On the difference between and root
1 Introduction
In traditional approaches to morphology, the notion of root tends to be described in terms
of form, namely as “what is left when all morphological structure has been wrung out
of a form” (Aronoff 1994). In syntax, such morphologically defined roots mostly coincided
(until recently) with the traditional lexical categories. However, with the progress
of structural decomposition, our conception of lexical categories began to change rather
profoundly. In particular, what happened is that most grammatical properties of the traditional
lexical categories were transferred from the lexical head onto functional projections.
When this happened also to the categorial label (so that the functional categories
a, v, n, p appeared where the earlier tradition had A, V, N and P), all grammatically relevant
information has been dislocated to functional heads (Marantz 1995, Borer 2005,
Ramchand 2008).
The consequence is that in a number of current approaches, the root has turned into a
special kind of a syntactic object. Apart from having no grammatical properties (as these
are now seen as properties of functional heads), a large part of the literature proposes that
it also lacks phonological and semantic properties. The reasons for this position go in
part back to Zwicky’s (Zwicky 1969, Zwicky and Pullum 1986) Principle of Phonology
Free Syntax, which says that “[i]n the grammar of a natural language, rules of syntax
make no reference to phonology” (Miller et al. 1997:68). For instance, there is no rule
1
such that if a verb begins with a labial, it moves to T, etc. If roots (like functional heads)
have no phonology in syntax, the Principle of Phonology Free Syntax simply follows
from this architecture.
A similar observation has been made for concepts associated to roots. For instance,
whatever the difference between cat and dog on the conceptual side, this difference does
not trigger differential syntactic behaviour. In sum, as Marantz (1995) puts it “[n]o
phonological properties of roots interact with the principles or computations of syntax,
nor do idiosyncratic Encyclopaedic facts about roots show any such interactions.”
The result is that a number of present-day approaches use in syntactic trees the symbol
, which is an object devoid of syntactic, phonological and semantic properties,
and works as a pure placeholder for the insertion of the morphological root. Within
this largely consensual position, there is a debate concerning the number of such s in
syntax. Some authors argue that there is only a single (Marantz 1996, Borer 2014,
De Belder and Van Craenenbroeck 2015), while others propose that there is potentially
an infinity of s, individuated through the use of numerical indices in narrow syntax
(Pfau 2000, 2009, Harley 2014).
The reasons for proposing the (potentially infinite) number of roots has to do with
root suppletion. As has been pointed out already in Marantz (1995), if root suppletion
exists, then roots have to be somehow individuated in syntax (the argumentation leading
to this conclusion will be presented in section 2). An initial hypothesis for the singleproposal
has thus been that root suppletion does not exist. However, Haugen and Siddiqi
(2013) and Harley (2014) have argued, convincingly to our mind, that root suppletion
does in fact exist, which then necessitates the individuation of s in syntax. In order to
2
both individuate roots in syntax, and at the same time avoid assigning them phonology or
meaning, Harley (2014) follows Pfau (2000, 2009) and proposes that roots in syntax are
differentiated by a unique numerical index. (The common notation dog is a version
of the same proposal.)
The consequences of this proposal are sub-optimal. While the spirit of Phonology
Free Syntax is preserved (since e.g. 95 has no phonology in narrow syntax), the architecture
of the system is such that syntactic rules can still in principle differentiate
between cat and dog, since they have different numerical indices. The proposal thus
does not go all the way to making the phonological and conceptual properties of individual
roots irrelevant to syntax in the same way as the single- hypothesis.
In this paper, we make the single- approach compatible with root suppletion. We
achieve this by introducing phrasal spellout into the theoretical landscape. In particular,
what all the late insertion literature by and large has continued to take for granted is that
there is an isomorphy between the traditional roots (i.e., the spellouts of nonfunctional
lexical items such as book, smart, etc.), henceforth roots, and roots in narrow syntax,
henceforth . Once this assumption is dropped, a single- theory becomes viable again.
2 The nature of
The question whether there is only a single in narrow syntax or an infinity of them
is intimately related to the issue of suppletion. In presenting the argument, we shall
limit ourselves to the empirical domain of adjectival degrees (positive, comparative,
superlative). We do so for the reason that this domain has been recently thoroughly investigated
in Bobaljik (2012) both with respect to suppletion as well as the functional
3
heads involved. The core of the proposal in Bobaljik’s (2012) book is that in order to
capture certain facts about suppletion (specifically the *ABA generalisation), the positive
degree of an adjective must be contained inside the comparative, which in turn must
be contained inside the superlative. A version of this proposal is shown in (1).
(1) a. superlative
sprlP
sprl cmprP
cmpr aP
a
b. comparative
cmprP
cmpr aP
a
c. positive
aP
a
In these trees, we go slightly beyond the letter of Bobaljik’s proposal in that we have
decomposed his lowest projection, A, reminiscent of the traditional lexical categories,
into a node and a ‘little a’ node, as customary in the literature where s are used.
However, the reasoning that we will develop here for the node would carry over,
mutatis mutandis, to Bobaljik’s A, which too would have to come in an infinite variety
(like ). Finally, this reasoning carries over also to approaches that dispense with s
all together, and adopt a view according to which there are ‘functional heads all the way
down,’ as in Ramchand (2008). These approaches apparently also have to introduce an
infinite variety of such heads, a conclusion that defeats the very purpose of the enterprise.
