der indogermanischen Bestandteile des armenischen
Wortschatzes. Wien: Mechitharisten-Buchdruckerei.
Solta, Georg R. 1965. Palatalisierung und Labialisierung. Indogermanische Forschungen 70.
276–315.
Sommer, Ferdinand. 1914. Handbuch der lateinischen Laut- und Formenlehre. Eine
Einführung in das sprachwissenschaftliche Studium des Lateins. Heidelberg: Carl Winter
Universitätsverlag.
Specht, Franz. 1944. Die Ursprung der indogermanischen Deklination. Göttingen:
Vandenhoeck & Ruprecht.
Stang, Christian S. 1966. Vergleichende Grammatik der Baltischen Sprachen. Oslo / Bergen /
Tromsö: Universitetsforlaget.
Steensland, Lars 1973. Die Distribution der urindogermanischen sogennanten Gutturale.
Uppsala: Almqvist & Wiksel.
Stempel, Reinhard. 1994. Zur Vertretung der drei indogermanischen Gutturalreihen im
Armenischen. HS (KZ) 107. 298–309.
Stevens. Allan M. 1968. Madurese phonology and morphology. New Haven: American
Oriental Society.
Stempel, Reinhard. 1994. Zur Vertretung der drei indogermanischen Gutturalreihen im
Armenischen. Historische Sprachforschung 107/2. 298–309.
Stifter, David. 2017. The Phonology of Celtic. In: Klein, Jared / Joseph, Brian / Fritz / Matthias
(eds). Handbook of Comparative and Historical Indo-European Linguistics (HSK 41.2).
Berlin / Boston: Walter de Gruyter, 1188–1202.
Stiles, Patrick V. 2017. The Phonology of Germanic. In: Klein, Jared / Joseph, Brian / Fritz /
Matthias (eds). Handbook of Comparative and Historical Indo-European Linguistics
(HSK 41.2). Berlin / Boston: Walter de Gruyter, 888–912.
Stolz, Friedrich. 1894. Historische Grammatik der Laterinischen Sprache. Erster Band.
Einleitung, Lautlehre, Stammbildungslehre. Erste Hälfte. Einleitung und Lautlehre.
Leipzig: B. G. Teubner.
Strand, Richard F. 1973. Notes on Nūristāni and Dardic Languages. Journal of the American
Oriental Society 93/3. 297–305.
289
Streitberg, Wilhelm. 1974. Urgermanische Grammatik. Einführung in das vergleichende
Studium der altgermanischen Dialekte. Heidelberg: Carl Winter.
Stuart-Smith, Jane. 2004. Phonetics and Philology. Sound Change in Italic. Oxford: Oxford
University Press.
Sturtevant, Edgar H. 1930. The Gutturals in Hittite and Indo-European. Language 6. 213–228.
Sturtevant, Edgar H. 1933a. Archaism in Hittite. Language 9. 1–11.
Sturtevant, Edgar H. 1933b. A Comparative Grammar of the Hittite Language. Philadelphia:
Linguistic Society of America/University of Pennsylvania.
Sturtevant, Edgar H. / Hahn, E. Adelaide. 1951. A Comparative Grammar of the Hittite
Language. New Haven: Yale University Press.
Sukač, Roman. 2013. Introduction to Proto-Indo-European and Balto-Slavic Accentology.
Newcastle upon Tyne: Cambridge Scholars Publishing Szemerényi, Oswald. 1957. The
Problem of Balto-Slav Unity. A Critical Survey. Kratylos 2.2. 97–123.
Szemerényi, Oswald. 1964. Syncope in Greek and Indo-European and the Nature of IndoEuropean
Accent. Napoli: Istituto universitario orientale di Napoli.
Szemerényi, Oswald. 1967. The New Yook of Indo-European. Reconstruction and Typology.
Phonetica 17. 65–99.
