Retroflex fricatives in Slavic languages
Author(s): Silke Hamann
Source: Journal of the International Phonetic Association , June 2004, Vol. 34, No. 1
(June 2004), pp. 53-67
Published by: Cambridge University Press
Stable URL: https://www.jstor.org/stable/44526947
JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide
range of content in a trusted digital archive. We use information technology and tools to increase productivity and
facilitate new forms of scholarship. For more information about JSTOR, please contact support@jstor.org.
Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at
https://about.jstor.org/terms
Cambridge University Press is collaborating with JSTOR to digitize, preserve and extend access
to Journal of the International Phonetic Association
This content downloaded from
147.251.68.69 on Wed, 14 Feb 2024 14:25:14 +00:00
All use subject to https://about.jstor.org/terms
Retroflex fricatives in Slavic languages
Silke Hamann
ZAS (Centre of General Linguistics), Berlin
silke@zas.gwz-berlin.de
The present study explores the phonetic and phonological grounds on which postalveolar
fricatives in Polish can be analysed as retroflex, and considers whether postalveolar fricatives
in other Slavic languages are retroflex as well. Velarization and incompatibility with front
vowels are introduced as articulatory criteria for retroflexion, based on cross-linguistic data.
According to these criteria, Polish and Russian have retroflex fricatives (i.e., /§/ and IzJ),
whereas Bulgarian has a laminai palatoalveolar fricative (/J/). In addition, it is illustrated that
palatalization of retroflex fricatives in Slavic languages (and in general) causes a phonetic
and phonological change to a non-retroflex fricative.
1 Introduction
The postalveolar series of fricatives in Polish is traditionally described as apical postalveolar
and referred to with the symbols /s/ and 111 (e.g. Dłuska 1 950, Gussmann 1 980, Wierzchowska
1980, Rubach 1984). Biedrzycki (1974), Dogil (1990), Jassem (2003), and Spencer (1986)
use the IPA symbols /J/ and /3/ for these respective sounds. Nevertheless, the postalveolar
fricatives in Polish differ significantly from those usually designated with /J/ and /3/ in English,
for example, as the latter are clearly laminai, whereas the Polish phonemes seem to have the
apex involved in their articulation (Biedrzycki 1974, Catford 1988). Furthermore, the Polish
sounds yield a different perceptual impression than English postalveolars (Hume 1 994). These
facts led Keating (1991), Ladefoged & Maddieson (1996), and Hamann (2003a) to claim that
the postalveolar fricatives in Polish are retroflex, though their articulation does not involve
the characteristic curling backwards of the tongue tip found, for example, in Dravidian
retroflex stops. A phonological argument that Polish fricatives are retroflex was made by Hall
(1997b) and Hamann (2003a). Apart from Polish, Keating (1991: 35) classifies only Russian
and Serbian as Slavic languages with retroflex fricatives, but merely on phonetic grounds.
The present article proposes a retroflex analysis for the postalveolar fricatives in Polish
and Russian. Furthermore, it tackles the question whether only this subgroup of the Slavic
languages has retroflex fricatives or whether other subgroups have this segmental class, too.
In order to do so, phonetic and phonological characteristics for retroflex sounds in general
and retroflex fricatives in particular from typologically diverse language groups such as
Scandinavian, Indo- Aryan, Dravidian, and Australian are presented in section 2. Section 3
compares these characteristics with the postalveolars in Polish, Russian, Bulgarian, and Czech
to test the retroflex nature of these sounds. In line with former proposals, it will be argued that
Polish and Russian have a retroflex fricative. Bulgarian will be shown to have phonetically
and phonologically a laminai postalveolar. The data collected for Czech was not sufficient to
Journal of the International Phonetic Association (2004) 34/1 © International Phonetic Association
001:1 0.1 01 7/S00251 00304001 BD4 Printed in the United Kingdom
This content downloaded from
147.251.68.69 on Wed, 14 Feb 2024 14:25:14 +00:00
All use subject to https://about.jstor.org/terms
54 S. Hamann
a b
Figure 1 (a)
make a
provide
from a
Slavic. S
involve
a palata
2 Retr
The pho
as these
unifyin
logically
(1993), a
This sec
behaviour of retroflexion.
2.1 Articulatory definition
Retroflexes are often defined as sounds articulated with the tongue tip curling in a posterior
direction at the postalveolar region (e.g. Trask 1996: 308). However, the class of retroflex
sounds shows considerable articulatory variation. Two x-ray tracings of retroflex plosives, one
from Hindi (cf. figure la), and one from Tamil (cf. figure lb) - both based on x-ray tracings
in Ladefoged & Maddieson (1996: 27) - exemplify this variation. In this and the following
descriptions of x-ray tracings it should be kept in mind that the shape of the tongue and the
placement of its tip depend very much on the vocal tract anatomy of the speaker, and that
there is large variability in anatomy from speaker to speaker. The tracings are represented in
similar dimensions and with the same palate and velum in order to facilitate comparisons.
Hindi retroflex stops (figure la) are articulated with the tongue tip against the region
behind the alveolar ridge, whereas Tamil retroflex stops (cf. figure lb) involve a placement of
the underside of the tongue tip against the postalveolar region. These two types of retroflexion
can be interpreted as two extremes on a spectrum of possible retroflex articulations.