In the Distributed Morphology (DM) framework, which Bobaljik (in large parts)
adopts, suppletion is accounted for by the specific form of Vocabulary Items (VIs), which
insert phonology under terminal nodes. To account for different forms of in different
environments (e.g. positive vs. comparative), the VIs that apply at the node are sensitive
to the presence or absence of a comparative (cmpr) node higher up in the structure:1
4
(2) a. ⇔ bett- / ] a ] cmpr ]
b. ⇔ good
Under certain circumstances, these VIs are in competition with each other. Not in the
positive degree (recall (1c)), as there is no cmpr head here, so that only (2b) meets the
structural description, and good will be inserted. In the comparative, given in (1b),
however, a competition between (2a) and (2b) will arise, since the structure generated in
syntax meets the strucutural description of both rules. The outcome of that competition
is determined by the Elsewhere Principle (Kiparsky 1973, Halle 1997:428), which states
that a more specific rule takes precedence over a more general one. (2a) thus wins in
the competition in the comparative, since it is a more specific VI than (2b). As a result,
bett is inserted in the comparative. The cmpr head is subsequently spelled out as -er,
yielding the form bett-er, little a being silent.
Now the VI in (2b) as currently formulated is just a fragment of the English Vocabulary.
If left on its own, it will insert good under any terminal node. One way of
extending our fragment will therefore be to add more roots:
(3) ⇔ good, nice, happy, small, intelligent, bad, …
What this (extended) rule achieves is that there is a free choice of insertion of a variety of
roots in the positive degree under . But now a problem arises with respect to (2a): since
it is more specific than (2b), it is also more specific than (3) (which is in relevant respects
like (2b)). The result is that bett- will be inserted under in any comparative structure,
an obviously wrong result. This problem in the analysis of root suppletion was pointed
out in Marantz (1996), and it is a consequence of the format of the rule (2a): it basically
5
says that just about any has the form bett- in the context of a comparative. Before
we discuss possible solutions, notice finally that this problem is actually independent of
whether we have a node, a lexical A node, or a functional F node at the bottom of the
tree; the whole reasoning carries over to these proposals as well.
In an attempt to turn the apparently wrong prediction to the framework’s advantage,
Marantz (1997) takes the radical view that root suppletion does not exist. If that is so,
the apparent counterexamples like bad-worse (or go-went and others) must involve the
functional vocabulary. This entails that unlike in the case of cat and dog, there is some
feature in the syntax that distinguishes functional adjectives like good–bett and bad–
worse from lexical adjectives like happy etc., which realise the head. If this is so,
then bett- is not a comparative of just any , but the comparative of one particular
functional feature or functional head, and the problem disappears.
However, Harley (2014) argues against this position on empirical grounds (cf. Haugen
and Siddiqi 2013), because her facts suggest that suppletion targets not only functional
heads, but also s. In order to deal with the problem pointed out by Marantz, and
without giving up on Phonology/Concept Free Syntax, she claims that s are individuated
in the syntax, i.e., prior to vocabulary insertion, by means of a numerical index.
Once bett- is not a comparative of just any , but a comparative of one particular
with a unique index, the problem also disappears.
Let us show how this works for both Marantz’ and Harley’s proposal by considering
the VIs for the suppletive pair good–bett given in Bobaljik (2012):
(4) a. good ⇔ bett- / ] a ] cmpr ]
b. good ⇔ good
6
In Marantz’ idea, good is interpreted as a reference to a functional feature that may
be part of the numeration, and as such be the input for syntax. Nonsuppletive adjectives
are inserted by the free-choice rule (3) (except that it would no longer contain good as
a choice). But the VIs in (4) only compete with each other, not with (3), since their
structural descriptions apply to different syntactic heads.
In Harley’s idea, good in (4) would be written more accurately as 93 (or any
other kind of index that would uniquely identify this root among all others). The presyntactic
lexicon in this view contains an infinity of different, individuated, s (see also
Pfau 2000, 2009). Free choice of a root is then not exercised at the point of insertion (as
in Marantz’ approach), but at an earlier point, namely in the selection of items for the
numeration, i.e., when the elements that will serve as the input to the syntactic computation
are selected. At the point of insertion, the competition is consequently restricted
to the two VIs in (4), modulo the replacement of good by 93.
In sum, in order to deal with suppletion, both approaches use different means to
achieve the same goal: they must somehow identify the unique lexical item that undergoes
suppletion, and limit the competition to those VIs which stand in a suppletive
relation to this particular item. Marantz (who works with just a single ) has no other
option but to make suppletive items unique by placing them in the class of ‘functional’
heads. Evidence against this position has been presented in Harley (2014), who argues
that suppletive verbs in Hiaki include verbs with rich lexical meanings, for which an
analysis in terms of functional heads is unlikely. To deal with the issues, she proposes
that the number of s is infinite, and that they are individuated by an index. But if s
really lack any constant substantive property that allows the syntax to identify them, one
7
needs to seriously wonder why they should be differentiated in narrow syntax at all. By
making them distinct, we allow for a theory where cat and dog have different syntax
after all: there is no conceptual reason why syntax should not be able to make reference
to this distinction in its inner workings.
3 Cyclicity and Phrasal spellout
In this section, we describe the main features of an account that allows for root suppletion
with just a single in syntax (or without any at all, if s are to be eliminated, as in
Ramchand 2008). What makes such a theory possible is phrasal spellout. The reason
for this is that phrasal spellout allows for a suppletive comparative to spell out a different
node than the positive, which is impossible in the standard account.
