xvii col·loqui de gramàtica generativa xvii colloquium on ... pdf/all_abstracts_cgg.pdf · xvii...
TRANSCRIPT
XVII Col·loqui de Gramàtica Generativa
XVII Colloquium on Generative Grammar
XVII Coloquio de Gramática Generativa
Girona, 13-15 June, 2007
TALKS
Enoch O. ABOH & Roland PFAU ‘What’s a wh-word got to do with it?’ 1
Karlos ARREGI & Andrew NEVINS ‘A principled order in postsyntactic operations’ 4
Elena E. BENEDICTO ‘Obligation without representation’ 7
Paola BENINCÀ ‘Some descriptive notes on relative clauses’ 9
Judy BERNSTEIN ‘Reconsidering Romance l-forms’ 10
Valentina BIANCHI & Cristiano CHESI ‘Quantifier raising in a top-down grammar’ 12
Theresa BIBERAUER, Anders HOLMBERG & Ian ROBERTS ‘Structure and
linearization in disharmonic word orders’ 15
Ana Maria BRITO ‘Nominalizations derived from unaccusative verbs: argument
structure and aspect value’ 18
Norbert CORVER ‘Silent person with semi-lexical identity’ 21
Maia DUGUINE ‘On the syntax of possessors and structural case’ 22
Ángel J. GALLEGO ‘An L-syntax for adjuncts’ 25
Anna GAVARRÓ ‘A first approach to Catalan and Spanish SLI’ 29
Susana HUIDOBRO & Jonathan MaCDONALD ‘Two properties of Spanish
non-argumental clitic pronouns’ 32
Gergely KANTOR ‘Edge effects and comparative AdvP’ 35
Richard S. KAYNE ‘Some English (and Romance) Auxiliaries’ 37
Jorge LÓPEZ – CORTINA ‘Split questions, extended projections, and dialect variation’ 38
Núria MARTÍ ‘The among construction: Catalan (d’)entre’ 40
Jaume MATEU ‘On the L-syntax of Manner and Causation’ 43
Gabriela MATOS ‘Set-Merge and Pair-Merge in coordination and subordination’ 46
Marios MAVROGIORGOS ‘“Where did that clitic come from?”: the case
of non-subcategorized clitics in Greeek’ 49
Ía NAVARRO ‘On the argument structure of le-predicates in Mexican Spanish’ 52
Andrew NEVINS, Cilene RODRIGUES & Luis VICENTE ‘Preposition stranding
under sluicing in Western Romance’ 55
Rolf NOYER ‘Stem-adjacent blocking in morphological fission’ 58
Francisco ORDÓÑEZ & Esthela TREVIÑO ‘Unambiguous impersonal se’ 61
Susanna PADROSA TRIAS ‘On the nature of the Catalan prefix entre-’ 64
Cristina REAL PUIGDOLLERS ‘The nature of cognate objects in Romance’ 67
Cilene RODRIGUES ‘Middles: transforming events into states’ 70
Jason ROTHMAN & Michael IVERSON ‘Syntax-before-morphology in
L2 acquisition: semantic evidence from aspect in L2 Portuguese’ 73
Luka SZUCSICH ‘Agreement and case: Impersonal sentences in Slavic and
accusative and dative NPs’ 76
Juan URIAGEREKA & Ángel GALLEGO ‘Subjunctive dependents’ 79
Vidal VALMALA ‘The syntax of little things’ 82
POSTERS
Bakhta ABDELHAY & Ouoahmiche GHANIA ‘Consonant clusters and phonotactic
constraints’ 85
Roberta D’ALESSANDRO & Ian ROBERTS ‘Movement and agreement in Italian
past participles and spell-out domains’ 86
Dorothy BEZERRA SILVA DE BRITO ‘Reflexive se and agreement in Brazilian Portuguese’ 89
Daniel CARVALHO ‘Agreement weakening in Brazilian Portuguese: the complementizer se’ 90
Mirian SANTOS DE CERQUEIRA ‘Subject-verb agreemant in Brazilian and
European Portuguese: the case of the partitive constructions and complex DPs’ 92
XVII Colloquium on Generative Grammar – Girona, 13th-15th June 2007
1
What’s a wh-word got to do with it?
Enoch.O. Aboh ([email protected]) & Roland Pfau ([email protected])
University of Amsterdam
Wh-questions are traditionally taken to require a wh-phrase, which expresses question and focus
features [wh, focus] and moves at the relevant level to SpecCP (syntax vs. LF) to check these features
under C (Chomsky 1977; Hovarth 1986; Kiss 1998). This generalization appears robust and
empirically motivated.
This paper argues against the traditional view and demonstrates that the occurrence of a wh-phrase is
not a necessary condition for wh-questions. Instead, questions uniformly require a left peripheral
(c)overt question particle or operator in InterP. Put another way, Inter clause- types both yes-no and
wh-questions. Empirical support for this view comes from Indian Sign Language (IndSL), Sign
Language of the Netherlands (NGT), and Wan (spoken in Brazil).
IndSL, for instance, has only one monomorphemic question sign (G-wH), which obligatorily appears
in sentence-final position (1) (Zeshan 2003). In such questions, the questioned D-linked DP has to be
retrieved from context. In vague contexts, however, G-WH combines with various associate phrases
(e.g. PLACE, TIME) to express more specific meanings (e.g. ‘where’, ‘when’) (2). Based on these
observations, we argue that IndSL does not involve overt wh-phrases. Instead, we analyze G-WH as a
wh-question particle that merges under Inter within the C-system, which is licensed under Spec-head
relation (Rizzi 1997, 2001). This licensing requirement forces pied-piping of FinP into SpeclnterP,
hence G-WH occurs sentence-finally (3). In addition, we propose that the pied-piped FinP includes a
silent whphrase (a generic DP or a variable (Cheng 1991)), or else an associate phrase that may be
attracted by the focus head (Foc) prior to movement. This analysis extends to wh-questions with silent
wh-phrases in various SLs (e.g., the NGT examples in (4)). In those cases, too, the interpretation of the
question is context-dependent.
The discussion on IndSL is particularly insightful in regard to certain spoken languages with no proper
wh-words, as e.g. Wan (Everett & Kern 1997). In neutral Wan sentences, the word order is V-O-S-
(XP). Wh-questions are formed thanks to the sentence-initial question operator ma’ ‘that’ derived
from a demonstrative, a special inflection on the verb/predicate (co/ca), and an optional generic DP
(xec ‘day’) comparable to the associate phrase in IndSL (5). The combination of ma’ and INFL
determines the target of wh-questions. These generally follow the pattern in (6) for subject and object
DPs. We propose that IndSL and Wan show parallels in that the two languages lack proper wh-phrases
and resort to generic DPs bound by a higher element in Inter. However, while IndSL involves a
question particle that merges in Inter, we propose that Wan involves a question operator in SpeclnterP.
These data lead us to conclude that Inter is active in all question types. Extending this analysis to wh-
movement languages (e.g., English) we show that there is no direct correlation between interrogative
clause-typing and wh-expressions, and by implication between clause- typing and wh-movement. For
instance, wh-phrases occur in both interrogative and declarative contexts and do not always bear
interrogative force. We therefore argue that in apparently uncontroversial examples (e.g., What have
you prepared?), interrogative force is encoded by a (null) question marker in InterP, whose presence
can be detected (e.g., in English) through intonation or other morphosyntactic phenomena typically
found in questions (e.g., subject-auxiliary inversion). We conclude from this that wh-phrases occur in
questions for interpretive reasons. We suggest that wh-questions involve essentially two operations:
clause-typing and fixing the target of question (i.e., interpretation of the wh or indefinite phrase). The
two operations are properties of Inter, which clause-types and Foc, which hosts an operator that
assigns a range to the variable that represents the target of the question. Finally, the distribution of wh-
phrases is shown to derive from their internal make-up.
Examples
(1) a. BOOK BUY G-WH ‘Who buys/bought a book?’
b. CHILD CRY G-WH ‘Why does the child cry?’
(2) a. INDEX2 SLEEP [PLACE G-WH]
‘Where do/did you sleep?’ b. TRAIN ARRIVE [TIME G-WH] ‘When does/did the train arrive?’
(3) [InterP [FinP BOOK BUY] [Inter G-WH [ tFinP ] ] ] top wh
(4) a. SHOP INDEX3 INDEX2 BUY ‘What did you buy in this shop?’ wh
b. INDEX1 SUITCASE ‘Where is my suitcase?’
(5) a. Ma’ co pa’ nana hwam pain xec ca’ ne that.PROX.H M/F.RP/P kill 3PL.RP/P fish PREP.N day this.N REC.P ‘Who killed fish this morning?’
b. Ma’ carawa ca pa’ caca mon tarama’ that.PROX.H animal N.RP/P kill 3PL.M COLL man ‘What thing/animal did the men kill?’
(6) a. Q. . .(DP). . . .INFLOBJECT. . . .V-S-XP [object wh-question] b. Q INFLSUBJECT. . . .V-O—XP [subject wh-question]
References
Cheng, L. (1991). On the Typology of Wh-questions. Ph.D. dissertation, MIT [distributed by
MITWPL].
Chomsky, N. (1977). On Wh-movement. In: Culicover, P.W. (ed.), Formal Syntax. New York:
Academic Press.
XVII Colloquium on Generative Grammar – Girona, 13th-15th June 2007
3
Everett,, D.L. & B. Kern (1997), Wan’. The Pacaas Novos language of Western Brazil. London:
Routledge.
Horvàth, J. (1986). Focus in the Theory of Grammar and the Syntax of Hungarian. Dordrecht:
Foris.
Kiss, K. E. (1998). Identificational focus versus information focus. Language 74: 245-273.
Rizzi, L. (1997). The fine structure of the left periphery. In Elements of Grammar. Handbook in
Generative Syntax, Liliane Haegeman (ed.), 28 1-337. Dordrecht: Kluwer.
Rizzi, L. (2001), On the position “Int(errogative)” in the left periphery of the clause. In Current
Studies in Italian Syntax: Essays Offered to Lorenzo Renzi, G. Cinque & G. Salvi (eds.),
287-296. New York: Elsevier.
Zeshan, U. (2003), Indo-Pakistani Sign Language grammar: a typological outline. Sign Language
Studies 3: 157-2 12.
A Principled Order to Postsyntactic Operations
Karlos Arregi ([email protected]) & Andrew Nevins ([email protected])
University of Illinois at Urbana-Champaign& Harvard University
1. Overview. Although a suite of operations have been argued for within the post-syntactic component
of Distributed Morphology (e.g. obliteration, impoverishment, morphological metathesis, etc.), little
effort has been devoted to the question of the relative order of these operations.
Embick & Noyer (2001) present a notable exception, motivating a distinction between those
dislocation operations that are sensitive to hierarchical structure and those that operate purely on linear
sequences of phonological material. In this paper we examine the relative ordering of feature-sensitive
operations that are demonstrably post-syntactic but which occur prior to Vocabulary Insertion (i.e.
phonological realization). We argue that the necessary ordering of two operations in Basque follows
from a principled division of those which occur prior to the linearization of syntactic terminals and
those which occur after linearization (1), and we provide new empirical motivation for a feeding
relation between the two.
2. Ergative displacement (ED). Basque finite verbs have the structure in (2), illustrated in (3). (ABS,
DAT, ERG are absolutive, dative and ergative clitics.) Basque also has a word- internal Non-Initiality
requirement (4) which bans T-initial finite verbs (Laka 1993, Albizu & Eguren 2000). Cliticization of
the absolutive argument, as in (3a), yields a structure that satisfies Non-initiality. This structure is
derived in the syntax, by moving the clitics from their argument position to T. However, Basque third
person absolutive arguments do not undergo clitic doubling. As a result, one of two operations may
occur. Ergative Displacement (5), a metathetic operation whereby the ergative clitc is dislocated to
initial position, is one of these morphological operations, as illustrated in (6). This operation is limited
to the past tense. In the present, and in verbs without an ergative clitic, Non-initiality is ensured by a
last resort insertion of an epenthetic proclitic, realized as d- in the present (3b) (Laka 1993, Albizu &
Eguren 2000). Non-initiality, as well as ED and proclitic epenthesis, are clearly sensitive to linear
order: the constraint bans finite verbs whose initial morpheme is T, and the repair operations move or
insert material in initial position. Hence they apply after linearization. Another crucial property of
these operations is that they must occur before vocabulary insertion: when an ergative clitic is
displaced by ED, it takes the form of a proclitic (e.g. lSG n- in (6b)), as opposed to its normal
(phonologically unrelated) enclitic realization (e.g. -t in (3b)).
3. The g/z constraint. Many dialects do not allow certain combinations of 1PL with 2nd person clitics,
with microvariation in the exact conditions (Arregi & Nevins 2006). In Ondarru, the constraint is
instantiated as a ban on auxiliaries with 1PL.ABS and 2.ERG ((3a) is banned). This triggers obliteration
of 1PL.ABS (7). Both the constraint and its obliteration repair are sensitive only to abstract
XVII Colloquium on Generative Grammar – Girona, 13th-15th June 2007
5
morphosyntactic features: they affect certain Auxes with two nodes specified as [+Participant]. Note
that deletion of the ABS morpheme in (I) leaves T word-initial. subsequently triggering the context for
proclitic epenthesis. Unlike Non-initiality and ED, the relative linear order of the clitics subject to g-
/z- is irrelevant: various repairs exist; some delete the absolutive and others delete the ergative. Thus,
they apply before linearization.
4. g/z precedes ED. The current analysis predicts that obliteration (ordered before linearization) feeds
ED and proclitic epenthesis (ordered after linearization). This feeding relation is implicit in our
analysis of (7): obliteration of ABS due to g-/z- triggers proclitic epenthesis in the present. (8) shows
that obliteration feeds ED in the past: obliteration of ABS triggers proclisis of ERG. If ED/proclitic
epenthesis were ordered before obliteration, incorrect forms would be derived (9). The data thus
strongly support the existence of a derivational sequence of post-syntactic operations, those which
precede and follow linearization of hierarchical terminals.
Abbreviations: abs(olutive), dat(ive), EP(enthetic proclitic), erg(ative), prf (perfective), sg (singular),
pl(ural)
Examples
(1) Syntax → * Hierarchically-sensitive operations → * Linearization → * Linearly-sensitive operations → * Vocabulary Insertion (VI)
(2) Basque finite auxiliaries [T [T [T ABS T ] DAT ] ERG ] (Past auxiliaries also have an additional Past suffix at the end)
(3) a. Zu-k gu-Ø ikus-i g-aitu-zu. you-ERG us-ABS see-PRF ABS.1PL-T-ERG.2SG ‘You have seen us.’ b. Ni-k Jon-i liburu-a ema-n d-i-o-t. I-ERG Jon-DAT book-ABS.SG give-PRF EP-T-DAT.3SG -ERG.1SG ‘I have given Jon the book.’
(4) Non-initiality: T cannot be the first morpheme within a finite auxiliary.
(5) Ergative Displacement: [T T ERG ] → [T ERG T ]
(6) a. Zu-k liburu-a irakurr-i z-enu-en. you-ERG book-ABS.SG read-PRF ERG.2SG-T-PAST ‘You read the book.’ b. Ni-k Jon-i liburu-a ema-n n-i-o-n. I-ERG Jon-DAT book-ABS.SG give-PRF ERG.1SG-T-DAT.3SG-PAST ‘I gave the book to Jon.’
(7) Su-k gu-Ø ikus-i d-o-su. (Ondarru) you-ERG us-ABS see-PRF EP-T-ERG.2SG ‘You have seen us.’
(8) a. Su-k gu-Ø ikus-i s-endu-n. (Ondarru) you-ERG us-ABS see-PRF ERG.2SG-T-PAST ‘You have seen us.’
b. ABS.1PL - T - ERG.2SG - PAST Oblit. T - ERG.2SG — PAST ED . ERG.2SG - T - PAST VI . √ s - endu - n
(9) a. Wrong derivation for (7) ABS.1PL - T - ERG.2SG Epenth. Not Appl. Oblit . T - ERG.2SG - VI . * o – su
b. Wrong derivation for (8) ABS.1PL - T - ERG.2SG - PAST ED . Not Appl. Oblit. . T - ERG.2SG - PAST
VI . * endu –su - n
References
Albizu, Pablo, and Luis Eguren. 2000. An optimality theoretic account for “ergative displacement” in
Basque. In Morphological analysis in comparison, ed. Wolfgang U. Dressler, Oskar E. Pfeiffer,
Markus A. Pochtrager, and John R. Rennison, 1-23. Amsterdam: John Benjamins;
Arregi, Karlos, and Andrew Nevins. 2006. Obliteration vs. Impoverishment in the Basque g-/z-
constraint. In Penn Linguistics Colloquium Special Session on Distributed Morphology, (also
available at http://ling.auf.net/lingBuzz/000280). U. Penn Working Papers in Linguistics 13.1;
Laka, Itziar. 1993. The structure of inflection: A case study in X° syntax. In Generative studies in
Basque linguistics, ed. José Ignacio Hualde and Jon Ortiz de Urbina, 2 1-70. Amsterdam: John
Benjamins.
XVII Colloquium on Generative Grammar – Girona, 13th-15th June 2007
7
Obligation without Representation.
Elena E. Benedicto ([email protected])
Purdue University
Mayangna, a language of the isolated Misumalpan family, shows a variety of morphosyntactic structures
for expressing modality: (a) a stative form sip ‘be able to’ (for possibility modality) which can be used in
a serial construction or with a more traditional (inflected) infinitival dependent constituent; (b) a set of
items borrowed from English (nît, nû, laik,…) used with a dependent infinitival constituent; and (c) an
inflected infinitival, with no visible element indicating modality, which directly attaches to (main clause)
tense and negation, and which may or may not present sentence final particles; this structure encodes
universal quantificational force (necessity). The present work deals with this last case.
Let us consider an example illustrating type (c):
(1) yang wauhtaya yak ulnik (ki).
PRN.1s notebook in.P write.INF1sg PTC[3s?]
‘I must write in the notebook’
In the preceding example we can observe the following properties: the lexical verb ulnik appears in the
infinitival form, inflected with the morpheme corresponding to the 1st person singular (-ik) for non-finite
forms. The subject ‘yang’, on the other hand, appears in its Nominative form. Finally, the particle ki,
typically but not exclusively associated with stative main clauses with a 3rd person singular subject,
appears optionally. It is worth noting that the presence/absence of this sentence-final particle ki has
consequences for the choice of particular subtypes of modality: when present, it conveys the meaning of
strong obligation; when absent it conveys the meaning of strong intention or imminence. None of these
elements, however, can by themselves convey the meaning of necessity modality; they all appear in a
variety of structural contexts where no modality is encoded: infinitival constructions can appear as the
complements of verbs of language and thought, and the sentence final particle ki appears, as we
mentioned, in stative sentences (but only for 3rd person singular subjects).
It is the particular combination of these elements in a single structural unit that provides the modality
reading. Evidence for (1) constituting a monoclausal unit comes from, among other, its behavior with
respect to tense and negation (the main inflectional nodes associated with the clausal architecture). The
forms used for tense and negation are the ones used for non-verbal predicates and are directly attached to
the infinitival unit:
(2) yang wauhtaya yak ulnik awas (ki).
PRN.1s notebook in.P write.INF1sg Neg PTC
‘I must not write in the notebook’
(3) yang wauhtaya yak ulnik dai.
PRN.1s notebook in.P write.INF1sg PastPTC
‘I had to write in the notebook’
The hypothesis to be presented here will postulate that the non-finite morpheme in the lexical verb has the
effect of creating an intensional domain (but not necessarily a syntactic subordination) and that the
different types of sentence-final particles (phonologically realized or not) act as operators on that
intensional domain.
From a crosslinguistic point of view, this type of inflectionally built modality can be found in languages
like Tigrinya, in Eritrea, or Latin (with its gerundive and future participial forms, with meanings very
close to the ones described for Mayangna).
XVII Colloquium on Generative Grammar – Girona, 13th-15th June 2007
9
Some descriptive notes on relative clauses
Paola Benincà
Università di Padova
I will present some notes on relative clauses in early and modern varieties of Italian and compare some
of their characteristics with Middle and Modern English.
I will mainly deal with headless relatives and 'non-identifying' restrictive relatives.
The phenomena observed have to do with overt Case and Verb Agreement.