Ladefoged & Maddieson (1996: 15) introduce two different symbols for these two types
of retroflex stops: the Hindi-type of articulation, involving just the tongue tip, is transcribed
with a subscript dot beneath the symbol for the alveolar stop /t/, whereas the articulation
with the underside of the tongue, as in Tamil and other Dravidian languages, receives the
traditional IPA symbol for a retroflex stop l[l. Švarný & Zvelebil (1955) distinguish 'retroflex'
from 'cacuminal', where the former involves the use of the underside of the tongue, whereas
the latter involves the tongue tip only. In this article, only one term, 'retroflex', and one
symbol, the traditional IPA symbol for retroflex, /[/, are used for both kinds of articulation.
The problematic nature of defining retroflexes is reflected in the IPA symbols chart, where
'retroflex' is used along with terms referring to the place of articulation such as 'dental',
'alveolar', etc., though 'retroflex' can be interpreted as describing a gesture of articulation.
This content downloaded from
147.251.68.69 on Wed, 14 Feb 2024 14:25:14 +00:00
All use subject to https://about.jstor.org/terms
Retroflex fricatives in Slavic 55
a b
Figure 2 (a)
The lite
as a valu
(1994) ap
gesture
Variati
find con
(1990) de
of articu
lead to l
further
Unifyin
the acti
and a su
is caused
tongue
tongue.
for pala
Further
a movem
Retracti
to move
tongue b
of retro
For the
articulat
a large f
the tong
Ladefog
language
far mor
on Lade
in the a
fricativ
the two
alveolar
for retro
It is int
within t
fricative
Tamil h
A retro
stop in
This content downloaded from
147.251.68.69 on Wed, 14 Feb 2024 14:25:14 +00:00
All use subject to https://about.jstor.org/terms
56 S. Hamann
Figure 3 English palatoalveolar fricative /J7.
retroflex fricatives do not involve as much curling backwards of the tongue tip as retroflex
stops. Other retroflex manners such as nasal, lateral, and rhotic, behave like the stop, as shown
in Hamann (2003a). Furthermore, the x-ray tracing of the Tamil retroflex in figure 2b does
not show a clear apical articulation, the tongue tip seems to be withdrawn into the tongue
body. Nevertheless, both retroflex fricatives share the postalveolar place of articulation, and
the sub-lingual cavity with retroflex stops, and also show a retraction of the tongue body. The
difference in articulation between retroflex fricatives and stops is explainable by the different
requirements that hold for stops and fricatives. Whereas stops involve a simple ballistic
movement and allow some degree of sloppiness in their articulation, fricatives and especially
sibilants like retroflex and alveolar fricatives require very precise articulatory gestures that
yield a turbulent airstream. Summing up the articulatory criteria for retroflex fricatives, they
are all articulated behind the alveolar ridge, show a sub-lingual cavity, are articulated with the
tongue tip (though this is not always discernible in the x-ray tracings), and with a retracted
and flat tongue body.
In order to determine whether the Slavic fricatives in question are retroflexes or laminai
postalveolars, we will now compare retroflex to laminai postalveolar articulations to formulate
criteria that differentiate between them. The x-ray tracing in figure 3 illustrates an English
laminai postalveolar (based on Laver 1994: 246). This type of sound will henceforth be
referred to as palatoalveolar.
In contrast to the retroflex fricatives in figure 2, the palatoalveolar in figure
clearly laminai and shows a fairly long constriction in the alveolar and postalveolar regio
Furthermore, the tongue has a raised and fronted tongue body behind the constriction.
this reason the tongue shape of palatoalveolars is sometimes called 'domed' by, for examp
Ladefoged & Maddieson (1996: 148). Lass (1976) and Catford (1977) infer from this to
shape that the palatoalveolar is inherently palatalized. The present study follows these auth
in this assumption.
The palatoalveolar differs from the retroflex in its tongue shape, it is raised and front
instead of flat and retracted, and in its constriction length, which is far longer than the retro
constriction. Both articulations share the postalveolar place and a sub-lingual cavity. The
sound classes cannot be distinguished systematically by their active articulator: alth
palatoalveolars are always laminai, retroflex fricatives are not always clearly discern
as apical (cf. the Tamil retroflex in figure 2b). In the following, a postalveolar plac
articulation combined with a retracted, flat tongue body is therefore used as the definin
articulatory property for retroflex fricatives.
Reference to the shape of the tongue as a defining criterion for a sound class is
new in phonetics; Ladefoged & Maddieson (1996: 29f.), for example, propose to distinguis
between the four coronal stops and nasals in the Australian language Eastern Arrernte by
articulator shape instead of the place of articulation.
When defining retroflex sounds as retracted postalveolars, the Australian language Lard
seems to pose a problem, as it is said to have phonetically and phonologically a non-velar
retroflex fricative (Hall 1997a, 2000, and Wilkinson 1988). As shown in Hamann (20
2003a), these assumptions are phonetically incorrect and also unnecessary phonologic
This content downloaded from
147.251.68.69 on Wed, 14 Feb 2024 14:25:14 +00:00
All use subject to https://about.jstor.org/terms
Retroflex fricatives in Slavic 57
since an alternative representation for the Lardil data can be given without stating that the
retroflex sounds in this language are [-back], i.e. non-retracted.
2.2 Acoustics of retroflexes
The acoustic criteria for retroflex fricatives presented in this section are not going to be applied
to all Slavic fricatives in question, but yield some supplementary data.
The following acoustic description is restricted to voiceless fricatives; voiced fricatives
usually only differ in having lower intensity than voiceless ones and an additional voicing bar.
