Allomorphy in the Italian Passato Remoto:
A Distributed Morphology Analysis
Andrea Calabrese
Calabrese, Andrea(2012), “Allomorphy in the Italian Passato Remoto: A Distributed
Morphology Analysis,” Language & Information Society 18. Many Italian verbs display stem
alternations with highly idiosyncratic vocalic and consonantal allomorphy in the paradigm
of the simple past. Due to their complexity, these alternations are often used by linguists as
evidence for models assuming rote memorization of stem alternants and endings (see the
recent work by Maiden 2005, 2010 for example). In this paper, I will show that these alternations are characterized by basic regularities and that a rather simple analysis of them can be
formulated in the framework of Distributed Morphology (DM) (Halle and Marantz 1993;
Embick 2010; Embick and Marantz 2008). In particular, this analysis involves notions such
as roots, Vocabulary Items and Readjustment Rules as predicted by DM. There is no evidence
for memorized stem alternants. Instead, the allomorphy we see in the Italian Passato Remoto
is readily accounted for by providing an appropriate morphosyntactic analysis of the forms and
by deriving the irregular alternants from single underlying roots by means of Readjustment
Rules. This paper also shows that it is crucial that the rules accounting for this allomorphy
obey a strict definition of locality requiring linear adjacency. This explains why the presence of
I would like to thank Morris Halle, David Embick, Diego Pescarini, three anonymous reviewers
and especially Jonathan Bobaljik for their comments and helpful suggestions on a previous draft of
this paper.
University of Connecticut
Allomorphy in the Italian Passato Remoto: A Distributed Morphology Analysis | 1
the Thematic Vowel prevents their application. In the irregular forms where these rules apply,
the Thematic Vowel is removed by a special pruning rule. Crucially, Impoverishment removes
a diacritic index associated with the application of this pruning rule in certain morphosyntactic contexts (in the 1st and 2nd pl., and 2nd sg.), thus the regular basic form of the root will
appear in these contexts. Impoverishment in this case is motivated by a general Markedness
principle that disfavors complex exponence in morphologically marked environments.
keywords: allomorphy, stem alternations, Distributed Morphology, roots, vocabulary items,
readjustment rules, locality, impoverishment, morphological markedness, Italian
verbal morphology, Italian stress system
0. Introduction
Many Italian verbs display stem alternations with highly idiosyncratic vocalic and
consonantal allomorphy in the paradigm of the simple past, what is traditionally
called Passato Remoto in Italian grammars—in this paper I will use this term to refer
to this tense. Due to their complexity, these alternations are often used by linguists
as evidence for models assuming rote memorization of stem alternants and endings
(see the recent work by Maiden 2005, 2010 for example). In this paper I will show
that these alternations are characterized by basic regularities and that a rather simple
analysis of them can be formulated in the framework of Distributed Morphology
(Halle and Marantz 1993; Embick 2010; Embick and Marantz 2008). I will show
that the best synchronic analysis of the Italian Passato Remoto morphology involves
notions such as roots, vocabulary items and readjustment rules as predicted by DM.
In particular, I will propose that the irregular allomorphy we see in the Italian Passato
Remoto is readily accounted for by providing an appropriate morphosyntactic analysis of the forms and by deriving the alternants from single underlying roots by means
2 | 언어와 정보 사회 제18호
of Readjustment Rules, phonological operations operating under morphological conditions. No memorization of stem alternants and endings is needed.
A sample case of the allomorphy found in the irregular forms of the Passato
Remoto is given in (1) where we see alternations between 2nd and 3rd person plural
forms.
(1) perdeste
persero
‘lose-Past-2nd pl./3rd sg.’
In the 2nd pl., there is a thematic form of the verb—i.e., a thematic vowel (TV) follows the root1)—and the form of the root regularly appearing with other tenses is
found (cf. perd-e-t-e ‘pr.Ind -2nd pl.’, perd-e-v-a-te ‘Imp.Ind-2nd pl.’, perd-e-re ‘inf’).
In contrast, in 3rdpl., we have an athematic form of the verb—no thematic vowel is
present after the root and the stem undergoes changes in phonological shape, specifically addition of the consonant /s/ (cf. spensero, vissero)—actually an exponent of the
past Tense, as argued later—with subsequent deletion of the final root consonant /d/
(i.e., perd-s-e… → per-s-e-…). The 1st pl. and 2nd sg. behave like the 2nd pl.; in contrast, the 1st and the 3rdsg. behave like the 3rd pl. Therefore, we have the paradigmatic
alternations in (2):
(2) Singular
1
2
persi
perdesti
3
perse
Plural
1
perdemmo
2
perdeste
3
persero
‘lose’
There are 200 verbs in Italian that share this basic pattern, let’s call it a regularity
1) To understand alternations such as that in (1) an important feature of Italian verbal (and also nominal) morphology must be considered: the presence of a vowel, the so-called Thematic Vowel, after
the verbal root and other functional projections. This important aspect of Italian (and of Latin and
other Romance languages) will be discussed in detail below.
(i) perd-e-v-a-no/perd-e-ss-i-mo ‘lose-imperf3pl./ imperfSubj1pl.’
(schematically: [[[[perd]root -eTV-]V -v-T-aTV-T]- no]AGR/ [[[[perd]root -eTV-]V -ss-T-iTV-]T -mo]AGR)
Allomorphy in the Italian Passato Remoto: A Distributed Morphology Analysis | 3
in the irregularity: in the 1st and 3rdsg. and 3rdpl., they have an athematic form of
the verb and the root undergoes the idiosyncratic phonological changes characteristic of this tense. In the 1st and 2nd pl. and 2nd sg., there is a thematic form of the
verb and the form of the root regularly appearing with other tenses is found. Stress
shifts are also associated with these alternations. The irregular forms receive stress
on the root—the traditional literature calls them rhizotonic. The thematic forms
instead receive stress on the thematic vowel—these forms are traditionally called
arhizotonic.2)
(3) Irregular: [[[[pérd]root ___]V -s-]T i]AGR
pérsi
Athematic
vs.
Regular: [[[perd]root-éTV-]V -sti]T+AGR3)
perdésti
Thematic
Notice that despite this unifying regularity, the surface shape of irregular forms may
be quite different: specifically, the exponents of the past tense in the irregular forms
may be different, and different morphophonological operations may also apply in
their context. I provide a sample of the alternations found in the irregular forms of
the Passato Remoto in (4).
(4) i. Exponent /s/:
No change:
valeste/valsero ‘be worth’ (Root: val)
Deletion of final coronal stop:
perdeste/persero ‘lose’ (Root: perd)
Assimilation of final consonant: viveste/vissero ‘live’ (Root: viv)
Assimilation of final consonant + degemination:
speɲɲeste/spensero ‘turn off’ (Root: speɲɲ)
Changes in vowel quality:
espelleste/espulro ‘expell’ (Root: espell )
2) Traditional accounts assume that stress position is fundamental to accounting for the distribution
of regular/irregular stem forms. In section 5, I demonstrate that stress in verbal forms is predictable
and that its distribution is in part related to the presence /absence of Thematic Vowels. I defer discussion of stress to that section. I will not mark stress in the examples unless an accent is required
by the Italian orthography.
3) I assume that there is fusion between Tense and AGR in the regular form of the verb. See section 3
below for discussion.
4 | 언어와 정보 사회 제18호
ii. Exponent /Ø/:
a. With gemination of the final
root consonant (the consonant
is also rounded if dorsal):4)
b. With changes in vowel quality:
veniste/vɛnnero ‘come’ (Root: ven)
tat∫este/takkwero ‘be silent’ (Root: tak)
faceste/fet∫ero ‘make, do’ (Root: fak)
videro/vedeste ‘see’ (Root: ved)
The paper provides an account for how the special allomorphs characterizing the
irregular forms are derived. Here is a brief sketch of how the allomorphs are derived
(see section 3 for more details). As mentioned above, the irregular stem allomorphs
are athematic as in (5). The thematic vowel that is always present in regular morphology is removed by a pruning rule—which I assume to be a Readjustment Rule—as
in Embick and Halle’s (2005) analysis of irregular forms in the Latin Perfect, the
ancestor forms of the athematic forms in (5). The exponents for the past morpheme
are given in (6). As will be shown in section 3, (6a) causes gemination and secondary rounding of the preceding consonant, where rounding is preserved only in
dorsal stops (see (4iia)). The surface phonological shape of the roots is accounted for
through the application of Readjustment Rules. If one disregards vowel changes5)
and some other minor phonotactic adjustments, two simple phonological operations
account for the surface forms of the roots: one is coronal stop deletion and the other
one is consonantal assimilation, where coronal stop deletion applies before the other
rule. Simplified derivations involving these rules are given in (7). (The symbol ^ indicates that the two elements must be linearly adjacent for the rule to apply (see section
1)). The reasons for having a diacritic index on the root will be discussed in section 6.
4) In the analysis developed below, this exponent is not null but actually involves a skeletal position
associated with [+round] secondary articulation.
5) Vowel changes in the root require special rules of ablaut. For example, one implements vowel
fronting in /fat∫/:
(i) /fat∫/-/Ø/-/e/ ↑ vowel fronting ↑ [fet∫e]
Allomorphy in the Italian Passato Remoto: A Distributed Morphology Analysis | 5
(5) [[[[perd]root ]V -s-]T -e ]AGR
+pst 3sg.
X
↔ [+past]T / RootL ^__ , when RootL = ven, tak, dʒak, etc.
|
Labial
|
[+round]
b. s ↔ [+past]T / RootS ^__ , when RootS = mett, viv, muov, perd, etc.
c. Ø ↔ [+past]T / RootA ^__ , when RootA = fat∫, ved, etc.
(6) a.
(7) a. /ven/-/Xw/-/e/ → gemination ↑ vennwe → removal-rounding → [venne]
b. /tak/-/Xw/-/e/ → consonant gemination → [takkwe]
c. /perd/-/s/-/e/ → coronal stop deletion → [perse]
d. /prend/-/s/-/e/ → coronal deletion → [prense] → coronal deletion → [prese]
e. /viv/-/s/-/e/ → consonant assimilation → [visse]
The phonological shape of the large majority of the irregular forms of the Italian
Passato Remoto can be readily derived by means of theoretical primitives and rules
like those mentioned above.
In addition to showing how the allomorphs of the irregular forms are derived,
another important goal of this paper is accounting for why there are the alternations
according to person-number in (1)-(2). Following Embick (2003, 2010, 2011), I
assume that rules dealing with contextual allomorphy (both Vocabulary Item Rules
and Readjustment Rules) apply under strict locality conditions. In fact, the rules
needed to account for the allomorphy in the Italian Passato Remoto require linear adjacency between the root and past tense. This readily explains why no special
allomorphy occurs in the forms in which there is a thematic vowel. The intervening
thematic vowel disrupts the adjacency that is required for the application of these
rules. The unmodified form of the root therefore appears since only regular morphology vocabulary items can be inserted in this case and, in addition, no Readjustment
6 | 언어와 정보 사회 제18호
Rule can apply. One of the contributions of this paper is then to provide evidence for
morphological rules that require linear adjacency to apply. In section 1, I address the
question why not all morphological rules appear to undergo this requirement.
Another contribution made by this paper regards a novel use of the operation
of Impoverishment (Bobaljik 2003; Bonet 1991; Halle 1997; Halle and Marantz
1993; Harley 2008; Nevins 2011; Noyer 1992, 1998). The issue is to explain why
the 1st and 2nd pl. and 2nd sg. Passato Remoto forms are thematic, or more precisely,
to explain why the rule that prunes away the Thematic Vowel does not apply in
their case. I will account for this fact by proposing the following: 1) the application
of Readjustment Rules requires the presence of special diacritic indices. In particular,
a root diacritic index is necessary for the application of the Thematic Vowel Pruning
Rule. 2) Impoverishment deletes this diacritic index in 1st and 2nd pl. and 2nd sg.
forms. The Pruning Rule therefore does not apply. The Thematic Vowel stays and
regular morphology will be inserted as discussed above.
The operation impoverishing the diacritic, however, is not arbitrary. In fact, I
will hypothesize that the ultimate (diachronic) cause of this operation is a markedness principle: Brøndal’s (1940, 1943) “Principle of Compensation”. This principle
states that marked categories tend not to combine. In Calabrese (2011) I interpret
this principle as stating that marked categories cannot have idiosyncratic exponence
when in combination with other marked categories. Special Vocabulary Items
and Readjustment Rules indeed create idiosyncratic exponence. Now the Passato
Remoto, and persons such as the 1st and 2nd plural are marked morphological categories. Therefore, according to the Principle of Compensation, idiosyncratic exponence
in the 1st and 2nd plural of the Passato Remoto should be avoided (see section 8 for
a proposal regarding the 2nd sg.) I propose that this is obtained by impoverishing the
diacritic index required for the application of the Thematic Vowel Pruning Rule; the
Thematic Vowel therefore remains with the consequences discussed above. In secAllomorphy in the Italian Passato Remoto: A Distributed Morphology Analysis | 7
tion 5, I will discuss other cases from the Present and other tenses that show the same
markedness effect in the 1st and 2nd pl. As discussed in this section, the emergence of
regular morphology in the 1st and 2nd pl. in the Passato Remoto and these other
cases appears to be part of a general pattern characterizing Italian and Italian dialects. In fact, in many of these varieties, idiosyncratic exponence is avoided in the 1st
and 2nd plural by means of replacement of exponents (syncretism) or even removal/
omission of exponence (cf. in subject clitic systems (Calabrese 2011), in object clitics
(Calabrese 1995), in verbal inflections (Calabrese 2002)).
1. Distributed Morphology
The theory of Distributed Morphology (Embick 2010; Embick and Marantz
2008; Halle and Marantz 1993; Harley and Noyer 1999, among others) proposes
a piece-based view of word formation in which the syntax/morphology interface is
as transparent as possible. Distributed Morphology posits that there are two types
of primitive elements in the grammar that serve as the terminals of the syntactic
derivation, and, accordingly, as the primitives of word formation. These two types
of terminals correspond to the standard distinction between functional and lexical
categories (Embick and Halle 2005):6)
(8) a. Abstract Morphemes: These are composed exclusively of nonphonetic features, such
as [past] or [pl], the features that make up the terminal nodes of the syntax.
6) Abstract morphemes receive phonological features, i.e., exponents, by Vocabulary Insertion in
the morphological component (see below). Roots, instead, already have exponents and therefore
do not undergo Vocabulary Insertion. However, Root suppletion (see section 8 for an example)
is usually accounted for by assuming that suppletive roots undergo Vocabulary Insertion (see
Bobaljik 2012 for recent discussion). Some models of DM (see Marantz 1995) assume that all
roots undergo Vocabulary Insertion like abstract morphemes. The issue of whether or not roots
undergo Vocabulary insertion is not relevant for the analysis developed in this paper and will not
8 | 언어와 정보 사회 제18호
b. Roots: These make up the open-class vocabulary. They include items such as the
verbal roots /ven/, /mett/, /prend/, etc., which are sequences of complexes of phonetic
features, along with abstract indices (to distinguish homophones) and other diacritics
(e.g., class features).
Distributed Morphology conceives of the architecture of the grammar as sketched
in (9), in which Morphology refers to a sequence of operations that apply during the
PF derivation, operations that apply to the output of the syntactic derivation. This
theory is in its essence a syntactic theory of morphology, where the basic building
blocks of both syntax and morphology are the primitives in (8). There is no Lexicon
distinct from the syntax where word formation takes place; rather, the default case is
one in which morphological structure simply is syntactic structure.
(9) The Grammar
(Syntactic Derivation)
|
Morphology
PF
LF
The derivation of all forms takes place in accordance with the architecture in (9).