In early Italian (Florentine) case distinctions are still preserved in interrogative [+human] wh's; since
these pronouns are used in headless relatives, we are able to ascertain from where the wh pronouns get
Case in this kind of relative. The presence of morphological case seems to allow headless relatives to
have a Case on the wh which does not match– or is morphologically distinct from - the case of the
antecedent. This will be shown by comparing early Italian with Middle and Modern English and
Modern Italian.
From the available data, it appears that headless relatives are the first structures where the so-called
'doubly filled Comp Filter' is violated in the history of both English and Italian varieties.
Case seems to indirectly play a role with respect to interesting properties of relatives whose antecedent
is the predicate of a copular sentence (such as John is a man who loves his children). As a predicate,
the antecedent does not get nominative case directly but inherits it from the subject of the relative ("I
am a man who love1st.sg. his children".
This kind of relative seems restrictive, but has special characteristics: in early Italian, for example, one
can observe that in relatives whose antecedent is the predicate of a copular sentence with a 1. or 2.
person subject, the relative agrees not with its antecedent but with the subject of the copular sentence.
This means that the relative is connected in some way to the subject, which probably contributes to the
identification of the antecedent.
Reconsidering Romance l- Forms
Judy B. Bernstein ([email protected])
William Paterson University
Romance third person clitic pronouns are isomorphic with definite articles in many Romance
languages, often displaying word-initial l-. The examples in (1) and (2) illustrate this for Spanish. This
(near) isomorphism between definite article and clitic pronoun suggests a parallel internal structure. If
article and clitic l- forms both correspond to D, the difference may reduce to a lexicalized NP
complement in the case of articles (la mesa) and an empty NP complement in the case of clitics (la Ø;
see Uriagereka 1995).
In this talk, I suggest that the relevant parallelism between the clitics and articles is not
definiteness, but rather ‘person’. In other words, what the clitic and article l- forms share is an overt
marker for (third) ‘person’. Retaining the idea that l- forms occupy D, I argue more generally that D is
not inherently relevant to definiteness, but rather to ‘person’ (see Bernstein 2006, Longobardi 2006).
The relationship between pronoun and determiner in (1) and (2) recalls work by Postal (1966).
Postal argued for English that cases of [pronoun + noun] as in (3a) correspond to those of [determiner
+ noun] as in (3b) (see also Roehrs 2005). Retaining Postal’s idea about the parallelism between
pronoun and determiner, I take the relevant feature to be ‘person’, not definiteness. In other words,
what unifies the examples in (3) is ‘person’, the underlying feature of D. What the definite article in
(2), (3b), and below contributes to the DP is (third) ‘person’, which is not otherwise expressed with
common nouns in English and Romance languages.
The very idea that ‘definiteness’ could be the relevant feature of definite articles (and also of
D) is problematic because definite articles in Romance languages do not always receive or contribute a
definite interpretation. In fact, the definite article appears in many contexts where definiteness does
not obtain at all, such as with generics (as in (4)), nominalized verbs (as in (5)), and non-specifics (as
in (6)). Interestingly, a definite article may even appear in contexts where the interpretation is
possessive, as in (7). I claim that the definite article in (4)-(7) encodes third ‘person’.
In still other contexts, the appearance of the definite article is somewhat mysterious. In several
Romance languages, for example, the definite article appears with proper names, as in (8) (discussed
by Longobardi 1994 and others). What is interesting here is that proper names like Maria are
inherently definite and referential and so the definite article should not be necessary, resulting in labels
like ‘expletive article’. I claim instead that the definite article la in (8) is the expression of third
person, never expressed directly on proper names in Romance languages.
The analysis builds on the novel idea that a grammatical property (‘person’) and not a
semantic one underlies D. If on the right track, the approach suggests a sharper division between more
fundamental grammatical features of noun phrases (e.g., ‘person’, ‘number’, ‘gender’) and interpretive
properties such as ‘definiteness’ and ‘reference’.
XVII Colloquium on Generative Grammar – Girona, 13th-15th June 2007
11
Examples
(1) Juan la compró. (Spanish) ‘Juan bought it (F.SG.).’
(2) La mesa está en el comedor. (Spanish) ‘The (F.SG.) table is in the dining room.’
(3) a. vous linguistes (French) ‘you linguists’
b. les linguistes ‘the linguists’
(4) La zorra es celebrada por su astucia. (generic) (Spanish) ‘The fox is celebrated for its cunning.’
(5) El leer es fundamental. (verbal noun) (Spanish) ‘Reading is fundamental.’
(6) Nunca fui a la playa en México. (non-specific) (Spanish) ‘I never went to the beach in Mexico.’
(7) Veronique veut sortir à manger avec la soeur. (possessive) (French) ‘Veronique wants to go out to eat with her sister.’
(8) La Maria parla italiano. (Italian) ‘(The) Mary speaks Italian.’
References
Bernstein, Judy B. 2006. “English th- Forms,” to appear in Alex Klinge and Henrik Müller
(eds.) Essays on Nominal Determination. Amsterdam: John Benjamins.
Longobardi, Giuseppe. 1994. “Reference and proper names: A theory of N-movement in
syntax and logical form,” Linguistic Inquiry 25: 609-665.
Longobardi, Giuseppe. 2006. “Reference to Individuals, Person, and the Variety of Mapping
Parameters,” to appear in Alex Klinge and Henrik Müller (eds.) Essays on Nominal
Determination. Amsterdam: John Benjamins.
Postal, Paul M. 1966. “On So-Called ‘Pronouns’ in English,” in Francis P. Dineen (ed.)
Monograph Series on Languages and Linguistics 19. Washington, DC: Georgetown
University Press, 177-206.
Roehrs, Dorian. 2005. “Strong pronouns are determiners after all,” In Marcel den Dikken and
Christina Tortora (eds.), The Function of Function Words and Functional Categories.
251-285. Amsterdam: John Benjamins.
Uriagereka, Juan. 1995. “Aspects of Clitic Placement in Western Romance,” Linguistic
Inquiry 26: 79-124.
Quantifier Raising in a top-down grammar
Valentina Bianchi ([email protected]) & Cristiano Chesi ([email protected])
University of Siena
The covert movement operation of Quantifier Raising cannot cross a finite clause boundary ((1); see
Reinhart 1997 on the exceptional behaviour of indefinites), though it can cross certain non-finite
clauses (e.g. (2), Cecchetto 2004). In the standard bottom-up grammar, this means that QR cannot
access the edge of a finite embedded CP phase, so that it cannot undergo cyclic movement into the
matrix phase. Cecchetto’s elegant account of (1)-(2) rests on the assumption that QR cannot have
semantically unmotivated intermediate steps, unlike feature-driven instances of movement. QR thus
remains a “unicum” (though a well-behaved one w.r.t. the PlC):
(a) it is covert;
(b) it is not feature-driven;
(c) it cannot be freely cyclic.
In this paper we argue that the unique properties of QR can be properly understood from a top-down
derivational perspective. From this perspective, leftward movement is a head-first dependency, in
which the syntactic system computes the displaced head first, and then the rest of the chain, looking
for a selected position where the displaced element can be reintegrated (Chesi 2004); on the contrary,
QR is an operation which computes a selected position and removes from it an element (QP) which
cannot be interpreted there. It follows that QR is actually a rightward movement (Fox/Nissenbaum
1999). Specifically, we propose that QR:
(a) stores a QP in a dedicated memory buffer of the current phase (Chesi 2004, Schlenker
2005);
(b) integrates a coindexed variable in the corresponding argument position;
(c) when the top-down computation of the current phase is concluded, the QP function is
retrieved from the Q-buffer and takes scope over the structure.
The elements retrieved from memory buffers are (typically) not spelled out (Chesi 2004); hence QR is
covert, complying with a weak LCA. CP and DP phases are computed top-down, and a phase is closed
off after its lexical head is computed; the recursive right branch constitutes the next sequential phase.
E.g., in the computation of (3), CP is the first phase computed and QP2 constitutes the sequential
phase; the subject QP1 is instead a nested phase, which is computed while CP is still incomplete. The
nested QP1 is stored in the Q-buffer of the containing phase (Phi); the sequential QP2 is stored in the
Q-buffer of the immediately previous phase (again Phi). The two QP functions are stored in the same
Q-buffer and can be retrieved in either order, yielding scope ambiguity. The clause-boundedness of
QR follows from the computational sequencing of phases. The computation of(1) is illustrated in (4).
The matrix subject QP1 is stored in the Q-buffer of the containing phase Phi, while a variable x1 is
inserted in the argument position; this Q-buffer is also inherited by the sequential phase Ph3. The
XVII Colloquium on Generative Grammar – Girona, 13th-15th June 2007
13
embedded subject QP2 too is stored in the Q-buffer of Ph3, but it cannot “get into” the Q-buffer of the
previously computed Phi. As a result, QP2 will only have scope over the embedded CP2, whereas QP
1 will have scope over both CP1 and CP2. Thus, the phase boundaries determined by our top-down
model, though not corresponding to complete subtrees, derive the clause-boundedness of QR,
subsuming one instance of the mysterious “right roof constraint”. As for (2), we critically re-examine
Cecchetto’s (2004) data, and argue that long QR is only possible out of non-finite (or subjunctive)
clauses which do not constitute independent phases.
Finally, the top-down derivation also accounts for the Leftness Condition. We assume that a pronoun
can be Q-bound only by being coindexed with a variable; since the variable is inserted by the
application of QR, coindexing will be impossible whenever the pronoun is computed before the QP, as
is the case in a WCO configuration like (5) (cf. Schlenker 2005, Shan & Barker 2006,). However, Q-
binding will be possible whenever the computation of the pronoun follows the application of QR, even
if the variable position fails to c-command the pronoun (cf. Bianchi 2001). We conclude that the top-
down derivational perspective allows for a more principled account of the exceptional, yet
indispensable, operation of QR.
Examples
(1) Someone expected [CP that every Republican would win]. (•>•.; *•>•)
(2) Someone expected [IP every Republican to win]. (•>•.; •>•)
Ph1 Ph3 (3) [CP [QP1 A student] admires] [QP2 every professor]. Ph2
Ph1 Ph3 (4) [CP1 [QP1 Someone] expected x1] [CP2 that [QP2 every republican] would win x2
Ph2 Ph4
QPhl: QP1 QPh3: QP1 , QP2
Ph1 Ph3
(5) * [CP1 [DPi His wife] loves] [QP2 every man]. *•x: x’s wife loves x Ph2
Selected references
Bianchi, V. 2001. Antisymmetry and the Leftness Condition: Leftness as anti-c-command.
StudiaLinguistica 55, 1-38.
Bianchi, V. & C. Chesi. 2006. Phases, left branch islands, and computational nesting. U Penn Working
Papers in Linguistics 12.1, 15-28.
Cecchetto, C. 2004. Explaining the locality conditions of QR: Consequences for the theory of phases.
Natural Language Semantics 12, 345-397.
Chesi, C. 2004. Phases and Cartography in Linguistic Computation. Doct. Diss., University of Siena.
Fox , D.& Nissenbaum, J. 1999. Extraposition and scope: A case for overt QR. Proceedings of
WCCFL 18, 132-144.
Reinhart, T. 1997. Quantifier scope: how labor is divided between QR and choice functions.
Linguistics and Philosophy 20, 335-397.
Sauerland, U. 2005. DP is not a scope island. Linguistic Inquiry 36, 303-314.
Schlenker, P. 2005. Non-redundancy: Towards a semantic reinterpretation of Binding Theory. Natural
Language Semantics 13, 1-92.
Shan, C. & C. Barker. 2006. Explaining crossover and superiority as left-to-right evaluation.
Linguistics and Philosophy 29, 91-134.
XVII Colloquium on Generative Grammar – Girona, 13th-15th June 2007
15
Structure and Linealization in Disharmonic Word Orders
Theresa Biberauer ([email protected] ), Anders Holmberg ([email protected] ) &
Ian Roberts ([email protected] )
Cambridge University & Newcastle University
A. ‘Mixed’/ ‘disharmonic’ word-order systems have created particular difficulties for parametrically
based accounts of word-order variation. Here we formulate and motivate a generalisation about
disharmonic systems which we conjecture has universal validity for word-order typology, and which
we derive from current notions of cyclicity and linearization.
B. Holmberg (2000) proposes the Final-over-Final Constraint (FOFC) in 1 and schematized in 2.
Evidence for FOFC includes i.a. the following:
1) Old and modem Germanic varieties exhibit a mix of head-initial and head-final orders in VP
and IP, with all permutations of Aux, V and Object attested except one (den Besten &
Edmondson 1983, Hróarsdóttir 2002) — cf. 3. The unattested order (f) violates FOFC. Finnish,
which also permits mixed orders, shows the same pattern.
2) Sentence-final complementisers are not found in VO languages (Hawkins 1990). This is
since both [CP [TP [VP V O] T] C] and [CP [TP T [v V O]] C ] violate FOFC. Thus, there is no way
for VO languages to have TP C order.
3) In the nominal domain, Finnish has mixed projections too: it has both pre- and postpositions
and N-Complement (N-O) as well as Complement-N order. All permutations of P, N and 0 are
found except N-O-P, the FOFC-violating order.
C. However, a range of constructions apparently violate FOFC:
a) Head-final VP may immediately dominate head-initial DP/PP in OV Germanic — cf. 4.
b) Many VO & I VP languages have clause-final “force” particles (e.g. Mandarin). If these
particles are C-elements, they violate FOFC.
c) Many VO languages have clause-final negation. If the negation is a head taking a VP, TP or
CP-complement, these are potential counterexamples to FOFC.
d) Although SVOAux is extremely rare globally (Julien 2000), some Central Sudanic languages
(Logbara, Mamvu [Tucker & Bryan 1966] and Mande [Kastenholz 2003]) allow it; and VO
order with clause-final tense/aspect particles is also attested in some East Asian languages
(Thai, Chinese).
D. The central ideas in our account of FOFC are:
(i) any head may have an EPP-feature,
(ii) the LCA, and
(iii) linearization is cyclically determined by phase-heads according to the (strict version of) the
PlC so that material sent to Spellout upon completion of a phase can no longer be affected by
any subsequent linearization process, including movement.
Against this background, we propose the preliminary generalisation in 5, which, applied to the vP-
phase, delivers 6. The FOFC violation in 1 (*VOAux) thus falls under 6d: for [VPVO] to precede Aux,
it must move to or through the specifier of vP, i.e. v must have an EPP-feature, while V does not, in
violation of 5. If auxiliaries are in v, VOAux violates 6d directly; if auxiliaries are in T, 6d is an
intermediate stage of the derivation. VOAux is therefore underivable. The order in 2 cannot be derived
via intermediate step 6d. The only option is thus pied-piping an entire head-initial vP (i.e. 6c) to
SpecTP. However, in this case, the object must appear in clause- final position, having been “frozen in
place” by the PlC as VP is spelled out upon completion of the vP phase. The nominal cases of FOFC
are identical to 6d with the single difference that the categories are n/N.
E. The central observation regarding the exceptions to FOFC is that there is a categorial distinction
between the phase-head and the moved category. This is very clear for German a: PP/DP and v are
distinct categories. 5 should thus be reformulated as 7, which is very much in the spirit of Chomsky’s
(2005) idea that phase-heads determine many of the properties of the heads in their phrasal domain,
including their ability to act as probes and/or movement triggers. 7 implies the configuration of FOFC-
obeying and -violating cases in 8 (where n/N and v/V represent categorially like/unlike heads).
Counterexamples a-d all fall under 8c,d. The German case is clearly 8d. Following Holmberg (1986),
we assume that C may be nominal or verbal in nature, while T is always verbal. We thus predict FOFC
violations where C is nominal. This appears to be true for the Mandarin final force particles in b (Li
2006) and the final negative elements (polarity markers) in c. Lastly, the final “auxiliaries” in the
putative VOAux orders in various African languages have independently been argued to be nominal
(Fabb 1992, Kastenholz 2003), making these orders cases of 8c. Our proposal that some Aux in these
languages = n leads to the prediction that when the relevant nEPP selects N, N will have an EPP-feature
(cf. 6a). We thus expect rigid head-final order in “pure” nominal phases, which seems to be true.
F. Studying FOFC, both its violations and non-violations, in addition to its obvious typological
interest, may provide an empirical basis for distinguishing categories that are phase-heads from those
that are not.
Data
(1) Final-over-Final Constraint If α is a head-initial phrase and β is a phrase immediately dominating x, then β must be head- initial. If α is a head-final phrase, and β is a phrase immediately dominating c then β can be head- initial or head-final. (cf. also Julien 2000)
(2) * [Head1’ [Head2P Head2 Compl] Headl ]
(3) a. AUX V O b. AUX O V c. O V AUX d. O AUX V e. V AUX O f. *V O AUX
XVII Colloquium on Generative Grammar – Girona, 13th-15th June 2007
17
(4) a. Johann hat [VP [DP den Mann] gesehen] [German] Johann has the man seen
b. Johann ist [VP [PP nach Berlin] gefahren] Johann is to Berlin driven
(5) If a phase-head PH has an EPP-feature, all the heads in its complement domain must have an EPP-feature.
(6) a. vEPP VEPP � [ [VP O V ] v ] (consistent head-final order) b. v VEPP � [ v [VP O V ] ] (disharmonic non-FOFC-violating order) c. v V � [ v [VP V O ] ] (consistent head-initial order) d. *vEPP V � [ [VP V O ] v] ] (FOFC-violating order)
(7) If a phase-head PH has an EPP-feature, all the heads in its complement domain with which it Agrees in categorial features must have an EPP-feature.
(8) a. nEPP N FOFC violation b. vEPP V FOFC violation c. nEPP V allowed d. vEPP N allowed
References
Den Besten, H. & J. Edmondson (1983). The Verbal Complex in Continental West Germanic. In:
Abraham, W. (ed.). On the Formal Syntax of the Westgermania. Amsterdam: Benjamins, 155-
216.
Chomsky, N. (2005). On phases. To appear in: Otero, C. et al. (eds.). Foundational Issues in
Linguistic Theory.Cambridge, Mass.: MIT Press.
Fabb, N. (1992). Reduplication and object movement in Ewe and Fon. Journal of African Language
andLinguistics 13: 1-39.
Hawkins, J. (1990). A parsing theory of word-order universals. Linguistic Inquiry 21: 223-261.
Holmberg, A. (1986). Word order and syntactic features in the Scandinavian languages and English.
Ph.D. dissertation: University of Stockholm.
Holmberg, A. (2000). Deriving OV order in Finnish. In: Svenonius, P. (ed.). The derivation of
VOandOV. Amsterdam: Benjamins, 123-152.
Hróarsdóttir, Th. (2000). Word order change in Icelandic: from OV to VU. Amsterdam: Benjamins.
Julien, M. (2000). Syntactic heads and word formation: a study of verbal inflection. Oxford: OUP.
Kastenholz, R. (2003). Auxiliaries, grammaticalization and word order in Mande. Journal of African
Languages and Linguistics 24(1): 3 1-53.
Li, B. (2006). Chinese Final Particles and the Syntax of the Periphery. Ph.D. dissertation: Leiden.
Tucker, & Bryan (1966). Linguistic Analyses: the non-Bantu Languages of Northeast Africa. Oxford:
OUP.
Nominalizations derived from unaccusative verbs: argument structure and aspect values
Ana Maria Brito ([email protected])
Universidade do Porto
As nominalizations derived from unaccusative verbs (in particular verbs of appearance / disappearance
and movement) have an internal argument but no external argument, it is expected that the argument
cannot be expressed by a referential adjective (Kayne 1984, Giorgi & Longobardi 1991, Picallo 1991,
Picallo 1999, Bosque & Picallo 1996, a.o.). Bosque & Picallo 1996 and Picallo 1999 proposed that, if
these nominals admit a referential adjective, stopping the projection of the internal argument by a
genitive, as in:
(1) a. the American entrance in Iraq b. the Soviet arrival in Afghanistan
that excludes the event reading and permits only the result reading.
The main goal of the paper is to reconsider this proposal, mainly analysing European Portuguese.
In the classical bibliography about this matter, it has been proposed that de-verbal nominalizations
may have two readings: event / process and result (Milner 1982, Grimshaw 1990, Picallo 1991,
Alexiadou 2001, a. o.). Van Hout 1991, Brito & Oliveira 1997, Shoen 2006 a. o. have shown, from
different points of view, that the ambiguity of the deverbal nominalizations cannot be reduced to two
readings. In particular, starting from the ambiguity of the notion “result”, Brito & Oliveira 1997
distinguish three readings in the nominalizations derived from transitive verbs: process (2a), result
(conceived as the culmination of a process) (2b) and entity or individual (2c):
(2) a. A Maria ainda está a fazer a encomenda de livros (Mary is still doing the order of books)
b. A encomenda de livros enriqueceu a nossa biblioteca (The order of books enriched our library)
c. Vou mandar esta encomenda (I am sending this parcel)
Accepting this general idea, the process, the result will be here considered parts of the complex event
structure.