A problem in describing acoustic characteristics of fricatives is the great discrepancies among
the spectra of a given fricative as spoken by different speakers (Ladefoged & Maddieson 1 996:
172). The descriptions given here are generalizations across the data and might not hold for
every single instance of a retroflex fricative.
In general, the cavity under the tongue in palatoalveolars and retroflexes has a lowering
effect on the acoustic resonance, as Keating (1991) points out. For the retroflex fricative the
ffication noise usually is lower than for the palatoalveolar and shows an energy maximum in
the area of the second and third formant of the adjacent vowels, i.e. between 1600 Hz and
2400 Hz. These values are similar to those of the low second and third formant transitions
observed for retroflex stops (see Hamilton 1996 and Stevens 1998).
In contrast to the spectral peak of retroflex fricatives, palatoalveolars show a more evenly
spread energy up to roughly 5000 Hz (Ladefoged & Maddieson 1996: 174f.), resulting in a
flat, plateau-like spectrum. According to Cruttenden (1994: 164), English /J/ has continuous
noise in the 2000-7000 Hz region. The peaks discernable in the spectrum are higher than those
of retroflexes, namely around 1 800 Hz for the peak corresponding to the second formant, and
2600 Hz for that corresponding to the third formant (Stevens 1998: 410).
Despite these differences, [J] and [§] are very similar acoustically when compared to
fricatives articulated in the velar or alveolar region. Rounding of the postalveolar fricative
enhances the similarity, as lip rounding enlarges the front cavity and results in a lowering of
the formant frequencies, especially the third (Johnson 1997: 118). As a result, [Jw] and [g] are
perceptually hard to distinguish.
2.3 Phonological behaviour of retroflexes
Cross-linguistically, retroflex sounds occur infrequently in a close - or mid-front vowel context,
which might result in a change of the vowel in retroflex context. In Norwegian we
can observe a phonological process of /e/-lowering in retroflex context: The segment lei has
an allophone [ae] before a retroflex, exemplified in (1). The first column contains words with
/e/, (la), or /ei/, (lb), followed by a dental, the second column contains words with the same
vowels followed by a retroflex, where the vowels are realized as [ae] (Kristoffersen 2000:
105f.).
(1) a. vett [vet] 'intelligence' vert [uaef] 'host'
helg [helg] 'weekend' helg [haetj] 'weekend', less formal register
b. hœl [heil] 'heel' hœl [haeit] 'heel', less formal register
A similar process of front vowel avoidance can be found in Mandarin (Yip 1996). Here, the
retroflex series /t§, t£h, §1 is in complementary distribution with the palatal series /tç, tçh, ç
The retroflex segments occur in the context of the open or back vowels, e.g. [fëai] and [fêu
Before close front vowels, however, the retroflex segments are replaced by the palatal series,
[tei] and [tev], but not *[[§i] and *[fèY],
According to Himmelmann (1991), the Austronesian language Tolitoli spoken in
Indonesia shows an alternation between an alveolar lateral approximant /1/ and a retroflex
This content downloaded from
147.251.68.69 on Wed, 14 Feb 2024 14:25:14 +00:00
All use subject to https://about.jstor.org/terms
58 S. Hamann
lateral flap /[/. The retroflex surfaces only after back vowels (cf. (2a)), the lateral /1/ occurs in
all other surroundings (cf. (2b)).
(2) a. mo[J]ogo 'wash hands' b. membembe[l]an 'to tremble'
u[l]ag 'snake' [l]abia 'sago'
to[Uito[l]i 'Tolitoli' kiki[l]o 'firefly'
A possible explanation for the phonological process of retroflexes avoidi
vowel context (with both possible outputs, a change in the vowel and a ch
to non-retroflex) is their retracted tongue body, which is articulatorily v
fronted tongue body and the close jaw of close front vowels. This articula
the presented avoidance strategies of close front vowel and retroflex seq
2003a and 2003b for a detailed discussion of this point and further exam
3 Postalveolar fricatives in Slavic languages
The postalveolar sounds investigated in this study are illustrated with w
Russian, Bulgarian, and Czech in (3)-(6). The upper row of each lang
voiceless sound, the lower row, the voiced sound, in word-initial, medial
In all these languages final devoicing occurs, thus the final column
remains empty in each case. The palatoalveolar symbols were used for th
throughout the phonetic transcription, in order to avoid any implications
flexion.
WORD-INITIAL WORD-MEDIAL WORD-FINAL
(3) Polish
szal [Jal] 'scarf kasza [kaja] 'groats' lekarz [lekaj] 'physician'
żal bal] 'grief gaża [ga3a] 'fee'
(4) Russian
sag [Jak] 'step' pošel [pAJol] 'went' naš [naj] 'ours'
žar bar] 'heat' xožu [xa'3u] 'I go'
(5) Bulgarian
šal [Jal] 'scarf kaša [kaja] 'groats' loš [loj] 'bad'
žar bar] 'heať lože [lo3e] 'bed'
(6) Czech
šál [Jail] 'scarf šašek [Ja[ek] 'fool' váš [va:J] 'your'
žal bal] 'grief lože [lo3e] 'bed'
Slavic postalveolars are usually transcribed with the symbols /S/ and Izl in trad
literature (e.g. Dukiewicz & Sawicka 1995 and Wierzchowska 1980 for Polis
for Russian, Simeonova 1988 for Bulgarian, and Kučera 1961 for Czech). O
of the IPA symbols /J, 3/ can be found, as in, for example, Spencer 1986, Dog
Jassem 2003 for Polish; Bolla 1981 and Jones & Ward 1969 for Russian; Sk
and Dankovičová 1997 for Czech; and Ternes & Vladimirova-Buhtz 2000 for
In order to determine their retroflex status, we will examine, in section 3.1,
postalveolars in these languages are articulated with the tongue tip and a flat, re
body. Furthermore, some acoustic cues and perceptual judgments on the sounds
will be discussed there. In section 3.2, the phonological behaviour of the Slavic
is investigated. It will be shown that some of the languages show an incompatib
postalveolars with close front vowels, similar to the processes given in section 2.
speculates on possible explanations why some Slavic languages developed a retro
whereas others did not.