In order to present this idea in an accessible way, we will switch to the suppletive
pair bad—worse, which has been treated by non-terminal spellout also in Bobaljik
(2012). We shall then return to good—bett-er in the following section. The relevant
lexical entries (with the required containment relation) are given in (5). While bad is
just a regular (see (5a)), worse spells out a non-terminal node containing the , the
category-defining head a and the cmpr head (see (5b)).
(5) a. ⇔ bad b. cmprP
cmpr aP
a
⇔ worse
Independent support for (5b) comes from the fact that worse lacks the regular cmpr
marker -er, which is captured if its lexical entry is as in (5b), because such an entry on
8
its own pronounces the terminal where -er gets usually inserted.
We will get to the technical details of non-terminal insertion shortly, but the main
intuition is this: when syntax builds just the aP (corresponding to the positive degree),
only bad will be inserted, because its lexical entry provides an exact match for the syntactic
tree. The lexical item for worse, in contrast, is not an exact match: it is too big.
‘Too big’ may be understood either in an absolute sense (it is not a candidate for insertion
at all), or in a relative sense (it is a candidate, but it is ‘too big’ relative to bad).
When syntax builds cmprP, only worse is an exact match and will be inserted, this time
because bad is too small (again, this can be interpreted either in the absolute or in the
relative sense).
There are several ways of formalising the phenomenon that an exact match gets inserted,
and not a lexical item which is either too big or too small. For instance, one way
is to rely on the Subset Principle augmented by Radkevich’s (2010) Vocabulary Insertion
Principle VIP), a strategy adopted in Bobaljik (2012). On this account, the Subset
Principle makes sure that worse is too big for the positive, and the VIP makes sure that
bad is too small for cmprP. The second available option, which we develop and explain
later, adopts the Superset Principle (Starke 2009). For now, the main point is that no
matter how the ‘too big/too small’ difference gets encoded, we initially run up against
the same conundrum as the terminal-based proposal in section 2. In order to see that,
let us once again turn to the fact that there are a number of roots, all in free competition
for the node, as shown in (3) above, repeated here as (6):
(6) ⇔ good, nice, happy, small, intelligent, bad, …
9
Again, the problem is that once syntax builds the cmprP, all of these are going to be
‘too small’ compared to worse. The problem still resides in the fact that the lexical
entry in (5b) says that whenever the node is combined with a and cmpr, worse can
spell out such a constituent, and since other lexical entries are ‘too small,’ worse wins,
independently of how the ‘too big’/‘too small’ distinction is to be implemented.2
However, in the new setting based on phrasal spellout, a new type of solution to this
problem becomes available, if one more ingredient is added into the mix. The needed
addition is that spellout proceeds bottom-up, as in Bobaljik (2000, 2002), Embick (2010)
or Starke (2009, 2018). We phrase this as (7), noting that (7) need not be seen as an
axiom, but rather the consequence of two proposals, which are given in (8).
(7) Bottom-up spellout: If AP dominates BP, spell out BP before AP.
(8) a. Merge proceeds bottom up.
b. Spellout applies after every Merge step.
The consequence of (7) is that the node will always be spelled out before any phrase in
which it is contained, including, of course, cmprP. In other words, when spellout applies
at cmprP, this happens after spellout has applied at the node.
These two proposals—namely the bottom-up nature of spellout and the fact that it
targets non-terminals—bring us to where we wanted to go: to a single- syntax that
can accommodate root suppletion. In order to see this, consider the fact that spellout (in
order to be usable at all) must keep track of what it has done at lower nodes, so that it
ships this information to PF at some relevant point.3
Because of this, it is enough that we
now require that the phrasal lexical item (5b) can apply at cmprP only if the node has
10
been spelled out by bad before. Equivalently, worse is inapplicable if (by free choice of
) we have chosen a different lexical entry than bad as the pronunciation of .
In order to reflect this, let us rewrite the entry for worse as in (9), where instead of
the node, we write bad. Following Starke (2014), we refer to this device as a pointer.
The entry reads as follows: insert worse at cmprP, overwriting everything inside it, but
only if at a previous cycle, the sister to a was spelled out as bad.
(9) cmprP
cmpr aP
a bad
⇔ /worse/
Let us briefly reflect on the overwriting property of bottom-up spellout. When spellout
is successful at a given node, this means that a matching lexical entry has been found.
This, however, does not mean that this lexical entry is immediately shipped to PF. It is
remembered, and it will eventually be sent to PF; but if later on, a lexical item matching
a higher node is found, then the first (lower) candidate is not sent to PF at all: only the
higher spellout survives.
From the perspective of a single- theory, the problem with suppletive lexical items
like (5b) was that they could overwrite just any root. The pointer device just introduced
is here to restrict unlimited overwriting: worse can only overwrite bad. Caha et al. (to
appear) encode this by the so-called Faithfulness Restriction:
(10) Faithfulness Restriction (FR, preliminary)
A spellout α may override an earlier spellout β iff α contains a pointer to β
To conclude, let us stress the crucial point, which is that we now have a way to account
for root suppletion with just a single (or a single A, or, potentially, just functional
11
heads all the way down; the relevant point is that there is free choice of insertion at the
node which is at the bottom of the hierarchy). To do so, we need a bottom-up phrasal
spellout. In this kind of system, we can restrict worse to be the comparative of bad
without placing an index on in the syntax. The way we do this is by relying on the
so-called pointer. The pointer allows overwriting between two different entries, which
is otherwise disallowed. This is encoded in the Faithfulness Restriction.