The text analyses in particular the nominalizations derived from unaccusative verbs, specially those
referring movement, in order to verify if their argument may be expressed by a referential adjective
associated to an event reading.
The main factors that imply the process and the result and the individual readings will be reviewed.
Here are some of them:
(i) When the nominalization is the subject of verbs like prolongar-se (to prolong), durar (to last) and
co-occurs with temporal / aspectual adverbs or prepositional locutions like durante (during), no meio
de (in the middle of), ao fim de (at the end of), the process reading is obtained:
(3) A entrada americana (no Iraque) prolongou-se durante dias. (the American entrance (in Irak) prolonged during several days)
XVII Colloquium on Generative Grammar – Girona, 13th-15th June 2007
19
(ii) When the nominalization is the subject of a predicate like ser lenta, ser prolongada, (to be slow, to
be prolonged), also the process reading is obtained:
(4) A entrada americana (no Iraque) foi lenta. (the American entrance (in Irak) was slow.
When these type of nominalization (entrada, saída, vinda, etc.) is the subject of verbs like ocorrer (to
occur), dar-se, ter lugar (to take place) the reading is dependent from the nature of the temporal /
aspectual adverbs that modify the sentence: as example (5) shows, with localization temporal adverbs,
the main reading may be considered the event (as a whole), not the process:
(5) A entrada americana (no Iraque) ocorreu há três anos. (the American entrance (in Irak) occurred three years ago)
From examples like (3) and (4) we see that nominalizations derived from unaccusative verbs with a
referential adjective as an argument may occur in contexts where the process reading is available,
contrary to what was proposed by Bosque & Picallo 1996 and Picallo 1999.
Something different happens with nominalizations with the individual reading, where no referential
adjectives are expected:
(6) A entrada do edifício foi destruída (The entrance of the building was destroyed)
(7) * A entrada portuguesa foi destruída. (The Portuguese entrance was destroyed) (Portuguese may be ok as a classificatory A, what is irrelevant for our discussion).
The second part of the text will be dedicated to the study of the syntax of these constructions.
Following Picallo 1991, Alexiadou 2001 and the Distributed Morphology, the nominalizations are
considered to be build in Syntax. For event / process nominalizations an analysis inspired in
Alexiadou’s work will be defended, according to which their structure is very similar to the structure
of the sentence, with a functional category of Aspect; the lexical head is unspecified for syntactic
category and the nominal nature is given by the nominal environment, specifically by the functional
category DP. The structure of nominalizations with the individual reading have no Aspect node. But if
one accepts the view that nominalizations may have not two but three or even more readings, then
Alexiadou’s proposal must be somehow reviewed: both in the process reading (as in (3) and (4)) and
in the result reading the Aspect node is justified.
References
Alexiadou, A. 2001 Functional Structure in Nominals. Nominalization and ergativity, Amsterdam, John
Benjamins.
Bosque, I. & Picallo, C. 1996 “Postnominal adjectives in Spanish”, Journal of Linguistics 32, 1996, pp.
349-385.
Bottari, P. 1992 “Romance ‘Passive Nominals’”, Geneva Generative Papers, 0.0., pp. 66-80.
Brito, A. & Oliveira, F. 1997 “Nominalization, aspect and argument structure” in Matos, G. et alii (eds.)
Interfaces in Linguistic Theory, APL, Lisboa, pp. 57-80.
Giorgi, A. & Longobardi, G. 1991 The Syntax of Noun Phrases, Cambridge, Cambridge University Press.
Grimshaw, J. 1990 Argument Structure, Cambridge, Mass., MIT Press.
Kayne, R. 1984 Connectedness and Binary Branching, Dordrecht, Foris Publications.
Milner, J.C. Ordres et raisons de langue, Paris, Seuil.
Miguel, M. 2004 O Sintagma Nominal em Português Europeu, Diss. de Doutoramento em Linguística
Portuguesa apresentada à Universidade de Lisboa em 2004.
Picallo, C. 1991 “Nominals and nominalizations in Catalan”, Probus, 3, pp. 279-316.
Picallo, C. 1999 “La estructura del sintagma nominal: las nominalizaciones y otros sustantivos com
complementos argumentales”, in Bosque, I. and Demonte, V. (eds.) Gramática Descriptiva de la
Lengua Española, Madrid, Espasa, Vol. I, pp. 363-393.
Schoen, L. 2006 Les nominalizations déverbales et leurs arguments, Mémoire de Maîtrise sous la
direction de Petra Sleeman, University of Amsterdam.
Van hout, A. 1991 “Deverbal nominalization, object versus event denoting nominals: implications for
argument and event structure”, Linguistics in the Netherlands 8, pp. 71-80.
XVII Colloquium on Generative Grammar – Girona, 13th-15th June 2007
21
Silent Person with Semi-lexical Identity
Norbert Corver ([email protected])
Utrecht Institute of Linguistics-OTS / Utrecht University
The Dutch grammatical marker –s is well-known for its occurrence in possessive noun phrases such as
Piets auto (Piet-s car, ‘Piet’s car’) and vaders hoed (father-s hat, ‘father’s hat’), where –s is attached to
a proper name and a kinship term, respectively. As has been observed by various Dutch traditional
grammarians, the marker –s also shows up in a great variety of Dutch dialects in what appear to be
non-possessive contexts (cf. Van Haeringen 1947, Overdiep 1940). An example of this quite
remarkable phenomenon is given in (1), which represents Alblasserwaard Dutch:
(1) We kwamen Anna’s tegen
We met Anna-s PRT
‘We met Anna.’
The question, obviously, arises how to interpret here the occurrence of –s on the proper name, which
seems to function as an argument within the main clause. The approach taken by traditional
grammarians is to analyze –s as an accusative case morpheme. Another possible analysis would be to
say that –s is an enclitic article, i.e. an article suffixed to the noun, as familiar from e.g. Romanian
constructions like copil-ul (child-the, ‘the child’), where ul is attached to a common noun. In this
paper I will reject the case-analysis and the enclitic article-analysis. I will propose an alternative
analysis according to which a linguistic expression like Anna’s in (1) is, in fact, a ‘hidden’ possessive
construction, with -s as a possessive marker occupying a functional head position. More specifically,
Anna is a possessor which enters into a possessive relationship with an unpronounced (i.e. silent) noun
which stands for ‘person’. Thus, Anna’s is actually: [Anna + -s + PERSON]. If this is the correct
analysis, this would provide another instance of a nominal construction featuring an element which is
syntactically and interpretively active, but yet not pronounced (cf. Kayne 2003). It will be shown that
silent PERSON also has an overt equivalent, i.e. persoon. This lexical item turns out to display a
grammatical behavior that is deviant from that of ‘regular’ lexical nouns such as jongen (boy), vriend
(friend), et cetera. On the basis of this ‘special’ grammatical behavior, persoon and PERSON will be
identified as semi-lexical heads in the sense of Emonds (1985) and Corver and Van Riemsdijk (2001).
Having identified a silent semi-lexical noun PERSON in (1), I will further show that a number of other
nominal constructions featuring –s also contain a silent semi-lexical noun, among which PLACE,
MANNER, and TIME.
On The Syntax of Possessors and Structural Case
Maia Duguine ([email protected])
EHU-U. Basque Country & U. Nantes-Naoned
Overview. In this talk, I establish a generalization on the syntax of possessors (A), that makes
manifest the relation between possessor-agreement (la, 2a), -drop (1b, 2b) and -extraction (1c, 2c). I
also propose that (A) has a principled basis, dependent on structural case. Some consequences of the
results obtained are discussed, among them the plausibility of (B).
(A) Generalization on possessors: if in a language the noun agrees in φ-features with its possessor,
the possessor DP can be both extracted and silent.
(B) Structural case is the trigger for agreement, pro-drop and extraction of DPs.
Correlations in the syntax of possessors. The point of departure of the talk is the existence of two
correlations that are poorly understood in the theory of grammar:
(i) the correlation between the pro-drop and agreement phenomena: subjects seem to be able to be
silent in languages in which agreement has a rich morphological expression (cf. Taraldsen
1981, Rizzi 1986), and
(ii) the correlation between agreement displayed by DPs on functional heads and their ability to be
extracted/moved (cf. Szabolcsi 1983, 1994, Sportiche 1998, Gavruseva 2000, Rackowski 2002,
Chung & Ladusaw 2004).
Here, I propose to take the DP as a test domain to establish the strength of both these correlations. In
order to establish the generalization A, I will discuss the patterns of possessors in a sample of eleven
languages that display agreement with possessors in the DP: Southern varieties of Quechua,
Hungarian, Chamorro, Fijian, Nahuatl, Jacaltec, Itzaj Maya, Tzotzil, Tzutujil, Mohawk, and Wan’. I
show that both correlations hold in all these languages (cf. (1) and (2)), which leads me to formulate
the generalization (A).
In order to strengthen the generalization, I also give negative evidence, centering the discussion on
Quechua, a language with strong typological variation among its dialects. I will show that crucially, in
Imbabura Quechua, where the possessive suffixes are said to have been ‘lost’ (Cole 1985: 115),
possessor DPs cannot be silent nor can they be extracted (compare (3) to (2)). That is, this variety
displays the negative of the correlation found in Southern Quechua, showing that the generalization is
very tight.
Towards a principled account. The second part of the talk is a discussion of the principle behind (A).
The proposal that I will put forward in this talk is that agreement, dropping and extraction of the
possessor are operations that are made possible by virtue of bearing structural case (as opposed to
inherent case). The evidence in favor of this analysis is related to the hypothesis that DP and IP/CP are
structurally parallel (cf. Szabolcsi 1983, 1994, Abney 1987). I will show that in the languages in
question (i) the case on the possessor is morphologically the same as the case present on one of the
XVII Colloquium on Generative Grammar – Girona, 13th-15th June 2007
23
arguments of the verb, typically the subject (compare (4a) and (2a)), and (ii) the clausal argument
marked by the same case as the possessor (i.e. typically the subject) can also be dropped and extracted
(4a, b).
Finally, I will discuss the possibility of extending the generalization (A) to the clausal level, i.e. the
plausibility of a principle like (B). First, I will present evidence reported in the literature that will show
that agreement, pro-drop, and extraction are related. Among them are the impossible extraction for
non-agreeing DPs in Chamorro (Chung & Ladusaw 2004) and Tagalog (Rackowski 2002), and the
impossible pro-drop for direct objects in Italian that do not trigger agreement (Rizzi 1986). Then, the
relevance of structural case will be highlighted. I will show that agreement is not a necessary condition
on extraction and pro-drop, and that in fact agreement itself is subject to a condition on case: (i)
structural case is what licenses pro in Italian, be it an agreeing pro or not (cf. Rizzi 1986), (ii) non-
agreeing arguments can be extracted and dropped in Basque (cf. Ortiz de Urbina 1989) (iii) only
structural case-marked objects trigger agreement in French (cf. Sportiche 1990 (based on Kayne
1989), Belletti 2001), (iv) structural case-marked subjects but not quirky subjects trigger agreement in
Icelandic (Taraldsen 1995).
Examples
Wan’ (Everett & Kern 1997) (1) a. xiricon Xijam house.3SGM Xijam
‘Xij am’s house’ b. pro xiricon
house.3 SGM ‘his house’
c. Naroin’ jujun carawa co jam’ curucucun. deny 2P1.MOOD. 3PLM animal INFLM/F.MOOD tired body. 3PLM
‘Deny food to lazy (people)!’ (lit. ‘Deny food to those whose bodies are tired).’
Southern Quechua (Sánchez 1996) (2) a. Maria-q wasi-n Maria-GEN house-3
‘Maria’s house’ b. pro wasi-n house-3 ‘her/his house’
c. Pi-qpa-ta reqsi-nki [t tura-n-ta]? who-GEN-ACC know-2SG brother-3SG-ACC
‘Whose brother do you know?’
Imbabura Quechua (Cole 1985) (3) a. pay-paj wasi
he-GEN house ‘his house’ b. #pro wasi house ‘his house’
c. * pi-paj-taj riku-rka-ngui alku-ta? who-GEN-INTER see-PAST- 2 dog-ACC
‘Whose dog did you see?’
Southern Quechua (Sánchez 1996) (4) a. [(Maria-q) papa-ta ranti-sqa-n-ta] muna-ni Maria-GEN potato-ACC buy-NOM-3P-ACC want-1P ‘I want Maria/pro to buy the potato.’
b. Pi-qpa-ta muna-nki [t platanu-ta ranti-na-n-ta] who-GEN-ACC want-2P banana-ACC buy-NOMFUTURE-3P-ACC] ‘Who do you want to buy bananas?’
References
Belletti, A. 2001. (Past-)participle agreement. Ms. Università di Siena.
Chung, S. & W. Ladusaw, 2004. Restriction and Saturation, Cambridge: MIT Press.
Cole, P. 1985, Imbabura Quechua, London: Croom Helm.
Everett, D. L. & B. Kern, 1997. Wan’. The Pacaas Novos Language of Western Brazil. London and
New York: Routledge.
Gavruseva. E, 2000. On the syntax of possessor extraction”, Lingua 110, 743-772.
Kayne, R. 1989. “Facets of Romance Past Participle Agreement”, P. Benincà, ed., Dialect Variation
and the Theory of Grammar, Foris Publications, Dordrecht.
Rackowski, A. 2002. The Structure of Tagalog: Specificity, Voice, and the Distribution of Arguments.
Ph.D. dissertation, MIT.
Rizzi. L., 1986. Null Objects in Italian and the Theory of pro”. Linguistic Inquiry 17:50 1-557.
Sánchez. L. 1996. Why does Southern Quechua agree in person nominally’?”, ms., Carnegie Mellon
University.
Sportiche, D. 1990. Movement. Case and Agreement”. ms., UCLA.
Szabolcsi. A. 1983. The possessor that ran away from home”. The Linguistic Review 3: 89-102.
Szabolcsi, A. 1994. The noun phrase”. In Kiefer & Kiss (eds.), Syntax and Semantics 27. the Syntactic
Structure of Hungarian, New York: Academic Press, 179-274.
Taraldsen, K. 1995. On agreement and nominative objects in Icelandic’. In Studies in Comparative
Germanic Syntax, eds. H. Haider, S. Olsen and S. Vikner, 307-327. Dordrecht: Kluwer.
XVII Colloquium on Generative Grammar – Girona, 13th-15th June 2007
25
An L-Syntax for Adjuncts
Ángel J. Gallego ([email protected] )
Universitat Autònoma de Barcelona
This paper proposes an l-syntactic (in [11]’s sense) approach to adjuncts that extends the dyadic
pattern of (1) to traditional VP modifiers (e.g. happily with the key etc.) thus supporting a Figure-
(Ground-based approach to predication (see [9, 11, 18]). In particular and assuming a (neo-)
Davidsonian treatment of adjuncts as predicates of the event ([22]), we claim that prepositions relate
VPs qua ‘subjects’ (formally specifiers of P) to adjuncts qua ‘predicates (formally complements of P)
as shown in (2). Such a structure follows straightforwardly in the case of bona fide PPs (e.g. in Boston
by John etc.) for the P is overt but what about other adjuncts most notably adverbs (e.g. peacefully
now etc.)?. We argue that (2) can be extended to them under [18]’s hypothesis that adverbs decompose
into a non- relational element (typically a noun or Root) and a relational one (typically a P) (see [9,
13]). What this amounts to is that the VP-Adjunct relation in the examples in (3) resorts to the same l-
syntax (3b) being decomposed as in (4b) thus (3a) and (3b) can be roughly paraphrased as ‘the event
of John speaking is (placed) on Monday/clear’. As can be seen we crucially assume that the external
argument is first-Merged as Spec-V (see [11, 15, 23]).
Technically, though, the analysis poses a categorial problem: in (2) we have a PP not a v*P/YP. This
goes against one robust empirical property of adjuncts namely that they do not modify the
(syntactic/semantic) nature of the object they attach to (see [2, 4, 10, 12, 16, 22, 24]). This fact has
been taken particularly by [4] to indicate that adjuncts occupy a parallel plane’ being introduced by a
dedicated operation that yields predicate composition: pair-Merge. We hold that such move is not
forced upon us: although adjuncts and arguments behave differently on both semantic (and syntactic)
grounds we do not need to invoke an independent operation instead all we need is a different syntax
for a different semantics to follow. For concreteness, we claim that the main difference between
arguments and adjuncts concerns labeling, as (5) shows: arguments require it, whereas adjuncts do not
(see [4, 12, 24]) a possibility we relate to the asymmetric nature of argument taking, which requires a
formal way of encoding the predicate-argument asymmetry —this distinction could of course be
achieved through different means (e.g. diacritics, lexical marking, theta grids, relational templates,
etc.) but labeling suffices and furthermore respects [3] ‘s inclusiveness. If so, the categorial problem
goes away since adjuncts give raise to unlabeled syntactic objects, which capture the ‘parallel’
behaviour of these dependents (see [2, 4, 8, 12,19, 24]). Suppose in this vein that clustered adjuncts of
the sort of (6a) conform to this specific syntax with no scope dependencies arising —if no label is
projected c-command cannot be calculated. As [12, 24] plausibly argue, this may be responsible for
the scopeless interpretation of adjuncts endorsed by the neo-Davidsonian framework (see [10, 12, 16,
22, 24]). Simply put: we argue that the syntax of (2)-(5b) captures the paratactic semantics of adjuncts,
the one encoded in (6b). If so, (6a) should be analyzed as in (7), with the adjuncts on Mondqy and
from Boston taking the VP John called Mary as their subject ‘in parallel (that is, both PPs take the VP
as its specifier).
There are more consequences: if c-command between VP and adjuncts fails, binding should fail too,
but this is empirically false, as ( shows (see [20]), which leaves us with two possible ways out: binding
obtains either at SEM (after [4]’s Simplification) or by raising adjuncts to a specifier in a cascade-like
structure a la [20]. We adopt the latter solution, identifying the landing site as an outer-Spec-V. If
correct, then binding from subjects goes through unproblematically, as these end up occupying Specv*
(or Spec-T); as for binding from objects, binding is possible if, as [17] convincingly argues, these
dependents move to a Case checking position higher than Spec-V: be it Spec-AgrOP (as [17] contends)
or Spec-TOP (see [21]); importantly, it cannot be Spec-V (contra [5]) due to anti-locality (see [1]).
In sum this paper argues for a configurational approach to adjuncts that builds on [l1]’s l- syntactic
system, claiming that traditional VP modifiers start off as PPs that establish a birelational dependency
between a Figure/Subject (P’s would-be specifier: the VP itself which we take to embody the event
position) and a Ground (P’s complement, either conflated or not). In order to solve the categorial
problem this analysis raises, we propose that when VP and PP undergo external-Merge, no label
projects, and so the adjunct does not participate in regular phrase structure relations (being in a
‘parallel plane’). Nevertheless, evidence from binding (also NPI licensing, topicalization, etc.; see [12,
16, 17, 20]) indicates that adjuncts manifest themselves as regular dependents too through labeling
either as complements (see [16]) or as specifiers (see [6 7 141); we have supported the second
alternative: adjuncts can optionally become specifiers by means of internal-Merge. These two options
are depicted in (9). The resulting scenario is interesting in various respects: first, it captures the
paradoxical nature of adjuncts (see [20]) without extra operations (e.g. pair-Merge, Simplification,
etc.), nor massive remnant movements (see 6, 7]); second, it fits with [5]’s observation that {X(P
Y(P)} structures created by external-Merge must be destroyed —either by expelling adjuncts to a
parallel plane (this happens when no label projects) or else by raising adjuncts to a specifier (where
labeling is obligatory; see [3]); and third, the duality in (9) reinforces [9]’s idea that Subject-Predicate
dependencies, though configurational, are non-directional, being expressed in both predicate-
complement (see (9a)) and predicate-specifier (see (9b)) fashions.