Szemerényi, Oswald. 1990. Einführung in die vergleichende Sprachwissenschaft. Darmstadt:
Wissenschaftliche Buchgesellschaft.
Szemerényi, Oswald. 1996. Introduction to Indo-European Linguistics, Oxford: Clarendon
Press.
Tischler, Johann. 1990. Hundert Jahre kentum-satem Theorie. Indogermanische Forschungen
95. 63–98.
Tedesco, Paul. 1968. Sanskrit dehi ‘give’. Language. 44. 1–24.
Thompson, Rupert. 2017. Greek. In: Kapović, Mate (ed.). The Indo-European Languages. New
York: Routledge, 287–316.
Thurneysen, Rudolf 1909. Handbuch des alt-irischen Grammatik, Texte und Wörterbuch. I.
Teil: Grammatik. Heidelberg: Carl Winterʼs Universitätsbuchhandlung.
Townsend, Charles E. & Laura A. Janda. 1996. Common and Comparative Slavic: Phonology
and Inflection with Special Attention to Russian, Polish, Czech, Serbo-Croatian,
Bulgarian. Columbus: Slavica Publishers.
Trautmann, Reinhold. 19742
. Die altpreussischen Personennamen. Göttingen: Vandenhoeck &
Ruprecht.
Trubetzkoy, Nikolai S. 1939. Grundüge der Phonologie. Prag: Travaux du Cercle linguistique
de Prague.
Turner, Ralph L. 1964. Sanskrit buddhi-. In: Indo-Iraniaca: Mélanges présentés à G.
Morgenstierne à l´occasion de son soixante-dixième anniversaire. Wiesbaden :
Harrasowitz. 174-176.
Turner, Ralph L. 1966. A Comparative Dictionary of the Indo-Aryan Languages. London:
Oxford University Press.
Uhlenbeck, Christianus C. 1894. Etymologisches. 3 Flëhtan. Beiträge zur Geschichte der
deutschen Sprache und Literatur 19. 517–519.
Untermann, Jürgen. 2000. Wörterbuch des Oskisch-Umbrischen. Heidelberg: Carl Winter.
Vaillant, André. 1950. Grammaire comparée des langues slaves, tome I, phonétique. Lyon &
Paris: IAC.
Vasmer, Max. 1953–1958. Russisches etymologisches Wörterbuch 1–3. Heidelberg: Carl
Winter-Universitätsverlag.
Vath, Bernd / Ziegler, Sabine. 2017. The Documentation of Celtic. In: Klein, Jared / Joseph,
Brian / Fritz / Matthias (eds). Handbook of Comparative and Historical Indo-European
Linguistics (HSK 41.2). Berlin / Boston: Walter de Gruyter, 1169–1188.
290
Vendryès, Joseph. 1908. Grammaire du Vieil-Irlandais. Paris: Librairie orientale & américaine.
Verner, Karl. 1878. (rev. of) Zur lautverschiebung. Von J F Kräuter. Strassburg, Trübner 1877.
[viii und] 154 ss. Anzeiger für deutsches Altertum und deutsche Literatur 21. 333–342.
Vey, Marc. 1931a. Slave st provenant ďi.-e- *pt. Bulletin de la Société de linguistique 32. 65–
67.
Vey, Marc. 1931b. [Procèses-verbaux des séances:] V. sl. nestera. Bulletin de la Société de
Linguistique 32. XV.
Vey, Marc. 1953. Les noms de l̕ «autour» en slave. Bulletin de la Société de linguistique 49
(1). 24–40.
Vey, Marc. 1958. [Procèses-verbaux des séances :] Hypothèse sur l̕etymologie de vieux russe
Stribogъ. Bulletin de la Société de Linguistique 53 (1). XLI.
Voyles, Joseph B. 1992. Early Germanic Grammar. Pre-, Proto-, and Post-Germanic
languages. San Diego/New York: Academic Press.