4 The nature of roots
From the perspective of Phonology/Concept Free Syntax, the zero theory of s is that
there is only a single root. This was clearly argued for in the early work in DM (Marantz
1997), and Harley’s retreat from this position requires justification. Here, we argue
that such a retreat is not necessary if cyclic phrasal spellout is adopted. However, the
success of this approach depends in part on how it applies in cases where suppletion
is accompanied by regular morphology, e.g. in pairs such as good—bett-er: it seems
impossible to have bett- both spell out cmprP (as required by the single -theory), and
at the same time, leave cmpr available for the insertion of -er.
In this section, we suggest a solution to this problem. The solution is based on the
observation (to be illustrated more extensively in sections 5 and 6 below) that in cases
where suppletion co-occurs with overt marking, the overt marking tends to be ‘reduced’,
often a substring of a different, non-reduced marker.
To see this on an example, let us turn back to English. Here we have -er and more for
the comparative, and -est and most for the superlative. Clearly, -er and -est are morphologically
reduced compared to more/most, if only because they are affixes while more
12
and most are free-standing items. Further, if mo-re and mo-st decompose in the manner
suggested by the dashes, then one can in fact speculate that er and est are in fact
component parts of the non-reduced markers.
Now that we have established what we mean by full and reduced markers, consider
the observation that suppletive adjectives in English only occur with the reduced markers
(i.e., -er/-est) and never with the full markers (i.e., more/most), as observed by Bobaljik
(2012). Theoretically, we can understand ‘reduction’ if we decompose the single cmpr
node into two heads, C1 and C2, as shown in (11).
(11) C2P
C2 C1P
C1 aP
a
Once cmpr is decomposed, ‘full’ comparative marking can be analysed as expressing
both C1 and C2, as in (12), while reduced marking would spell out only C2, as in (13).
The two different realisations of the comparative structure would be causally linked to
two classes of roots. If the root can spell out the aP only, non-reduced marking is found.
If the root can spell out the whole C1P, reduced marking must be used.4
(12) C2P
C2
C2
C1P
C1
C1
aP
a
root
(13) C2
C1P
C1 aP
a
C2
-er
root
more



13
How do the two scenarios relate to suppletion? In a theory with a single , the tree
in (12) (with non-reduced marking) cannot (correctly) give rise to suppletion. That is
because the root in (12) express a constituent (aP) that corresponds to the positive. And
since suppletion (on the single- theory) requires that suppletive roots each spell out a
different constituent, (12) is incompatible with suppletion (QED).
This prediction distinguishes the single- theory from the ∞ approach in Harley
(2014). In Harley’s approach, as in DM in general, suppletive roots spell out the same
terminal as their nonsuppletive counterparts. Suppletion arises because the relevant terminal
(say 93) is placed in different environments. Therefore, suppletive roots should
be in principle compatible with non-reduced marking (incorrectly, in this case).
Turning now to (13), this scenario allows for suppletion on our account, but does not
force it. We first show how suppletion works, and then we turn to non-suppletive roots
that combine with the reduced marker. Suppletive roots like bett will have an entry like
(14), with a pointer to a different root.
(14) C1P
C1 good
⇔ bett
In this case, good first spells out aP, as shown in (15), which is a stage of the derivation
that feeds the insertion of bett- at C1P. This C1P is subsequently merged with C2,
yielding the full comparative structure in (16).
(15) aP
a
good
(16) C2
C1P
C1 aP
a
C2
er
bett
14
We now turn to non-suppletive roots that combine with -er. In order to show how they
are accounted for, we shall diverge from our reliance on a broad spectrum of conceivable
approaches to phrasal spellout, and focus on one particular version, due to Starke
(2009, 2018). The specific component of this theory, which we now need, is a matching
procedure based on the so-called Superset Principle (17).
(17) The Superset Principle (Starke 2009):
A lexically stored tree L matches a syntactic node S iff L contains the syntactic
tree dominated by S as a subtree
The principle says that if there is an entry like (18), then it can also spell out a C1P as
well as an aP (because aP is contained in it).
(18) C1P
C1 aP
a
⇔ root
If a root has such an entry, it can be used both in the positive (i.e., as an aP), and, at the
same time, appear with reduced marking in the comparative. In English, the adjectives
old or nice would be examples of such roots. The possibility of entries like (18) is what
leads us to say that if a root spells out C1P (and thus occurs with reduced marking in the
comparative), it does not have to be necessarily suppletive.
To sum up, the theory allows for two kinds of roots with reduced comparative marking:
suppletive ones (bett-) and non-suppletive ones (old), depending on whether they
have a pointer in their lexical entry or not.