Examples
(1) DYADIC STRUCTURE (X=PREP0SITI0N)
[XP ZP (Subject/Figure) [x’ X YP (Predicate/Ground) ] ]
(2) DYADIC STRUCTURE APPLIED TO ADJUNCTS
[PP VP (Subject/Figure) [P’ P YP (Adjunct/Ground) ] ]
(3) a. John speaks on Monday.
b. John speaks clearly.
(4) a. [PP [VP John speaks] [P’ on Monday] ] (= (3a))
XVII Colloquium on Generative Grammar – Girona, 13th-15th June 2007
27
b. [PP [VP John speaks] [P’ √ clear ] ] (= (3b))
(5) a. Argument-taking. [x XP YP (Argument) ]
b. Adjunction. [ XP YP (Adjunct) ]
(6) a. John called Mary on Monday from Boston.
b. [•e: call(e) & Agent(John,e) & Theme(Mary,e) & on (Monday,e) & from (Boston,e)
[ Spec [PP on Monday (Ground1)] ]
(7) [VP John called Mary] (Figure)
[ Spec [PP from Boston (Ground2)] ]
(8) a. I saw two meni on each otheri’s birthdays.
b. [For which of hisi crimes] z do you think every gangsteri has been put into jail tz ?
(9) a. [VP (Subject/Figure) [PP P YP (Adjunct/Ground) ] ] Adjunct in parallel
b. [VP [PP P YP (Adjunct/Ground) ] [VP (Subject/Figure) ti]] Adjunct as specifier
References
[1] Abels, K. (2003): Successive Cyclicily, Anti-Locality, and Adposition Stranding, Ph.D.
Dissertation, U.Conn.
[2] Chametzky, R. (2000): Phrase Structure: from GB to Minimalism, Maiden: Blackwell.
[3] Chomsky, N. (1995): The Minimalist Program, Cambridge (Mass.): MIT Press.
[4] Chomsky, N. (2004): “Beyond Explanatory Adequacy”, in Beiietti, A. (ed.), Structures and
Beyond. The Cartography of Syntactic Structures (vol. 3), Oxford (New York): Oxford University
Press, 104-13 1.
[5] Chomsky, N. (2005): “On Phases”, Ms., MIT.
[6] Cinque, G. (1999): Adverbs and Functional Heads. A Cross-Linguistic Perspectiva, Oxford:
Oxford University Press.
[7] Cinque, G. (2004): “Issues in Adverbial Syntax”, Lingua 114, 683-7 10.
[8] Collins, C. (2002): “Eliminating Labels”, in Epstein, S. & T. Seeiy (eds.), Derivation and
Explanation in the Minimalist Program. Malden (Mass.): Blackwell, 106-132.
[9] Den Dikken, M. (2006): Relators and Linkers. The Syntax of Predication, Predicate Inversion, and
Copulas, Cambridge (Mass.): MIT Press.
[l0] Ernst, T. (2002): The Syntax of Adjuncts, Cambridge: Cambridge University Press.
[11] Hale, K. & S. Keyser (2002): Prolegomenon to a Theory of Argument Structure, Cambridge
(Mass.): MIT Press.
[12] Hornstein, N., J. Nunes, & P. Pietroski (2006): “Some Thoughts on Adjunction”, Ms.,
UMD/USP.
[13] Jackendoff, R. (1977): X’Syntax: A Study of Phrase Structure, Cambridge (Mass.): MIT Press.
[14] Kayne, R. (1994): The Antisymmetry of Syntax. Cambridge (Mass.): MIT Press.
[15] Koopman. H. & D. Sportiche (1991): “The position of subjects”, Lingua 85, 211-258.
[16] Larson, R. (2004): “Sentence-final adverbs and ‘scope’ “, in Wolf, M. & K. Moulton (eds.),
Proceedings of NELS 34, U.Mass.: GLSA, 23-43.
[17] Lasnik, H. (1999): Minimalist Analysis, Oxford: Blackweii.
[18] Mateu, J. (2002): Argument Structure: Relational Construal at the Syntax-Semantic Interface,
Ph.D. Dissertation, UAB.
[19] Moro, A. (2000): Dynamic Antisymmetry, Cambridge (Mass.): MIT Press.
[20] Pesetsky, D. (1995): Zero Syntax. Experiencers and Cascades. Cambridge (Mass.): MIT Press.
[21] Pesetsky, D. & E. Torrego (2004): “Tense, Case, and the Nature of Syntactic Categories”, in
Guéron, J. & J. Lecarme (eds.), The Syntax of Time, Cambridge (Mass.): MIT Press, 495-537.
[22] Pietroski, P. (2005): Events and Semantic Architecture, Oxford: Oxford University Press.
[23]Sigurôsson, H. (2003): “Case: Abstract vs. morphological”, in Brandner, E. & H. Zinsmeister
(eds.), New perspectives on Case theory, Stanford (Calif.): CSLI Publications, 223—268.
[24] Uriagereka, J. (2003): “Pure Adjuncts”, Ms., UMD.
XVII Colloquium on Generative Grammar – Girona, 13th-15th June 2007
29
A first approach to Catalan and Spanish SLI
Anna Gavarró ([email protected])
Universitat Autònoma de Barcelona
Specific Language Impairment affects the process of language acquisition in isolation from any other
impairment and in recent years has received much attention as a source of information on the workings
of the faculty of language. There are, however, relatively few works on Romance SLI; in this paper we
undertake the comparison of SLI in Catalan and Spanish, two quite closely related languages, and seek
to find an explanation for the cross-linguistic variation encountered.
In particular, we compare the results provided by De la Mora (2004) on Spanish with new
results based on the corpus analysis of two Catalan-speaking children with SLI (J and A, at MLU 1.5
and 4.7 and 1.3 and 3.1 respectively – recordings available in the Childes database), and we compare
the results to those available for normally developing children. The phenomena taken into
consideration are those for which the literature on other languages, mainly English, indicates deficits
in the productions of children with SLI: the production of articles, finiteness (Rice, 2003), and object
clitics (Jakubowicz et al. 1998). Despite the similarity in many respects of the grammars of Catalan
and Spanish, there is evidence that SLI in one respect displays quite different manifestations in these
two languages.
The Spanish data provided by De la Mora come from 10 Mexican Spanish-speaking children,
with MLU ranging between 1.84 and 3.66; the mean MLU is 2.88, which is close to the mean MLU
for the Catalan-speaking children, 2.69, rendering comparison possible. The results are summarised in
tables 1 and 2. Production of definite articles becomes quite accurate, both in Catalan and Spanish,
after the earlier period in which omission is high; this is reminiscent of a short-lived period of article
omission found in some normally developing Catalan-speaking children (Guasti et al. 2004). On the
other hand, clitic omission is not homogeneous in De la Mora’s and the original Catalan results here:
omission is found to some extent in Spanish SLI, but it is not comparable to that in Catalan, around
66% at the stage when J and A are producing enough clitic contexts to evaluate it. The difference
between article production and clitic production corroborates the finding by Jakubowicz and others
that, contra Leonard (1998), the phonological robustness of a constituent does not determine whether it
is spared or not: Catalan articles and 3rd person clitics are phonologically identical.
The results for clitic production in SLI do not coincide with those found for normally
developing children, for which there is optional omission in Catalan, but no omission in Spanish
(Wexler et al. 2004); however, the contrast between Catalan and Spanish remains. If we follow Wexler
(1998), Wexler et al. (2004) in considering object clitic omission as a manifestation of the Unique
Checking Constraint (UCC), subject to the specificity of clitics in a particular language, our results are
predicted: we expect clitic omission to be much more prominent in a participle-agreement language
such as Catalan than in a language without it like Spanish. The predictions go beyond these two
languages: within Romance we expect optional clitic omission in SLI in French and Italian (indeed
attested in the literature for French), but a much less significant rate of omission in languages without
participle agreement such as Romanian – a prediction still unverified for SLI.
Turning to the last SLI marker, the presence of an optional infinitive stage, this is not found in
Catalan: only a 1.25% root infinitives is found in the corpora of J and A. This result can be accounted
for under the UCC as well, and matches the non-existence of root infinitives in the child null subject
Romance languages. We can conclude that neither clitic omission nor root infinitives can be used as
universal diagnostics for SLI. More importantly for the study of the language faculty, SLI appears as a
grammatically based disruption, the result of a protracted UCC-stage.
Target-like article
prod
* article omission 3 clitic pronoun 3 clitic omission
J, 1st transc. 12 (25%) 35 (75%) 0 1 (100%)
J, 2nd transc. 265 (96.4%) 10 (3.6%) 4 (28.6%) 10 (71.4%)
A, 1st transc. 18 (47.4%) 20 (52.6%) 1 (100%) 0
A, 2nd transc 98 (90.7%) 10 (9.3%) 5 (35.7%) 9 (64.3%)
Table 1: Results for Catalan
Target-like article
prod
* article omission 3 clitic pronoun 3 clitic omission
85% Not availabl 225(53%) 39 (9%)
Table 2: Results for Spanish, adapted from De la Mora (2004) (remaining percentage in
clitic production: full DP)
References
De la Mora, J. (2004) Direct object clitics in Spanish-speaking children with and without Specific
Language Impairment. Master’s thesis, University of Alberta.
Guasti, M. T., J. de Lange, A. Gavarró and C. Caprin (2004) ‘Article omission: across child languages
and across special registers,’ in J. van Kampen and S. Baauw, eds., Proceedings of GALA 2003,
LOT Occasional Series, vol. 1, Utrecht, 199–210.
Jakubowicz, C., L. Nash, C. Rigaut & C. L. Gerard (1998) ‘Determiners and clitic pronouns in French-
speaking children with SLI’. Language Acquisition 7, 113–160.
Leonard, L. (1998): Children with Specific Language Impairment. Cambridge, MA.: MIT Press.
XVII Colloquium on Generative Grammar – Girona, 13th-15th June 2007
31
Rice, M. (2003) ‘A unified model of Specific Language Delay: grammatical tense as a clinical marker
of unexpected variation,’ in Y. Levy & J. Schaeffer, eds., Language Competence Across
Populations. Lawrence Erlbaum, Mahwak, NJ.
Wexler, K. (to appear) ‘The Unique Checking Constraint as the explanation of clitic omission in SLI
and normal development’, in Jakubowicz, C., L. Nash and K. Wexler, eds., MIT Press,
Cambridge, Mass.
Wexler, K., A. Gavarró, V. Torrens (2004) ‘Feature checking and object clitic omission in child
Spanish and Catalan,’ in R. B. Bennema, B. Hollebrandse, B. Kampers-Manhe, P. Sleeman,
eds., Romance Language and Linguistic Theories. John Benjamins, Amsterdam/Philadelphia.
Two properties of Spanish non-argumental clitic pronouns
Susana Huidobro ([email protected]) & Jonathan MacDonald ([email protected])
Stony Brook University - University of the Basque Country
This talk focuses on Spanish non-argumental reflexive clitic pronouns (non-argumental reflexives).
We argue that this reflexive is the complement of a null P°. We appeal to recent minimalist
assumptions regarding features to account for a restriction it places on the direct object. This in turn
allows for a natural account of Spanish sentential subjects, which show the same restriction.
Moreover, we see that this null P° account offers a new insight into the inability of these reflexives to
double.
Note three facts regarding the presence of this non-argumental reflexive:
1. It requires the presence of a direct object (1a);
2. The direct object must be quantized (a la Krifka 1989) (1b);
3. The predicate is interpreted as telic; thus, the durative phrase is incompatible (Zagona 1996) (1c).
Observe that the English verb particle exhibits the same three properties:
1. It requires a direct object (2a);
2. The direct object must be quantized (2b);
3. It’s presence results in a telic predicate (Svenonius 2004) (2c).
On a par with previous analyses of verb particle constructions (Svenonius 2004 etc.) we claim that the
Spanish reflexive is the complement of a null P° that merges low with the verb as in (3). The presence
of the P° itself is responsible for the telic interpretation of the predicate as prepositions often are
(Borer 2005). Nevertheless, the required quantized nature of the direct object is not directly related to
telicity (cf. Nishida 1994, Zagona 1996). For, observe in Romanian that the durative phrase is
incompatible independently of the presence of the reflexive (4a), and a non-quantized direct object
results in ungrammaticality (4b).
We claim that the null P° bears a feature QTZ such that the P° and the verb must predicate of a
quantized direct object (i.e. the subject of the predicate). This brings the proposed structural
configuration in line with Spanish sentential subjects, which are also required to be quantized NPs (5).
The head of the projection hosting the subject bears the QTZ feature accounting for the restriction. Not
all T°s or all (null) P°s bear QTZ. This is in line with Chomsky’ s (2004) assumptions regarding the
EPP feature.
Now let us consider the well-known fact about these Spanish reflexives in (6); they cannot double. On
the one hand, the lack of doubling is related to their reflexive status, for note that non-argumental non-
reflexives can double (7). On the other, the lack of doubling is related to their non-argumental status,
for note that argumental reflexives can double (8). Now observe a crucial difference between
reflexive and non-reflexive doubling. Doubled reflexives require mismo “self’ (9a), while non-
reflexives can optionally appear with mismo when argumental (9b). Interestingly though, when non-
XVII Colloquium on Generative Grammar – Girona, 13th-15th June 2007
33
argumental, the non-reflexive can no longer appear with mismo (10). If we adopt an analysis from
MacDonald (2006) in which these non-argumental non-reflexive clitic pronouns are analyzed as the
complement of a null P° (i.e. the same structure as the reflexives from (3), but with a null P° with
distinct properties), then we can draw the conclusion the lack of reflexive doubling is related to the
combination of two properties: (i) the obligatory presence of mismo when doubled and (ii) the
presence of the null P° itself Mismo is not licensed as the complement of a null P°. Since the reflexive
always requires the presence of mismo (see 9a), it can never double as the complement of a null P°;
while the non-reflexive can double as the complement of the null P° because it does not require the
presence of mismo to double (see 9b).
In this talk, we argue that the Spanish reflexive is the complement of a null P° that bears a QTZ
feature, the presence of which forces the presence of a quantized direct object via predication. The
presence of QTZ on some heads and not others reflects current minimalist assumptions regarding
features such as the EPP feature (Chomsky 2004). Furthermore, the null P° is directly responsible for
the reflexive’s lack of doubling, as the reflexive requires the presence of mismo to double, but mismo
is not licensed as the complement of this null P°.
Examples
(1) a. # Juan se comió. � out on non-argumental reflexive interpretation Juan REFL ate ‘Juan ate himself’
b. Juan se comió *(la) paella. Juan REFL ate the paella
‘Juan ate the paella.’ c. Juan se comió la paella #durante 10 minutos.
Juan REFL ate the paella for 10 minutes ‘Juan ate the paella for 10 minutes.’
(2) a. * John ate up. b. b. John ate *(the) soup up. c. John ate the soup up #for 10 minutes.
(3) … [vP v [VP DO V [PP P-ø] ] ]
(4) a. Eu (mi-)am mâncat mãrul *timp de zece minute. I REFL-have eaten apple-the time of ten minutes ‘I ate the apple in ten minutes.’
b. Eu (*mi)am mâncat supã I REFL-have eaten soup ‘I ate soup.’
(5) Tráfico hizo mucho ruido. Traffic made much noise ‘Traffic made a lot of noise.’
(6) Juan se comió la paella (*a sí mismo). Juan REFL ate the paella to him self ‘Juan ate the paella (on himself).’
(7) Yo le comí la paella (a él). I NON-REFL ate the paella on him ‘I ate the paella (on him).’
(8) Juan se afeitó (a sí mismo). Juan REFL shaved to him self) ‘Juan shaved (himself).’
(9) a. Juan se afeitó a sí *(mismo) Juan REFL shaved to him self ‘Juan shaved.’
b. Yo le afeité a él (mismo). I NON-REFL shaved to him self) ‘I shaved him.’
(10) Yo le comí la paella a él (*mismo). I NON-REFL ate the paella to him self ‘I ate the paella on him.’
References
Borer, B. 2005. Structuring Sense II: The Normal Course of Events. Oxford; New York: Oxford
University Press.
Chomsky, N. 2004. ‘Beyond Explanatory Adequacy. In A. Belletti (ed.) The Cartography of Syntactic
Structures. Vol. 3, Structures and Beyond. Oxford: Oxford University Press.
Krifka, M. 1989. ‘Nominal Reference, Temporal Constitution and Quantification in Event Semantics.’
In R. Bartsch, J. van Benthem, and P. van Emde Boas (eds.) Semantics and Contextual
Expression, 75-115. Dorcfrecht, Holland/Providence, Rhode Island: Foris Publications.
MacDonald, J. 2006. ‘Null Directional Prepositions in Romanian and Spanish.’ In J.P. Montreuil and
C. Nishida (eds.) New Perspectives in Romance Linguistics Vol. 1: Morphology, Syntax,
Semantics, and Pragmatics. Selected Papers from the 35th Linguistics Symposium on Romance
Linguistics. Amsterdam and Philadelphia: John Benjamins.
Nishida, C. 1994. The Spanish reflexive citic seas an aspectual class marker. Linguistics 32: 425-45 8.
Svenonius, P. 2004. Spatial P in English.’ Ms. University of Tromso Zagona, K. 1996.
Compositionality of Aspect: Evidence from Spanish Aspectual Se. In C. Parodi, C. Quicoli, M.
Saltarelli, and M.L. Zubizarreta (eds). Aspects of Romance Linguistics, 475-488. Washington
DC: Georgetown UP.
XVII Colloquium on Generative Grammar – Girona, 13th-15th June 2007
35
Edge effects and comparative AdvP
Gergely Kantor ([email protected])
Eotvos Lorand University
If the head of a head-initial phrase is modified by a pre-head modifier, this modifier cannot be
extended from its head towards the modified head, as can be seen below:
(1) John is a proud (*of his wife) man.
(2) The TU-144 faster (*than the Concorde) flew.
This phenomenon, named by Haider (2000, 2003) as edge effect, was intended to serve as crucial
evidence in the debate whether adverbs are base-generated in adjunct positions (cf. Ernst 2002,
Svenonius 2002) or in functional specifiers (cf. Cinque 1999, 2004).
In this article, I aim at answering the following questions: first of all, it is important to investigate
in what circumstances are edge effects encountered; and second, it should also be checked whether it
is a feasible step to state that edge effects undermine the functional specifier analysis of adverbs, as
argued by Haider (2000, 2003), who bases the latter assumption on the fact that edge effects can also
be characterized as side effects of adjunction in head-initial projections. In other words, the
phenomenon was intended to be regarded as a language- and projection-specific problem, typical of
adjuncts only. In order to test the validity of these hypotheses, I applied functionally extended
adverbial comparative constructions, as they may easily be targeted by the possible problems invoked
by edge effects.
As far as the first research question is concerned, I would like to prove that edge effect is a
universal generalization, and not a language- and projection-specific, violable constraint linked solely
to adjunction. This is so because there are instances of XPs in specifier and pre-head complement
positions also affected by edge effects. The validity of such a generalization can easily be proven by
the obligatory extraposition of post-head material of modifiers preceding the modified head. The
influential approaches to extraposition are evaluated, and it will be argued that extraposition is a PF-
operation – mainly due to the direction of movement and its being unmotivated in a strict minimalist
model (in line with Truckenbrodt 1995). As a consequence, edge effects could be placed in such a
theoretical framework in which it is irrelevant whether a modified head is preceded a modifier XP –
for example, an AdvP – base-generated in a functional specifier or an adjoined position, or – applying
Kayne’s (1994) restricted phrase structure approach – even in a complement position.
Second, it is also pointed out why in certain languages the edge effect generalization can have
such an evident role, whereas in other languages this is not so. It is claimed that this fact does not
weaken the universality of the generalization, but these differences are due to the clearly language-
specific nature of headedness, as in certain languages (e.g., English) the interposed complements must
undergo extraposition (as in 1-2 above), whereas in other languages (such as German) the edge effect
generalization is satisfied vacuously. This is the reason why it can be said about the –not yet
extensively examined – German AdvP and its functional extension that they are head-final, similarly
to German VPs. This is supported by the following example as well:
(3) Der schneller als der Portlandzement erhärtender Zement the faster than the Portland cement hardeningA cement the cement that hardens faster than the Portland cement
The greatest advantage of the theoretical solution outlined above can be captured with respect to
economy: a universal generalization besides an already existent and working system (headedness) is
more economical than providing an ad hoc, language- and projection-specific constraint, which purely
attempts, and eventually fails to describe empirical data.
Selected references:
Cinque, Guglielmo (1999) Adverbs and Functional Heads. A Cross-linguistic Perspective. Oxford,
UK: Oxford University Press.