Roots and abstract morphemes are combined into larger syntactic objects, which are
moved when necessary (Merge, Move).
The Latin verbal form in (10) illustrates how a complex verbal form is derived.
This form has the syntactic structure in (11), where as proposed by Oltra-Massuet
(1999), Oltra-Massuet and Arregi (2005), Embick and Halle (2005), every functional/lexical projection has a Thematic Vowel. Thematic Vowels (TV) are special morphological elements adjoined to certain functional heads in morphological structure.
be discussed further here.
Allomorphy in the Italian Passato Remoto: A Distributed Morphology Analysis | 9
(10) laud - a: + u-i + s + s-e: + mus
‘praised-Perf-Past-Subj-1st Pl’
|
(11)
|
MOOD
|
V7)
|
√Root
laud
V
|
|
ASP
|
|
MOOD
T
|
| ASP | T
TV ASP
|
|
T
|
|
AGR
|
|
MOOD
MOOD
TV
TV
TV
Structures like those in (12) determine the terminal nodes to which phonological
realization is provided by the vocabulary items. Each terminal node consists of syntactico-semantic features drawn from the set made available by Universal Grammar.
In the simplest case, Morphology linearizes the hierarchical structure generated by
the syntax by adding phonological material to the abstract morphemes in a process
called Vocabulary Insertion. Therefore, the (abstract) morphosyntactic representation
is the input to a morphological component that assigns phonological realizations
to the terminal nodes (Late Insertion). During Vocabulary Insertion, individual
7) I assume, following Marantz (1997, 1999), that roots have no category and that in the syntax, they
are merged with category-giving functional heads. In the verbal domain, this head is v (Harley
1995; Chomsky 1995). This head is responsible for the verbal properties of the verbal complex.
In nonverbal environments, this head is n for nouns and a for adjectives (see Marantz 1999;
McGinnis 2000).
(i)
V
/ \
Root v
N
/ \
Root
n
A
/ \
Root a
However, as discussed by Embick (2010, 2011), this category-giving functional head, especially if
phonologically empty as in all of the cases discussed in this paper, is systematically transparent to
Vocabulary insertion and Readjustment rules—perhaps it is automatically pruned away once the
lexical category cycle (V/N/A) cycle has run through; it therefore does not play any role in defining
adjacency or locality for them. For this reason and also for the sake of representational simplicity I
will omit these functional heads in the morphosyntactic representation of verbal form in this paper.
10 | 언어와 정보 사회 제18호
Vocabulary Items—rules that match a phonological exponent with a morphosyntactic
context—are consulted, and the most specific rule that can apply to an abstract morpheme applies (in the so-called Elsewhere (Subset, Paninian) ordering). Vocabulary
Items are essentially instructions that insert phonological material in a terminal node
given certain specific feature configurations in the terminal node and its adjacent
environment. Abstract morphemes are thus said to be spelled out during Vocabulary
Insertion.
Thus, morphological pieces such as those in (11) are filled in by exponents via the
Vocabulary Items in (12):8)
(12) a. Root:
Laud
Vocabulary items:
TV
TV
TV
TV
b. [+perf]
[+past]
[+subj]
↔
↔
↔
c. [+auth, +part, +pl.]
↔
↔
↔
↔
a/
i/
Ø/
e/
v
s
s
/ [+past] ___
Root V____
[+perf]___
[+past] ___
[+subj] ___
↔
mus
8) The person features I adopt in this paper are given in (i):
(i) Person Features (see Bobaljik 2008 for a different system of person features)
2nd
3rd
1st
Author of speech act
+
Participant of speech act
+
+
Allomorphy in the Italian Passato Remoto: A Distributed Morphology Analysis | 11
(13)
MOOD
MOOD
|
V
|
√Root
laud
V
ASP
|
ASP
|
|
TV ASP
| |
a
v
|
|
TV
|
i
T
|
T
|
|
AGR
|
T
TV
s
Ø
| |
|
| ||
MOOD
|
MOOD
s
|
TV
e
mus
Vocabulary Insertion proceeds cyclically from the inside out, beginning with the
root node: Inner nodes can’t see outer nodes’ phonology, but outer nodes can see all
the features of inner nodes. (Halle and Marantz 1993; Bobaljik 2000; Embick 2010).
In addition, it is assumed that morphological operations operate under strict conditions of Locality; in certain configurations, a trigger of a given rule may be too far
away to trigger that rule.
Notice at this point that the phonological information contained in the exponents inserted by Vocabulary Items is not sufficient to ensure that in all cases the
correct phonological output will be generated. In addition to the Vocabulary Items,
a number of further rules that alter the phonology of the exponents are required.
Such rules are called Readjustment Rules in Distributed Morphology. Readjustment
Rules are phonological rules. Their distinguishing property is, however, that they are
conditioned by both morphosyntactic and Root/morpheme-specific information. In
this way, Readjustment Rules differ from other rules of the phonology that require
no reference to morphosyntactic environments; nor to diacritics specifying whether a
given root/morpheme undergoes or triggers them.
An example of a Readjustment Rule applying in the Passato Remoto of irregular
12 | 언어와 정보 사회 제18호
verbs is the rule that changes the quality of the root vowel in a form such as 3rd pl.
videro (cf. vedesti 2nd pl) ‘see’. The rule is given in (14). Crucially this must be restricted to apply to the root vowel of roots such ved ‘see’, met ‘put’, fond ‘fuse’ which will
be marked by a special diacritic (see section 6 for discussion of the use of diacritics):
(14)
N
|
X
[-cons] → [+high] / ___ C]Rooti ^ Past
(e, o → i, u)
Rooti = ved, mett, fond
Locality will be of particular importance in the analysis developed here. An important issue is how locality is defined. Nevins (2010), along the lines of Rizzi (1990,
2001), proposes that Locality is not absolute but must always be relativized according
to some given parameter that establishes what is relevant in defining it. I adopt this
idea here. An example can be provided by long distance operations in Phonology
when analyzed in terms of Visibility theory (Calabrese 2005).9) According to this
theory, processes in search of a target may disregard/“not see” certain sets of features,
specifically non-contrastive or unmarked features, and achieve locality in this way.
Consider lateral dissimilation in Latin. According to this process, the /l/ of the suffix
/-alis/ dissimilates to /r/ when preceded by another /l/ (cf. semin-alis vs. aliment-aris
unless an /r/ intervenes between them (cf. littor-alis). The rule is defined as operating
only on contrastive specification for [lateral]. Therefore, the non-contrastive [-lateral]
specifications of non-sonorant or nasal consonants are disregarded by the process (cf.
aliment-aris, milit-aris). We can then say that the process applies locally between the
contrastive feature specifications [+lateral] on the lateral plane disregarding the non-
9) Visibility theory proposes an alternative account for facts that were previously accounted for in
phonology by assuming feature underspecification. Feature underspecification has been shown to
lead to various problems and has been abandoned in recent models of phonology (see Clements
2001; Calabrese 2005; Dresher 2009; Mohanan 1991; Nevins 2010, among others for discussion).
Allomorphy in the Italian Passato Remoto: A Distributed Morphology Analysis | 13
contrastive specifications [-lateral] (those of [m], [n] and [t]) which are shaded in (16)
and (17).
(15) [+lateral] > [-lateral] / [+lateral] ___ in the suffix -alis
(Parameter: Access only contrastive features.)
(16) a l i m e n t -a l i s > (by (15)) > a l i m e n t -a r i s
| |
| |
|
| |
| |
|
+lat -lat -lat -lat +lat
+lat -lat -lat -lat -lat
However, if the contrastive specification [-lateral] intervenes between them, the process can no longer apply.
(17) l i tt o r - a l i s
|
| | |
+lat -lat -lat +lat
Morphological operations can be defined in the same way so that certain aspects of
the morphological representations may be disregarded to achieve locality. The obvious question is that of the parameters governing locality in morphological operations.
In this paper I will focus on a basic parameter that appears to be governing morphological operations: featural locality vs. linear adjacency, and I will not investigate
other additional parameters that may intervene.
Morphemes have a bipartite structure in which a phonetic index—an exponent—is associated with a set of morphological (syntactic/semantic/lexical) features.
Morphological rules, then, may refer to these two components of the morphemes:
the features or the exponents. Different definitions of locality follow from this. I propose that there are rules that look at the featural organization of the string, and others
that look at the string’s composition in terms of exponents.10) The latter rules require
10) If one assumes that this holds only for overt exponents, morphemes with null, non-overt exponents will be disregarded by rules of this type.
14 | 언어와 정보 사회 제18호
surface linear adjacency as a condition for their application; the former look only for
adjacency between feature containing nodes or, better, between nodes containing
relevant features. These rules may disregard surface exponents lacking features or,
better, relevant features.
The Vocabulary Insertion Rules characterizing irregular morphology in the
Passato Remoto are a clear instance of rules requiring linear adjacency: as we will
see, they apply only if there is this type of adjacency between the location of their
application and the irregular root. If this adjacency is disrupted by the presence of a
Thematic Vowel, only regular unmarked morphology will occur. For example, the
special exponent /-s-/ for the[+past]Tense of some of the irregular Passato Remoto
forms is inserted by the Vocabulary Item in (18). This rule applies only when Tense
is linearly adjacent to the Root (Remember that this requirement is represented here
by the symbol ^.):
(18) VI rule for the Passato Remoto Tense morpheme (provisional)
/s/ ↔ [+Past]T / Roots ^____ (where Roots = corr, val, perd, met, etc.)
Given the linear adjacency requirements, (18) can apply only when the Thematic
Vowel is missing. So it can apply in the structure in (19) but not in (20) where the
TV is present:
(19) a. [[[[ corrS ]root
b. [[[[ corrS ]root
c. cor-s-i
] +PAST TV ]T +part, -+auth, -pl] AGR
]
s
]T i ] AGR
(20) a. [[[[ corrS ]root TV] +PAST TV ]T +part, +auth, +sg] AGR
b. [[[[ scriv ]root e]
s
]T i ] AGR
c. corr-e-s-i
A different type of vocabulary insertion rule is found in the case of the Latin Perfect
Allomorphy in the Italian Passato Remoto: A Distributed Morphology Analysis | 15
AGR suffix. Embick (2011) shows that the special perfect 2nd sg. AGR ending /-sti/
of Latin appears only when no other Tense or Mood morpheme appears between
AGR and [perf] Aspect. Otherwise, as can be seen in (21), the Agreement morphemes are those found elsewhere in the system (i.e., /-s/↔2s):
(21) amā-v-i-stī vs. amā-v-e-rā-s, amā-v-e-rĭ-s, amā-v-i-s-s-e-s ‘love-2Sg.’
Perf. Ind. Pluperf. Ind.
Perf. Subj. Pluperf. Subj.
Cf. amā-s amā-b-ā-s
‘love-2Sg.’
Pres. Ind.
Imperf. Ind.
The problem is that a theme vowel /-i-/ appears between the Perfect and AGR.
Therefore there is no linear adjacency between these morphemes. I propose that in
this case the Vocabulary Item is inserted by looking at the closest relevant feature
without requiring linear adjacency:
(22) /-stis / ↔ 2p / [perf.] __
In the form /amā-v-i-stī/, the feature [Perfect] is the closest feature to AGR insofar as
the Theme vowel does not contain any features. Locality is therefore met although
not in terms of linear adjacency.
We can now turn to the issue of allomorphy generated by Readjustment Rules. As
discussed below, the Readjustment Rules accounting for the allomorphy we observe
in the Passato Remoto obey linear adjacency. But this is only one of the parametric
options available for morphological rules. So there can be Readjustment Rules that
obey a different type of locality. In fact, Embick (2010, 2011), following Kiparsky
(1996), provides examples of Readjustment Rules that do not appear to obey linear
adjacency. An example is metaphony in Italian dialects:
16 | 언어와 정보 사회 제18호
(23) Metaphony triggered by AGR in the dialect of Ischia, Campania
Pr. ind.
Impf. ind.
1s kand-ə
kand-a-v-ə
‘sing’
2s kɛnd-ə
kand-ɛ-v-ə
3s kand-ə
kand-a-v-ə
The most adequate analysis of these alternations involves a Readjustment Rule
raising the stressed vowel triggered by 2nd sg. AGR, which is realized as [ə] as the
other AGR suffixes. The rule is fully morphologized in also having exceptions and
lexical conditioning. However, this readjustment rule applies over intervening
Imperfective tense morpheme -v in kand-ɛ-v-ə (Root-TV-T-AGR). Nonetheless,
although this rule violates linear adjacency, it can still be considered as applying
locally if we see locality in relativized terms. The point is that the metaphony rule
must be defined as applying to a stressed vowel. If we assume that the target of
metaphony must be contained in the “closest” morpheme that satisfies the rule
description, we have a simple account for how adjacency is violated. The “intervening” morpheme here contains only a consonant which is not a possible target of metaphony. Therefore, this morpheme can be disregarded.11) If a morpheme contains a
vowel, it cannot be skipped. Another similar case reported in the literature is from Zulu, where the palatalization of labial consonants triggered by the passive suffix -w skips the intervening causative morpheme -is (Carstairs 1987; Carstairs-McCarthy 1992):
11) If we look at the rule from the point of view of the trigger, one can say that morphological
rules apply to the ‘closest’ morpheme that can undergo the rule. If, on the contrary, we look at the
rule from the point of view of the target, the rule is triggered by the closest morpheme that contains a possible trigger. Therefore in a situation like that in (i):
(i) target(m)-X-trigger(m)
trigger-target interactions are possible only when X is neither a possible undergoer nor a trigger.
Allomorphy in the Italian Passato Remoto: A Distributed Morphology Analysis | 17
(24) Zulu palatalization of labials
Active
a. bamb-a ‘catch’
boph-a ‘tie’
b. bamb-is-a ‘cause to catch’
boph-is-a ‘cause to tie’
Passive
banj-wa ‘be caught’
bosh-wa ‘be tied’
banj-is-wa ‘be caused to catch’
bosh-is-wa ‘be caused to tie’
The idea proposed above can be easily extended to this case once we observe that the
target can only be a labial consonant. Therefore, the Readjustment Rule of Zulu targets the closest morpheme containing a labial.
The Italian Passato Remoto rules are not of this type, though. They indeed require
linear adjacency. As discussed above, this follows from a different parametric choice:
These rules are defined as requiring linear adjacency. Given this requirement, an
intervening TV blocks their application.
It is an open question whether or not there are principled reasons for the selection
of the different parametric choices defining locality in allomorphic processes. I do not
believe so. As in the case of the different parametric options defining locality in the
case of phonologic rules, I have the feeling that the different choices may have their
causes in diachronic idiosyncracies. Further research must establish if this is indeed
correct and investigate what happens in the historical development of these processes.
2. Basic Properties of Italian Verbal Morphosyntax
The basic morphosyntactic structure of Italian verbs is given in (25):12)
12) A morphosyntactic change occurred in the development of the Romance languages as can be seen
in (i) where I compare the Latin pluperfect subjunctive in (ia) (= (11)) with the form that historically derived from it in Italian, i.e., the Imperfect subjunctive (ib):
18 | 언어와 정보 사회 제18호
(25)
T
V
T
T
|
T
ROOT
AGR
The structure in (25) then undergoes Thematic Vowel insertion. Thematic Vowels
are adjoined to V and Tense heads as in (26):
(26)
T
V
T
|
V
|
TV1
ROOT
T
T
AGR
TV2
(26) accounts for the morphological structure of the imperfect forms in (27) (and for
the imperfect subjunctive in (23b)).