This content downloaded from
147.251.68.69 on Wed, 14 Feb 2024 14:25:14 +00:00
All use subject to https://about.jstor.org/terms
Retroflex fricatives in Slavic 59
a b
Figure 4 Tw
3.1 Phon
The foll
Bulgaria
The disc
well.
In figur
are give
respectiv
with a f
is not vi
tongue f
the cons
tongue b
(1974: 2
sound a
as lamin
and Keat
of the t
agreeme
for retr
characte
1991) or
The acou
to Dogil
(1965) co
postalveo
as that o
[J], tho
& Madd
consider
Keating
retrofle
The Rus
voiceless
(cf. figu
hollowed
explicitl
Ward (1
/J7 and
Russian
fricative
tongue t
This content downloaded from
147.251.68.69 on Wed, 14 Feb 2024 14:25:14 +00:00
All use subject to https://about.jstor.org/terms
60 S. Hamann
a b
Figure 5 Tw
a b
Figure 6 Tw
a b
c
Figure 7 Three realizations of the Czech postalveolar.
(Jones & Ward 1969, Bolla 1981). Concerning perception, the Russian sound has a 'darker'
or more 'hollow' quality than the English postalveolar, according to Jones & Ward (1969:
134), which is probably due to a large sub-lingual cavity. Altogether, the Russian postalveolar
is retroflex according to the previously established criteria.
Two articulations of the Bulgarian postalveolar are given in figure 6 (a and b), based
on x-ray tracing of the voiceless sound by Bojadžiev (1982: 87) and Stojkov (1955: 81),
respectively. The tongue shape of the postalveolar in Bulgarian differs very much from that
of the corresponding Russian and Polish sounds, as it has a raised tongue body, similar to the
English palatoalveolar. The Bulgarian fricative is described in the literature as 'prepalatal'
(Simeonova 1988: 174) or 'weakly palatalized alveopalatal' (Scatton 1983: 58). We can
conclude from this that the Bulgarian postalveolar is not retroflex but palatoalveolar.
For the Czech postalveolar, three x-ray tracings are given in figure 7 (all three tracings
of the voiceless sound), as they show a considerable difference in articulation. The tracing in
figure 7a is based on Skaličková (1974: 104), the one in figure 7b on Pálková (1994: 229),
and the one in figure 7c on Polland & Hála (1926: 23). The articulation of the Czech sound in
This content downloaded from
147.251.68.69 on Wed, 14 Feb 2024 14:25:14 +00:00
All use subject to https://about.jstor.org/terms
Retroflex fricatives in Slavic 61
figure 7a looks similar to the equivalent Russian or Polish sounds, as it has a flatter and more
retracted tongue body than the English palatoalveolar. The articulations of the same sound
in figures 7b and 7c, however, show a domed tongue body. Judging from figure 7a alone, the
Czech sound could be classified as retroflex, as it is articulated in the same way as the Russian
and Polish sounds, and fulfils the criteria of retroflexion developed earlier on. Its retroflex
nature is confirmed by the fact that the Czech sound is usually described as apical (Kučera
1961: 30). Furthermore, Harkins (1953: 6) describes the Czech postalveolar as 'less palatal'
and as having a 'deeper timbre' than the English sound. The x-ray tracings in figure 7b and
7c, however, show a domed and high tongue body, which can be compared to the articulation
of the respective Bulgarian sound in figure 6 above.
According to the criteria developed in this article, it cannot be clearly determined whether
Czech has a retroflex fricative or not, as the articulations taken as bases for judgment differ
too much. Future articulatory research has to clarify this point. It has to be noted that the
Czech sound is classified as a postalveolar and is referred to with the symbol /J/ in very recent
phonetic descriptions, such as Dankovičová (1997).
In sum, if one assumes postalveolar articulation combined with retraction as defining
criteria for retroflexion, the articulatory analysis shows that Polish and Russian postalveolar
fricatives are clearly retroflex. This was predicted by previous phonetic studies (such as
Keating 1991, etc.) although the grounds on which the Polish and Russian sounds were
claimed to be retroflex were not all clear. The two very different articulations of the Czech
fricative are surprising, since such a variation has not been mentioned in the literature before.
No clear statement on the retroflex status of this sound can be given at this point. The
palatoalveolar nature of the fricative in Bulgarian, however, is very obvious and does not
come as a surprise, as no suggestion to the contrary has previously been made.
3.2 Phonological behaviour
In this section, phonological evidence for and against the retroflex status of the postalveolar
fricative in Polish, Russian, Czech, and Bulgarian will be discussed. We will consider the
question whether these sounds undergo rules comparable to those involving retroflexes in
other languages, as discussed in section 2.3. The phonological investigation will start with
Polish, moving on to Russian, Czech, and Bulgarian.