Cinque, Guglielmo (2004) Issues in Adverbial Syntax. Lingua 114:683-710.
Ernst, Thomas (2002) The Syntax of Adjuncts. Cambridge, UK: Cambridge University Press.
Haider, Hubert (2000) Adverb Placement – Convergence of Structure and Licensing [online version].
Available: http://www.sbg.ac.at/spr/people/hubert_haider/dl/adv.pdf; last accessed: 10 May
2006. Also appeared in: Theoretical Linguistics 26: 95-134.
Haider, Hubert (2003) Pre- and Postverbal Adverbials in OV and VO [accepted draft version]. To
appear in: Lingua.
Kantor, Gergely (to appear) On the Left Periphery of Clausal Comparative Complements. In:
Interlingüística 17.
Kayne, Richard S. (1994) The Antisymmetry of Syntax. Cambridge, MA: MIT Press.
Svenonius, Peter (2002) Subject positions and the placement of adverbials. In: Svenonius, Peter ed.
Subjects, Expletives, and the EPP. Oxford, UK: Oxford University Press. pp. 199-240.
Truckenbrodt, Hubert (1995) Extraposition from NP and prosodic structure. In: Proceedings of NELS
25, edited by Jill N. Beckam. 503-517. Amherst, MA: UMass.
XVII Colloquium on Generative Grammar – Girona, 13th-15th June 2007
37
Some English (and Romance) Auxiliaries
Richard S. Kayne
New York University
The absence of a direct counterpart in Romance languages and in other Germanic languages of
English We are/were to leave immediately must be taken as a salient fact about those languages that
calls for an explanation. The one I propose brings into play the absence in all those other languages of
W-verb infinitival ECM and the absence of complementizer for. The limitation of the English
construction to finite contexts can be related to Holmberg's generalization via the postulation of a
silent passive past participle EXPECTED that needs to move. The type of account proposed may
extend to other cases of verbal constructions limited to finite contexts, including several from
Romance.
Split questions, extended projections, and dialect variation
Jorge López-Cortina ([email protected])
Seton Hall University
Spanish presents a type of two-part question where the first part resembles a wh-question and the
second part, separated from the first by an intonational break, consists of a proposed answer to the
question posed by the first part, as in (1a,b). In some dialects the wh-word is always qué ‘what’ but it
can assume the value of any other wh-word (2a,b).
Such constructions have occasionally been described for different dialects of Spanish (Lope Blanch,
1972; Py, 1971; Vigara Tauste, 1992). They are also standard in Catalan (López-Cortina, 2003) and
might appear in English and Basque (Arregi, 2006). Analyses differ on whether they consist of two
clauses (Arregi, 2006) or one (Camacho, 2002; López-Cortina, 2003; Lorenzo Gonzalez, 1994).
Among the latter, there are different views about the nature of the wh-word and its relationship to the
proposed answer.
In this paper a new analysis is presented, on the basis of new data from Chilean Spanish and
Northwestern Peninsular Spanish. These dialects show restrictions in the wh-movement of the first
part of the construction that are different from the restrictions found in regular (i.e. non-split) wh-
questions. Namely, it can be shown that, in cases where it has the same reference as an argument, the
wh-word does not behave like an argument, which could be extracted from a wh-island or a complex
NP (Torrego, 1984), but rather like an adjunct, which can not. Additionally, negative split questions
are not allowed, at least in the dialects described, which also matches the restrictions on wh-movement
for adjuncts.
These facts would support a single clause analysis (against Arregi. 2006). and also provide the basis
for a new analysis of the construction in which the wh-word is the specifier of an “extended
projection” (Grimshaw, 1991, 2005) of the phrase that serves as an answer. The head of that extended
projection can be pronounced in Chilean Spanish as an interrogative particle.
The structure given provides a straightforward explanation for the differences between the dialects
studied and suggests that the apparent contradictions in the description of the construction found in the
literature might be due to microparametric variation between the dialects described by each author. An
initial classification of these dialects is presented in light of the possibilities allowed by the proposed
analysis.
XVII Colloquium on Generative Grammar – Girona, 13th-15th June 2007
39
Examples
(1) a. Qué libro leyó Juan, El Quijote? (Arregi 2006) what book read Juan Don Quixote ‘What book did Juan read, Don Quixote?’ b. Quién leyó este libro, Juan? who read this book, Juan ‘Who read this book, Juan?’
(2) a. Qué llegó , Juan? (López-Cortina 2003) what arrived-3SG, Juan? ‘Who arrived? Was it Juan?’ b. Qué vas, a Oviedo? what go-2SG, to Oviedo? ‘Where are you going? To Oviedo?’
References
Arregi, K. (2006). Syntax and Semantics of Split Questions. Paper presented at the Linguistic
Symposium on Romance Languages 36, Rutgers University.
Camacho, J. (2002). Wh-Doubling: Implications for the Syntax of Wh-Movement. Linguistic Inquiry,
33(1), 157-164.
Grimshaw, J. (1991). Extended Projection.Unpublished manuscript, Brandeis University.
Grimshaw, J. (2005). Words and structure. Stanford, California: CLSI Publications.
Lope Blanch, J. M. (1972). Estudios sobre el español de México. México: Universidad Nacional
Autónoma de México.
López-Cortina, J. (2003). The Structure of Split Interrogatives. In P. Kempchinsky & C.E. Piñeros
(Eds.). Theory, Practice, and Acquisition (pp. 140-155). Somerville. Mass.: Cascadilla Press.
Lorenzo Gonzalez, G. (1994). Qué expletivo en preguntas dislocadas. Archivum Ovetensis, XLI V-XLV
(1), 423-446.
Py, B. (1971). La interrogación en el español hablado de Madrid. Bruxelles: Aimav.
Torrego, E. (1984). On inversion in Spanish and some of its effects. Linguistic Inquiry, 15, 103-129.
Vigara Tauste, A. M. (1992). Morfosintaxis del español coloquial. Madrid: Gredos.
The among construction: Catalan (d’)entre
Núria Martí Girbau ([email protected])
Universitat Autònoma de Barcelona
Indefinite nominals containing a PP headed by among (named from now on as the among
construction = AC) look very similar to partitives, as shown in (1). However, as noted in the literature,
the AC is much less restricted. In contrast with partitives, in the AC there is:1
(i) no need of identity between the elements of the subset and the set (2),
(ii) no need of gender agreement between Q and N (3),
(iii) no restriction on the Q and different kinds of determiner allowed (4), and
(iv) no restriction on the internal nominal with respect to the type of determiner allowed (5).
In fact, the among PP behaves like any N modifier, which is in general terms the conclusion reached in
the literature (though Cardinaletti & Giusti (2006) consider it as a type of partitive PP).
However, a closer look at data reveals that the among PP does not behave as an ordinary N
modifier inside the indefinite nominal but as a secondary predicate or the predicate of a small clause
instead, this is to say, it is outside the indefinite nominal which actually acts as its subject. We will
base our analysis on Catalan data, which interestingly shows the possibility of entre (‘among’) to be
preceded by de (‘of’), yielding d’entre (see 6). The presence of de is not optional and its role in the
structure will shed some light on the analysis of the entre PP, which can be generalised to other
languages.
In Catalan, the entre PP, when not preceded by de:
(i) is not accepted in sentences where the verb selects a nominal argument, not a predication (7),
(ii) it is not part of the nominal antecedent of a clitic (8), and
(iii) it is not accepted in cleft sentences as part of the nominal (9).
In contrast, the entre PP is perfectly gramatical as the predicate of a small clause (10), and it can only
licensed inside a nominal if preceded by de (11).
This behaviour of entre immediately reminds of a property of Catalan locative prepositions such
as sobre (‘on, above’), sota (‘under, below’), davant (‘in front of’), darrere (‘behind’), etc.: they too
can only be licensed as N modifiers if preceded by de (12). What these prepositions have in common
and differs from other prepositions is that they can be preceded and followed by other prepositions and
even be preceded by the definite article, as shown in (13). So they have a status close to a N and even
in some contexts they can behave as actual nouns (14). The obvious analysis for the requirement of de
to be licensed as N modifiers is that they need Case, like any other N that modifies a N. Although
entre is never preceded by the definite article, it does show the possibility of being preceded by other
prepositions (15) and we will claim that in Catalan it requires an open Case marker de.
1 Examples from (2) to (5) are from Lorenzo (1995: 215ff). For Italian examples illustrating the AC properties in comparison to partitives, see Cardinaletti & Giusti (2006: §3.1).
XVII Colloquium on Generative Grammar – Girona, 13th-15th June 2007
41
We will assume that in languages such as English or Italian where no of appears preceding among
or any locative preposition the licenser is required as well, the difference being that of is not overt (in
the line of Kayne 2005 for similar contrasts related to quantity words: Eng. few vs. Fr. peu de). So
even when among looks as being inside the nominal like any other N modifier, it is actually in a more
complex structure, within an OF phrase that licenses it.
To sum up, the conclusion reached in this paper is that among is a locative preposition in parallel
to inside, under, etc. which projects into a PP that usually functions as the predicate of a small clause
and which can only be a N modifier inside a nominal if licensed by of. There is a parametric difference
in the overt/covert character of the licenser of: some languages have an unpronounced counterpart OF.
Examples
(1) a. one among those books AC b. one of those books partitive
(2) a. Muchos ancianos entre estas personas. AC many old-men among these people b. * Muchos ancianos de estas personas. partitive many old-men of these people
(3) a. Muchos entre estas personas. AC manymasc among these peoplefem
b. *muchos de estas personas partitive manymasc of these peoplefem
(4) a. aquellos entre estos niños AC those among these children
b. *aquellos de estos niños partitive those of these children
(5) a. Éstos entre muchos niños. AC these among many children b. *Éstos de muchos niños. partitive these of many children
(6) a. moltes novel·les d’entre els llibres que et van deixar ‘many novels among the books that were lent to you’ b. dos entre els teus alumnes ‘two among your students’
(7) *[Tres novel·les entre els llibres recomanats] són de Paul Auster. three novels among the books recommended are of Paul Auster ‘Three of the novels among the recommended books are by Paul Auster.’
(8) * [Tres novel·les entre els llibres recomanats]i, ja lesi he llegides. three novels among the books recommended already them I-have read
(9) *Són tres novel·les entre els llibres recomanats que són de Paul Auster.
(10) Hi ha [tres novel·les de Paul Auster] [entre els llibres recomanats]. there is three novels by Paul Auster among the books recommended
(11) [Tres novel·les d’entre els llibres recomanats] són de Paul Auster.
(12) a. Hi ha un sobre dins el calaix/sota la cadira/darrere el llum/sobre la taula. ‘There’s an envelope in the drawer/under the chair/behind de lamp/on the table.’
b. *El sobre dins el calaix/sota la cadira/darrere el llum/sobre la taula és gros. ‘The envelop in the drawer/under the chair/behind de lamp/on the table.’ c. El sobre de dins el calaix/de sota la cadira/de darrere el llum/de sobre la taula és gros.
(13) a dins del calaix/a sota de la cadira/al darrere del llum/a sobre de la taula
(14) el dins del calaix/el sota de la cadira/el darrere del llum the inside of the drawer/the bottom of the chair/the back of the lamp
(15) Sortien d’entre els arbres / Passejaven per entre les flors. they-came-out from among the trees / they walked by among the flowers
References
Cardinaletti, Anna and Giuliana Giusti (2006) ‘The Syntax of Quantified Phrases and Quantitative
Clitics’. In M. Everaert, H. van Riemsdijk, R. Goedemans & B. Hollebrandse (eds.) The
Blackwell Companion to Syntax. Oxford: Blackwell. Vol. IV, Ch. 71 (p. 23-93).
Kayne, Richard (2005) ‘Some Notes on Comparative Syntax, with Special Reference to English and
French’. In R. Kayne, Movement and Silence, Oxford University Press (p. 277-333).
Lorenzo, Guillermo (1995) Geometría de las estructuras nominales. Sintaxis y semántica del SDet. PhD
Dissertation, Universidad de Oviedo.
XVII Colloquium on Generative Grammar – Girona, 13th-15th June 2007
43
On the L-Syntax of Manner and Causation
Jaume Mateu ([email protected])
Universitat Autònoma de Barcelona
In this paper, we argue that so-called ‘manner conflation’ (Talmy 2000) is a local process whose
semantic interpretation is syntactically determined; in particular, our main goal is to show how our
revision of H(ale) & K(eyser)’s (2005) analysis of constructions like (1) naturally leads us to analyze a
variety of causative constructions from a more radical syntactically-driven perspective than theirs.
Interestingly, H&K (2002, 2005) have shown that the reason whereby an intransitive alternant is
possible for (1a), but not for (1b), is related to the different properties of the ‘manner feature’ inherent
in the semantics of the roots involved: it is the case that (2a) is grammatical since splash involves a
manner feature which is ‘linked’ to the internal argument mud, while (2b) is ungrammatical since the
manner feature associated to smear can only be linked externally: crucially, (2b) is ruled out since
there is no agent to license such a feature. In other words, the manner feature is patient-oriented in
(1a), but agent-oriented in (1b). This said, some important remarks are in order: H&K’s analysis does
not appear to capture the relevant fact that splash is not locally conflated in (2c) (vs. cf. (2a)). In fact,
notice that they posit the same l-syntactic analysis for (1a) and (1b): see (2c)-(d). Our proposal is to
analyze the l-syntactic argument structure of splash verbs as similar to that of deadjectival verbs on the
basis that both typically enter into the causative alternation and both have a patient-oriented root: cf.
(3). Indeed, we will take pains to show that there is a crucial difference concerning their formation:
while deadjectival verbs are formed via incorporation of A into V (H&K 2002), splash verbs involve a
syntactic conflation of their root with the inner verb via a ‘plug-in’ device (Mateu & Rigau 2002,
McIntyre 2004, Harley 2005, and Zubizarreta & Oh 2007). [NB: Mateu & Rigau’s (2002) syntactic
plug-in operation has been appropriately referred to as ‘welcome invasion’ by Hirschbühler (2006)].
Indeed, as emphasized by these authors, the insertion of the root splash and smear should not be taken
as a trivial process (contra H&K’s (2002, 2005) simple analyses in (2)) since the syntactic formation
of these complex verbs involves adjunction of a root onto a light verb -a causative one (upper V) or a
transitional one (inner V)-. Accordingly, we argue that H&K’s l-syntactic analyses in (2) should be
recasted as in (4): the fact that now in (4) conflation is locally represented (e.g., cf. (4c-d) vs. (2c-d)),
and the fact that conflation is represented via a syntactic plug-in device contribute to showing a more
syntactically transparent semantic interpretation of the manner component.
On the other hand, we show why the parametrized operation of ‘welcome invasion’ (Mateu &
Rigau 2002; Zubizarreta & Oh (2007)) involved in these examples allows us to account for Talmy’s
(1985, 2000) typological predictions: i.e., Manner verbs (of which those ones in (1) are only a
particular case) are more abundant in Germanic than in Romance (Slobin 1996f). We show that
examples like (1) (but crucially not their Romance counterparts) are in fact complex resultative(like)
constructions where the P head is in fact to be decomposed into a complex one, whereby the visible on
is combined with an abstract TO: indeed, we will show the correctness of adopting H&K’s (2002:
chap. 7) proposal that terminal coincidence relations are more complex than central coincidence ones
(see Svenonius (2006), i.a., for an insightful syntactic recasting of these notions as Path and Place
heads). Accordingly, complex resultative constructions like those in (5a) and (5c) can also be analyzed
as involving conflation of the ‘welcome invasion’ kind: e.g., like smear, kick and push are agent-
oriented roots (H&K 2002, 2005) whereby the l-syntactic analysis in (6) seems to be appropriate (cf.
(4b)-(4d)). On the other hand, parallel examples to the splash case analyzed above where ‘welcome
invasion’ is carried out internally can interestingly be found as well in another lexical semantic area:
e.g. cf. causative constructions where the Theme can be said to move in a particular manner. Indeed,
we show that our present syntactic analysis of strict local conflation naturally leads us to analyze
causative constructions like (7a) from a different, more syntactically-driven perspective than the one
adopted by Folli & Harley (2006): while they argue that both (7a) and (7b) have the very same
syntactic argument structure where the root {√RUN/√WHISTLE} is inserted under a causative v, their
relevant differences being then not syntactically/configurationally represented, we argue, in contrast,
that conflation applies in a more local way whereby the syntactic locus of ‘welcome invasion’ is
different in (7a) from (7b): cf. (8a)-(8b). By using syntactic tests like the causative alternation (H&K
2002; but see Harley 1995), we will be arguing for the hypothesis that two verbal heads are
syntactically represented for (1a) and (7a), but not for (1b), (5a) or (7b): cf. (3a)/(8a) and
(4d)/(6a)/(8b), respectively.
Examples
(1) a. The kids splashed mud on the wall. b. The kids smeared mud on the wall. H&K (2002; 2005)
(2) a. [V Mud [V splash [P on the wall]]] (cf. Mud splashed on the wall) b. *[V Mud [V smear [P on the wall]]] (cf. *Mud smeared on the wall) c. [V splash [P mud [P on the wall]]] (cf. (1a)) d. [V smear [P mud [P on the wall]]] (cf. (1b))
NB: The external argument is not present at l-syntax (H&K 1993; 2002) H&K (2005: 19-21)
(3) a. ([V [V Ø]) [V Mud [V [V √SPLASH –V] [P on the wall]]]]] b. ([V [V Ø]) [V the sky [V V [A√CLEAR]]]] (cf. The strong winds cleared the sky / The sky cleared)
(4) a. [V Mud [V [V √SPLASH –V] [P on the wall]]] b. *[V Mud [V [V √SMEAR –V] [P on the wall]]] c. [V [V Ø] [V mud [V [V √SPLASH –V] [P on the wall]]]]] d. [V [V √SMEAR –V] [P mud [P on the wall]]]
XVII Colloquium on Generative Grammar – Girona, 13th-15th June 2007
45
(5) a. The kids kicked the ball into the kitchen. b. *The ball kicked into the kitchen. c. John pushed the car into the garage. d. *The car pushed into the garage.
(6) a. [V [V √KICK –V] [P the ball [P to [P in- the kitchen]]]] (cf. (4d)) b. *[V The ball [V [V √KICK –V] [P to [P in- the kitchen]]]] (cf. (4b))
(7) a. He ran the rats through the maze. (cf. The rats ran through the maze) b. Mary whistled Rover to her side.
Folli & Harley (2006)
(8) a. [V [V Ø]) [V the rats [V [V √RUN –V] [P through the maze]]]]] b. [V [V √WHISTLE –V] [P Rover [P to her side]]]
Selected references
Folli, R. & Harley, H. (2006). “On the Licensing of Causatives of Directed Motion: Waltzing Matilda
all over”. Studia Linguistica 60.2: 121-155.
Hale, K. & Keyser, S. J. (2005). “Aspect and the Syntax of Argument Structure”. In Erteschik-Shir, N.
& Rapoport, T. (eds.). The Syntax of Aspect. Deriving Thematic and Aspectual Interpretation.
Oxford: OUP.
Heidi, H. (2005). “How Do Verbs Get Their Names? Denominal Verbs, Manner Incorporation, and
the Ontology of Verb Roots in English”. In Erteschik-Shir, N. & Rapoport, T. (eds.). The
Syntax of Aspect. Deriving Thematic and Aspectual Interpretation. Oxford: OUP.
Hirschbühler, P. (2006) “The oblique Locatum in the locative alternation”. Talk delivered at UAB,
Bellaterra, Barcelona.
Mateu, J. & Rigau, G. (2002). “A Minimalist Account of Conflation Processes: Parametric Variation
at the Lexicon-Syntax Interface”. In Alexiadou, A. (ed.). Theoretical Approaches to Universals.
Amsterdam: John Benjamins.
McIntyre, A. (2004). “Event paths, conflation, argument structure, and VP shells”. Linguistics 42.3:
523-571.
Zubizarreta, M. L. & Oh, E. (2007). On the Syntactic Composition of Manner and Motion. Cambridge,
MA: MIT Press.
Set-Merge and Pair-Merge in coordination and subordination
Gabriela Matos ([email protected])
Universidade de Lisboa / Onset-CEL
1. Coordinate and subordinate adverbial sentences show a different behaviour, suggesting that they
have different syntactic structures. However, coordination is not a unitary phenomenon and may
exhibit configurations resulting from set-Merge as well as from Pair Merge. Thus, differences between
coordination and subordination should mainly rely on the features of Conj vs C(omp).