(i)
a. laud - a: + u-i
+s
b. lod -a+ss-i-
+ s-e:
+ mus
+ mo
‘praise-PlprfSubj1pl’
‘praise-ImpSubj1pl’
In Italian, functional categories such as aspect, tense and mood are no longer represented as
independent morphological pieces as they were in Latin. Instead, a single morpheme appears in
their place. This morphosyntactic development and its semantic consequences are irrelevant to
my analysis here. I will simply assume that the Asp, Tense and Mood nodes are fused together in
Italian (i.e., Tense in (25) = Aspect+Tense+Mood). Another issue that I will put aside is whether or
not V-to-V movement is needed to derive (25).
Allomorphy in the Italian Passato Remoto: A Distributed Morphology Analysis | 19
(27) Italian imperfect indicative
AMARE ‘love’:
am-a-v-o am-a-v-i am-a-v-a
BATTERE ‘beat’:
batt-e-v-o batt-e-v-i batt-e-v-a
PARTIRE ‘leave’:
part-i-v-o part-i-v-i part-i-v-a
1
2
3
Singular
am-a-v-a-mo am-a-v-a-te am-a-v-a-no
batt-e-v-a-mo batt-e-v-a-te batt-e-v-a-no
part-i-v-a-mo part-i-v-a-te part-i-v-a-no
1
2
3
Plural
In the Present Tense (see (28)), there is no evidence for an independent Tense morpheme. The same situation is found in Spanish, and to account for this fact, OltraMassuet and Arregi (2005) propose a process of fusion merging Tense and AGR into
a single node. I propose that the same process of fusion applies in Italian, changing
(25) into (29):
(28) Italian present indicative
AMARE ‘love’:
am-o am-i am-a BATTERE ‘beat’:
batt-o batt-i batt-e PARTIRE ‘leave’:
part-o
part-i part-e 1
2
3
Singular
am-ia-mo am-a-te am-a-no
batt-ia-mo batt-e-te batt-o-no
part-ia-mo 1
Plural
part-i-te 2
part-o-no
3
(29) Fusion between Tense and AGR:
T
V
|
V
|
ROOT
TV
20 | 언어와 정보 사회 제18호
T
|
T+ AGR
Observe that the Fusion process must apply before the insertion of the Tense
Thematic Vowel, otherwise the Thematic Vowel position would interfere with the
fusion process. Furthermore, as discussed later at the end of section 3, a cyclic implementation of fusion and TV insertion appears to be required.
The following vocabulary items account for verbal inflections in the Italian present
and imperfect tenses of regular verbs. In the case of the Vocabulary Items in (30) the
Head can include a root or the head of a functional projection (cf. (32)):
(30) TV13)
→
(31) AGR Suffixes:
a. /-mo/
b. /-te/
c. /-no/
d. /-o/
e. /-i/
f. /Ø/
↔
↔
↔
↔
↔
↔
/-a-/
/-e-/
/-i-/
Head a ____
Head e ____
Head i ____
[+author, +plural]AGR
[+participant, +plural]AGR
[+plural]AGR
[+author]AGR / [-subjunctive]T _____
[+participant]AGR / [-subjunctive]T _____
[-participant]AGR
13) According to (30) we should expect the forms in (i) for the 1st pl. present of all conjugations and
the 3rd pl. present of the /e/ and /i/ conjugations:
(i) a. am-a-mo
b.
batt-e-mo part-i-mo
batt-e-no part-i-no
We instead have the forms in (ii):
(ii) a. am-ya-mo batt-ya-mo par-ya-mo
b.
batt-o-no part-o-no
Further vocabulary items accounting for the surface form of the Thematic Vowels we see in (ii) are
given in (iii) and (iv).
(iii) TV ↔ ya / [ ___ ]TV^[PRES, +part, +auth, +pl]AGR+T
(TV is [ya] in the 1st pl. of present tense of all conjugation.)
(iv) TV ↔ o / Roote & i [ ___ ]TV ^[PRES, -part, -auth, +pl]AGR+T
(TV is [o] in the 3rd pl. of present tense of II and III conjugation.)
Allomorphy in the Italian Passato Remoto: A Distributed Morphology Analysis | 21
(32) Tense Exponents (the subscript a indicates that the imperfect Thematic Vowel is /a/ by
(30)) :
/-v a -/
↔
[+imperfect]tense
A phonological rule that is very important in accounting for the surface distribution
of the Thematic Vowels is that in (33) which deletes Thematic Vowels before vowelinitial suffixes. Sample applications of this rule are given in (34). Given rule (33) the
verb, Thematic Vowel in the present indicative and the Tense Thematic Vowel in
the imperfect indicative are deleted before the suffix /-o/.
(33) V
→
Ø / [TV ___ ] + [V
(34) [[[am]-a]-o]
[[[[am]-a]-v-a-o]
→
→
amo
amavo
‘love-PresInd-1sg.’
‘love-ImperfInd-1sg.’
3. The Passato Remoto Forms of Regular Verbs
Before dealing with the allomorphy characterizing the irregular forms of the
Passato Remoto let us consider the regular forms of this tense in Italian.
22 | 언어와 정보 사회 제18호
(35) Passato Remoto of Regular Verbs
AMARE ‘love’:
am-a-i14) am-a-sti am-ò
BATTERE ‘beat’:
batt-e-i batt-e-sti batt-è
PARTIRE ‘leave’:
part-i-i part-i-sti part-ì
TEMERE ‘be afraid’
tem-e-i tem-e-sti tem-è
1
2
3
Singular
am-a-mmo
am-a-ste am-a-ro-no
batt-e-mmo
batt-e-ste batt-e-ro-no
part-i-mmo
part-i-ste part-i-ro-no
tem-e-mmo
1
Plural
tem-e-ste 2
tem-e-ro-no
3
If we compare the regular forms of the Passato Remoto in (35) to those of the indicative present (28) and the imperfect (27), this Tense appears to have only 3 different
pieces like the present, not 5 like the imperfect (see (26)). One could assume a null
exponent for Tense as in (36):
(36) [[[[part]root i]V -Ø-]T ste]AGR
But in the structure in (36) we would also expect to find a Tense Thematic Vowel.
However, it is missing. The best account for this fact is to assume that the regular
forms of the Passato Remoto do not have the structure in (26) but actually that in
(29), repeated here as (38), as in the Present. Therefore, also in the Passato Remoto
of the regular verbs, Tense and AGR undergo fusion (see Oltra-Massuet and Arregi
2005 for a similar proposal for Spanish):
(37) a. tem-e-ste
b. tem-e-te
c. tem-e-v-a-te
14) The 1st sg. of the Passato Remoto displays a systematic exception to (33). In this person, the verb’s
Thematic Vowel is never deleted before the following vowel initial suffix /-i/. Rule (33) applies as
usual in the 3rd sg. (am-a-ò → am-ò).
Allomorphy in the Italian Passato Remoto: A Distributed Morphology Analysis | 23
(38)
T
V
|
V
|
ROOT
T
|
T+ AGR
TV1
This analysis accounts for the fact that neither a tense morpheme nor a related
Thematic Vowel are present in the regular forms of the Passato Remoto. Evidence
for this analysis is provided by the distributional properties of AGR suffixes. Notice
that the other past forms in the verb conjugation share AGR suffixes that are different from those found in the regular forms of the Passato Remoto (35). Consider for
example the 3rd sg. and pl. of the imperfect subjunctive and of the conditional where
we have the suffixes: /-e/, /-e-ro/:
(39) Imperfect Subjunctive
am-a-ss-e
‘love-3sg.’
am-a-ss-e-ro
‘love-3pl.’
(40) Conditional
am-er-ebb-e
am-er-ebb-e-ro
These suffixes also appear in the irregular forms of the Passato Remoto:
(41) venn-e
venn-e-ro
‘come-3sg’
‘come-3pl’
The non-appearance of these suffixes in the regular forms of the Passato Remoto can
be accounted for by saying that the suffixes found in the regular forms have a more
restricted distribution than those found in (39), (40) and (41). In fact, notice that
24 | 언어와 정보 사회 제18호
some of these suffixes require Thematic Vowel information like the 3rdsg. /-ò/ of the
/-a/-conjugation, (the so-called 1st conjugation), the 3rd sg. preaccenting /-Ø/ of the
/e/ and /i/-conjugations (the 2nd and 3rd conjugations), and also the augment /no/ for
the 3rd plural which appears only in the Passato Remoto of the regular verbs. Now, if
we assume the locality requirement—in any of its forms: either linear or featural adjacency—for morpheme-morpheme interactions, access to Thematic Vowel information would be impossible for a Vocabulary Item realizing AGR since the Tense node,
which contains a relevant [+past] feature, intervenes between them. However, if we
assume fusion between Tense and AGR, adjacency between the insertion site of the
special Vocabulary Items and the Thematic Vowel is achieved. Therefore, the presence of special suffixes requiring Thematic Vowel information in the regular forms
of the Passato Remoto provides independent evidence for the fusion between Tense
and AGR in (38).
The vocabulary items for the AGR terminal node are given in (42). Notice that,
in the 3rd sg. of the irregular forms, one finds the 3rd person exponent /Ø/ in (31f),
which is also characteristic of other tenses. I assume that the ending /-e/ we see in
(41) is actually the Tense Thematic Vowel (see (44) below). The symbol ] appearing
in (42) is a prosodic bracket causing stress assignment on the suffix (i.e., amò ‘lovePast-3rd sg’) or on the preceding TV (battè ‘beat-Past-3rd sg’, temè ‘fear-Past-3rd’, partì
‘leave-Past-3rd’) (see section 5 for discussion of the stress algorithm in Italian verbs).
The rules in (42a-b) apply only in regular morphology. This is indicated by the context TVa ]V_____ (The presence of the Thematic Vowel is in fact the characterizing
feature of regular morphology, as discussed above.). The other rules apply to all verbs:
Allomorphy in the Italian Passato Remoto: A Distributed Morphology Analysis | 25
(42) VIs for the AGR morpheme
]
a. /-o/
↔ [-part, +past]T+AGR / TVa ]V^ _____
]
b. /-Ø /
↔ [-part, +past]T+AGR / TV] V^ _____
c. /-mmo/ ↔ [+author, +plural]/ [+past] ____
d. /-ste/
↔ [+participant, +plural]/ [ +past] ____
e. /-sti/
↔ [+participant, -author]/ [ +past] _____
f. /-i/
↔ [+author]/ [+past] _____
g. /-ro/
↔ [-participant, +plural]/ [+past] _____
The suffix /-no/ that appears in the 3rd pl. of regular Passato Remoto forms is
accounted for in the following way: there is fission of [+plural] in [-part, +plural,
+past] in the context TV]V___; then application of (30c) leads to the insertion of
/-no/ (see Halle 1997; Noyer 1992 on morphological fission).
Above, I proposed that in the Passato Remoto of the regular verbs Tense and
AGR undergo fusion like in the present, as in (38), to account for the fact that no
overt tense morpheme appears in them. However, unlike the regular forms of the
Passato Remoto, irregular forms—which are athematic, as discussed below—do
show such an overt morpheme for this tense. In fact, forms like those in (43) argue
for a morphological segmentation in which /s/ is the exponent of the Past with the
additional application of the Readjustment Rules discussed in section 6.
(43) val-e
ett∫ɛll-e korr-e
speɲɲ-e
pɛrd-e
voldʒ-e
val-s-e
ett∫ɛll-s-e [ett∫ɛlse]
korr-s-e [korse]
speɲɲ-s-e [spese]
perd-s-e [perse]
voldʒ-s-e [volse]
‘be worth’
‘excel’
‘run’
‘turn off’
‘lose’
‘turn’
I assume that these forms have the constituent structure in (44) (after Verb TV pruning (see below)):
26 | 언어와 정보 사회 제18호
(44)
T
T
T
V
|
ROOT
korr
(cf. korr
T
TV
s
s
e
e
|
|
AGR
|
Ø (3rd sg.)
ro (3rd pl.))
The crucial aspect of the structure in (44) is the absence of the Verb Thematic
Vowel. Hence I propose that the Tense+AGR fusion is restricted to Thematic constructions as in (45):15)
(45) T
|
+past
AGR
→
T+AGR
|
+past
/
TV]V
______
Notice that once we assume the structure in (44), we have an account for why the
default AGR suffix /Ø-/ (31f) appears in the 3rd sg. As mentioned above, the special
items listed in (42a) appears only when linearly adjacent to the verb Thematic Vowel
(am-a-ò (surface am-ò)), and therefore not to the Tense Thematic Vowel, like the
AGR suffix in (31f).
The Vocabulary Item for the Passato Remoto forms in (44) is given in (46); other
two VIs will be discussed later. The reasons for having a diacritic index on the root
will become clear in section 6.
15) It follows that Fusion in (45) must apply after TV pruning in (49) which I assume to be a readjustment rule. A consequence of this is that (45) should also be considered a Readjustment Rule.
Allomorphy in the Italian Passato Remoto: A Distributed Morphology Analysis | 27
(46) VI for the Passato Remoto Tense morpheme (provisional)
/s/ ↔ [+Past]T / Roots^ ____ (where Roots = corr, val, perd, met, etc.)
I will close this section by turning again to the relative ordering of Fusion and
Thematic Vowel insertion. It is obvious that the Fusion process must apply before
the insertion of the Tense Thematic Vowel: otherwise the TV would prevent the
fusion process. However, as proposed above, the presence of the Root TV is crucial
for the application of Fusion in the past. This calls for a cyclic implementation of TV
insertion and fusion: first, at the root node a TV is appended, and then (39) applies
at the root node (if its conditions are met, i.e., the presence of a special diacritic index
(see section 4). Then, on the next cycle (Tense) either Fusion applies, or, if its conditions are not met, then TV is appended to (unfused) Tense.
4. The Passato Remoto Forms of Irregular Verbs
Let us turn to the irregular forms of the Passato Remoto. Sample cases are given in
(47):
(47) Singular
1
valsi
persi
misi
scrissi
fusi
venni
takkwi
2
valeste
perdesti
mettesti
scrivesti
fondesti
venisti
tat∫esti
3
valse
perse
mise
scrisse
fuse
venne
takkwe
Plural
1
valemmo
perdemmo
mettemmo
scrivemmo
fondemmo
venimmo
tat∫emmo
2
valeste
perdeste
metteste
scriveste
fondeste
veniste
tat∫este
3
valsero
persero
misero
scrissero
fusero
vennero
takkwerò
‘be worth’
‘lose’
‘put’
‘write’
‘fuse’
‘come’
‘be silent’
As observed earlier, if we look at the paradigm in (47) a striking regularity can be
28 | 언어와 정보 사회 제18호
observed. In the 1st and 3rd sg. and 3rd pl., we have an athematic form of the verb and
the root appears in an idiosyncratic irregular form. In the 1st and 2nd pl. and 2nd sg.,
there is a thematic form of the verb and the form of the root regularly appearing with
other tenses is found.
I will account for the alternations between the athematic and thematic forms in
(47) in section 6. For now, let us consider the phonological changes affecting the
athematic forms of the verbs. The first issue to address is the absence of Thematic
Vowels in these verbs. We saw that the irregular forms of the Passato Remoto are
athematic:
(48) Regular
[[[batt]V -e]TV-i]T+AGR
vs.
Irregular16)
[[[corr]V -s]T -i]AGR / [[[perd]V-s]T -i]AGR
The absence of the Thematic Vowel is accounted for by a rule of pruning in (49).
Rule (49) removes the Thematic Vowel after certain roots, those that have irregular
forms in the past. I propose that these roots are assigned a special diacritic [+TV-pruning].
The reason for assuming a special diacritic index for these roots will be discussed in
section 6. Observe that only roots of the e- and i- conjugations undergo Thematic
Vowel Pruning. Given that the rule in (49) is sensitive to root-specific information, I
assume it to be a Readjustment Rule:
(49)
V
V
TV / Root-e/i[+TV-pruning]^___ Past, Root[+TV-pruning]= korr, prend, met, ven, etc.