Dogil (1990: 5) states that the alveolo-palatal fricative /ç/ and the retroflex /§/ are in
complementary distribution in one context in Polish. Whereas the alveolo-palatal /ç/ does
not occur before the centralized, close vowel /i/, the retroflex /g/ does. On the other hand, the
retroflex does not occur before close front /i/, where the alveolo-palatal /ç/ does. Examples
for this restriction with the verbalizing suffix /i/ are given in (7).
(7) kos+i+c [çi] 'mow' inf.
towarzysz+y+c [gi] 'accompany' inf.
Thus, Polish has a phonotactic constraint which bans the sequence of a retroflex f
the close front vowel /i/, and which leads to a change towards the central vowel
two segments are joined at morpheme boundaries (this change is incorporated in
orthography). Rubach (1984: 44) terms this phonotactic constraint the Retraction
this rule comprises the retraction of the close front vowel after all non-palataliz
'hard' coronal consonants in Polish. Following Hall (1997b), the phonotactic constr
be taken as phonological evidence for the retroflex status of the postalveolar fricat
In Russian, the postalveolar also shows an incompatibility with the close f
/i/. Only the central vowel /i/ is allowed after this sound (Hamilton 1980; c
pronunciations in the second column of (8) as opposed to the impossible pronunc
the third column):
(8) šil [gil] *[gil] 'sewed' masc.
žil fcil] *[zjl] 'lived' masc.
This content downloaded from
147.251.68.69 on Wed, 14 Feb 2024 14:25:14 +00:00
All use subject to https://about.jstor.org/terms
62 S. Hamann
Drawing on these data, I assume a phonotactic co-occurrence restriction for Russian that
disallows sequences of retroflex fricatives and close front vowels. Hence, in Russian, there is
phonological evidence for the retroflex status of its postalveolar fricative just like in Polish.
Interestingly, both the Russian and the Polish sounds have a palatalized counterpart in front
vowel context (see discussion of these sounds in section 4 below), which further supports the
non-palatal nature of these sounds.
Czech has neither a close central vowel nor a phonotactic constraint comparable to the
one in Polish and Russian. The Czech language allows sequences of close front vowels and
postalveolar fricatives, as exemplified in (9).
(9) širší [JirJi:] 'wider'
úžina tutina] 'narrowness'
Kučera (1961: 26) mentions that the phonemes /i, ii, e/ have slightly raised allophones in
the pronunciation of most speakers when preceded by /V/ /dV /nV or /j/. As the postalveolar
fricative is not included in the context where vowel raising occurs, it can be concluded that this
fricative does not behave like a typical palatalized segment. Recall from section 2.1 that the
palatoalveolar 11/ is assumed to be inherently palatalized, following, for example, Lass (1976)
and Hall (1997b). On the other hand, Czech has palatalized counterparts for the phonemes
/d, t, n/ only (Carlton 1990: 26), the postalveolar does not have a palatalized variant, and
therefore no explicit evidence for its retracted/retroflex nature is given either. Altogether,
there is no phonological evidence for or against the retroflex status of the Czech sound
clarifying the phonetic findings.
In Bulgarian, no process of vowel change or phonotactic restriction for the sequence of
postalveolar fricative and front vowel exists. The postalveolar can occur with the front vowel
[i], as the words in (10) indicate.
(10) tišina [tijina] 'silence'
strašilo [strajilo] 'monster'
According to Scatton (1983: 54), the postalveolar fricative even further closes the fro
vowels lil and /e/, i.e. it behaves like a palatalized consonant. Furthermore, the postalveola
fricative and affricate series /J, 3, tj, d3 / does not have any palatalized counterparts, wher
the alveodental series does. This is a phonological argument that the sounds in question ar
palatoalveolars in Bulgarian.
In sum, the phonological regularities give further evidence for the phonetically attest
retroflex status of postalveolars in Polish and Russian, and for the non-retroflex status of t
palatoalveolar in Bulgarian. The Czech sound does not undergo any phonological proc
that could support its retroflex status, but does not show any counter-evidential behaviou
either.
3.3 Emergence of retroflexes in Slavic languages
This section speculates on the question of why some Slavic languages developed a retroflex
fricative. Diachronically, the postalveolar sounds (both retroflex and laminai palatoalveolar)
in Polish, Russian, Czech, and Bulgarian stem from a postalveolar fricative in Proto-Slavic.
The Proto-Slavic postalveolar came into existence via two processes. The first one was part
of the so-called First Velar Palatalization in Proto-Slavic (Stieber 1969, Carlton 1990, Mareš
1999), where the velar fricative lxi was fronted towards a postalveolar fricative if followed
by front vowels or the palatal glide (cf. (Ila) below). The second development involved
a retraction of the palatalized dental fricative towards the postalveolar fricative, and was
context-independent (ibid.; cf. (1 lb)).
(11) a) X J7_i,e,j
b) s' -► J"
This content downloaded from
147.251.68.69 on Wed, 14 Feb 2024 14:25:14 +00:00
All use subject to https://about.jstor.org/terms
Retroflex fricatives in Slavic 63
Table 1 Slavic coronal fricative systems (allophones are given in parentheses, question mark indicates that
the status of the fricative pairs has to be further investigated).
Dental/Alveolar Palatalized dental/Alveolar Retroflex Postalveolar Alveolopalatal
Polish s, z §, z, (J, 3) ç, z
Russian s, z sj, zj §, J, 3 (ç, z)
Bulgarian s, z J, 3
Czech s, z ? J, 3
Serbian s, z ? §,
When established as a phoneme, t
Carlton 1990: 116).