2. Coordinate and subordinate structures exhibit different syntactic properties (e.g. Ross 1967, Quirk
& al. 1985, Haspelmath 2004, Matos 2006): in contrast with concessive subordination, in (1),
adversative coordination in (2) does not allow displacement of any conjunct (2b), may connect
constituents below the sentence level (2c), favours across-the-board extractions of constituents inside
the conjuncts (2d) and, in sentence coordination, it is not sensitive to finiteness (2e). These differences
apparently indicate the existence of a divergence in their structural configurations. In Principles and
Parameters Theory, it has classically been suggested that adverbial clauses are adjuncts. On the other
hand, it is currently accepted that coordinate structures should be analysed in terms of the specifier-
head-complement configuration, (3) — Kayne 1994, Johannessen 1998.
3. Yet, a distinction between coordination and subordination on the basis of the adjunction vs.
specifier-head-complement structure is not convincing: sentence complementation may also be
analysed in terms of the latter configuration; besides, coordination is not a uniform phenomenon and
presents cases that are best viewed as instances of Adjunction, to be accounted for in terms of Munn’s
Adjunction hypothesis, (4). This is the case of appositive coordination, a construction usually not
mentioned, (5).
The contrasts in (6) and (7) show that appositive coordination differs from its non-appositive
counterpart. In particular, appositive sentences block c-command from a relevant expression outside of
it. Thus, in (7a) the appositive coordinate sentence the pronoun bound-variable reading is not allowed;
in (7b) the negative polarity of the underspecified polarity expression is not fixed; in (7c) Principle C
violations are blocked. I assume that these contrasts are partially a consequence of the type of Merge
involved: while the non-appositive coordination results from Set-Merge, appositive coordination
involves Pair-Merge.
4. Since sentence coordinate do not differ from subordination in the building structure operations
involved, what determines their different behaviour, cf. (1), (2)? I assume that the crucial difference
relies on the underspecified nature of Conj. As opposed to C(omp), Conj lacks specific categorial
features and does not impose any restriction on the categorial nature of its arguments. Its categorial
features are fixed by the conjuncts (Gazdar et al. 1985, a.o). Thus, the whole coordinate structure is
understood as a segment of the category of one of these conjuncts: in Spec-Head-Compl
XVII Colloquium on Generative Grammar – Girona, 13th-15th June 2007
47
configurations, typically the specifier (Joahnnessen 1989, a.o.), (8a), in appositive coordination,
presumably the Compl of Conj, (8b).
5. Finally, how does Adjunction accounts for the contrasts between appositive vs. non- appositive
coordination, cf. (6)-(7)? As for relative clauses, Lebaux 1988 claimed that they were exempt from c-
command due to Late Adjunction. Yet, some adjunct clauses present ccommand effects (Chomsky
2004, 2005), see (15). So, an alternative should be devised for appositive sentences, which
systematically block c-command. Relying in the parenthetical status of an appositive (sentence
coordinate) adjunct, I assume that its head exhibits the feature [+parenthetical], which is interpreted by
the computational system as a clue for its autonomy with respect to the constituent it adjoins to. Since
the derivation proceeds bottom up and Transfer applies as early as possible, this Phase is transferred to
SEM before the Phase it is inserted in, preventing c-command effects inside the appositive adjunct.
Examples
(1) a. Ele saiu, embora estivesse cansado. He went out, although he was tired b. Embora estivesse cansado, c/c saiti. Although he was tired, he went out c. ?/*Muitas canetas embora poucos livros caírarn no chão. Many pens although few books fell down on the floor. d. 0 que fez ele, embora estivesse cansado? What did he make, although he was tired? e. Ele afIrmou sair embora {*estar estivesse cansado} He claimed to go out although {be/ (he) was tired}
(2) a. Ele saiu, mas estava cansado. He went out, but he was tired b. *Mas estava cansado, ele saiu. But he was tired, he went out c. Muitas canetas mas poucos livros caíram no chão. Many pens but few books fell down on the floor d. *0 que fez ele, mas estava cansado? What did he make, but he was tired? e. Ele afIrmou {sair mas estar cansado / que saía mas estava cansado } He claimed to go out but to be tired / that he would go out but that he was tired}
(3) Spec-Head-Cornpl Hypothesis: [ConjP XP [Conj’ [Conj ] YP] (cf. Kayne 1994)
(4) Adjunct Hypothesis: [XP XP [ConjP [Conj] YP] ] In Munn ConjP=BP (Boolean Phrase)
(5) O Pedro, e ele dorme rnuitopouco, nunca está cansado. (Appositive Coordination) Pedro, and he hardly sleeps, is never tired.
(6) Non Appositive coordination a. Toda a mulheri gosta de urn homem e confia em sii própria. Every women likes a man and thrusts herself. b. Nenhum irmão o procurou ou um só amigo fez qualquer esförço nesse sentido. None of his brother looked for him nor any friend made any effort in that sense. c. *Elai e a filha daMariai são as minhas melhores amigas.
She and Mary’s daughter are my best friends.
(7) Appositive coordination a. *Todo o hornemi, e uma mulher confia nelei, está por vezes deprimido.
Every man, and a women thrusts him, is sometimes depressed. b. *Nenhum irmão, ou urn só amigo fez qualquer esförço nesse sentido, o procurou.
None of his brothers, or any friend made any effort in that sense, looked for him. c. Acho que elai, e a Mariai, é a rninha rnelhor arniga, é uma pessoa encantadora.
I think that she, and Mary is my best friend, is a charming person.
(8) a. [ConjP = TP Ele saiu [ [Conj = T mas ] [TP estava cansado ] ] ] (cf. (2a)) b. [DP [DP O Pedro] [ConjP = CP/TP [Conj = C/T e ] [CP/TP ele dorme muito pouco ] ] ] ... (cf (5))
(9) [CP [TP [TP Ele saiu] [CP embora estivesse cansado] ] ] (cf. (la))
(10) Hei asked which picture [that John*i liked ] Mary bought t. (Chomsky 2004)
XVII Colloquium on Generative Grammar – Girona, 13th-15th June 2007
49
‘Where did that clitic come from?’: the case of non-subcategorized clitics in Greek
Marios Mavrogiorgos ([email protected])
University of Cambridge
It is commonly assumed that unergatives do not normally transitivize ‘lexically’ (of Comrie 1981) as
opposed to unaccusatives, with any apparent counterexamples analyzed as involving causativization of
an ‘unaccusative’ result state (see e.g. Levin & Rappaport-Hovav 1995). In this paper I present
empirical evidence from Modem Greek (henceforth, Greek) which shows that this correlation between
unergativity and intransitivity is not absolute, and that ‘lexical’ causativization/transitivization may
involve more than one sub-type (including transitivization of purely unergative/intransitive V-roots),
depending on the morpho-syntactic properties of the internal argument (clitic vs. non-citic) and the
grammatical/derivational level at which this argument is merged (root/lexical vs. functional/syntactic).
I review sub-classes of unergative predicates in Greek (mainly manner of motion and bodily
function verbs) which may transitivize with varying degrees of productivity (see e.g.
Theophanopoulou-Kontou 2003; Chondrogianni 2004 and Roussou & Tsimpli to appear). These
predicates may belong to one of two subtypes (or both, of (1 )-(2)), depending on the obligatory (type
1) vs. non-obligatory (type 2) presence of an accusative direct object clitic (the ‘causee’), as well as on
a number of other interrelated properties (e.g. presence vs. absence and type of causation;
interpretation of the causee). Type 2 predicates can be further divided into two smaller subgroups
based on the meaning of the transitivized predicate (subgroup (2a): meaning shift/new use; subgroup
(2b): novel causative reading, formation of a new anticausative form). I further show that:
a) type 1 predicates are much more uniform th their properties than type 2 ones, which form a
continuum;
b) type 1 predicates are nevertheless less common than type (2a) predicates, which in turn are less
common than type (2b) predicates;
c) the internal argument of type 1 predicates is merged higher than the VP/thematic domain (but
within the vP) and is not a cognate object, while in type 2 predicates it is merged lower (within the
lexical domain).
More generally, type 1 predicates behave more like unergatives, while type 2 predicates behave more
like unaccusatives.
Based on these and other properties, I propose that type 1 predicates are derived in the syntax
(or the functional vP phase layer, if no Lexicon is presupposed; see Marantz 1997), as the result of
merging an unergative/intransitive V with a causative/agentive/transitive v*. Accordingly, the non-
selected accusative clitic merges directly on v* (or, alternatively on a non-thematic functional head
containing the object φ-features of v*), valuing v*’s object-phi features (and realizing its case), a
possibility linked to the fact that clitics are morpho-syntactically and semantically underspecified φ-
feature bundles that may incorporate onto their host at any stage of the derivation (under certain
morpho-syntactic conditions). The mixed agentive/theme interpretation of the clitic-causee is
accounted for as the combined result of a feature sharing mechanism inherent to the incorporation
process, which allows the agent theta-features of v* to be copied onto the citic, and of accusative case
assignment. Type (2a) predicates are, on the other hand, analyzed as instances of lexical
causativization, whereby a transitive/causative (but not necessarily agentive) v* merges with an
intransitive V in the lexicon. creating in this way a (new) theme position (see Theophanopoulou-
Kontou 2003), in which the citic/DP merges (or, in a non-lexical model, as the result of coercion at the
root level, where the clitic merges at the root and then moves to a higher functional head, which is
crucially lower than the one of type 1 predicates). Finally, I argue that type (2b) predicates are genuine
unaccusatives selected by a v head (possibly reanalyzed former unergatives), with the clitic/DP
merged in the theme position (or the root level in a non-lexical model). Merging with a v* head will
yield the causative/transitive variant. At the end of the paper I briefly discuss some of the theoretical
implications of this analysis (such as e.g. the relation between theta-assignment and case, the notion of
internal argument, the syntax-Lexicon interface, and the nature of cliticization and clitic doubling).
Examples
Type 1 predicates (obligatorily cliticized): (1) a. O proponitis ?*(tus) etrexe tus pektes gia pano apo dio ores the coach-NOM them-CL ran-3SG.IMP the players-ACC for over than two hours
mesa sto krio. in the cold
‘The coach was running the players for over two hours ii the cold weather.’ b. ?*(Ton) gonatise ton Iosif sti mesi tu domatiu.
him-CL kneeled-ACC the Joseph-ACC in-the middle the room-GEN ‘He made Joseph kneel in the middle of the room.’
c. ?Tu edose mia klotsia ke ton anastenakse. him-CL.GEN gave-3SG a kick-ACC and him-CL sighed-3SG ‘He gave him a kick and made him sigh.’
Type 2 predicates (non-obligatorily citicized): (2) a. Perpatisa to skilo stin paralia gia deka lepta. walked-1SG the dog-ACC at-the sea-shore for ten minutes
‘I walked the dog at the sea-shore for ten minutes. b. I Eleni etrekse ta windows. the Helen-NOM ran-3SG the windows-ACC ‘Helen ran the Windows.’ c. Ta pola eksoda gonatisan ton patera tu Andrea. the many expenses-NOM kneeled-3PL the father-ACC the Andrea-GEN ‘Andrea’ s father suffered under the bulk of the expenses.’
References
Chondrogianni, V. (2004) Comparing causative and inchoative alternations in Greek and Turkish.
Unpublished ms., University of Cambridge.
Comrie, B. (1981) Language Universals and Linguistic Typology. Oxford: Blackwell.
XVII Colloquium on Generative Grammar – Girona, 13th-15th June 2007
51
Levin, B. and M. Rappaport-Hovav (1995) Unaccusatives. At the syntax-lexical semantics interface.
Cambridge, Mass.: MIT Press.
Marantz, A. (1997) No escape from syntax: Don’t try morphological analysis in the privacy of your
own lexicon. University of Pennsylvania Working Papers in Linguistics 4.
Roussou, A. and I. M. Tsimpli (to appear) Clitics and Transitivity. Proceedings of the Workshop on
Modern Greek Morpho-syntax, LSA, Harvard, July 2005.
Theofanopoulou-Kontou. D. (2003) Τα ρήµατα κίνησης της Νέας Ελληνικής και η µεταβιβαστικη
τους χρήση. In D. Theofanopoulou-Kontou, C. Laskaratou, M. Sifianou, M. Georgiafentis, and
V. Spyropoulos (eds.) Σΰγχρονες Τάσεις στην Ειρήνη Φιλιππάκή – Warburton, 236.255. Αθήνα:
Πατάκης.
On the Argument Structure of le-predicates in Mexican Spanish
Ía Navarro ([email protected])
Universitat Autònoma de Barcelona
We will propose an analysis for the argument structure of the complex le-predicates, i.e.,
Mexican Spanish complex predicates composed by a verbal base and a non co-referential le clitic that
canonically is a 3rd Person, Singular, Dative Pronoun but in le-predicates it is a formally defective
syntactic argument (cf. (1)-(3)). We affirm that le-predicates involve both event modification and
unergative complex predicate formation that occur at the syntactic-semantic and the lexical-syntactic
interface.
Our proposal concerns two main assumptions: the first assumption is linked to the following three
linguistic facts
(i) le has suffered a weakening of its referential and anaphoric status (Torres Cacoullos 2002,
Company Company 2004);
(ii) this weakening is concomitant to the grammaticalization of a (goal) thematic role (Company
Company 2004); and
(iii) these processes result in a very productive formation of both idiomatic expressions and
complex predicates with le (Kany 1951, Torres Cacoullos 1999, 2002, Company Company
2004).
The second assumption is that le-predicate formation results from a pseudo-incorporation process
(Dayal 2003, Espinal 2006, Espinal & McNally in press), in which a non instantiated argument le with
<e,t> semantic type becomes part of the predicate by event modification. This is possible because le is
a syntactic argument φP (Dechaine & Wiltschko 2002) but semantically defective, then it does not
denote an individual but a thematic property.
We will propose a preliminary classification of Spanish predicates based on the relevant
characteristics that are related with the complex le-predicates. These are:
a) the type of subject, internal or external (Levin & Rappaport 1995, Demonte 2002 (cf. (3));
b) the event structure of the predicates as composed by state (statives) or process (activities)
subevents, and having simple or complex event structure (Pustejovsky 1995, Levin & Rappaport
1995); and finally,
c) the l-syntax projection.
Based on our preliminary classification we say that the formation and the semantic
composition of le-predicates depend on three argument structure requirements and on two constraints.
The requirements include:
(i) an internal argument available for syntax saturation (an Accusative, a Cognate or PP locative),
(ii) an Agentive (external) subject, and so a v external argument projection, and
(iii) a process subevent involved in the event structure.
XVII Colloquium on Generative Grammar – Girona, 13th-15th June 2007
53
The two constrains are:
(a) the ditransitive verbal bases cannot derive le-predicates since they have Dative selection
lexicalized (then le it is always instantiated), and
(b) the locative predicates (but not locatum) have already a goal incorporated (a la Halle & Keyser),
then locative verbs cannot derive le-predicates either.
Hence the possible le-predicate verbal bases are (mono) transitive (cf. (1)), unergative (cf. (2)) and
agentive unaccusative verbs (cf. (3)), and the impossible le-predicate verbal bases are ‘pure’
unaccusative (cf. (4)), stative (cf. (5)), ditransitive (cf. (6)), locative (cf. (7)) and all no Agentive
verb’s variants.
Our analysis argues that le-predicates involve event modification by the syntactic argument le
(φP) that is legitimated into a unergative-like argument structure. The unergative argument structure
satisfy the requirements of le-predicates formation in the next way: in the unergative argument
projection there is an activity structure that ensures the internal argument position for le, also triggers
the external subject interpretation for the Agent, and finally, this involves the necessary process
subevent that is the target of the goal modification in the semantic derivation. Furthermore, in the case
of change of state transitive verbs (with a change of state subevent subordinated to an activity), le
pseudo-incorporation causes both intransitivization and the blocking of a semantic argument for the
change of state subevent. We take the right dislocation of the erstwhile Accusatives as syntactic
evidence of this blocking (cf. (1)).
We will demonstrate that le-predicates argument structure is similar to the unergative
argument structure, even if the verbal base is either transitive, unergative or (agentive) unaccusative.
Le-predicates are similar to unergative predicates in that they have an Agentive external subject, they
denote processes (activities) and they have a (le) syntactic argument that is not an affected object (but
an event modifier).
Examples:
(1) Te dije que no lei abrieras a la puerta*i/j (Transitive base) You.DAT (I)-told that NEG LE open to the door.OBL. “I told you not to open le at the door.”
(2) Vamos a bailarle un poquito más para allá. (Unergative base) Let’s-go to dance-LE a little-bit more for there “Let’s go dance le a little bit further”
(3) ¿Le sales tú o le salgo yo? (Unaccusative base) LE go-out you or LE go-out I “Are you going out le or do I?”
(4) *Los claveles no le florecen. (‘Pure’ unnacusative base) The carnations NEG LE flower *“The carnations does not flower le”
(5) *Le estoy cansada (Stative base) LE I-am tired. *”I le am tired”
(6) *¡A regalarle! (Ditransitive base) To gift-LE *“Let’s gift le”
(7) *Ayúdame a enjaularle. (Locative base) Help-me to cageLE *“Help me to cage le”
Selected Bibliography:
Company Company (2004), “Gramaticalización por Subjetivización como Prescindibilidad de la
Sintaxis”, Nueva Revista de Filología Hispánica, LII, 1:1-27
Dayal, V (2003), A semantics for pseudo-incorporation, Ms. Rutgers.
Demonte. V (2002), “Preliminares de una clasificación léxico-sintáctica de los predicados
verbales del español”, en Sybille Grosse y Axel Schönberger (eds.), Ex oriente lux:
Festchrift für Eberhard Gärtner zu seinem 60. Geburtstag. Frankfurt am Main: Valentia.
Espinal, M.T. (2006), “Clitic incorporation and abstract semantic objects in idiomatic
constructions”. Ms. Universitat Autònoma de Barcelona.
Levin, B. & M. Rappaport (1995), Unaccusativity, Cambridge, The MIT Press.
Torres Cacoullos, R. (2002), “Le: From Pronoun to Verbal Intensifier”, Linguistics 40.2: 285-318.
XVII Colloquium on Generative Grammar – Girona, 13th-15th June 2007
55
Preposition stranding under sluicing in Western Romance
A. Nevins ([email protected]), C. Rodrigues ([email protected]) & L. Vicente
Harvard University / Unicamp - Brazil / Leiden University
Intro - Merchant (2001:92) claims that, for any given language, preposition stranding under sluicing is
possible only if it is also possible in non-elliptical environments. In this talk, we focus on two
languages (Spanish and Brazilian Portuguese) that appear to falsify Merchant’s generalisation: while
they are non-P-stranding languages (1), they do allow P-drop under sluicing (2). Our claim is that,
despite initial appearances, these two languages are not counterexamples, and that sluicing cannot
salvage P-stranding violations (contra Almeida & Yoshida 2007). Therefore, grammatical P-drop
sluices like (2) must stem from a structure that doesn’t involve a P-stranding violation in the first
place.
Analysis — We propose that, in both Spanish and BP, P-drop sluices are derived from underlying
cleft-like structures. For BP, we claim that the source structure is a cleft composed of a relative clause,
as relative clauses in BP independently allow for prepositions to be omitted (3a). Hence, no P-
stranding violation occurs (3b). However, this analysis is not extensible to Spanish, given that it
doesn’t allow P-drop in relative clauses (4a). Rather, we propose that the source structure for Spanish
is a copular clause functioning as a cleft. The wh-phrase that functions as the subject of the copula
need not be introduced by a preposition (4b), thus circumventing the P-stranding violation (4c). Thus,
the conclusion is that there is no unique way to derive P-drop sluices: simply, each language makes
use of whatever strategies are available in order to produce the same result.
Support — An interesting restriction is that, although both Spanish and BP allow for multiple sluicing
with prepositional wh-phrases, it is impossible in these cases to drop either one or both of the
prepositions (5). This restriction follows from our analysis: the pivots of clefts and relative clauses can
only accommodate one constituent. Therefore, since multiple sluicing must stem from a regular
interrogative, it will incur a P-stranding violation.
Other arguments are language specific. For instance, in Spanish, P-drop sluices are incompatible with
modification by más ‘else’ (6a), which follows from the exhaustivity requirement on the pivots of
clefts. We give (6b) as a control to show that else modification is otherwise allowed in regular
interrogatives.