Thus given (50), we obtain (51) due to the application of (49):
16) Actually, the forms below contain an underlying Tense Thematic Vowel that is deleted before the
suffixal V by (33): [[[corr]V -s-e-]T -i]AGR / [[[perd]V-s-e-]T -i]AGR. Remember that the 1st sg. of the
regular Passato Remoto is an exception to rule (33).
Allomorphy in the Italian Passato Remoto: A Distributed Morphology Analysis | 29
(50)
|
V
|
V
|
ROOT[+TV-prn]
(51)
T
|
TV1
T
|
|
T
|
AGR
T
T
T
| |
AGR
V
|
ROOT[+TV-prn]
In addition to being athematic, however, irregular verbs undergo various phonological changes. For example, coronal [+anterior] stops /t, d/ and /n/ are deleted
before /s/. This accounts for metteva/mise (root: mett ‘put’), kyudeva/kyuse (root: kyud
‘know’), utt∫ideva/utt∫ise (root: utt∫id ‘kill’). Another rule that is crucial to account
for the allomorphy we see in the irregular past forms is the rule of /s/- assimilation.
It applies to obstruents (both stops and fricatives) and to nasals but not to liquids:
condut∫eva/condusse (root: konduk ‘conduct’), skonfiddʒeva/skonfisse (root: skonfigg
‘defeat’), diridʒeva/dirisse (root: dirig ‘direct’), komprimeva/comprɛsse (root: komprim
‘compress’), kwɔtƒeva/kɔsse (root: kwɔt∫ ‘cook’), mwɔveva/mosse (root: mwɔv ‘move’),
viveva/visse (root: viv ‘live’), skriveva/skrisse (root: skriv ‘write’). Another rule deletes
the root nasal in the Passato Remoto: rompeva/ruppe, interrompeva/interruppe. Other
rules change the quality of the root vowel: fat∫eva/fet∫e, vedeva/vide, rompeva/ruppe.
In Calabrese (2012), I provide a detailed account of these rules and how they apply.
Here, I will focus only on another change characterizing the irregular morphology:
30 | 언어와 정보 사회 제18호
gemination. Consider sample forms like those in (52):
(52) Imperf.
vɛniva
kadeva
voleva
not∫eva
tat∫eva
Passato Remoto
vɛnne kadde
volle
nokkwe
takkwe
Root
vɛn
kad
vol
nok
tak
‘come’
‘fall’
‘want’
‘harm’
‘be silent’
Notice that the gemination of the final consonant in these forms indicates the
presence of an extra skeletal position. The rounding observed on the velar stop in
takkwe, nokkwe also indicates the presence of a labial [+round] feature that is otherwise
lost. To account for the gemination we observe in these forms, I propose to postulate
the exponent in (53a) for the Passato Remoto in addition to that already mentioned
in (46) (= (53b)). Another exponent of the Passato remote is /Ø/ which is inserted by
(53c) after the roots /fak/ ‘do’ (fet∫ -Ø- i) ‘I did’ and /ved/ ‘see’ (vid-Ø-i ‘I saw’).
(53) a. X ↔ [+past]T / RootL ^____ (RootL = nok, tak, dʒak, etc.)
|
Labial
|
[+round]
b. s
↔ [+past]T / RootS ^ ____ {RootS= scriv, muov, etc.}
c. Ø ↔ [+past]T / RootØ ^ ____ {RootØ= fak, ved, etc.}
(53a) consists of an empty skeletal position including, however, a floating nondesignated labial [+round] articulator, a secondary articulation (see Halle 1995 for
the notion of designated articulator). The skeletal position is filled in by the preceding consonant. The secondary labial articulation is attached to the place node only
when the designated articulator of the first consonant is dorsal, otherwise its attachment is blocked because of an active Marking Statement disallowing labial secondary
articulation with non-dorsal consonants. In this case, the secondary articulation is
Allomorphy in the Italian Passato Remoto: A Distributed Morphology Analysis | 31
deleted.17)
(54) v e n
XXX
|
-
i
-X
X
[+cons]
|
Place
|
=
Coronal
|
|
→
|
-
X
[+cons]
|
Place
|
Dorsal
|
[+dorsal]
|
|
Labial
|
[+round]
[+cons]
|
Place
|
Coronal
|
[+coronal]
i
-X
tak
XXX
[+coronal]
[+anterior]
[+distributed]
(55) t a k
XXX
ven
XXX
→
|
|
n
-X
i
-X
Labial →Ø
|
[+round]
kw
- X
i
-X
[+cons]
|
Place
|
Labial
|
[+round]
Dorsal
|
[+dorsal]
Labial
|
[+round]
17) The Passato Remoto of the irregular verb parere ‘seem’ seems to require the postulation of an
additional exponent /-v-/ in addition to those listed in (53) (cf. 3rd pl. parvero /par-v-e-ro/ (vs. 2nd
pl. par-e-ste). This exponent would be restricted only to this verb. Here I prefer to assume in the
case of this form that we are dealing with an instance of the exponent in (53a) which after the root
/par/ undergoes the special readjustment rule in (i) assigning features such as [+consonantal, -sonorant, +cont] to the empty skeletal position characterizing this exponent. The feature [+round] is
deleted because of the restriction of [+round] to dorsal stop consonants:
(i)
X
→
[
Labial
|
[+round]
32 | 언어와 정보 사회 제18호
X
/ par ^ ____
|
+cons
-son
|
Labial [+cont]
]
|
[+round]
Derivations for the 3rdpl. forms in (47) are given in (56):
(56) Rules:
Thematic Vowel Deletion (TV): (49)
Ablaut Rules (A) changing root vowel quality
Coronal Stop Deletion (CD)
Consonantal Assimilation (CA)
Nasal Deletion (ND)
a.
TV
VI
Output:
b.
TV
VI
CD
Output:
c.
TV
VI
A
CD
CD
Output:
d.
TV
VI
CA
Output:
e.
TV
VI
A
CD
ND
Output:
[[[[ val ]root TV] +PAST TV ]T -part, +pl ] AGR
[[[[ val ]root ] +PAST TV ]T +part, +pl] AGR
[[[[ val ]root ]
s e ]T ro
] AGR
valsero
[[[[ perd]root TV] +PAST TV ]T-part, +pl] AGR
[[[[ perd ]root ] +PAST TV ]T-part, +pl] AGR
[[[[ perd ]root ] s
e ]T ro
] AGR
per
s
e ro
persero
[[[[ mett ]root TV] +PAST TV ]T +-part, +pl] AGR
[[[[ mett ]root ] +PAST TV ]T-part, +pl ] AGR
[[[[ mett ]root ]
s
e ]T ro
] AGR
mitt
s
e
ro
mit
s
e
ro
mi
s
e
ro
misero
[[[[ skriv]root TV] +PAST TV ]T-part, +pl] AGR
[[[[ skriv ]root ] +PAST TV ]T-part, +pl] AGR
[[[[skriv ]root ]
s
e ]T ro
] AGR
skris
s
e
ro
skrissero
[[[[ fond]root TV] +PAST TV ]T-part, +pl] AGR
[[[[ fond ]root ] +PAST TV ]T-part, +pl] AGR
[[[[ fond ]root ]
s
e ]T ro
] AGR
fund
s
e ro
fun
s
e ro
fu
s
e ro
fusero
Allomorphy in the Italian Passato Remoto: A Distributed Morphology Analysis | 33
f.
[[[[ ven ]root TV ] +PAST TV ]T-part, +pl] AGR
[[[[ ven ]root
] +PAST TV ]T-part, +pl] AGR
[[[[ ven ]root
] Xw e ]T ro
] AGR
w
venn
e
ro
venn
e
ro
vennero
g.
[[[[ tak ]root TV] +PAST TV ]T-part, +pl] AGR
TV
[[[[ tak ]root ] +PAST TV ]T-part, +pl] AGR
VI
[[[[ tak ]root ]
Xw e ]T ro
] AGR
w
Gemination takk
e ro
Output:
takkwero
TV
VI
Gemination
Derounding
Output:
5. Brief Discussion of Stress in Verbal Forms
The alternations in (47) are also associated with stress shifts. Most other accounts
of them—some which will be critiqued later—pay much attention to the stress patterns of the alternating forms. In this section, I briefly consider the stress properties
of the Passato Remoto. I argue that the position of stress simply follows from the
morphosyntactic structure of the verb forms. Therefore, it cannot have any role in
determining them.
Italian nouns display stress on any of the three final syllables. The stress falls on the
penultimate if it is heavy. Otherwise, it can occur either on the penultimate or the
antepenultimate. Here I assume that penultimate stress is lexically marked, and the
antepenultimate is the default. I will not deal with stress on nouns here.
In verbs, any of the last four syllables may be stressed, as shown in (57). If we
include forms with clitics we even find stress on the fifth-last syllable (see (57)).
34 | 언어와 정보 사회 제18호
(57) a. Final syllable
auguró ‘s/he wished’
b. Penultimate syllable
auguráva ‘s/he was wishing’
c. Antepenultimate syllable
áuguro ‘I wish’
d. Preantepenultimate
áugurano ‘they wish’
áuguraglielo ‘Wish it to him/her!’
As observed by Oltra-Massuet and Arregi (2005), the morphosyntactic structure of
the verbal form is important to determine stress in Spanish and Catalan. Specifically,
according to Oltra-Massuet and Arregi, in these two languages stress falls on the
vowel preceding the Tense node, in particular on the verbal Thematic Vowel. Most
stress patterns in Italian verbs follow this generalization, as we can see in (58):
(58) Stress on the Thematic Vowel (in bold and underlined) before Tense
augurávo
[[[ root TV] T TV ] AGR ]
augurávano
[[[ root TV] T TV ] AGR ]
augurássi
[[[ root TV] T TV ] AGR ]
However, there are some differences. First of all, in Italian the Thematic Vowel
immediately preceding the 1st and 2nd pl. suffix stress has a special status and always
receives stress so that in this case stress is shifted rightwards from the Thematic Vowel
before T. In Spanish this does not occur and stress always falls on the Thematic
Vowel before T (Spanish: cantábamos, cantábais ‘sing’ vs. Italian cantavàmo, cantavàte
‘sing’).18)
18) An exception is the imperfect subjunctive where stress falls on the verb Thematic Vowel and not
on the Thematic Vowel preceding the suffix of 1st:
Allomorphy in the Italian Passato Remoto: A Distributed Morphology Analysis | 35
(59) auguravàmo (cf. augurávo)
[[[ root TV] T TV ] AGR ]
auguraváte
[[[ root TV] T TV ] AGR ]
There is then another difference more complex to describe since it has to do with
exceptions to the generalization that stress is assigned to the V before Tense. In both
Spanish and Italian, present indicative and subjunctive forms, stress falls on the root
and not on the Thematic Vowel19),20) as would be expected given Oltra-Massuet and
Arregi’s generalization. Take for example the 3rd pl. cánten in Spanish. According to
this generalization, stress should be on the thematic /e/ as in cantén instead of being
on the root syllable. To account for this fact, Oltra-Massuet and Arregi propose
that there is a special rule that retracts stress from the final vowel onto the preceding
vowel. However, this account cannot hold in Italian due to the following: (i) the suffix of the 3rd plural in Italian ends in a vowel; the consequence of this is that we cannot use this retraction rule to account for the root stress in this person. (ii) In addition, if stress falls on the root in Italian and the root contains more than one syllable,
stress can be on the last root syllable as in (61), or crucially on the penultimate root
syllable as we can see in (60) and (61), although never on a syllable preceding this
one (see (62)). The retraction rule would always give stress on the last root syllable in
Italian, contrary to the facts.
(i)
augurássimo
19) One could hypothesize that this fact is related to the fusion between Tense and AGR in this tense:
namely, once Tense is fused with AGR, it is no longer a true Tense node, and one could hypothesize that only “true” Tense nodes can trigger stress assignment. However, this does not hold for
the regular forms of the Passato Remoto where we also have fusion between Tense and AGR but
Tense still assigns stress. This hypothesis must then be rejected and the rule must be formulated as
not applying in the present.
20) As mentioned earlier, the 1st and 2nd pl. behave differently and stress is assigned to the TV that
precedes the AGR suffix in this case.
36 | 언어와 정보 사회 제18호
(60) Stress on the root
áuguro
[[ root TV] T + AGR ]
áugurano
[[ root TV] T + AGR ]
(61)
1sg
3sg
3pl
a.
prédico ‘preach’
prédica
prédicano
(62) teléfono ‘phone’
comúnico ‘communicate’
b.
adóro ‘worship’
adóra
adórano
télefono
cómunico
Note that regardless of the position of stress in the present, in the other tenses verbs
follow the preceding generalization: stress falls on the Thematic Vowel before Tense,
on the Thematic Vowel before 1st and 2nd pl., or on exceptional final endings, e.g.,
predicò (see below for discussion):
(63) predicáva, predicavámo, predicò
adorávo, adoravámo, adorò
‘preach-ImpInd-3sg./1pl./past-3sg.’
‘worship-ImpInd-3sg./1pl./past-3sg.’
I will simply propose that the rule assigning stress to the TV before Tense does not
apply in the present (see note 20). The default stress rule of Italian applies in this
case, assigning antepenultimate stress unless the penultimate vowel is heavy or is
assigned an idiosyncratic lexical accent: (prédic-o vs. adór-o). A detailed discussion
of Oltra-Massuet and Arregi’s analysis is not possible here due to space limits so I
will simply go on and propose a modification of Oltra-Massuet and Arregi’s theory
that can account for the Italian stress pattern. Following their lead, I assume an
account within the framework of the metrical grid (see, e.g., Liberman and Prince
1977; Prince 1983; Halle and Vergnaud 1987) and adopt the implementation of
this framework proposed in Idsardi (1992) and Halle and Idsardi (1995), which can
Allomorphy in the Italian Passato Remoto: A Distributed Morphology Analysis | 37
be summarized as follows. Stress is not a phonetic feature. Rather, it is a phonetic
means for marking certain groupings of linguistic elements. Furthermore, not all
phonemes in the string are capable of bearing stress. This fact is implemented by
having the elements capable of bearing stress project an abstract mark on a separate
plane, the metrical plane. The sequence of abstract marks projected by stressable elements constitutes the line 0 of the metrical plane. These elements are grouped into
what are called feet. This grouping is done by inserting parentheses on the metrical
plane. Idsardi’s (1992) innovation is that only one parenthesis is necessary to group
grid marks: a left parenthesis groups all the marks to its right, and a right parenthesis
groups all the marks to its left. Parentheses are projected either at the edges of the
stress domain (Edge-Marking rules) or at the edges of certain syllables (heavy syllables, a specific syllable in an accented vocabulary item, etc.). Within each foot, a grid
mark (the rightmost or the leftmost one) is designated as the head and is projected
onto the next line in the grid. A new grouping and a relative head are assigned on this
line and thus word stress is determined.
As in Oltra-Massuet and Arregi’s proposal, Idsardi’s notion of Edge Marking
is extended by allowing the projection of metrical parentheses from the syntactic
structure. In particular, as in their analysis of Spanish, the projection of parentheses
is determined by the morphosyntactic structure of the string. The set of rules that
derive stress placement in verbs in Italian is the following:
38 | 언어와 정보 사회 제18호
(64) Stress algorithm:
a. Project a line 0 mark for each syllable nucleus
(3rd pl. suffix -no and enclitics do not project).
b. On line 0, insert a left parenthesis to the left of the mark projected by the stressable
element preceding a [-present] Tense.
c. On line 0 insert a left parenthesis to the left of mark projected by the stressable element preceding [+part, +plur]AGR.
d. Place a right parenthesis to the left of the right-most element on line 0.
e. Insert a parenthesis every two elements starting from the right-most element.
f. Project the leftmost mark of each line 0 foot onto line 1.
g. Insert a right parenthesis to the right of the rightmost mark on line 1.
h. Project the rightmost mark of each line 1 foot onto line 2.