Waanders, Frederik M. J. 1992. Greek [numerals]. In: Gvozdanović, Jadranka (ed.). IndoEuropean
Numerals. Berlin/New York: Mouton de Gruyter, 369–388.
Wackernagel, Jakob. 1896. Altindische Grammatik. Band I. Lautlehre. Göttingen:
Vandenhoeck & Ruprecht.
Walde, Alois. 1897. Die Verbindungen zweier Dentale und tönendes z im Indo-germanischen.
KZ 34: 461–536.
Wallace, Rex. 2017. Greek. In: Kapović, Mate (ed.). The Indo-European Languages. New
York: Routledge, 317–351.
Watkins, Calvert. 1966. Italo-Celtic Revisited. In: Puhvel, Jaan / Birnbaum, Henrik (eds.).
Ancient Indo-European Dialects. Berkeley / Los Angeles: University of California Press.
29–50.
Watkins, Calvert. 1970. A Further Remark on Lachmannʼs Law. Harvard Studies in Classical
Philology 74. 55–65.
Weisgerber, Leo. 1935. Sprachwissenschaftliche Beiträge zu frührheinischen Siedlungs- und
Kulturgeschichte, I. Rheinisches Museum für Philologie, Neue Folge, 84/4. 289–359.
Weiss, Michael. 2009. Outline of the Historical and Comparative Grammar of Latin. Ann
Arbor / New York: Beech Stave Press.
Weiss, Michael. 2012. Italo-Celtica: Linguistic and Cultural Points of Contact Between Italic
and Celtic. In: Jamison, Stephanie W. / Melchert, H. Craig / Vine, Bret (eds.).
Proceedings of the 23rd Annual UCLA Indo-European Conference. Bremen: Hempen:
151-73.
Werba, Chlodwig H. 2016. Ur(indo)arisches im Nūristānī. Zur historischen Phonologie des
Indoiranischen. In: Miller Andrew M./DeLisi Jessica/Wenthe Mark (eds.). Tavet Tat
Satyam. Studies in Honor of Jared S. Klein on the Occasion of His Seventieth Birthday.
Ann Arbor/New York: Beech Stave Press.
WH LEW = Walde, Alois / Hoffmann, Johann Baptist. 2007–20086
. Lateinisches
etymologisches Wörterbuch. Erster Band A – L. Zweiter Band M – Z. Registerband
zusammengestellt von Elsbeth Berger. Heidelberg: Carl Winterʼs
Universitätsbuchhandlung.
Whatmough, Joshua. 1937. The development of the Indo-European labiovelars with special
reference to the dialects of Ancient Italy. In: Mélanges linguistiques offerts à M. Holger
Pedersen : à l'occasion de son soixante-dixième anniversaire, 7 avril 1937. København:
Levin & Munksgaard. 45–56.
Whitney, William D. 1879. A Sanskrit Grammar, including both the classical language and the
older dialects of Veda and Brāhmana. Leipzig: Breitkopf and Härtel.
Whitney, William D. 1885. The roots, verb-forms and primary derivatives of the Sanskrit
language: a supplement to his Sanskrit grammar. Leipzig: Breitkopf and Härtel.
291
van Wijk. Nicolaas. 1931. Geschichte der altkirchenslavischen Sprache. Erster Band. Lautund
Formenlehre. Berlin / Leipzig: De Gruyter.
van Windekens, Albert J. 1976. Le Tokharien confronté avec les autres langues indoeuropéennes.
Volume 1. La phonétique et le vocabulaire. Louvain: Centre International
de Dialectologie Générale.
van Windekens, Albert J. 1986. Notes sur le traitement des dentales indo-européennes en
tokharien. Zeitschrift für vergleichende Sprachforschung 99/1. 151–160.
Windfuhr, Gernot L. 1971. Diacritic and Distinctive Features in Avestan. Journal of the
American Oriental Society 91/1: 104–124.
Winter, Werner. 1954. Problems of Armenian Phonology I. Language 30/2, 197–201.