In BP, on the other hand, an additional argument comes from the verb conhecer ‘to be acquainted
with’, which requires P-drop under sluicing of its complement (7a). This restriction follows from our
analysis once we factor in the observation that conhecer cannot select a CP. Hence, it must be the case
that its complement is not an interrogative clause, but a (free) relative clause headed by a wh-phrase
(8a). Furthermore, in Spanish, the equivalent verb conocer doesn’t allow sluicing of its complement,
independently of whether the preposition is dropped or not (7b). This contrast supports our claim that,
in Spanish, P-drop sluices stem from a copular clause (i.e., a clausal constituent), and not from a
relative (8b).
Extensions — Our analysis preserves Merchant’s original intuition that, unlike strong island
violations, P-stranding violations cannot be circumvented via ellipsis. This asymmetry suggests a non-
unified analysis of these two phenomena. In cross-linguistic terms, we note that P-drop sluicing is not
a quirk of Spanish and BP: rather, it is attested in a wide range of languages, such as Finnish, Dutch,
Czech, Polish, or SerboCroatian. The strongest hypothesis is that all these languages also make use of
alternative sources for sluicing that do not involve P-stranding.
Examples
(1) a. *Quem que a Maria dançou [com __]? [BP] who that the Maria danced with
b. *¿Quién ha bailado María [con __]? [Sp] who has danced María with
(2) a. √ A Maria dançou con alguém, mas eu não sei quem [BP] the Maria danced with someone but I not know who
b. ?/√ María ha bailado con alguien, pero no sé quién [Sp] María has danced with someone but not know who
(3) a. A pessoa (com) que a Maria dançou na festa estava bêbada the person with that the Maria danced in-the party was drunk b. … mas eu não sei quem [é a pessoa que ela dançou] [≈ (2a)] but I not know who is the person that she danced
(4) a. La persona con que María bailó en la fiesta estaba borracha the person with that María danced in the party was drunk
b. ¿Quién es la persona con quién María bailó en la fiesta? who is the person with who María danced in the party c. … pero no sé quién [es la persona con quién María bailó] [≈ (2b)] but not know who is the person with who María danced
(5) a. Ela falou sobre alguna coisa para alguém [BP] she talked about some thing to someone
… mas eu não sei *(sobre) o que *(para) quem but I not know about the what to who
b. Ella habló sobre algo con alguien [Sp] she talked about something with someone
… pero no sé *(sobre) qué *(con) quién but not know about what with who
(6) a. Juan bailó con una chica, pero no sé *(con) qué chica más Juan danced with a girl but not know with what girl else
b. ¿Con qué chica más bailó Juan? with what girl else danced Juan
(7) a. Maria dançou com alguém, mas eu não conheco (*com) quem [BP] Maria danced with someone but I not know with who
b. *Maria bailó con alguien, pero no conozco (con) quién [Sp] Maria danced with someone but not know with who
XVII Colloquium on Generative Grammar – Girona, 13th-15th June 2007
57
(8) a. ... eu não conheco [DP quem [é a pessoa que ela dançou] ] [≈ (7b)] I not know who is the person that she danced
b. ... no conozco [CP quién [es la persona con quién ella bailó] ] [≈ (7b)] not know who is the person with who she danced
References:
Merchant, 2001 The syntax of silence, Oxford University Press
Almeida & Yoshida, 2007 “A problem for the preposition stranding generalisation”, LI 38.2.
Stem-adjacent blocking in morphological fission
Rolf Noyer ([email protected])
University of Pennsylvania
In Distributed Morphology (Halle & Marantz 1993), the mapping between abstract syntactic terminals
and their phonological exponents or ‘spell-out’ is by default one-to-one; that is, each syntactic
terminal has as its phonological image a single ‘piece’ or vocabulary item (1a). Positional
disjunctivity, that is, the blocking of a ‘general’ vocabulary item by another ‘more specific one’ (as for
example English irregular plural -en blocking regular -s), thus follows directly from the natural
assumption that a terminal can be spelled out only once. Nevertheless, exceptions to the simple one-to-
one relation are not infrequent; an additional mechanism of morpheme fission (Noyer 1997, Halle
1997) in particular permits a single syntactic terminal to have several vocabulary items as its
phonological image (1b).
As Noyer 1997 observed, however, if morpheme fission is permitted in theory, the problem of
positional blocking reemerges, since if a morpheme may ‘split’, there is no longer any one single
position for which affixes compete. For example in the classical Arabic prefix conjugation (2) the
agreement features of person, number and gender and mood are parcelled out into both prefixes and
suffixes (3). For example, feminine plural together appear as suffixal -na, while feminine without
plural is prefixal t-. Dual number is suffixal -ā, but plural is combined with 1st person as prefixal n-.
Nevertheless, at most one prefix and at most one suffix may attach to the verb stem. The default prefix
y- and default suffix -u appear where no other can be spelled out were it not for positional disjunctivity
the grammar would not be constrained from generating prefix combinations such as *y-t- or suffix
combinations such as *-ā-u.
In response to these problems, Noyer 1997 resorted to a diacritic on morphemes which controlls
whether they could ‘split’ into pieces or not, and in addition postulated that ‘positions of exponence’
could be stipulated in language-specific word templates. Halle 1997, on the other hand, introduced ad
hoc morpheme-splitting rules prior to vocabulary insertion, as well as stipulations on the blocking of
default affixes.
The present paper rejects both of these complications of the theory, and takes as its starting point the
empirical observation that positional blocking and fission co-occur only when spell-out comprises both
prefixes and suffixes at once (a ‘circumfixal’ distribution), as in Arabic (2) or in Georgian (Anderson
1982). More specifically, blocking is observed in the positions immediately left-adjacent to the stem
(‘prefix’) and immediately right-adjacent (‘suffix’) as schematized in (4a). Blocking is never attested
among vocabulary items which attach via fission but non-adjacent to the stem (4b). This restriction is
derived automatically if positional blocking results only from local selectional restrictions on
vocabulary items. In Arabic, for example, in t-aktub-na ‘you (fem. pl.) write’, default y- and -u are
XVII Colloquium on Generative Grammar – Girona, 13th-15th June 2007
59
specified as insertable only immediately adjacent to the stem ( Root+v in syntax) as in (5): when more
specified t- and -na are inserted adjacent to the stem, then y-- and -u are blocked (6).
Further evidence that the stem-adjacent position is the only locus of blocking in fission comes from
mixed systems in which a class of stem-adjacent affixes participate in blocking, but non-stem-adjacent
affixes do not. For example, in the post-thematic conjugation in Huave (7), preterite -t, 1st person
exclusive -n, and 2nd person -r are stem-adjacent only and block default -m; the remaining suffixes do
not have strict selectional requirements and do not participate in blocking. For example, the default
plural suffix -on is not limited to a ‘position’ one affix to the right of the verb stem; instead it may
immediately follow stem-attaching -m, -r or -t, or more peripherally still after -os (1st person when -n
is blocked by -t) or -ear (2nd person when -r is blocked by -t).
Examples
(1) a. Ordinary spell-out b. Morpheme fission:
syntax: [ [ ... ] X°] [ [ ... ] X°]
phonology: stem + affix stem + affix1 + affix2 + affix3 + …
(2) ‘to write’ singular dual plural 3 masc y-aktuh-u y-aktuh-d y-aktuh-ü 3 fem t-aktuh-u t-aktuh-d y-aktuh-na 2 masc t-aktuh-u t-aktuh-d t-aktuh-ü 2 fem t-akuth-i t-aktuh-d t-aktuh-na 1 2-aktuh-u n-aktuh-u
(3) lpl ←→ n- plural ←→ -ū
1 ←→ Ȥ- fem sg ←→ -ī 2 ←→ t- fem ←→ t- fem pl ←→ -na (elsewhere) ←→ -u dual ←→ -ā (elsewhere) ←→ y-
(4) a. Morpheme fission with b. Morpheme fission with non-local positional blocking blocking: unattested
syntax: [ [ ... ] X°] [ [ ... ] X°]
phonology: affix + stem + affix stem + affix + affix ↑ ↑ ↑
blocking blocking blocking
(5) 2 ←→ t / ___ [ plural fem ←→ na / ] ___ (elsewhere) ←→ u / ]V ___ (elsewhere) ←→ y / ___ [V
(6) [ [ aktub V ] Agr ]
(*y-) t- [ aktub V ] - na (*-u)
(7) ‘to rise’ atemporal preterite [-plural] [+plural] [-plural] [+plural] 1 excl witi-n witi-n-on witi-t-os witi-t-as-on 1 incl witi-m-or witi-m-o:ts witi-t-or witi-t-o:ts 2 witi-r witi-r-on witi-t-ear witi-t-ear-on 3 witi-m witi-m-oh witi-t witi-t-oh
References
Anderson, Stephen R. 1984. ‘On representation in morphology: case agreement and inversion in
Georgian.’ NLLT 2: 157-218.
Halle, Morris. 1997. ‘Distributed morphology: Impoverishment and fission.’ MITWPL 30:425-449.
Halle, Morris & Alec Marantz. 1993. ‘Distributed morphology and the pieces of inflection.’ In The
View from Building 20, ed. Ken Hale and S. Jay Keyser. MIT Press, 111-176.
Noyer, RoIf. 1997. Features, positions and affixes in autonomous morphological structure. Garland
Publishing, New York.
XVII Colloquium on Generative Grammar – Girona, 13th-15th June 2007
61
Unambiguous SE
Francisco Ordóñez, & Esthela Treviño
SUNY Stony Brook & UAM Iztapalapa
In this presentation, we will analyze only the non-paradigmatic instances of the clitic SE, those
included in what has been called Imp(ersonal)-SE. We will assume that the single non- paradigmatic
use of SE occurs in contexts like those in (1).
(1) a. Jamás SE calumnió al gobernador. b. SE supo que vendría el embajador de Irán. c. Se rechazó un sinnúmero de solicitudes.
Contexts in which distinct meanings of reflexivity such as the syntactic (‘Me vestí’), the inherent
reflexive (‘Te atreviste’), the inchoative (‘Nos alborotamos’), as well as utterances with the
benefactive (‘Se construyó una casa’), and the aspectual (‘Me leí un cuento’), reading of SE, all admit
the paradigmatic series of personal pronouns of which the third person is realized as SE.
Since Cinque’s 1988 seminal work on SE, it is assumed, e.g. Dobrovie-Sorin 1998, D’Alessandro
2004, that Imp-SE is ambiguous, (+ACC in 1b,c; -ACC in la), in languages of the Catalan, Spanish
and Italian type. Acc-SE frees T to AGREE with the DP-object, (1b,c); NOM-SE entails that the DP-
object checks ACC case, and T shows (default) 3rd Agr, (1a). However, the ambiguity hypothesis
makes the wrong predictions. A double syntactic characterization of SE cannot explain the fact that the
single DP argument unequivocally acts like an object in active sentences, differing crucially from its
behavior in passive sentences. First, unlike passives, (2b), Non-Par(adigmatic)-SE rejects NOM
subject pronouns, (2a); only the reflexive interpretation in (3) is permitted. The impersonal reading
obtains in (4) only if the pronoun emerges as a regular object: preceded by the so-called personal-a,
and doubled by the corresponding CL(itic).
(2) a. *Ayer SE arrestaron ellos. b. Ellos fueron arrestados.
‘They were arrested’
(3) SE veían-3pl ellos. ‘They saw themselves/each other’
(4) SE los-3M-PL/les-3PL veía a ellos-3M-PL. ‘They were seen’
Second, determinerless DPs are licensed as arguments of SE constructions, (5a), but they are excluded
in passives (5b):
(5) a. Se encontró plata. b. * Fue encontrada plata.
Third, a null argument is fully licensed in passives, but illegitimate in Non-Par-SE contexts:
(6) En las elecciones fue castigado pro. (Passive) ‘The candidate was punished’
(7) *En las elecciones SE castigó pro. ‘The candidate was punished’
The contrast shows a subject pro-dropped in (6), while object pro-drop is shown in (7). Unlike
reflexive-SE (the only possible reading in 7), Non-Par-SE requires the overt presence of the object
argument in situ, be it a D/NP (8a), or a CL in the context of CLLD or extraction:
(8) a. SE castigó al candidato. b. Al candidato SE le castigó / Al candidato que se (le) castigó.
The explanation for (4, 7 and 8) comes from the important role that animacy plays in Spanish, which is
preserved in impersonal constructions as opposed to Italian and Romanian; these languages do not
represent animacy in Non-Par-SE-contexts. It is clear that the Spanish animacy feature is not
eliminated from the VP thus, it bars an empty pronoun in null object contexts, (7), and requires the
presence of personal-a in (4). Evidence that animacy is preserved comes from the fact that in many
Spanish varieties (excluding some from Argentina), that feature surfaces as the DAT clitic form in non
paradigmatic SE constructions, as seen in (10). This DAT represents pure Animacy; otherwise, ACC
clitics are used:
(9) A Juan lo vi contento. (Mexican, Colombian, Peruvian Spanish) (10) A Juan se le vio contento. (Mexican, Colombian, Peruvian Spanish)
Proof that this DAT does not involve a process of dativization but is rather the spell out of the animacy
feature, is offered by instances where LE in Non-Par-SE contexts, does not show doubling with overt
postverbal objects or with extracted wh-questions:
(11) a. */? ¿A quién SE le intentó visitar? b. *¿A qué SE le leyó en voz alta?
Inanimates, on the contrary, do no permit the presence of the clitic, as observed in (12).
(12) a. Esto lo dices en la reunion y te ahorcan. b. *Esto SE lo/le dice en la reunion y te ahorcan.
The contrast in animacy and the possibilities of expressing the clitic, recalls the clitic doubling
paradigm observed with objects in situ in Rio Platense Spanish (Suñer 1988, Jaeggli 1982). Inanimates
disfavor clitic doubling, but animates (almost) do require it. Thus, we find that all Spanish varieties
follow the same pattern regarding the interaction between animacy and clitics, whenever the object is
moved to the left in the Non-Par-SE constructions. The parallelism with clitic doubling in Argentinian
Spanish is further confirmed by the fact that this Spanish variety takes the same set of clitics in
doubling, and in Non-Par-SE contexts with left movement of the object; it further shows the same
contrasts in animacy:
(13) A Juan SE lo ye contento. (Clitics in Argentinian Spanish)
(14) Lo vi a Juan
(15) *Esto SE lo dice
(16) *Lo dices esto
As expected, inanimates can be salvaged if the personal-a is reintroduced in specific syntactic
environments, as shown in (17). This is a pattern found in clitic doubling with inanimates in
Argentinian Spanish as well:
XVII Colloquium on Generative Grammar – Girona, 13th-15th June 2007
63
(17) A esa película SE la/le otorgó el Oscar. (All Spanish varieties)
Under our analysis, SE is radically defective. It does not deactivate the animacy feature expressed in
transitive constructions. This property triggers a process of animacy recoverability by means of clitic
doubling in all Spanish varieties. SE is also defective with respect to agreement features on the verb
(Raposo and Uriagereka 1996), thus, Agr on the verb is free to probe into the object overtly (1c) or
covertly (1a). Furthermore, it is not surprising to find overt Agr probing into the object with personal-
a; this is indeed quite common in different varieties of Spanish (18).
(18) a. [...] desde el inicio . . .SE han-3PL detenido a 398 personas-PL. (La Nación, Argentina) b. Con qué tipo de testimonios SE condenaron-3PL a estas personas-PL? (Nexos, Mexico)
Moreover, SE —under constrained conditions and in certain varieties— may appear in syntactic
passives with an overt internal argument which gets NOM and Agr markings; see (19) taken from
(Reforma, Mexico).
(19) [...] a quien SE le fue negada-SG-FEM la visa-SG-FEM [...] SE+Passive+object
We claim that the simplest assumption regarding case-marking, is that Non-Par-SE is caseless. As it
will be discussed, there is a further obstacle against the NOM/ACC Case hypothesis; this is exposed
by examples with passives in which the notional object is moved to the left as in (20), and remains as a
syntactic object; notice that it triggers agreement on the passive verb.
(20) A las brujas-FEM-PL se les-PL era-SG sometidas-FEM-PL a un cruel interrogatorio.
Summing up: we claim that SE is radically underspecified. It does not eliminate anymacy in Spanish;
it does not trigger agreement; it does not absorb nominative nor accusative case, since it is compatible
with passives, unergatives and unaccusatives, all of which have no source for ACC case.
On the nature of the Catalan prefix entre-
Susanna Padrosa Trias ([email protected])
Universitat Autònoma de Barcelona
1. Introduction. I first show that the Catalan prefix entre- ‘between’ is a non-head, i.e. a modifier,
when found in compounds like entreobrir (to-half-open) ‘to open halfway’. Secondly, I argue for a
unification of the three apparent different interpretations of the prefix, namely reciprocity (1a),
location (1b) and incompletion (1c) (cf. Gràcia et al. 2000, Déchaine & Tremblay 2006) in Lieber’s
(2004) framework of lexical semantic representation. Finally, I claim that entre- selects for some
notion of ‘plurality’.
(1) a. entrematar-se (between-kill-SE) ‘to kill each other’ b. entre/interposar (between-put) ‘to put (something) between (something else)’ c. entreobrir (between-open) ‘to open halfway/not completely’
2. Head vs. non-head status. I argue that entre- is a modifier of the head it attaches to, i.e. the prefix
semantically subordinates into the V (cf. e.g. Di Sciullo 1997). Some evidence for this view comes
from the unchanged argument structure of the prefixed V with respect to its base (2) (cf. Mateu 2001).
In addition, if entre- were the head, it would be difficult to explain existing prefixed and unprefixed
verbs with the same meaning (3). The nonhead status of entre- can explain the ‘optionality’ of this
element.
(2) a. L’Adrià va [obrir]V la porta. ‘Adrià opened the door’ b. L’Adrià va [entreobrir]V la porta. ‘Adrià opened the door halfway/not completely’
(3) a. Els dos enemics es maten. ‘The two enemies kill each other’ b. Els dos enemics s’entrematen. ‘The two enemies kill each other’
3. Lieber’s (2004) framework and interpretations of the prefix. According to Lieber, the lexical
semantic representation has two parts: the skeleton and the body. The skeleton decomposes the
representation into those aspects of meaning that are relevant for syntax (cf. Jackendoff’s (1990),
Levin & Rappaport Hovav’s LRH (2005) level of lexical conceptual structure). The body is the
encyclopaedic, holistic and idiosyncratic part of the representation (the ‘constant/root’ in LRH’s
terms). The skeleton consists of functions and arguments. Functions are decomposed into features
which are active across a number of categories. The two relevant features here are [Loc] (for
“Location”) and [IEPS] (for “Inferable Eventual Position or State”), which are used to distinguish
between locations and paths (e.g. English [+Loc] prepositions: in and [+IEPS] prepositions: to).
Prepositions have skeletons composed of the features [Loc] and [IEPS], and arguments. The
XVII Colloquium on Generative Grammar – Girona, 13th-15th June 2007
65
preposition between would be characterized as in (4) and so would the Catalan preposition entre
‘between’ and its bound counterpart, the prefix.
(4) between: [+Loc ([ ], [ ])]
I take the locational interpretation of the preposition as the basic reading of the prefix (1b), on which
the other readings are based (1a, 1c). My claim is that the different readings associated with entre-
prefixed Vs arise from a single affixal skeleton (4) interacting with the syntactic and semantic
properties of different bases. (5) shows the composed skeleton of the derived word (viz. the prefix plus
the base). Co-indexation of prefix arguments with base arguments explains the creation of a single
referential unit out of two distinct semantic skeletons.
(5) a. interposar (between-put): [+Loc ([ ]i, [ ]j [+dynamic, +IEPS ([ ]x, [ ]y, [ ]i/j)])] b. entrebesar-se (between-kiss-SE): [+Loc ([ ]i, [ ]j [+dynamic, ([ ]i/j, [ ]j/i)])] c. entreobrir (between-open): [+Loc ([ ]i, [ ]j [+dynamic, +IEPS ([ ], [ ]i/j)])]
The concept of location can be physical, as in (5a) (see translation in (1b)), and metaphorical, as in
(5b) and (5c). The reciprocal V entrebesar-se (5b) treats the two people performing the action of
kissing as locations; the kisses then go from one person/location to the other. As for incompletion
predicates (5c), I claim that the base denotes a scalar quality. That is, the predicate projects a scale
which may manifest different degrees of the gradable property at different times. Degrees can be seen
as a set of points/locations on the scale (e.g. OPENING) (cf. Hay et al. 1999). Concerning
incompletion predicates (5c) I make the prediction that only those predicates that can be viewed as
having scalar properties can be prefixed with entre- and have the incompletion reading. This
prediction is borne out by the data (*entremorir (between-die) vs. entreobrir (between-open)).