The basic rule that derives stress placement in finite tenses is (64b). Note that it
makes crucial reference to the syntactic node Tense, not to its phonological realization; that is, it ensures that stress precedes Tense, no matter what the realization of
Tense is.
As an example, here I give the forms of the imperfect, both the indicative and the
subjunctive. All of these forms have the stress pattern in (65), where stress falls on the
vowel realizing the verbal theme position.
(65) Stress in the imperfective
[[[ root TV] T TV ] AGR ]
ImpInd . . . . . . á v o
ImpSbj . . . . . . á ss i
As shown in (66), the application of stress algorithm (64) gives the right result. First,
(64a) projects stressable elements onto the grid (line 0). Second, (64b) places a left
parenthesis to the left of the mark projected by the stressable element preceding
Tense. In this case the element is the Thematic Vowel. Since, by (64f), feet on line 0
are left-headed, the result is that stress falls on the Thematic Vowel preceding Tense.
Allomorphy in the Italian Passato Remoto: A Distributed Morphology Analysis | 39
(66) a. 1Sg imperfective indicative (verb cantare ‘sing’)
Line 2
x
Line 1
x)
Line 0
x (x)
x
String
c a n t - a - v - _21)- o
Syntax
[[[ root TV] T TV ] AGR ]
b. 1Sg imperfective subjunctive
Line 2
x
Line 1
x)
Line 0
x (x)
x
String
c a n t -a -ss -i - Ø
Syntax
[[[ root TV] T TV ] AGR ]
Rule (64c) accounts for the stress patterns found in the case of the 1st and 2nd plural.22)
(67) 1Pl imperfective indicative
Line 2
x
Line 1
x
x)
Line 0
x (x
(x)
x
String
c a n t -a -v -a -m o
Syntax
[[[ root TV] T TV ] AGR ]
(68) 2Pl present indicative (verb telefonare ‘phone’)
Line 2
x
Line 1
x
x)
Line 0
x(x x (x)
x
String
t e l e f o n -a
-t e
Syntax
[[ root TV] T +AGR ]
The stress pattern found in the other forms of the present indicative can now be
derived. The rule inserting a left parenthesis on the Thematic Vowel before Tense
21) The TV is deleted before the vowel initial suffix by (33).
22) An exception to (64) is the imperfect subjunctive where the TV preceding the AGR of 1st and 2nd
pl. is not stressed; instead stress is assigned to the TV preceding Tense.
40 | 언어와 정보 사회 제18호
does not apply. Application of (64d) and (64e) constructs a left-headed foot starting
from the right edge as in (69). Observe that the suffix /-no/ of the 3rd pl. does not
project. The stress pattern constructed in (70) is also found here.
(69) 1Sg present indicative
Line 1
x)
Line 0
x (x x )
x
String
t elefono
Syntax
[[ root
TV] T+AGR ]
(70) 3Pl present indicative
Line 1
x)
Line 0
x(x x) x
String
t elefon a
no
Syntax
[[ root
TV] T+AGR ]
The last vowel of the root of the verb adorare (/ador/) ‘worship’ is idiosyncratically
marked as projecting a left parenthesis. This vowel will get word stress as shown in
(71). The pattern shown in (71) does not change in the 3rdpl. in (72) because /-no/
does not project on line 0 as discussed above:
(71) 1Sg present indicative
Line 1
x)
Line 0
x (x)
x
String
a dor -_
-o
Syntax
[[ root TV] T+AGR ]
(i)
1Pl. imperfective indicative
Line 1
x)
Line 0
x
(x
x)
x
String
c a n t - a ss i
mo
Syntax
[[[ root TV ] T TV ] AGR ]
Allomorphy in the Italian Passato Remoto: A Distributed Morphology Analysis | 41
(72) 3Pl present indicative
Line 1
x)
Line 0
x (x) x
String
a dor- a _
-no
Syntax
[[ root TV] T+AGR ]
Notice that the left parenthesis inserted before the AGR suffix of 2nd(and 1st) pl. will
trigger rightwards stress shift both in the case of predicate (see (68)) and of adorate
(73):
(73) 2Pl present indicative
Line 2
x
Line 1
x x)
Line 0
x ( x ) (x)
x
String
a d o r -a -te
Syntax
[[ root TV] T+AGR ]
In the regular Passato Remoto, rule (64) will insert a left parenthesis on the Thematic
Vowel preceding Tense–fused with AGR by (45). This will generate the stress pattern shown in (74). Remember that the 3rd pl. /-no/ does not project on line 0 so we
get the same pattern as in (75):
(74) 1Sg Passato Remoto
Line 1
x)
Line 0
x (x) x
String
cant-a- i
Syntax
[[ root TV] T+AGR ]
(75) 3Pl Passato Remoto
Line 1
x)
Line 0
x ( x) x
String
c a n t -a- r o -no
Syntax
[[[ root TV] T+AGR ]+fissioned [+plur] (see Sect. 3)
42 | 언어와 정보 사회 제18호
This also holds for the 3rd sg. of the II and III conjugations that have a zero-suffix in
this case. The same analysis as discussed above can be given for the 3rd pl: partírono
‘veave’, battérono ‘leave’):
(76) 3Sg Passato Remoto (verb partire ‘leave’)
Line 1
x)
Line 0
x (x) x
String
p a r t -i -Ø
Syntax
[[[ root TV] T+AGR ]
(77) 3Pl Passato Remoto (verb battere ‘beat’)
Line 1
x)
Line 0
x (x)
x
String
ba t t - e
-Ø
Syntax [[[ root TV] T+AGR ]
To account for the 3rd person of the past form where stress falls on the final suffix
instead of the verbal thematic as expected by (64), I assume that, as in Oltra-Massuet
and Arregi’s analysis of similar cases in Spanish, the vocabulary item for this morpheme is exceptionally specified to project a line 0 left parenthesis to its left, as shown
in (78) (where the thematic vowel /-a-/ is deleted before /-ó/ by (33)):23)
(78) 3Sg Passato Remoto
Line 2
x
Line 1
x
x)
Line 0
x (x) (x
String
cant- _
-o
Syntax
[[[ root TV] T+AGR ]
The stress pattern of the athematic forms of the Passato Remoto is now easy to
23) I assume that also the endings of the future (ameró, amerà, etc. ‘love’) and of the conditional
(ameréi, etc.) are exceptionally specified to project a line 0 left parenthesis to their left (see OltraMassuet and Arregi (2005) for an alternative analysis of these tenses in Spanish).
Allomorphy in the Italian Passato Remoto: A Distributed Morphology Analysis | 43
account for. Since no verbal Thematic Vowel is present, the left parenthesis assigned
by Tense will fall on the last vowel of the root. This accounts for why the athematic
roots have root stress, i.e. using the traditional term, for why they are rhizotonic.
(79) 3Sg Passato Remoto (verb mettere ‘put’)
Line 1
x
Line 0
(x)
x
String
mi s _
i
Syntax
[[[ root ]T TV ] AGR ]
(80) 3Pl Passato Remoto
Line 1
x
Line 0
(x
x) x
String
m i s e ro
Syntax
[[[ root ] T TV ] AGR ]
6. An Account of the Alternations in the Passato Remoto
Irregular Forms
We can now provide an account of why the 1st and 2nd pl. and 2nd sg. in (47) display regular morphology. As a reminder, I list here the different forms of the Passato
Remoto of the verbs perdere ‘lose’, scrivere ‘write’ and venire ‘come’:
(81) Singular
1
persi
scrissi
venni
2
perdesti
scrivesti
venisti
3
perse
scrisse
venne
Plural
1
perdemmo
scrivemmo
venimmo
2
perdeste
scriveste
veniste
3
persero
scrissero
vennero
As was proposed earlier, accounting for the allomorphy we observe in the Italian
Passato Remoto requires assuming that the relevant Vocabulary Items and
44 | 언어와 정보 사회 제18호
Readjustment Rules apply under conditions of linear adjacency. All these rules make
reference to the root exponent and the tense node. This implies that there must be
linear adjacency between them. This explains why no special allomorphy occurs in
the forms in which there is a Thematic Vowel. The presence of the Thematic Vowel
interferes with the adjacency that is required for the application of these rules. The
unmodified form of the root, therefore, appears since the rules introducing irregular
allomorphy cannot apply.
For example, the Thematic Vowel interrupts the adjacency required for the application of the Vocabulary Item in (53b), repeated in (82). This accounts for why
forms like those in (83) where the past is realized with the special allomorph /s/ are
impossible:24)
(82) /-s-/
↔
[+past] / Roots ^ ____ , Roots = scriv, muov, etc.
(83) perdesisti
scrivesisti
The presence of a Thematic Vowel in the forms of the 1st and 2nd pl. and 2nd sg.
in (83) means that the Thematic Vowel has not been deleted. This means that the
pruning rule in (49) does not apply in the case of these persons. How can we account
for the non-application of the rule in this case? I propose that this is due to the
operation of Impoverishment (Bobaljik 2003; Bonet 1991; Halle 1997; Halle and
Marantz 1993; Harley 2008; Nevins 2011; Noyer 1992, 1998). As usually used in
24) This obviously holds for all the special Vocabulary Items and Readjustment Rules characterizing
the Passato Remoto, so none of the forms in (i) below could be derived given the presence of the
Thematic Vowel:
(i) mittesti
fecesti
ruppesti
compremesti
redagesti
Allomorphy in the Italian Passato Remoto: A Distributed Morphology Analysis | 45
DM, Impoverishment blocks the insertion of Vocabulary Items by deleting morphosyntactic features; here I propose that Impoverishment can also delete lexical diacritics. Specifically, the pruning rule in (49) does not apply because the lexical information required for its application has been “impoverished”.
Let us consider impoverishment more in detail. Insertion of phonological exponents is governed by the Subset Principle (Halle 1997), according to which the phonological exponent of a Vocabulary Item is inserted into a morpheme in the terminal
string if the item matches all or a subset of the grammatical features specified. Where
several Vocabulary items meet the conditions for insertion, the item matching the
greatest number of features specified in the terminal morpheme must be chosen in
the terminal morpheme. Impoverishment deletes or removes features from the morpheme in a terminal node. By doing this, it blocks the insertion of a more specific
Vocabulary item and allows the insertion of a less specified one.25) This hence results
in what we can call the retreat to the general case.
An example of use of impoverishment in DM is provided by Sauerland’s (1995)
analysis of Norwegian adjectival endings. Norwegian adjectival endings differ in
strong vs. weak syntactic positions, as can be seen in (84):
25) Most Distributed Morphology works see impoverishment as involving the permanent deletion
of a feature in a morphosyntactic representation. In Calabrese (2008) (see also Calabrese 1994),
I argued that morphosyntactic representations must be always fully specified, and propose that
when a feature is deleted, an insertion of the opposite feature value immediately follows (which
leads to a syncretic change). I also proposed that cases of impoverishment that cannot be analyzed
in terms of feature deletion + insertion actually involve the temporary unavailability of a feature
specification during vocabulary insertion—a procedure I called feature freezing. The issue is not
relevant in this paper (but see note 26) and I will put it aside.
46 | 언어와 정보 사회 제18호
(84)
-neuter
Strong
-plural
+neuter
Ø
-t
+plural
Weak
e
-plural
+plural
The issue is the distribution of the exponent /e/ in the system, in particular, the
default status of /e/ in the strong forms. The rules in (85) account for the fact it is not
accidental that /e/ is default in the plural in the strong context:
(85) /t/ ↔ [ _ , -plur, +neut] / Adj + _
/Ø/ ↔ [ _ , -plur, -neut] / Adj + _
/e/ ↔ elsewhere / Adj + _
The impoverishment rule in (86) deletes the feature [neuter] in weak contexts:
(86) [±neuter] Ø / in weak contexts
This rule blocks /–t/ and /-Ø/ from being inserted in weak contexts, since each has a
gender specification. Therefore, the default /e/ is inserted in all such cases.
Impoverishment can be used to account for cases of regularization or morphological leveling. For example, in the history of English we see cases of regularization of
strong past forms as in (87):
(87) clomb
crope
lough
yold
holpen
⇒
⇒
⇒
⇒
⇒
climbed
crept
laughed
yielded
helped
Allomorphy in the Italian Passato Remoto: A Distributed Morphology Analysis | 47
How can we account for these cases? A simplified sketch of the Vocabulary Items
and Readjustment Rules needed for the relevant verbal morphophonology of English
is given below. The following vocabulary items account for verbal inflections in
English. Here I am assuming that the roots that undergo the special treatments are
characterized by a special diacritic (X, Y, Z). The reason for adopting this strategy
will be discussed below.
(88) I (= fused T and AGR)
a. [+participle, +past] ↔ -n / RootX ^ ____
where RootX = go, beat, ^hew, ...
b. [+past]
↔ -Ø / RootY ^ ____
where RootY = beat, drive, bind, sing, ...
c. [+past]
↔ -t / RootZ ^ ____
where RootZ = dwell, buy, send, ...
d. [+past]
↔ -d
I will not consider all of the Readjustment Rules needed to account for stem allomorphy in English strong verbs (see Halle and Marantz 1993), but only the rule of vowel
backing that is relevant in the forms in (87):
(89) V →
[
]
+back / [C1____ C2]RootZ [+past]
+round
where RootZ = sell, tell
I propose that Regularization is obtained through making lexical information
required for the application of special Vocabulary Items and the Readjustment Rules
unavailable. This involves removing the special root index required for the application of the special Vocabulary Items and Readjustment Rules. Therefore, the regular
forms appear. Formally we can say that climb in (87) loses the lexical markings Y, Z
48 | 언어와 정보 사회 제18호
that are required for the application of the rules in (88) and (89): 26), 27)
(90) rootX ↑ impoverishment ↑ root
In this way, we have an analysis of the changes in (87).
Observe that the same result could be obtained by modifying the rule. In this case
the innovation would involve removing the information about the relevant root from
the rule. The reasons for not opting for this solution will become clear later.
Let us turn back to Italian. I propose that in the 1st and 2nd pl. and in the 3rd sg.,
the special root index required for the application of the pruning rule in (49), repeated here in (91), is impoverished as in (32) by rule (33). The pruning rule can no
longer apply. This has a cascading effect: neither the special Vocabulary Items in (42)
nor the Readjustment Rules discussed in section 4 can apply. Only regular unmarked
morphology will occur.
(91)
V
V
(= (49))
TV / Root-e/i[+TV-pruning]___ Past,
Root[+TV-pruning] = korr, prend, scriv, ven, etc.
(92) root[+TV-pruning] ↑ impoverishment ↑ root
(93) Delete diacritic [+TV-pruning rule] in the environment ___[+participant, +plural]AGR28)
26) The feature that is being deleted here is a diacritic. Therefore Calabrese’s (2008) requirement of
full specification of morphosyntactic features (see note 25) does not hold in this case and this feature could be permanently removed also in this approach.
27) In this way I want to capture the idea that the innovation involves forgetting/not accessing for
some reason (see section 8 for discussion) the information (the diacritic) needed for the application of the rule, which remains the same. Once the innovation occurs, the rule may indeed
change. This cannot be the case for Italian, as discussed below.
28) Notice that impoverishment cannot be subject to the linear adjacency requirement on locality.
Crucially, impoverishment applies before Vocabulary insertion and is an inward-looking rule (cf.