Notice, however, that the entry for adjectives like old in (18) is very similar to the
entry we have originally considered for worse, recall (5b). The problem with (5b) was
15
that it could spell out the comparative form of just about any root. But we have closed this
possibility by introducing the FR (10): overwriting at C1P only happens if the overwriter
has a pointer to the overwritee. But this raises another issue: in a bottom up cyclic
system, which we are proposing, the is always spelled out first. Here all lexical items
that contain the node are candidates, and we let free choice decide. Suppose we choose
an entry like old, with the entry (18). Since this entry contains the , the Superset
Principle is satisfied, and old gets inserted at . The next step is to merge little a with
the , forming aP, and we again try to spell it out.5
But the problem is that any item
inserted at aP will overwrite the item selected for insertion at the node, and hence,
this situation is regulated by the Faithfulness Restriction (10). So far, the condition says
that overwriting is only possible if the overwriter has a pointer to the overwritee, which is
not the case here. At the same time, we are not literally overwriting one entry by another,
since we want to insert at aP the very same entry that we inserted at the node. This
must be legal, otherwise an entry such as (18) would never get to use its lexicalisation
potential. In order to allow this, we augment the FR in the following way:
(19) Faithfulness Restriction (FR)
A spellout α may override an earlier spellout β iff
a. α contains a pointer to β
b. α = β
This formulation allows the entry (18) to keep overwriting itself (due to (19b)) all the
way to C1P. When C2 is merged, the whole C2P cannot be spelled out by (18), and so
-er is inserted under C2.
16
Let us now turn to aP-sized roots, whose entry is as in (20).
(20) aP
a
⇔ /good, intelligent, …/
These again come in two varieties. Those that are not pointed to by anything (intelligent)
combine with more in the comparative. Those that are pointed to by some other lexical
item (good is pointed to by bett-) switch to the suppletive form in the comparative.
Proposing that these adjectival roots spell out aP (as opposed to spelling out just
the ) has the advantage that we shall not need to worry about the exponence at little
a, which these adjectives lack. Cross-linguistically, there are, of course, examples of
morphologically complex positive-degree adjectives (e.g. English adjectives like slimy,
cheek-y, etc.), where the adjectival affix spells out the little a, so that the root cannot
be inserted as a full aP, but as a only.
The final theoretical option is that a root spells out the whole C2P, as in (21). This
root spells out both C1 and C2, and hence, it appears with no comparative marking
whatsoever. Such roots come again (in principle) in two flavours. One type of such
roots has a pointer to a different root, as in (21), and then the root works as a suppletive
counterpart of a positive root. We analyse worse like this.
(21) C2P
C2 C1P
C1 aP
a
root
(22) C2P
C2 C1P
C1 bad
⇔ worse
17
The other type is as in (23), without a pointer. English has no such adjectives, but we
find cases like this in other languages, to be discussed in section 5 below.
(23) C2P
C2 C1P
C1 aP
a
⇔ root
In sum, the approach sketched in this section allows for there to exist a variety of roots
in the morphological sense, while still maintaining a single in syntax. The variety
of roots that our theory makes available can be visualised as a set of concentric circles,
encompassing various sizes of structure, as shown in (24):
(24) C2P
C2 C1P
C1 aP
a P
root4
root3
root2
root1
(25) a. root1: appears with an overt
little a in the positive
b. root2: no overt little a, full
comparative marking
c. root3: no overt little a, reduced
comparative marking
d. root4: no overt little a, no
comparative marking
From the perspective of suppletion, we note that roots that reach up to the comparative
zone (namely 3, 4) may work as suppletive comparatives of positive roots (those of size
2). The crucial theoretical possibility allowed by the split cmpr system is the existence
of suppletive roots of size 3, corresponding to bett-: these can both work as suppletive
counterparts to positive-degree roots of size 2, and, at the same time, combine with a
‘reduced’ comparative marker, namely -er. In the remainder of the paper, we present
two case studies which illustrate this reduction effect under suppletion.
18
5 Czech
The first case study concerns the interaction between comparative marking and suppletion
in Czech. We start from the fact that the traditional descriptions recognise three
different allomorphs of the comparative (see Dokulil et al. 1986; Karlík et al. 1995; Osolsobě
2016). We give them in the first column of (26). The actual comparative marker
precedes the dash, the -í following it is an agreement marker. Note that (26b,c) represent
an increasingly ‘reduced’ realisations of the full marker -ějš-, seen in (26a).
(26) allomorph pos cmpr sprl gloss
a. ějš-í chab-ý chab-ějš-í nej-chab-ějš-í ‘weak/poor’
b. š-í slab-ý slab- š-í nej-slab- š-í ‘weak’
c. -í hez-k-ý hez-č- -í nej-hez-č- -í ‘pretty’
d. -í ostr-ý ostř -í nej-ostř -í ‘sharp’
On the first two lines, we illustrate the ějš-í and -š-í allomorphs with two adjectives that
are semantically and phonologically similar. We do so to show that the allomorphy is
not driven by phonology or semantics. Rather, the distribution is governed by arbitrary
root class: -ějš is the productive allomorph, while -š-í is restricted (occurring with 72
out of 5440 adjectives sampled in Křivan 2012).
On the third line, we illustrate the zero allomorph, and two facts should be noted.
First of all, the positive and the comparative are not homophonous: their morphological
identity is obscured by phonological interactions with the agreement markers. Specifically,
the agreement marker -í, found in the comparative, triggers the palatalisation of
the base (k goes to č), while the elsewhere agreement marker -ý does not palatalise the
19
base (see Caha et al. to appear for a discussion of the palatalisations). As a result, the
forms are distinct. The second fact to be noted is that in the standard language, this type
of marking only occurs after a particular little a marker, namely -k. This morpheme is
similar to the English -y in that it sometimes occurs after nominal roots (e.g., sliz-k-ý =
‘slim-y’) and sometimes after cranberry type of morphemes (e.g., hez-k-ý = ‘prett-y’).
Because of its limited distribution, it is not clear whether the ø allomorph needs to be
recognised as a separate marker, or perhaps dismissed as a special realisation of -š after
-k. We do, however, recognise the zero as a relevant allomorph to consider, because
in the dialects of North Eastern Bohemia (Bachmannová 2007), one finds it also after
non-derived adjectives, as shown on the last row.