(12) *Ju:ti: > *Jiti 'to sew'
The fronting of back vowels in postalveolar context is a clear indication for the palatalized,
non-retroflex status of the Proto-Slavic fricative. Why did this palatoalveolar sound then
change into a retroflex one in some Slavic languages but not in others?
One possibility of accounting for the difference between Polish and Russian, where it
did, and Bulgarian, where it didn't, can be found in the sub-classes of the Slavic language
family. Polish (and Czech) belong to West Slavic, Russian to East Slavic, and Bulgarian to
South Slavic. The hypothesis that might be proposed here is that only the fricative in South
Slavic remained palatoalveolar, whereas the same sound in East and West Slavic underwent
a change to a retroflex. Testing this hypothesis involves establishing the exact nature of the
Czech sound and further investigating members of all three families. If Keating's (1991) claim
is right that Serbian also has a retroflex fricative (recall the introduction), this might refiite the
hypothesis, as Serbian is a South Slavic language and should have a palato-alveolar according
to this hypothesis.
Another explanation for the different development of Russian and Polish versus Bulgarian
might be the markedness of the fricative inventory in the respective languages. Hall (1997a)
shows that the development of the Indo- Aryan retroflex fricative was caused by the emergence
of the alveolo-palatal in Indo-Iranian. He assumes the diachronic developments given in (13).
(13) Indo-European Indo-Iranian Old-Indo- Aryan
/sj/ /s, J, ç/ -> /s, §,ç/
According to Hall, the contrast /J7 vs. /ç/ is cross-lingu
share the same feature specification [-anterior, +distribut
of Indo-Iranian /s, J, ç/ is marked. This marked status
Old-Indo- Aryan inventory with a retroflex postalveolar.
Rochoń (2001) applies Hall's proposal to Slavic lang
grounds that the emergence of an alveolo-palatal fri
postalveolar to retroflex in Polish. The introduction
fricative system /s, J / therefore led to a change from
retroflex postalveolar /§/ (cf. first row in table 1 of the
reasoning might hold for the Russian fricative, too, wher
a palatalized allophone of the dental stop (Rochoń, p.c.; c
represented in the third row in table 1, did not underg
retroflex, as it does not have any alveolo-palatals eith
Bulgarian thus supports Rochoń 's hypothesis. The hypo
This content downloaded from
147.251.68.69 on Wed, 14 Feb 2024 14:25:14 +00:00
All use subject to https://about.jstor.org/terms
64 S. Hamann
Figure 8 Russian retroflex fricative (solid line) and palatalized postalveolar fricative (dashed line).
non-retroflex fricative (cf. the fourth row of table 1). The status of the Czech sound has to
be investigated in order to conclusively test this hypothesis. If Keating 's (1991) observation
can be attested that the Serbian postalveolar is retroflex, this would refute the hypothesis,
since Serbian has no alveolo-palatal to trigger a change from postalveolar to retroflex (cf. the
Serbian coronal fricatives in the last row of table 1). In sum, further investigation is necessary
to account for the diverging developments of the Slavic postalveolar fricatives.
4 Palatalization of retroflexes
Palatalization in traditional articulatory terms means a supplementary articulatory gesture
superimposed upon a labial, dental, alveolar or postalveolar consonant. However, as shown in
Scatton (1983) for Bulgarian and Cavar & Hamann (2002) for Polish, palatalization of dental
fricatives and affricates involves a change in primary place of articulation. It will be argued
in this section, following Keating (1993), that palatalization of the retroflex fricative in Slavic
languages causes a similar change in the major place of articulation, which has an impact on
its phonological representation.
Examples of Russian and Polish palatalized postalveolar fricatives are given in (14) and
( 1 5), respectively. In the phonetic transcription, I use the symbol [J(j)] for the palatalized sound,
since I argue below that this sound is not a retroflex but a palatoalveolar. The secondary
palatalization is parenthesized, because it is redundant according to the previously made
assumption that palatoalveolar fricatives are inherently palatalized (recall section 2.1).
(14) Russian
sci [fi(J)i] 'cabbage soup'
(15) Polish
a. masz je [maj^je] 'you have them'
b. Shiva [f(j)iua] 'Shiva'
In Russian, the palatalized postalveolar can only occur as a geminate. In Polish, two contexts
are possible for this sound: it can occur across word boundaries when a retroflex fricative is
followed by [i] or [j] (cf. (15a)), and within loanwords with the sequence retroflex plus close
front vowel (cf. (15b)).
In figure 8, x-ray tracings of the Russian retroflex fricative (solid line) and its palatalized
counterpart (dashed line) are given, based on Bolla (1981: 159). As we can see from this
figure, the palatalized sound does not involve a simple addition of an elevated tongue body
onto its non-palatalized counterpart, but has changed its primary articulation from an apical
to a laminai fricative. Furthermore, the palatalized sound is articulated with a long and
narrow constriction and an elevated tongue body, no retraction is involved. This is not in
accordance with the criteria for retroflex sounds posed in section 2.1; thus, we must conclude
that palatalization changes a retroflex into a (palatalized) palatoalveolar (see Hamann 2002
This content downloaded from
147.251.68.69 on Wed, 14 Feb 2024 14:25:14 +00:00
All use subject to https://about.jstor.org/terms
Retroflex fricatives in Slavic 65
for a more detailed argumentation and a discussion of languages with allegedly palatalized
retroflex sounds). Hume (1994) also claims that palatalization changes a retroflex into a
non-retroflex, but her claim is based solely on the featural representation of both sounds.