Bobaljik 2000). If one holds that linearization is part of vocabulary insertion, then linear adjaAllomorphy in the Italian Passato Remoto: A Distributed Morphology Analysis | 49
For example, we can account for the difference between the 1st sg. and the 1st pl. of
the verb /scriv/ in (81) as follows. In (94), scriv has the diacritic [+TV-pruning] (= X in
(94)) that triggers the application of the TV pruning rule in (91). The exponent /-s-/
for the Tense [+past] can be inserted by (53a). The Readjustment Rule triggering /s/assimilation can also apply so that we get (94c).29)
(94) a. [[[[ scrivX ]root TV] +PAST TV ]T +part, +auth, -pl] AGR
b. [[[[ scriv ]root
] +PAST TV ]T +part, +auth, -pl] AGR
c. [[[[ scris ]root
]
s
]T i
] AGR
scriss i
In (95) Impoverishment deletes the diacritic X as in (b). Thematic Vowel /e/ is not
deleted. AGR and Tense fusion in (45) applies ((96c)) and the exponents of the regular past tense morphology are inserted.
(95) a.
b.
c.
d.
[[[[ scrivX ]root TV] +PAST TV ]T +part, +auth, +sg] AGR
[[[[ scriv ]root TV] +PAST
]T +part, +auth, +sg] AGR
[[[[ scriv ]root TV] +PAST, +part, +auth, +sg
] T+AGR
[[[[ scriv ]root e ] mmo
]
] T+AGR
scrivemmo
Notice that the irregular verbs, scriv for example, can often be regularized throughout the Passato Remoto paradigm so that we get the forms in (96). In this case
Impoverishment of the diacritic index applies to all persons.
cency is (effectively) undefined prior to vocabulary insertion, and so cannot constrain it.
29) Notice that potentially the rule could also be prevented from applying by impoverishing the feature [+past]. However, this would lead to a problem since the feature [+past] is later needed for
the insertion of the suffix /-mmo/.
50 | 언어와 정보 사회 제18호
(96) scriv-e-i
scriv-e-sti
scriv-é
scriv-e-mmo
scriv-e-ste
scriv-e-rono
Also observe that under the approach that was just outlined, the surface allomorph [scriss] can appear only in athematic contexts so it cannot appear before the
Thematic Vowel /-e-/. At the same time, /scriv-/ requires the presence of a Thematic
Vowel in the Passato Remoto. This excludes Passato Remoto forms such as those in
(97) and (98):
(97) scriss-ei
scriss-etti
scriss-ette
(98) scriv-i
scriv-e
Before going on, let us consider a few alternative accounts for how to obtain preservation of the Thematic Vowel as in (95). First of all, instead of using impoverishment
we could complicate the rule of TV Pruning so that it applies only in the 1st sg., and
3rd sg. and pl. Then, the rule would not apply in a structure like that in (95) and
the same result as impoverishment could be obtained. However, the 1st sg. and 3rd
sg. and pl. do not form a natural class and therefore we would be forced to split the
pruning rule into two different rules, one applying in the 1st sg. and another one
in the 3rd person. These two rules would be quite complex and idiosyncratic. The
function of Impoverishment as used in Distributed Morphology is that of avoiding
the postulation of idiosyncratic Vocabulary Items. For example, one could always
add the feature [+strong] to the Vocabulary Item in (85) to account for the distriAllomorphy in the Italian Passato Remoto: A Distributed Morphology Analysis | 51
bution of the exponents in (84). Impoverishment allows a simpler, more general
and transparent statement of the Vocabulary Items. The same can be said for the
Readjustment Rules. So by assuming Impoverishment as hypothesized here, we can
assume the simpler pruning rule in (49).
In addition, by preventing the insertion of a given, more specific Vocabulary Item,
Impoverishment allows for the extension in the distribution of another more general
VI (see Nevins 2012 for more discussion). This is again what we observe in the alternations in (47). In the rule solution hypothesized above, the appearance of default
morphology in the 1st, 2nd sg. and pl. in the alternations in (47) would be treated as
accidental.
In the account proposed above, impoverishment deleted the diacritic index
associated with the root. Other standard approaches to exceptional behaviors of
Vocabulary Items and Readjustment Rules (see Halle and Marantz 1993, for
example) do not use diacritic indices but simply listing of roots in the relevant rules.
Suppose that this is the right way of dealing with the exceptional behavior of morphemes and that diacritics are not allowed. This would lead to major difficulties in
the treatment of the alternations in (47). How would we prevent access to idiosyncratic root information? Obviously we cannot delete the root in the morphosyntactic
representation. We would then be forced to operate on the rules and delete roots in
the listing associated with the rule. But having a rule applying on another rule synchronically would be nonsense from the point of view of the architecture of the theory. Rules can be changed only during grammar restructuring in historical innovations
or during grammar learning and not in the synchronic derivation. This option therefore is impossible. The only way out is to assume diacritic indices in the morphosyntactic representations and hypothesize that under some conditions these indices can
52 | 언어와 정보 사회 제18호
be impoverished as proposed here. In this way regular morphology appears.30), 31)
7. Alternative Accounts
Let us consider alternative accounts for the appearance of the default basic root in
these persons. As far as I know, they all assume that the allomorphy seen in the irregular forms of the Passato Remoto involves memorized alternants in a suppletive relationship (see Vogel 1994; Pirrelli and Battista 2000). Traditional analyses (Buchholtz
1889; Meyer-Lübke 1972; Lausberg 1976; Rohlfs 1966; Tekavcic 1980) assume
that these stem alternants must be stressed on the root–they must be rhizotonic. The
reason for this property, according to them, is to be found in the development of the
Perfect aspect exponent /-u-/ of Latin in postconsonantal position. In the traditional
account, this /u/ caused gemination of the preceding consonant after a stressed syllable, whereas it was lost in pretonic position:
30) Some regular verbs of the /e/ conjugation optionally display a special augment -tt- that is inserted
between the Thematic Vowel /e/ and the AGR suffix. The distribution of this augment is similar
to that of the irregular stems: pot-e-tt-i, pot-e-sti, pot-e-tt-e, pot-e-mmo, pot-e-ste, pot-e-tt-e-ro ‘can/
be able-Past’ instead of pot-e-i, pot-e-sti, pot-é, pot-e-mmo, pot-e-ste, pot-e-ro-no. Therefore this augment does not appear in the 1st pl. and 2nd pl. and sg. I assume that the presence of this index is
governed by a diacritic that can be impoverished as proposed for the one triggering the rule of TV
pruning.
31) Observe that a slightly different analysis of the alternations in (47) could also be proposed. In
the text, I assume that the Verb TV are inserted across-the-board both after regular and irregular
roots and then pruned away by (49) after irregular roots. However, one could also assume that
the Thematic Vowel is simply not inserted after irregular roots, or, in other words, that irregular
roots are an exception to the TV insertion rule. Given the hypothesis that idiosyncratic behavior
in morphology involves the assignment of diacritic index to roots/morphemes, the exceptionality of these irregular roots would be marked by a special diacritic [+Exception to TV insertion].
At this point the same analysis as in the text could be proposed: the diacritic [+Exception to TV
insertion] would be impoverished under the same conditions as the diacritic index [+TV Pruning]
and hence the Thematic Vowels would be regularly inserted, which would result in a disruption of
the linear adjacency required by the irregular allomorphy rules, as discussed in the text. The consequences of the adoption of such an analysis are not clear to me at this point and I leave it to future
research.
Allomorphy in the Italian Passato Remoto: A Distributed Morphology Analysis | 53
(99) vénui → vénni
venuísti → venísti
‘come’
cádui → cáddi
caduísti → cadísti
‘fall’
vólui → vólli
voluísti → volísti
‘want’
‘1sg.’
‘2sg.’
Thus the stems of the athematic Passato Remoto, when they were not stressed,
became identical to the basic stem allomorph found in the imperfect, in the infinitive, and other regular forms: 32)
(100) ven-ísti
ven-ívo
ven-íre
cad-ésti
cad-évo
cad-ére
vol-ésti
vol-évo
vol-ére
The verbs in which this occurred were not many (venire, cadere, volere, avere, sapere,
nuocere, tacere). To account for what happened to the other irregular verbs, the traditional account assumes that given the alternations in (100), a generalization was
postulated:
(101) If root unaccented → regular
If root accented → irregular
It is assumed that this generalization lead to a restructuring of all irregular Passato
Remoto forms by analogical levelling.
(102) piac-évo piac-ére
ved-évo ved-ére
scriv-évo scriv-ére
piac-ésti
ved-ésti
scriv-ésti
(vs. 1sg. piácqui)
(vs. 1sg. vídi)
(vs. 1sg. scríssi)
‘like’
‘see’
‘write’
32) Notice that if the changes accounting for the development of the forms in (100) were only
phonological, we should expect Italian cadisti, volisti (cf. venisti) with preservation of the vocalic
quality of the Latin Perfect Thematic Vowel, instead of the actual cadesti, volesti where we see
the appearance of the Verb Thematic Vowel. This can be explained only by assuming some form
of morphological restructuring. In traditional accounts, this requires resorting to analogy which
involves a complication of the purely phonological account that is assumed for the development of
the forms in (100). The presence of the Thematic Vowel /e/ in these forms is obviously predicted
in the account proposed in the preceding section.
54 | 언어와 정보 사회 제18호
There are various problems with this account (see also Maiden 2000):
(i) From a phonological point of view, it is unclear why there should be deletion of
the labio-velar glide after unstressed vowels. It is also unclear why the labio-velar glide
should cause gemination after stressed vowel but not unstressed ones.
(ii) Observe that in nouns we find gemination after unstressed vowels:
(103) januárius
manuária
→
→
gennáio ‘January’
mannáia ‘cleaver’
The evidence for lack of gemination is then provided only by the evolution of the
Passato Remoto forms in (99). The traditional explanation, therefore, is fundamentally circular.
(iii) The 1st pl. is stressed on the root vowel in Latin ((104)). Therefore there
should be gemination in this case in Italian, contrary to the facts.
(104) vénuimus
cáduimus
‘come-Perf-1pl’
‘fall-Perf-1pl’
>
>
vennimo/ vennemo
caddimo/ caddemo
(iv) Assuming that the phonological account assumed traditionally is correct, it is
unclear why a few alternations such as those in (100) could lead to a such a robust
generalization causing analogical restructuring in all other verbs. One can then wonder why analogical pressure did not work in the opposite way so as to remove the few
marginal alternations in (100) by extending to them the irregular stem form.
In conclusion, the traditional phonological explanation is rather weak and should
be rejected.33)
The most recent analysis of the allomorphy seen in the Italian Passato Remoto is
found in Maiden (2000, 2010), who also rejects the traditional account of alterna33) I will not consider the structuralist account of Dardel (1958) based on homonymity avoidance
(see Maiden 2000 for a critical assessment).
Allomorphy in the Italian Passato Remoto: A Distributed Morphology Analysis | 55
tions. He assumes that the defining property of PYTA stems—Maiden uses the
acronym PYTA from Spanish linguistics (Preterito Y Tiempos Afines) to refer to the
special allomorph characteristic of irregular forms of the Passato Remoto—is their
being stressed, rhizotonic, and proposes that this is simply a consequence of the fact
that these stems must be followed by unstressed desinences. As Maiden (2000, p.
165) puts it: “se recano sempre l’accento, e’ solo perche’ il loro signans e’ una desinenza atona (e quindi l’accento deve cadere sulla sillaba precedente). [Their always
carrying stress is only due to the fact that their signans is an unstressed desinence
(and therefore stress must fall on the preceding syllable)]” (where signans refers to the
special “associative” relationship that according to Maiden Italo-Romance developed
between an unstressed desinence, the “signans” and the preceding PYTA root, its
“intramorfologico signatum” “intramorphological ‘signatum’”).
The issue is then why PYTA stems became associated with unstressed desinences. Maiden, following traditional approaches, assumes the following pieces for the
Passato Remoto:
(105) PYTA stems:
scrìssUnaccented endings
-i
-e
-ero
Regular stem
scrivAccented endings
-èi
-èsti
-è
-èmmo
-èste
-èrono
‘write’
1st sg.
2nd sg.
3rd sg.
1st pl.
2nd pl.
3rd pl.
Given these pieces, the following combinations are theoretically possible with their
relevant stress patterns. Remember that according to Maiden, although the rule is
not made explicit, the “syllable before” the unaccented ending is stressed:
56 | 언어와 정보 사회 제18호
(106) a.
b.
c.
d.
PYTA + unaccented ending
scrìss-i
scrìss-e
PYTA + accented ending
scriss-èi scriss-èsti scriss-è scriss-èmmo scriss-èste
non-PYTA + unaccented ending
scrìv-i
scrìv-e
non-PYTA + accented ending
scriv-èi scriv-èsti scriv-è scriv-èmmo scriv-èste
scrìss-ero
scriss-èrono
scrìvero
scriv-èrono
Maiden suggests that the reason for the impossibility of non-PYTA root + unstressed
ending lies in a universal principle (the so-called ‘No Blur Principle’ elaborated by
Carstairs-McCarthy 1994; also Cameron-Faulkner and Carstairs-McCarthy 2000)
disfavoring absolute synonymy among inflectional affixes.
“Now if a non-PYTA root appeared before a stress-less suffix, this principle would be violated, since it would then be the case that some verbs had in the preterite the unstressed endings
and other had stressed endings, without any functional difference between the rival sets of endings. One might imagine, that is, that Italian could have scrìv-i, scriv-èsti, scrìv-e, scriv-èmmo,
scriv-èste, scrìvero ‘write’ [i.e., (106c) above (AC)] but ricev-èi, ricev-èsti, ricev-è, ricev-èmmo, ricevèste, ricev-èrono ‘receive’, showing the complete set of stressed preterite endings. In other words,
the distribution of the inflectional endings of the first person singular, third person singular ...,
and third person plural would be lexically unpredictable, and two perfectly synonymous suffixes
would coexist, non-optionally in the grammar ... What Italo-Romance varieties have done is to
hypercharacterize this interdependency by making the unstressed desinence a unique defining
characteristic of PYTA roots: no PYTA root without unstressed desinence, and no unstressed
desinence without a PYTA root. (Emphasis Mine, A.C.)”
(Maiden 2010: 199)
In my critique of Maiden’s analysis, I will assume the correctedness of the No Blur
Principle. Obviously if this principle is rejected—see Halle and Marantz (2008) for
why it should—Maiden’s analysis can no longer be maintained.34)
34) In developing my critique, I will put aside the issue of what “hypercharacterization of interdependence” between PYTA roots/stems and unstressed desinences means and is implemented. I
will also not consider the fact that this interdependence involves a surface, word-level requirement
Allomorphy in the Italian Passato Remoto: A Distributed Morphology Analysis | 57
First of all, observe that other solutions to the problem potentially posed by Italian
for the No Blur Principle are possible. A first possibility is simply to merge the two
subparadigms that are problematic for this principle either by replacing accented ending with unaccented ending (as in (107a) or vice versa by replacing the unaccented
endings with the accented ones, as in (107b).
(107) a. scrìv-i
ricèv-i
b. scriv-èi
ricev-èi
scriv-èsti
ricev-èsti
scriv-èsti
ricev-èsti
scrìv-e
ricèv-e
scriv-è
ricev-è
scriv-èmmo
ricev-èmmo
scriv-èmmo
ricev-èmmo
scriv-èste
ricev-èste
scriv-èste
ricev-èste
scrìv-ero
ricèv-ero
scriv-èrono
ricev-èrono
Both changes in (107) are possible ways of removing the forms that are problematic
for the No Blur Principle. However, whereas the change in (107b) indeed occurs,
although not frequently–it in fact simply corresponds to the regularization of the
Passato Remoto irregular forms discussed earlier, the change in (107a) is just unattested. Nothing in Maiden’s framework accounts for the absence of such possible
development.