Given that -š is a substring of -ějš, it is tempting to decompose -ějš into a C1 -ěj
and a C2 -š, as suggested by Caha et al. (to appear). Independent evidence for this
analysis comes from comparative adverbs, seen in the second column of (27). Here the
-š-part of the comparative adjective is systematically missing, while -ěj is preserved.
This confirms an analysis where -ěj and -š are independent morphemes.
(27) cmpr Adj cmpr Adv
chab-ěj-š-í chab-ěj-i ‘weak’
rychl-ej-š-í rychl-ej-i ‘fast’
červen-ěj-š-í červen-ěj-i ‘red’
Given our model with two comparative heads, the facts are easily captured if -ěj and -š
spell out C1 and C2 respectively. With aP-sized roots, both markers surface, see (28).
With roots of the size C1P, only -š appears, as in (29), and zero marking arises when
the root spells out all of the projections, as in (23). Recall that (23) was presented as a
20
logical option allowed by our system, and though it was not attested in English, we need
it to account for ostr-ý ‘sharp’ in (26).
(28) C2P
C1P
aP
a
C1
-ěj
C2
-š
root
(29) C2
C1P
aP
a
C1
C2
š
root
Given our theory, where comparative suppletion requires a root that is at least of the
size C1P (so that it applies at a different node than the positive, which spells out aP), we
arrive at a prediction. In particular, root suppletion should be incompatible with -ěj-š.
To verify this, the table (30) presents an exhaustive list of suppletive adjectives based on
Dokulil et al. (1986:379) and Osolsobě (2016). The table shows that the prediction is
borne out: all suppletive adjectives require the ‘reduced’ -š allomorph.
(30) pos cmpr gloss pos cmpr gloss
dobr-ý lep-š-í ‘good’ špatn-ý hor-š-í ‘bad’
velk-ý vět-š-í ‘big’ mal-ý men-š-í ‘little, small’
dlouh-ý del-š-í ‘long’
We submit these facts here as an important confirmation of the phrasal spellout model,
which says that when there are two or more ways of marking the comparative, suppletion
is incompatible with the full marker. With reduced markers, we find both suppletive and
regular cases, depending on whether the entry of the size C1P has a pointer or not.
21
6 Latin
Latin provides evidence for a correlation between reduced marking of the superlative
and suppletion. The regular marking of comparative and superlative is shown in (31a).
(31) pos cmpr sprl gloss marking in sprl
a. alt-us alt-i-or alt-i-ss-im-us ‘tall’ full marking
b. mal-us pe(i)- or pe- ss-im-us ‘bad’ sprl lacks -i
c. bon-us mel-i-or opt- im-us ‘good’ sprl lacks -i-ss
d. magn-us ma-i-or max- im-us ‘big’ sprl lacks -i-ss
e. parv-us min- or min- im-us ‘small’ sprl lacks -i-ss
f. mult-us plūs plūr- im-us ‘much’ sprl lacks -i-ss
Following De Clercq and Vanden Wyngaerd (2017), we segment the regular superlative
into five morphemes. The first morpheme is the root (alt), and the last one is agreement
-us. The reason for treating the three middle markers -i, -ss and -im as separate morphemes
is that they can be missing in the irregular forms in (32b-f). These represent an
exhaustive list of the suppletive cases given in Gildersleeve and Lodge (1903:46).
We analyse -i (the first of the post-root morphemes) as a comparative marker, i.e., as
a morpheme identical to the -i of the comparative alt-i-or. We treat -i in the same way
as the English -er, namely as the spellout of C2. Consequently, we analyse -or, which
follows -i in the comparative, as an agreement marker. We do so because the masculine
form alt-i-or ‘taller, m.sg’ alternates with the neuter alt-i-us. As a C2 marker, -i is
compatible with suppletion. In (31c), for instance, the positive degree root bon- realises
aP, the suppletive comparative root mel- realises C1P, and -i- is the marker of C2.
22
The remaining two morphemes mark the superlative, which we then split into S1
and S2, analogously to Cmpr. The structure of alt-i-ss-im-us thus looks as follows:
(32) S2P
S1P
C2P
C1P
C1 aP
a
C2
i
S1
ss
S2
im
alt
Against this background, consider the fact that the superlative marking with suppletive
roots is always reduced, see (33b-f). There is not a single suppletive root in Latin which
keeps all the three pieces in place, as indicated in the final column of (32). More specifically,
we see two classes of suppletive roots. The majority of suppletive roots lacks the
C2 -i as well as the S1 -ss, and we would thus analyse them as spelling out S1P. Pe- lacks
only the -i, which, on the assumption that -i is C2, leads to the proposal in (33).