5 Conclusion
The fricatives of Polish and Russian that are articulated in the postalveolar region are shown
to be more similar to retroflex fricatives than to postalveolar fricatives in English. This
similarity is based on articulation, in particular retraction, and on the segment's phonological
incompatibility with close front vowels, which is phonetically grounded in their retracted
tongue body. The fricatives of Bulgarian, on the other hand, are palatoalveolar, as they show
no phonetic or phonological retraction. Consequently, the Polish and Russian postalveolar
fricatives should be represented by the IPA symbols /§/ and /z/, whereas the Bulgarian
postalveolar fricatives are better transcribed as /J/ and /3/.
In the present study, Czech postalveolar fricatives could be classified as neither retroflex
nor palatoalveolar, since the phonetic descriptions of these sounds differ widely, and the class
behaves as phonologically neutral. Further research on the Czech postalveolar fricative should
be carried out in order to answer the question of its status. Although two possible explanations
were proposed for the difference in Slavic languages with respect to the postalveolar, neither
of them frilly accounted for it. Finding a plausible explanation for these differences in Slavic
languages thus remains another topic for future research.
With respect to the palatalization of the retroflex fricatives in Polish and Russian, it was
argued that these sounds change their primary articulation to a laminai palatoalveolar. Thus,
a representation of a secondary palatalized retroflex fricative should employ the IPA symbols
/JW and /3(jV instead of the articulatorily non-realizable symbols /^/ and /zj/.
Acknowledgements
This article is largely based on my paper 4 Postalveolar fricatives in Slavic languages as retroflexes',
published in the OTS Yearbook 2002 (edited by S. Baauw, M. Huiskes & M. Schoorlemmer). I want to
thank Katarzyna Dziubalska-Kołaczyk, John Esling, T. A. Hall, Peter Ladefoged, and an anonymous
reviewer for their helpful comments on the revised version. Errors remain my own.
References
Akishina, A. A. & Baranovskaja, S. A. (1980). Russkaja fonetika. Moscow: IzdatePstvo Russkij Jazyk.
Bhat, D. N. S. (1973). Retroflexion: an areal feature. Working Papers on Language Universais 13, 27-67.
Biedrzycki, E. (1974). Abriß der polnischen Phonetik. Warszawa: Wiedza Powszechna.
BojadZiev, T. (1982). Gramatika na suremennja bułgarski knižoven ezik. Sofia: Izdatelstvo na bulgarskata
Akademie na Naukite.
Bolla, K. (1981). A Conspectus of Russian Speech Sounds. Budapest: Hungarian Academy of Science.
Carlton, T. R. (1990). Introduction to the Phonological History of the Slavic Languages. Columbus:
Slavica.
Catford, J. C. (1977). Fundamental Problems in Phonetics. Edinburgh: University Press.
Catford, J. C. (1988). A Practical Introduction to Phonetics. Oxford: Clarendon Press.
Cavar, M. & Hamann, S. (2002). Polish velar and coronal palatalization - its perceptual background.
In Kosta, P., Blaszczak, J., Frašek, J., Geist, L. & Zygis, M. (eds.), Investigations into Formal Slavic
Linguistics , 31-48. Berlin: Peter Lang Verlag.
Cruttenden, A. (1994). Gimson s Pronunciation of English. London: Arnold.
DANKOVIČOVÁ, J. (1997). Czech. Journal of the International Phonetic Association 27, 77-80.
Dixit, R. P. (1990). Linguotectal contact patterns in the dental and retroflex stops of Hindi. Journal of
Phonetics 18, 89-201.
This content downloaded from
147.251.68.69 on Wed, 14 Feb 2024 14:25:14 +00:00
All use subject to https://about.jstor.org/terms
66 S. Hamann
Dłuska, M. (1950). Fonetyka polska . Kraków, Warszawa: Państwowe Wydawnictwo Naukowe (2nd edn.
1983).
Dogil, G. (1990). Hissing and hushing fricatives: a comment on non-anterior spirants in Polish. Ms.,
Stuttgart University.
Dukiewicz, L. & Sawicka, I. (1995). Gramatyka współczesnego języka polskiego: Fonetyka i fonologia.
Kraków: Instytut Języka Polskiego PAN.
Gnanadesikan, A. (1994). The geometry of coronal articulations. Proceedings of the North East
Linguistics Society 24, 25-139.
Gussmann, E. (1980). Studies in Abstract Phonology 28. Cambridge, MA: MIT Press.
Hall, T. A. (1997a). The historical development of retroflex consonants in Indo- Aryan. Lingua 101,
203-221.
Hall, T. A. (1997b). The Phonology of Coronals. Amsterdam: John Benjamins.
Hall, T. A. (2000). Typological generalizations concerning secondary palatalization. Lingua 110, 1-25.
Hamann, S. (2002). Retroflexion and retraction revised. In Hall, T. A., Pompino-Marschall, B. &
Rochon, M. (eds.), Papers on Phonetics and Phonology: The Articulation, Acoustics and Perceptio
of Consonants (ZAS Papers in Linguistics 28), 13-25. Berlin: ZAS.
Hamann, S. (2003a). The Phonetics and Phonology ofRetroflexes. Utrecht: LOT Press.
Hamann, S. (2003b). On the non-interaction of high front vowels and retroflex consonants. Ms., ZA
Berlin.
Hamilton, P. (1993). On the internal structure of the coronal node. In Kathol, A. & Bernstein, M. (eds
ESCOL 1993 Proceedings , 129-140. Cornell University.