Notice that an important step in Maiden’s argument that is actually not made
explicit is the assumption that the PYTA roots are disfavored, so that they should
tend to be removed. Only by this assumption can we get the configuration in (106c)
where violations of the No Blur Principle can be created. However, nothing in
Maiden’s framework excludes a historical scenario in which PYTA roots begin to be
used with stressed endings, so that we could also have the historical development of
the forms in (108):
(108) scriss-èi
scriss-èsti
scriss-è
scriss-èmmo
scriss-èste
scriss-èrono
that cannot be maintained in any theory in which morphemes are inserted inside out.
Notice by the way that the unstressed endings /-i/, /-e/ and /-ero/ are found also in other verbal
forms like the imperfect subjunctive and the conditional. The requirement of interdependency
does not hold for these forms.
58 | 언어와 정보 사회 제18호
However, despite the widespread preservation of these roots across Italo-Romance,
such a development is never found in these varieties. This is obviously predicted in
the account developed here where forms like those in (108) are simply impossible to
derive.
At this point observe that Maiden assumes that the endings of the regular forms
must be analyzed as involving unsegmentable wholes: -èi, -èsti, etc. However, even
in a framework like his, they can indeed, and perhaps even should, be segmented as
being composed of a Thematic Vowel followed by a person-number suffix: ricev-è-i,
ricev-è-sti, ricev-è (ricev-e+stress), ricev-è-mmo, ricev-è-ste, ricev-è-rono (imperfect: riceve-v-a, infinitoi); cf. part-ì-i, part-ì-sti, part-ì(part-i+stress), part-ì-mmo, part-ì-ste, partì-rono (imperf: part-i-va, inf. part-i-re). It follows that there is no violation of the No
Blur Principle in the forms in (106)c) insofar as the distribution of the inflectional
endings of the 1st and 3rd sg. and 3rd pl. would be predictable from the absence of
the Thematic Vowel. Therefore, there is no need to resort to PYTA roots to disambiguate the distribution of the Passato Remoto endings.
Notice furthermore that Maiden’s analysis is crucially based on the stress status of
the Passato Remoto endings. Now consider that if we put aside the case of stressed
/-ò/ of the I conjugation, saying that an ending is stressed in Italian simply means
that it includes a Thematic Vowel, which predictably get stressed, as discussed in
section 5. Lack of stress in a Passato Remoto ending can occur only if the Thematic
Vowel is missing. It follows that PYTA roots only occur in athematic contexts.
Therefore in Maiden’s approach also, the correct generalization is that PYTA roots
are athematic and an explanation of this fact must be provided. On the other hand,
as discussed earlier, regular, non-PYTA root forms must be thematic in the Passato
Remoto, which immediately explains why (106) is impossible. The analysis outlined
above provides a unified account of the allomorphic combinations we observe in
the Italian Passato Remoto. There is no problem in accounting for the absence of
Allomorphy in the Italian Passato Remoto: A Distributed Morphology Analysis | 59
forms such as those in (107) and (108)in the model developed here. These forms
simply cannot be derived: PYTA roots must be athematic and therefore followed by
unstressed desinences, and the non-PYTA roots must be thematic and therefore followed by stressed desinences.35), 36)
8. Reasons for Impoverishment of Diacritic Indices
Earlier I accounted for the complex alternations between irregular and regular
forms found in the paradigm of the Italian Passato Remoto by assuming that the
forms of 1st, 2nd pl. and 2nd sg. are thematic because in these forms the special root
index required for the application of the rule pruning the verb Thematic Vowel is
impoverished.
The issue I want to turn to now is why there is impoverishment of this index. In
particular, I would like to propose that this operation of impoverishment is due to a
markedness principle, specifically, Brøndal’s (1940, 1943) “principle of compensation” that states that marked categories tend not to combine (see Calabrese 2011).
This is in line with what was proposed recently by Nevins (2011), who, although not
invoking this specific principle, provides overwhelming evidence in support of the
idea that combinations of marked categories in morphology are removed by means
35) There is also a general theoretical point that I will not fully develop here. An analysis such as
Maiden’s crucially requires vocabulary insertion to be dependent on a word-level property such
as stress. As shown by Embick (2010), analyses like this are possible only in global models such
as parallel OT, exemplar phonology, word-and paradigm approaches where all types of surface
phonological and morphological properties can be accessed at the same time. Embick (2010) has
shown the problems that global models face and provided evidence for localist derivational models
where VI insertion occurs long before stress is assigned.
36) Note that the arguments put forth above can be leveled against any theory like Maiden’s in which
the allomorphy seen in the Italian Passato Remoto is not derived as discussed above but accounted
for by means of the juxtaposition of memorized surface alternants.
60 | 언어와 정보 사회 제18호
of Impoverishment.
Note, first of all, that the 1st and 2nd person plural forms display a special morphological behavior throughout Italo-Romance where idiosyncratic exponence for these
persons is avoided (see (109)). In Calabrese (1994, 2011) I proposed that this special
behavior is a consequence of the markedness of these persons.
(109) a. In many Italian dialects, the exponent of the 1st pl. oblique clitic was replaced by
the exponent of a locative. The same thing happened for the 2nd pl.
b. In other Italian dialects, the exponent of the 1st pl. oblique clitic was replaced by
the exponent of a partitive ( < Latin INDE).
c. In the Tuscan dialect of Lucca, the exponent of the reflexive/impersonal clitic /si/
has also become the exponent of 1st pl. oblique clitic.
d. In Tuscan the 1st pl. subject is replaced by the reflexive/impersonal clitic /si/. The
verb appears in the 3rd sg.
e. In the Campidanese dialect of Sardinian, the exponent of the reflexive clitic /si/
has also become the exponent of I and II Pl.
f. In many dialects, 1st and 2nd pl. subject clitics do not have idiosyncratic exponence. They are either replaced by other clitics or they are just omitted.
According to Brøndal (1940, 1943) there are two different ways of eliminating
marked categories in marked contexts: (i) by syncretism in which the exponent in
a marked category is replaced by the exponent of an unmarked category, or (ii)by
removing/omitting a marked morphological category. It is a fact about languages that
there is an asymmetry in behavior between the exponents of certain morphological
feature configurations and the exponents of others: the former are more likely to disappear and/or be replaced by other exponents through morphological processes such
as syncretism/omission.
In Calabrese (2011), I proposed that the variations we observe in (109) are due
to the application of these two operations. In particular, in certain morphosyntactic
positions, languages tend not to assign idiosyncratic exponence to marked feature
combinations. So morphemes with marked feature combinations may have the same
Allomorphy in the Italian Passato Remoto: A Distributed Morphology Analysis | 61
exponent (syncretism) as other morphemes or may be just deleted (defectiveness).
I proposed that these two operations—replacement of exponents (syncretism) or
removal of exponents—are induced by the principle of compensation. The principle
can be formulated as in (110). Notice that the principle refers to merged morphemes,
i.e., to co-presence inside the same word. The principle in fact seems to apply only
in the case of synthetic morphology, so after morphological categories undergo postsyntactic merger. In other words, it applies only to cases of “merged” morphemes—
word-internal properties. This is not surprising. In fact, only in this case can we talk
of word complexity and word-internal structure.37)
(110) Principle of Compensation
Marked categories should not have idiosyncratic exponence when merged with
other marked categories, where idiosyncratic exponence is a phonological shape
F generated by the insertion of a vocabulary item or by the application of a
Readjustment Rule.
This principle then prevents the accumulation of idiosyncratic exponence in words
containing marked categories. Now, Readjustment Rules create idiosyncratic exponence in the Passato Remoto, a marked morphological category. A crucial rule in
this case, as discussed earlier, is the rule pruning the Thematic Vowel. By removing
the lexical diacritic required for the application of the Thematic Vowel pruning rule,
impoverishment prevents the appearance of idiosyncratic exponence in the 1st and 2nd
plural, which are marked morphological categories.38) It follows that the emergence
37) It follows that analytical, periphrastic morphology is much less constrained. So whereas the use of
Dual affixes is restricted and fewer languages have it, no language restricts the expression of duality
analytically by using the number 2. It is always possible to use combinations of words in periphrastic constructions to express combinations of features that otherwise cannot be expressed by
affixal/synthetic morphology. To put it more generally, any concept or property can be expressed
through periphrasis or paraphrases.
38) See Calabrese (2011) for some discussion of the markedness of the 1st and 2nd plural across languages. Possible further evidence for the markedness of these persons comes from Veselinova’s
(2006) investigation of Verbal Suppletion, where a number of cases are discussed in which the 3rd
62 | 언어와 정보 사회 제18호
of regular morphology in the 1st and 2nd pl. in the Passato Remoto falls together with
the other cases in (109) where idiosyncratic exponence is prevented from appearing
by syncretism/omission.
However, there is an important issue that needs to be discussed at this point. The
removal of combinations of marked feature configurations investigated in Calabrese
(1994, 2008, 2011) leads to conspiracies; for example, the exponent of the 1st pl.
subject clitics may be replaced by either a 1st sg. in some dialects or by a 2nd pl. exponent in other dialects or simply deleted, the 1st pl. clitic object may be replaced by
either a reflexive, locative or genitive clitic, depending on the dialect. No such conspiracies are found in the phenomena studied in this paper. Potentially the complex
configuration involving a lexical diacritic on the root, the marked feature [+past],
and the marked features [+participant, +plural], could be dealt with by removing the
lexical diacritic, as argued here, or the other morphosyntactic features. We should
see impoverishment of the feature [+past] or of the feature [+part] or [+plur]. Now,
impoverishment of [+past] would result in the replacement of the idiosyncratic past
exponent that we see in the irregular forms with a default one. Let us assume it to
be Ø. One could say that this is actually what we observe in a form such as ven-iØ-mmo. But this is not correct. Impoverishing [+past] would have an immediate
effect on the AGR ending /-mmo/ which requires the feature [+past] to be inserted.
Therefore, impoverishing [+past] would also affect the exponence of the AGR suffix
and this is not what we see in forms such ven-i-mmo. On the other hand, no changes
in the exponence of the ending of 1st and 2nd pl. is observed across dialects in the
irregular Passato Remoto forms. This is what we would expect if impoverishment
could affect the marked person features.
Now the fact that we do not have this type of conspiracy could be simply due
person takes an idiosyncratic form, but 1st and 2nd persons both sg. and pl. take default morphology (p. 84).
Allomorphy in the Italian Passato Remoto: A Distributed Morphology Analysis | 63
to chance, a historical idiosyncracy. However, here I propose that there is actually
a principled reason for this fact. Specifically, diacritic indices are treated differently
from morphosyntactic features by the Principle of Compensation and their elimination is governed by a dedicated principle, i.e., (111):
(111) Avoid diacritic indices on a morpheme if a morpheme is merged with marked morphemes.
It is this principle that accounts for the impoverishment of root lexical diacritics in
the 1st and 2nd plural.
A problem that needs to be dealt with at this point is that of the 2nd sg. Whereas
evidence can be provided for the markedness of the 1st and 2nd plural (see note 38),
this cannot be said for the 2nd sg., which actually behaves quite differently than the
1st and 2nd pl. in the data discussed by Calabrese (2011) (see also below). Should the
idea of markedness be abandoned here? I do not believe so. Observe that the 2nd sg.
of the Passato Remoto is quite peculiar with respect to the other person of the singular in having the exponent /-ste/, which is unique among the endings of the singular
in having an initial consonantal cluster. I propose that it is the phonological complexity of its exponent that characterizes this person as marked for the purposes of the
Principle of Compensation.39)
39) Notice that the same complexity in exponency is found in the 1st and 2nd plural /-mmo/, /-ste/
where one could analyze it in terms of Jakobsonian isomorphism requiring marked exponents for
morphemes of marked categories. Notice that the condition of the stress algorithm requiring an
accent before these persons could also be seen as a sign of their markedness. Observe in this regard
that there are several Italian dialects having a 1st person ending /-imV/ where historically the /i/
is the TV of the Latin perfect tense. Irregular allomorphy is displayed before this suffix in these
dialects (dissimo ‘say’, fecimo ‘make, do’, venimo ‘come’) (cf. Rohlfs (1966-9). This indicates that
impoverishment did not apply in this case, if my analysis is correct. Regular morphology, however, appears in the 2nd sg. and pl. in these dialects where the exponents are the phonologically complex /-ste/, /-sti/, respectively. These dialects perhaps indicate that the phonological complexity of
the exponent may have a more important role than what assumed in the paper. More research is
needed to investigate this point.
64 | 언어와 정보 사회 제18호
(112) A morpheme may be marked because of the morphosyntactic complexity of its
featural content or because of the phonological complexity of the exponent.
There is evidence for this. In fact, observe that there are other allomorphic alternations which depend on person-number in Italian verbal morphology. Consider the
following forms of the present. To account for the special vowel quality we see in the
three persons of the singular and in the 3rd pl. we need ablaut rules of lowering and
fronting. The rules are given in (114):
(113) ‘ɔdo ‘ɔdi
‘ɔde
cf. udivano/udire ‘hear’
‘ɛsko ‘ɛ∫∫i
e∫∫e
cf. u∫∫ivano/u∫∫ire ‘go out’
‘devo ‘devi
‘deve
cf. dovevano/dovere ‘must’
(114) a.
[-cons] →
[-high ]
[-cons] →
[ ]
b.
-high
-back
u’djamo
u’dite
‘ɔdono
u∫∫amo
u∫∫ite
ɛskono
dob’bjamo
do’vete
‘devono
N
|
___
N
|
___
/\
]Rooto Pres, Rooto = ud
/\
]Roote Pres, Roote = u∫∫, dov
The rules do not apply in the 1st and 2nd pl. where the underlying quality of the root
shows up. The presence or absence of the Thematic Vowel here does not count.
These are thematic forms. I assume that the alternations are again due to impoverishment of the lexical diacritic necessary in the structural description of the rules in
(114). Given (111), the impoverishment of the lexical diacritic is due to the markedness of the 1st and 2nd pl. morphemes. Observe now that there is no impoverishment
in the 2nd singular which displays the phonologically simple exponent /-i/ as expected
by (112).
Note that in these cases we cannot just say that we are using the unmarked alloAllomorphy in the Italian Passato Remoto: A Distributed Morphology Analysis | 65
morph of the stem. In fact only the rule in (114) is prevented from applying in
(113), but not the Readjustment Rule palatalizing the velars or the Readjustment
Rule geminating a consonant before the palatal glide /y/ (needed for doʎʎamo/dolere,
dobbyamo/dovere, abbyamo/avere but not salyamo/salire, bevyamo/beveva). We need to
say that impoverishment targets the diacritic index associated with the rules of ablaut
but not with the diacritic indices associated with rules of palatalization and gemination.
Assuming the analysis proposed above we can now deal with other aspects of
Italian verbal morphology. If one assumes that the addition of the special root extension /isk/ in the present of certain verbs of the /i/-conjugation involves special diacritic indices, and these indices are disallowed in environments containing marked categories, we have an account for the alternations we observe in the paradigm in (116).
We can simply propose that the special diacritic index required for the insertion of
the suffix /-isk/ (see rule (115)) is impoverished when the root is in the environment
of marked [+part, +pl.]:
(115) Insert /-isk-/ / Root-i-isk __ ^ Pres
(116) The suffix -isk-:
1sg. fin-isc-o
2
fin-isc-i
3
fin-isc-e
1pl. fin-i-a-mo
2
fin-i-te
3
fin-isc-o-no
‘finish’
Embick (2011) also uses impoverishment in the environment of [+part, +pl.] to
account for the alternations in (116). However, he impoverishes the tense feature
[Present]. Notice, however, that the insertion of thematic /-ya-/ before the AGR
suffix of 1st pl. crucially requires the feature [Present] (see note 13). If this feature
66 | 언어와 정보 사회 제18호
is impoverished we should expect the appearance of default Thematic /i/ ( finimo),
which is incorrect. Therefore the alternation in (116) cannot be accounted for by
impoverishing [Present].