(33) S2P
S1P
C2P
C1P
C1 aP
a
C2
i
S1
ss
S2
im
opt
pe
alt
23
This picture has implications for the analysis of the comparative. Specifically, all
suppletive roots which spell out a projection larger than C1P should make -i disappear
not only in the superlative, but also in the comparative. This is true for the adjectives
min-or ‘smaller’ and plus ‘more’, as well as, arguably, pe-or ‘worse,’ where the glide
in the comparative pe[j]or results, on our analysis, from phonological factors (hiatus
filling). Note that plus lacks the agreement marker -or, and Gildersleeve and Lodge
(1903:46) analyse it as a neuter form, with the masculine cell left blank. Here we treat
plus as spelling out minimally S1P, lacking -i in the comparative and in the superlative
also -ss). We leave the reasons for the lack of agreement open to interpretation.6
The (c) and (d) cases of (31) warrant some further comment, since they have -i- in
the comparative but lack it in the superlative. This is because they instantiate an ABCpattern,
with two different suppletive roots: one of size C1P (explaining the presence of
-i- in the comparative), and another of size S1P (explaining the absence of both -i- and
-ss- in the superlative). These suppletive roots successively point to one another, e.g.
the lexical entry for opt contains a pointer to mel, which itself contains a pointer to bon.7
7 Conclusion
This paper proposes an approach that allows for root suppletion within a theory of narrow
syntax that is phonology/concept free, and which dispenses with indexed s. By
dispensing with indexed s, the theory is also compatible with approaches where s are
dispensed with all together (Ramchand 2008). What makes this type of theory available
is a bottom up phrasal spell out, where different roots apply at different nodes. Finally,
we have addressed the issue of how to maintain such a theory in the face of the fact that
root suppletion sometimes combines with overt marking.
24
References
Abels, Klaus, and Peter Muriungi. 2008. The focus particle in Kîîtharaka: Syntax and
semantics. Lingua 118:687–731.
Aronoff, Mark. 1994. Morphology by itself. Cambridge: MIT Press.
Bachmannová, Jarmila. 2007. Okrajové úseky severovýchodočeské nářeční oblasti z
pohledu celočeského (na materiálu korespondenční ankety úJČ). Naše řeč 90:7–19.
Bobaljik, Jonathan. 2000. The ins and outs of contextual allomorphy. University of
Maryland Working Papers in Linguistics 10:35–71.
Bobaljik, Jonathan. 2002. Syncretism without paradigms: Remarks on Williams 1981,
1994. In Yearbook of morphology 2001, ed. Geert Booij and Jaap van Marle, 53–85.
Dordrecht: Kluwer.
Bobaljik, Jonathan. 2012. Universals in comparative morphology. Cambridge, MA:
MIT Press.
Borer, Hagit. 2005. In name only. Oxford: Oxford University Press.
Borer, Hagit. 2014. The category of roots. In The syntax of roots and the roots of syntax,
ed. Artemis Alexiadou, Hagit Borer, and Florian Schäfer, 112–148. Oxford: Oxford
University Press.
Caha, Pavel. 2017. Suppletion and morpheme order: Are words special? Journal of
Linguistics 53:865–896.
Caha, Pavel, Karen De Clercq, and Guido Vanden Wyngaerd. to appear. The fine structure
of the comparative. Studia Linguistica .
De Belder, Marijke, and Jeroen Van Craenenbroeck. 2015. How to merge a root. Linguistic
Inquiry 46:625 – 655.
25
De Clercq, Karen, and Guido Vanden Wyngaerd. 2017. *ABA revisited: evidence from
Czech and Latin degree morphology. Glossa 2:69: 1–32.
Dékány, Éva. 2011. A profile of the Hungarian DP. the interaction of lexicalization,
agreement and linearization with the functional sequence. Doctoral Dissertation,
CASTL, University of Tromsø.
Dokulil, Miloš, Karel Horálek, Jiřina Hůrková, Miloslava Knappová, and Jan Petr. 1986.
Mluvnice češtiny 1. fonetika. fonologie. morfologie a morfemika. tvoření slov. Praha:
Academia.
Embick, David. 2007. Linearization and local dislocation: Derivational mechanics and
interactions. Linguistic analysis 33:303–336.
Embick, David. 2010. Localism versus globalism in morphology and phonology. Cambridge,
MA: MIT Press.
Gildersleeve, Basil L., and Gonzalez Lodge. 1903. Latin grammar. London: MacMillan
and Co.
Halle, Morris. 1997. Distributed morphology: impoverishment and fission. MIT Working
Papers in Linguistics 30:425–449.
Harley, Heidi. 2014. On the identity of roots. Theoretical Linguistics 40:225–276.
Haugen, Jason D., and Daniel Siddiqi. 2013. Roots and the derivation. Linguistic Inquiry
44:493–517.
Haugen, Jason D., and Daniel Siddiqi. 2016. Towards a restricted realization theory:
Multimorphemic monolistemicity, portmanteaux, and post-linearization spanning. In
Morphological metatheory, ed. Daniel Siddiqi and Heidi Harley, 343–385. Amsterdam:
John Benjamins.
26
Karlík, Petr, Marek Nekula, and Zdena Rusínová. 1995. Příruční mluvnice češtiny.
Praha: Nakladatelství Lidové noviny.
Kiparsky, Paul. 1973. ‘Elsewhere’ in phonology. In A festschrift for Morris Halle, ed.
Stephen Anderson and Paul Kiparsky, 93–106. New York: Holt, Rinehart & Winston.
Křivan, Jan. 2012. Komparativ v korpusu: explanace morfematické struktury českého
stupňování na základě frekvence tvarů. Slovo a slovesnost 73:13–45.
Marantz, Alec. 1995. A late note on late insertion. In Explorations in generative
grammar: A festschrift for dong-whee yang, ed. Young Sun Kim, Byung-Choon Lee,
Kyoung-Jae Lee, Kyun-Kwon Yang, and Jong-Kuri Yoon, 396–413. Seoul: Hankuk.