Hamilton, P. (1996). Phonetic constraints and markedness in the phonotactics of Australian Aborigin
languages. Ph.D. dissertation, University of Toronto.
Hamilton, W. S. (1980). Introduction to Russian Phonology and Word Structure. Columbus: Slavica.
Harkins, W. E. (1953). A Modern Czech Grammar. New York: King's Crown.
Himmelmann, N. P. (1991). Tonini-Tolitoli sound structures. In Sneddon, J. (ed.), Studies in Sulawes
Linguistics II, 49-70. Jakarta: NUSA.
Hume, E. (1994). Front Vowels, Coronal Consonants and their Interaction in Nonlinear Phonolog
London: Garland.
Jassem, W. (1965). The formants of fricative consonants. Language and Speech 8, 1-16.
Jassem, W. (2003). Polish. Journal of the International Phonetic Association 33, 103-108.
JOHNSON, K. (1997). Acoustic and Auditory Phonetics. Oxford: Blackwell.
Jones, D. & Ward, D. (1969). The Phonetics of Russian. Cambridge: Cambridge University Press.
Keating, P. A. (1991). Coronal places of articulation. In Paradis, C. & Prunet, J.-E (eds.), The Special
Status of Coronals : Internal and External Evidence , 29-48. San Diego: Academic Press.
Keating, P. A. (1993). Phonetic representation of palatalization versus fronting. UCLA Working Papers
in Phonetics 85, 6-21.
Kristoffersen, G. (2000). The Phonology of Norwegian. Oxford: Oxford University Press.
Kučera, H. (1961). The Phonology of Czech. 's-Gravenhage: Mouton.
Ladefoged, P. (1975). A Course in Phonetics. New York: Harcourt Brace Jovanovich.
Ladefoged, P. (2001). Vowels and Consonants. Oxford: Blackwell.
Ladefoged, P. & Maddieson, I. (1996). The Sounds of the World's Languages. Oxford: Blackwell.
Lass, R. (1976). English Phonology and Phonological Theory. Cambridge: Cambridge University Press.
Layer, J. (1994). Principles of Phonetics. Cambridge: Cambridge University Press.
Lin, Y.-H. ( 1 989). The retroflex as a complex segment. ESCOL 6 Proceedings , 1 82-1 93 . Cornell University.
Maddieson, I. (1984). Patterns of Sounds. Cambridge: Cambridge University Press.
Mareš, E V (1999). Diachronische Phonologie des Ur- und Frühslavischen. Frankfurt am Main: Peter
Lang.
Ohala, M. (1983). Aspects of Hindi Phonology. Delhi: Motilal Banarsidas.
Oliverius, Z. F. (1974). Fonetika russkogo jazyka. Praha: Státní Pedagogické Nakladatelství.
Pálková, Z. (1994). Fonetika a fonologie češtiny. Praha: Univerzita Karlova.
Polland, B. & Hála, B. (1926). Artikulace českých zvuku v roentgenových obrazech. Prague: České
Akademie Věd a Umění.
This content downloaded from
147.251.68.69 on Wed, 14 Feb 2024 14:25:14 +00:00
All use subject to https://about.jstor.org/terms
Retroflex fricatives in Slavic 67
Rochon, M. (2001). Coronal fricatives and affricates in Slavic languages. Paper presented at the Fourth
European Conference on Formal Description of Slavic Languages, Potsdam.
Rubach, J. (1984). Cyclic and Lexical Phonology: The Structure of Polish. Dordrecht: Foris.
Scatton, E. A. (1983). Bulgarian Phonology. Ohio: Slavica.
Simeonova, R. (1988). Grundzüge einer kontrastiven Phonetik Deutsch/Bulgarisch. Sofia: Nauka i
Iskustvo.
SkaliČková, E. (1974). Srovnávací fonetika angličtiny a češtiny. Praha: Academia.
Spencer, A. (1986). A non-linear analysis of vowel-zero alternations in Polish. Journal of Linguistics 22,
249-280.
Stevens, K. (1998). Acoustic Phonetics. Cambridge, MA: MIT Press.
Stieber, Z. (1969). Zarys gramatyki porównawczej języków słowiańskich: Fonologia. Warszawa
Państwowe Wydawnictwo Naukowe.
Stojkov, S. (1955). Uvod v balgarskata fonetika. Sofia: Balgarska Akademia na Naukite.
Švarný, O. & Zvelebil, K. (1955). Some remarks on the articulation of the "cerebral" consonants in
Indian languages, especially Tamil. Archiv Orientalni 23, 374-434.
Ternes, E. & Vladimirova-Buhtz, T. (2000). Bulgarian. Journal of the International Phonetic
Association 20, 45-47.
Trask, R. L. (1996). A Dictionary of Phonetics and Phonology. London: Routledge.
Wierzchowska, B. (1980). Fonetyka i fonologia języka polskiego. Wrocław: Zakład Narodowy imienia
Ossolińskich.
Wilkinson, K. (1988). Prosodie structure and Lardil phonology. Linguistic Inquiry 19, 325-334.
Yip, M. (1996). Lexicon optimization in languages without alternations. In Durand, J. & Laks, B. (eds.),
Current Trends in Phonology: Models and Methods , 354-385. Salford: ESRI.
This content downloaded from
147.251.68.69 on Wed, 14 Feb 2024 14:25:14 +00:00
All use subject to https://about.jstor.org/terms