Winter, Werner. 1955. Problems of Armenian Phonology II. Language 31/1, 4–8.
Winter, Werner. 1962a. Problems of Armenian Phonology III. Language 38/1, 254–262.
Winter, Werner. 1962b. Die Vertretung indogermanischer Dentale im Tocharischen.
Indogermanische Forschungen 67. 16–35.
Winter, Werner. 1978. The distribution of short and long vowels in stems of the type Lith. jsti
: vfsti : mfsti and OCS jasti : vesti : mesti in Baltic and Slavic languages. In: Fisiak, Jack
(ed.). Recent developments in historical phonology. The Hague: Mouton. 431-46.
Winter, Werner 1979: Zur “Überlänge” im Deutschen. In: Ezawa, Kennosuke – Rensch Karl.
H. – Bethge Wolfgang (eds.): Sprache und Sprechen. Tübingen: Max Niemeyer. 197–
200.
Winter, Werner. 1992a. Armenian [numerals]. In: Gvozdanović, Jadranka (ed.). Indo-European
Numerals. Berlin/New York: Mouton de Gruyter, 347–367.
Winter, Werner. 1992b. Tocharian [Numerals]. In: Gvozdanović, Jadranka (ed.). IndoEuropean
Numerals. Berlin/New York: Mouton de Gruyter, 97–164.
Winter, Werner. 2011. Vowel lengthening before distinctively voiced consonants. Tocharian
and IE Studies 12. 221–238.
Woodhouse, Robert. 1998. On PIE tectals. Indogermanische Forschungen 103. 40-60.
Woodhouse, Robert 2005. Assibilative palatalization of tectals in Phrygian and the adequacy
of bitectal frameworks for Proto-Indo-European. Indogermanische Forschungen 110.
205-234.
Young, Steven. 2017. Balto-Slavic. Baltic. In: Kapović, Mate (ed.). The Indo-European
Languages. New York: Routledge, 497–518.
Zinko, Christian. 2017. The Documentation of Anatolian. In: Klein, Jared / Joseph, Brian / Fritz
/ Matthias (eds). Handbook of Comparative and Historical Indo-European Linguistics
(HSK 41.1). Berlin / Boston: Walter de Gruyter, 239–249.
Zubatý, Josef. 1891. Slav. pastorъkъ. Archiv für slavische Philologie 13. 315–317.
Zupitza, Ernst. 1896. Die germanische Gutturale. Berlin: Weidmannsche Buchhandlung.
Zupitza, Ernst. 1899. Wortdeutungen. Zeitschrift für Vergleichende Sprachforschung 35. 265–
270.
293
Contents
1 On IE obstruents, examined clusters and used methods........................................................5
2 The development of the two-obstruent clusters from Indo-European into Indo-Iranian.......17
3 The development of the two-obstruent clusters from Indo-European into Baltic.................83
4 The development of the two-obstruent clusters from Indo-European into Slavic ..............107
5 The development of the two-obstruent clusters from Indo-European into Armenian.........127
6 The development of the two-obstruent clusters from Indo-European into Albanian..........141
7 The development of the two-obstruent clusters from Indo-European into Greek...............155
8 The development of the two-obstruent clusters from Indo-European into Italic languages173
9 The development of the two-obstruent clusters from Indo-European into Celtic...............195
10 The development of the two-obstruent clusters from Indo-European into Germanic.......213
11 The development of the two-obstruent clusters from Indo-European into Anatolian.......223
12 The development of the two-obstruent clusters from Indo-European into Tocharian ......231
13 The development of the two-obstruent clusters in the Indo-European languages: the
summary and conclusions..........................................................................................241
Appendix I: The comparative table of IE clusters plosive + t/s- in the given Indo-European
languages ..................................................................................................................257
Appendix II: The pan-chronic overviews of the given developments ..................................259
Literature............................................................................................................................269