Marantz, Alec. 1996. Cat as a phrasal idiom: consequences of late insertion in Distributed
Morphology. Ms., MIT.
Marantz, Alec. 1997. No escape from syntax: Don’t try morphological analysis in the
privacy of your own lexicon. In University of Pennsylvania working papers in linguistics,
ed. Alexis Dimitriadis, Laura Siegel, Clarissa Surek-Clark, and Alexander
Williams, volume 4, 201–225. University of Pennsylvania.
Matushansky, Ora. 2013. More or better: On the derivation of synthetic comparatives
and superlatives in English. In Distributed morphology today: Morphemes for Morris
Halle, ed. Ora Matushansky and Alec Marantz, 59–78. Cambridge, MA: MIT Press.
Merchant, Jason. 2015. How much context is enough?: Two cases of span-conditioned
stem allomorphy. Linguistic Inquiry 46:273–303.
Miller, Philip, Geoffrey Pullum, and Arnold Zwicky. 1997. The principle of phonologyfree
syntax: four apparent counterexamples in French. Journal of Linguistics 33:67–
90.
27
Moskal, Beata, and Peter W. Smith. 2016. Towards a theory without adjacency: hypercontextual
VI-rules. Morphology 26:295–312.
Osolsobě, Klára. 2016. Komparativ. In Nový encyklopedický slovník češtiny, ed.
Petr Karlík, Marek Nekula, and Jana Pleskalová, 839. Praha: Nakladatelství Lidové
noviny.
Pfau, Roland. 2000. Features and categories in language production. Doctoral Dissertation,
Johann Wolfgang Goethe-Universität, Frankfurt am Main.
Pfau, Roland. 2009. Grammar as processor: a distributed morphology account of spontaneous
speech errors. Amsterdam: Benjamins.
Radkevich, Nina. 2010. On location: The structure of case and adpositions. dissertation,
University of Connecticut, Storrs, CT.
Ramchand, Gillian. 2008. Verb meaning and the lexicon. Cambridge, Massachusetts:
MIT Press.
Starke, Michal. 2009. Nanosyntax: A short primer to a new approach to language.
Nordlyd 36:1–6.
Starke, Michal. 2014. Why you can’t stay away from clitics. Talk presented at GIST 7,
Ghent, June 2014.
Starke, Michal. 2018. Complex left branches, spellout, and prefixes. In Exploring
nanosyntax, ed. Lena Baunaz, Karen De Clercq, Liliane Haegeman, and Eric Lander,
239–249. Oxford: Oxford University Press.
Svenonius, Peter. 2012. Spanning. Ms., CASTL.
Taraldsen, Knut Tarald. 2010. The nanosyntax of Nguni noun class prefixes and concords.
Lingua 120:1522–1548.
28
Williams, Edwin. 2003. Representation theory. Cambridge, Massachusetts: MIT Press.
Zwicky, Arnold. 1969. Phonological constraints in syntactic descriptions. Papers in
Linguistics 1:411–453.
Zwicky, Arnold, and Geoffrey Pullum. 1986. The principle of phonology-free syntax:
introductory remarks. Working papers in Linguistics 32:63–91.
29
Notes
1
Bobaljik adopts the adjacency condition on suppletion, and so the cmpr node triggering the application
of the rule (2a) appears be too far away from the , being separated from it by the intervening a-head.
Nevertheless, due to certain intricacies of the ABC suppletion pattern, there is a way to make sensitive
to the cmpr node despite the non-adjacency even in Bobaljik’s system, as reviewed in Caha (2017:sect.
3.1). We do not elaborate on this issue here as it is orthogonal to the main point.
2
This problem is analogous to the one noted by Marantz (1995). Our discussion of this issue is indebted
to Michal Starke (p.c.).
3
This could be a phase or some larger chunk of structure relevant for the locality of suppletion, see, e.g.
Embick (2010), Merchant (2015), Moskal and Smith (2016).
4
We leave it open whether mo-re spells out C1 and C2 separately, or as a portmanteau. We also leave it
open as to how exactly spell out applies here. It could be that C1 and C2 are turned into a constituent,
either by head-movement (Matushansky 2013), Local Dislocation (Embick 2007) or by Complex-Spec
formation (Caha et al. to appear). Alternatively, if some version of Spanning is adopted (Williams 2003,
Abels and Muriungi 2008, Taraldsen 2010, Dékány 2011, Svenonius 2012, Merchant 2015, Haugen and
Siddiqi 2016), more can simply span C1 and C2, as shown in (32). Similarly, we leave it aside how the
surface order is derived in (33), as this would take us too far afield, and is orthogonal to our concerns.
5
Should spellout fail, then little a would have to be spelled out on its own as a terminal.
6
Note that plus and plur are two shapes of a single root, with s undergoing rhotacism in intervocalic positions,
which happens also in the comparative when inflected, cf. plur-is ‘more, gen.sg.’
7
We take s in the superlative maks-im-us ‘biggest’ to be a part of the root, given that it is not a geminate like
the superlative S1 marker. The comparative ma-i-or ‘bigger’ could arise from the root mag-, as suggested
in Bobaljik (2012), with the root final g first assimilating to j (yielding maj-j-or), which is then reduced
due to degemination. Bobaljik (2012) concludes from this that this adjective has a regular AAA pattern,
and hence, that is is irrelevant for suppletion. However, this move requires the parsing of the positive as
mag-nus, which we see little evidence for. We therefore treat it as an ABC pattern (magn–ma(g)–maks).
30