I end by discussing two cases involving suppletive roots. Another instance of a
special Readjustment Rule not applying in the 1st and 2nd Plural is found in the
Imperfect of essere where the regular exponent of the imperfect is missing except in
the 1st and 2nd pl.
(117) The imperfect indicative of the verb essere ‘be’ and parlare ‘speak’
1sg. a. er-o
b. parl-a-v-o
2
er-i
parl-a-v-i
3
er-a
parl-a-v-a
1pl.
er-a-v-amo
parl-a-v-amo
2
er-a-v-ate
parl-a-v-ate
3
er-ano
parl-a-v-ano
Following the analysis developed above, I assume that here again there is a diacritic
that is being impoverished. So I assume that the absence of the imperfect exponent is
due to a Readjustment Rule deleting Tense in the context of the suppletive root [er]
where this context is formalized as involving a root diacritic , i.e., rule (118).
(118) [+imperfect]T → Ø / Root/-er/____
It is the diacritic /er/ that is impoverished in the marked context of the 1st and 2nd pl.
thus preventing the application of (118).
Let us finally consider the allomorphic alternations in (119).
Allomorphy in the Italian Passato Remoto: A Distributed Morphology Analysis | 67
(119) Suppletive vad-/and- in the verb andare ‘go’
1sg. vádo
2
vái
3
vá
1pl. andiámo
2
andáte
3
vánno
Embick (2011) proposes the Vocabulary Items in (120) and uses impoverishment of
the feature [Present] in the environment of [+part, +pl.] to account for the alternation:
(120) vgo
vgo
→
→
va(d)/
and
___^ Present
As in the case of the alternations involving the suffix /-isk/, however, impoverishment
of [Present] creates problems with the allomorphy of the Thematic /-ya-/ before the
1st pl. suffix /-mo/. I propose that also in this case, we are dealing with impoverishment of a diacritic index. Therefore I assume that suppletion involves the presence of
a special diacritic index, as in (121), and it is this index that is removed by impoverishment in the 1st and 2nd pl.
(121) vgo[+sup]
vgo
→
→
va(d)/
and
___^ Present
9. Conclusions
Maiden (2005, 2010) among many others, including traditional neogrammarian and structuralist accounts (see also Pirelli and Battista 2000), assume that the
complex allomorphy involving regular/irregular stem forms in the so-called Passato
68 | 언어와 정보 사회 제18호
Remoto must be accounted for in terms of memorized stem alternants and endings.
This paper shows that the best analysis of the Italian verbal morphology involves
notions such as roots, Vocabulary Items and Readjustment rules as predicted by the
Distributed Morphology (DM) model (Halle and Marantz 1993; Embick 2010;
Embick and Marantz 2008). There is no evidence for memorized stem alternants.
Instead, the allomorphy we see in the Italian Passato Remoto is readily accounted
for by providing an appropriate morphosyntactic analysis of the forms and by deriving the irregular alternants from single underlying roots by means of Readjustment
Rules. This paper also shows that it is crucial that the rules accounting for this allomorphy obey a strict definition of locality requiring linear adjacency. This explains
why the presence of the Thematic Vowel prevents their application. In the irregular
forms where these rules apply, the Thematic Vowel is removed by a special pruning
rule. Crucially, Impoverishment removes a diacritic index associated with the application of this pruning rule in certain morphosyntactic contexts (in the 1st and 2nd pl.,
and 2nd sg.), thus the regular basic form of the root will appear in these contexts.
Impoverishment in this case is motivated by a general Markedness principle which
disfavors complex exponence in morphologically marked environments.
Allomorphy in the Italian Passato Remoto: A Distributed Morphology Analysis | 69
References
Bybee, Joan(1985), Morphology: A Study of the Relation between Meaning and Form,
Amsterdam: Benjamins.
Bobaljik, Jonathan David(2000), “The ins and outs of contextual allomorphy,” In
Kleanthes K. Grohmann and Caro Struijke (Eds.), Proceedings of the Maryland
Mayfest on Morphology 1999, 35–71, University of Maryland Working Papers
in Linguistics 10, College Park, MD: University of Maryland, Department of
Linguistics.
Bobaljik, Jonathan(2003), “Syncretism without paradigms: Remarks on Williams 1991,
1994,” In Geert Booij and Jaap van Marle (Eds.), Yearbook of Morphology 2002,
53-85, Dordrecht: Kluwer.
Bobaljik, Jonathan(2008), “Missing Persons: A case study in morphological universals,”
The Linguistic Review special theme issue Examples of Linguistic Universals 25(12), 203-230.
Bobaljik, Jonathan(2012), Universals in Comparative Morphology, Cambridge, MA: MIT
Press.
Bonet, Eulalia(1991), Morphology after Syntax: Pronominal Clitics in Romance, Doctoral
Dissertation, MIT, Cambridge, MA.
Brøndal, Viggo(1940), “Compensation et variation,” Deux principes de linguistique
générale Scientia 61, 101-109 [Reprinted in V. Brøndal (1943), Essais de linguistique générale, 105-116, Copenhagen: Munksgaard].
Brøndal, Viggo(1943), “Structure et variabilité des systèmes morphologiques,” In V.
Brøndal (1943), Essais de linguistique générale, 1-24, Copenhagen: Munksgaard.
Buchholtz H.(1889), “Lockere und straffe italienische Perfektformen,” Archiv für das
Studium der neuen Sprachen 82, 133-166.
Calabrese, Andrea(1994), “Syncretism phenomena in the clitic systems of Italian and
Sardinian Dialects and the notion of morphological change,” In J. Beckman
(Ed.), Proceedings of NELS 25, 151-173.
Calabrese, Andrea(2002), “On Impoverishment and fission in the verbal morphol70 | 언어와 정보 사회 제18호
ogy of the dialect of Livinallongo,” In Christina Tortora (Ed.), Studies on Italian
Dialects, 3-33, Oxford: Oxford University Press.
Calabrese, Andrea(2005), Markedness and Economy in a Derivational Model of Phonology,
Berlin: Mouton de Gruyter.
Calabrese, Andrea(2008), “On absolute and contextual syncretism. Remarks on the
structure of paradigms and on how to derive it,” In Andrew Nevins and Asef
Bachrach (Eds.), The Bases of Inflectional Identity, 156-205, Oxford: Oxford
University Press.
Calabrese, Andrea(2011), “Investigations on markedeness, syncretism and zero exponence in morphology,” Morphology 21, 283-325.
Calabrese, Andrea(Forthcoming), “The development of idiosyncratic patterns of allomorphy in Italian verbal morphology: linguistic changes as a window on the form of
linguistic competence,” ms., University of Connecticut.
Carstairs, Andrew(1987), Allomorphy in Inflexion, London: Croom Helm.
Carstairs-McCarthy, Andrew(1992), Contemporary Morphology, London: Routledge.
Carstairs McCarthy, Andrew(1994), “Inflection classes, gender, and the principle of
contrast,” Language 70, 737-788.
Cameron-Faulkner, Thea and Carstairs-McCarthy, Andrew(2000), “Stem alternants and
morphological signata: evidence for blur avoidance in Polish nouns,” ms.
Chomsky, Noam(1995), The Minimalist Program, Cambridge, MA: MIT Press.
Clements, George Nick(2001), “Representational economy in constraint-based phonology,” In T. Alan Hall (Ed.), Distinctive Features Theory, 71-146, Berlin: Mouton
de Gruyter.
Dardel, R. de(1958), Le pairfait fort en romane commun, Geneve: Droz.
Dresher, B. Elan(2009), The Contrastive Hierarchy in Phonology, Cambridge: Cambridge
University Press.
Embick, David(2003), “Locality, listedness, and morphological identity,” Studia
Linguistica 57, 143–169.
Embick, David(2010), Localism versus Globalism in Morphology and Phonology,
Cambridge, MA: MIT Press.
Allomorphy in the Italian Passato Remoto: A Distributed Morphology Analysis | 71
Embick, David(2011), McSIRG Lectures, McGill University, ms., University of
Pennsylvania
Embick, David, and Morris Halle(2005), “On the status of stems in morphological
theory,” In Twan Geerts, Ivo van Ginneken, and Haike Jacobs (Eds.), Romance
Languages and Linguistic Theory 2003, 59–88, Amsterdam: John Benjamins.
Embick, David, and Alec Marantz(2008), “Architecture and blocking,” Linguistic Inquiry
39, 1–53.
Glushan, Zhanna(2010), “Deriving case syncretism in differential object marking systems,” ms., University of Connecticut.
Halle, Morris(1995), “Feature geometry and feature spreading,” Linguistic Inquiry 26, 1-
46.
Halle, Morris(1997), “Impoverishment and fission,” PF: Papers at the Interface, In
Benjamin Bruening, Yoonjung Kang, and Martha McGinnis (Eds.), MIT
Working Papers in Linguistics 30, 425-450, Cambridge, MA: MITWPL.
Halle, Morris, and Alec Marantz(1993), “Distributed Morphology and the pieces of
inflection,” In Kenneth Hale and Samuel Jay Keyser (Eds.), The View from
Building 20: Essays in Linguistics in honor of Sylvain Bromberger, 111–176,
Cambridge, MA: MIT Press.
Halle, Morris, and Alec Marantz(2008), “Clarifying ‘blur’: paradigms, defaults, and
inflectional classes,” In Andrew Nevins and Asef Bachrach (Eds.), The bases of
Inflectional Identity, 55-72, Oxford University Press.
Halle, Morris, and Jean-Roger Vergnaud(1987), An Essay on Stress, Cambridge, MA:
MIT Press.
Halle, Morris, and William Idsardi(1995), “General properties of stress and metrical
structure,” In John A. Goldsmith (Ed.), The Handbook of Phonological Theory,
403-443, Oxford: Blackwell.
Harley, Heidi(1995), Subjects, Events, and Licensing, PhD dissertation, MIT, Cambridge,
MA.
Harley, Heidi and Rolf Noyer(1999), “Distributed morphology,” Glot International 4(4),
3-9.
72 | 언어와 정보 사회 제18호
Harley, Heidi(2008), “The importance of impoverishment,” In Daniel Harbour, David
Adger and Susana Bejar (Eds.), Phi Theory, 251-294, Oxford: Oxford University
Press.
Idsardi, William(1992), The Computation of Stress, PhD dissertation, MIT, Cambridge,
MA.
Kiparsky, Paul(1996), “Allomorphy or morphophonology?” In Rajendra Singh and
Richard Desrochers (Eds.), Trubetzkoy’s Orphan, 13–31, Amsterdam: John
Benjamins.
Lausberg, H.(1976), Linguistica Romanza, Milano: Feltrinelli.
Liberman, Mark, and Alan S. Prince(1977), “On stress and linguistic rhythm,” Linguistic
Inquiry 8, 249–336.
Magni, Elizabetta(1999), “Paradigm organization and lexical connections in the development of the italian passato remoto,” In Geert Booij and Jaap van Marle (Eds.),
Yearbook of Morphology 1999, 75-96.
Maiden, Martin(2000), “Din cambiamento morfologico: origini del tipo dissi dicesti
ecc,” Archivio Glottologico Italiano 85, 137-171.
Maiden, Martin(2005), “Morphological autonomy and diachrony,” Yearbook of
Morphology 2004, 137-175.
Maiden, Martin(2010), “Morphological persistence,” In M. Maiden, J.C. Smiths, and
A. Ledgway (Eds.), The Cambridge History of the Romance Languages, Vol. I.
Cambridge: Cambridge University Press.
Marantz, Alec(1995), “A late note on late insertion,” in Young-Sun et al., Expolrations in
Generative Grammar, 396-413, Seoul: Hankuk.
Marantz, Alec(1997), “No escape from syntax: Don’t try morphological analysis in
the privacy of your own lexicon,” In Alexis Dimitriadis, Laura Siegel, Clarissa
Surek-Clark, and Alexander Williams (Eds.), Proceedings of the 21st Annual Penn
Linguistics Colloquium, 201-225, UPenn Working Papers in Linguistics 4(2),
Philadelphia: University of Pennsylvania, Penn Linguistics Club.
Marantz, Alec(1999), “The pieces of derivation,” Paper presented at the First
International Symposium on Linguistics, Lyon Institute for Cognitive Science,
Allomorphy in the Italian Passato Remoto: A Distributed Morphology Analysis | 73
October 1999.
McGinnis, Martha(2000), “Event heads and the distribution of psych-roots,” In
Alexander Williams and Elsi Kaiser (Ed.), Current Work in Linguistics, 107144, UPenn Working Papers in Linguistics 6(3), Philadelphia: University of
Pennsylvania, Department of Linguistics.
Mohanan, K.P.(1991), “On the bases of radical underspecification,” Natural Language
and Linguistic Theory 9, 285-326.
Nevins, Andrew(2010), Locality in Vowel Harmony, Cambridge, MA: MIT Press.
Nevins, Andrew(2011), “Marked targets versus marked triggers and impoverishment of
the dual,” Linguistic Inquiry 42, 413-444.
Noyer, Rolf(1992), Features, Positions, and Affixes in Autonomous Morphological Structure,
PhD dissertation, MIT, Cambridge, MA.
Noyer, Rolf(1998), “Impoverishment theory and morpho-syntactic markedness,” In
Steven G. Lapointe, Diane K. Brentari & Patrick M. Farrell (Eds.), Morphology
and its relation to Phonology and Syntax, 264-285, Stanford: CSLI.
McCarthy, John(2011), “An introduction to harmonic serialism,” ms., University of
Massachusetts, Amherst.
Meyer-Lübke, W.(1972), Grammatica storica della lingua italiana e dei dialettissuccedanei,
Torino: Loerscher.
Pirrelli, Vito & Marco Battista(2000), “The paradigmatic dimension of stem allomorphy
in italian verb inflection,” Rivista di Linguistica, 12, 307-380.
Prince, Alan S.(1983), “Relating to the grid,” Linguistic Inquiry 14, 19–100.
Oltra-Massuet, Isabel, and Karlos Arregi(2005), “Stress by structure in Spanish,”
Linguistic Inquiry 36, 43–84.
Oltra-Massuet, Maria Isabel(1999), On the Notion of Theme Vowel: A New Approach to
CatalanVerbal Morphology, Master’s Thesis, MIT.
Pirrelli, Vito, and Marco Battista(2000), “The paradigmatic dimension of stem allomorphy in italian verb inflection,” Rivista di Linguistica 12, 307-380.
Rizzi, L.(1990), Relativized Minimality, Cambridge, MA: MIT Press.
74 | 언어와 정보 사회 제18호
Rizzi, L.(2001), “Relativized minimality effects,” In Mark Baltin and Chris Collins
(Eds.), The Handbook of Contemporary Syntactic Theory, 89-110, Oxford:
Blackwell.
Rohlfs, Gerhard(1966-9), Grammatica storica della lingua italiana, Torino: Einaudi.
Sauerland, Uli(1995), “The late insertion of germanic inflection,” ms., MIT.
Tekavcic, Pavao(1980), Grammatica storica dell’italiano: I Fonematica, II Morfosintassi, III
Lessico, Bologna: Il Mulino.
Veselinova, Ljuba N.(2006), Suppletion in Verbal Paradigms, Amsterdam: John
Benjamins.
Vogel, Irene(1994), “Verbs in Italian morphology,” In Geert Booij and JaapVan Marle
(Eds.), Yearbook of Morphology 1993, 219-254, Dordrecht: Kluwer.
Andrea Calabrese
University of Connecticut Department of Linguistics
E-mail: andrea.calabrese@uconn.edu
Received: April 30, 2012
Revised: August 10, 2012
Accepted: November 16, 2012
Allomorphy in the Italian Passato Remoto: A Distributed Morphology Analysis | 75