4. Selectional requirements of entre-. The prefix requires some notion of ‘plurality’, which is
satisfied in the reciprocal V (6a) and in the locational V (6b) by having plural marking, and in
incompletion Vs like (6c) by viewing the object placed somewhere in between a range of points on a
scale.
(6) a. La mare s’entrellaça les mans ‘The mother intertwines her (own) hands’ b. La noia es va interposar entre ells ‘The girl interposed herself between them’ c. La mare va entreobrir la porta ‘The mother opened the door halfway’
5. Conclusion. The Catalan prefix entre- is a modifier, not a head. Its three apparent different
interpretations of the prefix, namely reciprocity, location and incompletion, can be unified in Lieber’s
(2004) framework of lexical semantic representation. Its selectional requirement is that of plurality.
References
Déchaine, R.-M. and Tremblay, M. 2006. La préfixation en entre-: pluralité, réciprocité et valeur
aspectuelle. Paper presented at Diachro-III, Evolutions en français, Paris.
Di Sciullo, A. M. 1997. “Prefixed-verbs and Adjunct Identification”. In Di Sciullo, A. M. (ed.).
Projections and Interface Conditions. New York/Oxford: OUP. pp. 52-73.
Gràcia, L. et al. 2000. Configuración morfológica y estructura argumental: léxico y diccionario.
Zarautz (Guipúzcoa): Euskal Herriko Unibertsitateko Argitalpen Zerbitzua.
Hay, J. et al. 1999. “Scalar Structures Underlies Telicity in ‘Degree Achievements’”. In Mathews, T
and D. Strolovitch (eds.). SALT IX. Ithaca: CLC Publications. pp. 127-144.
Jackendoff, R. 1990. Semantic Structures. Cambridge, Mass.: MIT Press.
Levin, B. and Rappaport Hovav, M. 2005. Argument Realization. Cambridge/New York: CUP.
Lieber, R. 2004. Morphology and Lexical Semantics. Cambridge/New York: CUP.
Mateu, J. 2001. “Preverbs in Complex Denominal Verbs: Lexical Adjuncts or Core Predicates?”
Catalan Working Papers in Linguistics 9: 37-51.
XVII Colloquium on Generative Grammar – Girona, 13th-15th June 2007
67
The nature of Cognate Objects in Romance
Cristina Real-Puigdollers
Universitat Autònoma de Barcelona
Introduction. This work offers an analysis of the so-called C(ognate) O(bject) C(onstructions) (eg.
He sighed a weary sigh) in a theory of lexical syntactic decomposition based on the works of Mateu &
Rigau (2000) and Mateu (2002). In this paper I will explore the conditions upon which CO are
licensed. Specifically, I will compare Romance Languages with English. This study focuses on the
reasons why COCs are not attested in Romance in contrast with English [examples from Jones (1988)
(1) and Massam (1990) (1)b and (1)c]. Conversely, Romance prefers other strategies as the use of
cognate adjuncts (cf.(3)a and (3)b) or structures with a light verb and a nominal complement (cf. (3)c)
(1) COC IN ENGLISH a. Bill sighed a weary sigh b. Rosamond cried a good long cried c. Henleigh smiled a wicked smile
(2) COC IN ROMANCE a. *María sonrió una sonrisa malévola (Spanish) b. *Juan murió una muerte terrible
(3) OTHER STRATEGIES IN ROMANCE a. María sonrió con una sonrisa malévola (Spanish) b. Juan murió de una muerte terrible c. Juan tuvo una muerte terrible
The Puzzle. COCs have received a widespread attention in the lexical-syntax-semantics literature.
The discussion has been focused on the status of COs regarding the argument/adjunct dichotomy.
Accordingly, they have been analyzed as adjuncts (Jones 1988, Zubizarreta 1987, Moltmann 1989) or
as arguments (Massam 1990, Hale & Keyser 1997) or as both (Pereltsvaig 1998 or Nakajima 2006).
Crucially, Massam (1990) argues that two kind of constructions must be distinguished: (true) COC
and T(ransitivized) O(bject) C(onstruction), which display different properties.
(4) COC ADJUNCT-LIKE PROPERTIES a. *A silly smile was smiled [Passivization] b. *A silly smile, nobody smiled [Topicalization] c. *Maggie smiled a silly smiled and then her brother smiled [Pronominalization] d. *He smiled the smile for which he was famous [Definiteness Restriction] e. *What did he die? [Questioning] f. ?He died a death [Modifier obligatory] g. *He died a suicide/ a murder [Object necessarily cognate]
(5) TOC ARGUMENT-LIKE PROPERTIES a. The Irish jig was danced by Bernardette Dooley [Passivization] b. The Irish jig, nobody danced [Topicalization] c. I sang the aria then Tosca sang it [Pronominalization] d. Fred danced the slow number [Definiteness Restriction] e. What did he sing? [Questioning] f. She sang a song [Modifier obligatory] g. He sang an aria / a song [Object necessarily cognate]
Now, if we look at Romance languages the only cognate-like objects we may find fit into the TOC
[(3)].
(6) TOC IN ROMANCE a. El sueño Americano fue vivido con gran fervor [Passivization] b. María cantó una cancióni y luego lai cantamos nosotros [Pronominalization] c. Bailamos el baile que nos indicó la profesora [Definiteness Restriction] d. Juan vive la vida [Modifier obligatory] e. María baila el tango estupendamente [Object necessarily cognate]
In conclusion, despite argument and adjunct COs, we find another kind of construction, the so-called
TOC, where the object is in a hyponomous relation with the conflated verbal root. This construction
must be distinguished from the true argumental COC present in English and absent in Romance
Languages.
Proposal. I will show that real COCs do not exist in Romance and what seems COCs are in fact
TOCs. The differences in the spreading of such constructions in Romance and English are related to
the well-known elasticity of English verbs meaning, allowing unergatives to appear in transitive-like
constructions, such as resultatives. As shown in Mateu & Rigau (2000), the difference between verb-
frame languages (Romance) and satellite-frame languages (English) depicted by Talmy (1991, 2000)
is due to a process of lexical subordination of the manner into a transition relation. The contrast
between Romance languages and English regarding COs can be explained in similar terms. Thus, I
will propose that COC in English are derived from a process of lexical subordination of the manner
into a transitive structure similar to that of a light verb of creation and an object [(7)].
(7) COGNATE OBJECT CONSTRUCTION IN ENGLISH
V
V DP
V V a wicked smile
√SMILE V
In contrast, Romance Languages do not allow never lexical subordination processes. The cognate-like
constructions present in Romance are obtained by inserting the object in the root place after conflation
of the root into the verbal head, as proposed by Hale & Keyser (1997). This operation requires the
object to be in a hyponomous relation with the verbal root [(8)].
(8) TRANSITIVIZED ROMANCE CONSTRUCTION IN ROMANCE
V
V - -√BALLAR DP
una sardana
XVII Colloquium on Generative Grammar – Girona, 13th-15th June 2007
69
Conclusion. The impossibility of CO in Romance languages is due to their impossibility to have
manner incorporation, theoretically instantiated as the syntactic operation of lexical subordination (cf.
Mateu 2002). The few CO cases attested in Romance languages are obtained in two ways. By
adjunction, an option always available in most languages, or by inserting a DP, with an hyponomous
semantic relation with the verbal root, into the object position, leading a normal direct object, a TOC
in Massam’s term. Our proposal also predicts the correlation of resultative constructions and CO as it
may be proved to be the case.
Selected References
Hale, K & J. Keyser. 1997. “On the complex nature of simple predicators” Complex Predicates, 29-
65. Stanford, CA: CSLI Publications.
Jones, M. A. (1988). “Cognate objects and the case filter”. Journal of Linguistics 24, 89-110.
Massam, D. (1990). “Cognate objects as thematic objects”. Canadian Journal of Linguistics35(2),161-
190.
Mateu, J. (2002). Argument Structure. Relational Construal at the Syntax-Semantics Interface.
Unpublished Ph.D. dissertation, UAB, Bellaterra.
Middles: Transforming Events into States
Cilene Rodrigues ([email protected])
Unicamp/Brazil
I — The Implicit Agent of middles. With respect to their external argument, English middle
constructions behave semantically as passives, but syntactically as unaccusatives. Speakers have the
intuition that an agent is involved in the situation described by the predicate in passives and middles,
but not in unaccusatives (1). Nevertheless, the syntax of middles resemble that of unaccusatives in that
its alluded external argument has no syntactic effects, being unable to license subject-oriented adjuncts
and by-phrases (2). Thus, the leading question of the present presentation is:
(i) What yields the intuition that middles involve an Agent?
II— Previous analysis. Hoekstra and Roberts (1993) seek an explanation for (i), arguing that in the
underlying syntactic structure of English middles, the external argument position is occupied by a null
pronoun (pro). Stroik (1992) presents a similar proposal suggesting that the Agent is syntactically
represented by a null pronoun (PRO) which adjoins to VP. Keyser and Roeper (1984, 1992), on the
other hand, claim that similarly to Romance middles, middles in English has a syntactically null clitic
that absorbs the agent θ-role. However, all of these analyses wrongly predict that English behaves like
Romance, which, unlike English, allows subject-oriented adjuncts and by-phrases in middles (3).
III — New analysis. By comparing English with Romance, I suggest a syntax-semantics interface
analysis, according to which speakers know that middles involve an agent because cross-linguistically
the argument structure of middles is formed by a causative verbal head that conflates (in the sense of
Hale & Keyser 2002) into a stative root, as in (4). In English, the external argument position of the
causative vP is not syntactically saturated. As a result, the causative vP is semantically ill-formed, It
describes a causative event that is missing the causer/agent. However, the derivation converges due to
the conflation of the causative verb into the stative root. The conflation transforms an event with two
arguments into a state, with only one argument. The internal argument of the causative vP is then
promoted to the argument position of the stative VP. This results in the impossibility of independently
interpreting the causative vP at LF. Conversely, in Romance, since the middle marker SE is inserted
into spec of causative vP, the causative vP is semantically well-formed, even though it conflates with
the stative root. Thus, in Romance, but not in English, the causative vP is available for interpretation at
LF. That is, Romance middles can receive both a state and an event reading, whereas English middles
allow only a state reading.
XVII Colloquium on Generative Grammar – Girona, 13th-15th June 2007
71
IV - Evidence for the suggested argument structure.
(a) The causative P: As discussed in Hale & Keyser (1999) and Pesetsky (1995) psych verbs can be
divided into two classes: those whose experiencer is mapped into the object position (ObjExp - e.g
annoy, frightens, bore) and those whose experiencer is mapped into the subject position (SubjExp -
e.g. respect, envy, love). Only ObjExp verbs are formed via attachment of causative morpheme to the
verbal root (Cf. Pesetsky 1995), and, as predicted by our analysis above, only these can participate into
middle constructions (5)-(6).
(b) The stative root: As shown in (7)-(10), in English middles are incompatible with structures that S-
select events. Therefore, middles are stative constructions.
(c) The interpretation of the causative vP at LF: In Romance, but not, in English, middles are
compatible with ditransitive VPs. Under the assumptions that states take only one participant and that
all arguments are to be semantically integrated into the interpretation of the predicate containing them,
it follows that a ditransitive VP has to be interpreted as an event, not as a state. Thus, ditransitive VPs
can form middles in Romance, but not in English (11)-(12). Importantly, in Romance ditransitive
middles allows only an event interpretation.
Examples
(1) a. This wall paints easily → √ Implicit Agent b. This wall was painted → √ Implicit Agent c. The branch broke → * Implicit Agent
(2) a. This wall paints easily *by Johnldeliberately/*to protect it against the rain b. This wall was painted by John! deliberately/to protect against the rain c. This branch broke *by Johnl*deliberately/*to make a point
(3) a. Las manzanas se comen para adelgazar/voluntariamente the apples SE eat-3PL to lose weight/intentionally
(Spanish —Roberts 1987)
b. En général, ces débats s’enregistrent par Anne, qui est notre technicienne in general these debates SE register-3PL by Anne which is our technician
(Canadian French- Authier and Reed 1996)
c. Grego clássico se traduz pelo Junito, que é um especialista Greek classic SE translate-3SG by.the Junito who is a specialist
(Portuguës- Rodrigues 1998)
(4) The argument structure of middles in English.
VP → Stative event
Greek V’
[V [v translates+ v ] vP → Causative event
t VP
t tDP
(5) a. Politicians anger easily. (cf. The truth angers politicians) b. This colt frightens easily. (cf. Loud noises frighten this colt) c. Children bore easily. (cf. Adult talk bores children)
(6) a. *The truth respects easily. (cf. We respect the truth) b. *John’s talent envies easily. (cf. Everyone envies John’s talent) c. *French films love easily. (cf. My friends love French films)
(7) a. *John is knowing the answer. (Incompatibility with present tense) b. *The walls are painting easily.
(8) a. *What the box did was contain books. (Incompatibility with pseudoclefis) b. * What the bureaucrats do is bribe easily.
(9) a. *John exists every day at 6. (Incompatibility with iterative simple present) b. *Bureaucrats bribe every year at Christmas
(10) a. * I saw Mary knowing the answer. (Incompatibility with perception verbs) b. *1 saw the walls painting easily.
(11) *This car puts easily in the garage (English —Hale and Keyser 2002)
(12) Queste idée si insegnano/comunicano ai bambini con facilità these ideas SE teache/communicate to-the children with easy (Italian - Cinque 1988)
XVII Colloquium on Generative Grammar – Girona, 13th-15th June 2007
73
Syntax-before-Morphology in L2 Acquisition: Semantic Evidence from Aspect in L2 Portuguese
Jason Rothman & Michael Iverson
University of Iowa
The majority of available studies on the acquisition of [±perfective] aspect in adult second language
acquisition (L2A) have been undertaken from non-generative paradigmatic approaches to SLA such as
cognitive-perceptual, lexical/semantic and narrative perspectives (see Montrul & Salaberry 2003 for a
review). Moreover, in light of the questions they seek to investigate, most non-generative studies
(especially within L2 Romance) have primarily focused on the emergence and use of
Preterit/Imperfect morphology in performace (oral and written). However, assuming a syntax-before-
morphology position (e.g. Lardiere 1998, 2000; Prévost & White 2000; Schwartz 2003), which
anticipates possible performance-level differences between L2 and native [±perfective] aspect
morphological use (notwithstanding a competence that is fundamentally the same), it is not clear that
examing L2 morphological use alone achieves the intended goal of guaging underlying grammatical
competence. As a result, available generative studies have tested for L2 knowledge of associated
poverty-of-the-stimulus (POS) semantic entailments to determine underlying morphosyntactic
competence in this domain (e.g., Goodin-Mayeda & Rothman to appear; Montrul & Slabakova 2003;
Slabakova & Montrul 2003). This study continues this line of investigation, considering new data of
the L2 acquisition of [±perfective] aspect in L2 Portuguese.
Following Giorgi & Pianesi (1997) and others we assume that [±perfective] aspect —as seen in the
Preterit/Imperfect contrast in Romance languages— derives from the specification of relevant
morphosyntactic features associated with the functional category higher AspP, and therefore falls
under the phenomena accounted for by UG. Furthermore, stemming from the dichotomy of
[±perfective] aspect, sentences with the Preterit and Imperfect are subject to related POS semantic
entailments that restrict the range of interpretations. In light of this, we test for [±perfective] aspect
competence in advanced English learners of adult L2 Portuguese via their knowledge of a particular
POS semantic entailment, namely the associated [±accidental] interpretation restriction with adverbial
quantifiers.
As Lenci & Bertinetto (2000) have demonstrated for Italian, and Menéndez-Benito (2001, 2002) for
Spanish, both the Preterit and Imperfect contrast is not neutralized in adverbially quantified sentences
in Romance languages (as predicted by Bonomi’s (1997) theory) since these forms are not
interchangeable in context with, for example, expectative phrases such as in (1) and (2).1 That is, in
sentences with adverbial quantification like (1) and (2) the perfective necessarily denotes an accidental
generalization while the imperfective denotes non-accidental generalizations. Additionally, only
perfective sentences with adverbial quantifiers block the kind-referring reading of the subject DP,
1 In fact, she demonstrates that the Preterit and Imperfect are also not interchangeable with respect to generic adverbials (normalmente, a menudo) and durational phrases (durante cuatro semenas) as well.
which is otherwise available as a choice with definite DPs along with a group-denoting reading, as in
(3 and 4). Portuguese definite DPs can have either a kind-reading or a group-denoting reading. For
example, ‘As mulheres são inteligentes’, can be understood as women are intelligent in general (the
kind-reading) or a contextually determined group of individual women are intelligent. Whereas
imperfective adverbially quantified sentences, as in (3), retain both types of subject DP reading, only a
group-denoting reading is available for similar perfective sentences, as in (4). As a result, sentence (3)
can indicate the actions of a contextually determined group of specific students or students in general
whereas sentences like (4) can only refer to a particular group of students. Furthermore, only the
Imperfect can support the truth of counter factuals, therefore, (4) cannot support either (5) or (6).
However, the possible group-denoting interpretation of (3) supports the counterfactual in (5) whereas
the available alternative kind-reading supports the counterfactual in (6).
Employing a context-rich matching judgement task (scale -2 to 2), the empirical study tests 20
advanced English learners of L2 Portuguese as to their knowledge of the aboved discussed semantic
restrictions, as in (1-6). The results demonstrate L2 interpretations in line with both the native speaker
control (n=19) and the discussion above. It is argued that these data strongly support full access to UG
in adulthood (inclusive of access to crucial syntactic features not available from the L1, in this case the
[-perfective] feature). We discuss these results in terms of what they bring to bear on various L2A
theories.
Test sentences:
(1) Sempre que nós falamos da morte da Maria, eu comencei a chorar. always that we speak-1PPL-PAST-PFV about the death of Mary, pro start-1PSG-PAST-PFV to cry. ‘Whenever we spoke of Mary’s death, I ended up crying.’
(2) Sempre que nós falavamos da morte da Maria, eu començava a chorar. always that we speak-1PPL-PAST-IMP about the death of Mary, pro start-1PSG-PAST-IMP to cry. ‘Every time we spoke of Mary’s death, I would cry.’
(3) Sempre que os romanos precisavam mais, conquistavam novas terras. always that the Romans need-3PPL-PAST-IMP more, pro conquer-3PPL-PAST-IMP new lands. ‘Whenever the Romans would need something, they would conquer new lands.’
(4) Sempre que os romanos precisaram mais, conquistaram novas terras. always that the Romans need-3PPL-PAST-PFV more, pro conquer-3PPL-PAST-PFV new lands. ‘Whenever the Romans needed something, they wond up conquering new lands.’
(5) Se os romanos tivessem precisado mais terra durante aquele tempo, se teriam apoderado dela. ‘If the Romans had needed more land during that time, they would have seized it’
(6) Se você fosse romano e tivesse precisado mais terra, você se teria apoderado dela. ‘If you were Roman and you had needed more land, you would have seized it.’
XVII Colloquium on Generative Grammar – Girona, 13th-15th June 2007
75
References
Bonomi, A. (1997). Aspect, quantification and when-clauses in Italian. Linguistics and Philosophy,
20, 469-514.
Giorgi, A. & Pianesi, F. (1997). Tense and aspect: From semantics to morphosyntax. Oxford: Oxford
University Press.
Goodin-Mayeda, C.E. and Rothman, J. (to appear). The acquisition of aspect in L2 Portuguese and
Spanish: Exploring native/non-native performance differences. In Baauw, S., Dirjkoningen, F.
and Pinto, M. (eds.) Romance languages and linguistic theory. Amsterdam: John Benjamins.
Lardiere, D. (1998b). Dissociating syntax from morphology in a divergent L2 end-state grammar.
Second Language Research, 14, 359-375.
Lardiere, D. (2000). Mapping features to forms in second language acquisition. In Archibald, J. (Ed).
Second language acquisition and linguistic theory (pp.102-129). Maiden, MA: Blackwell.
Lenci, A. & Bertinetto, P. M. (2000). Aspect, adverbs and events: Habituality and perfectivity in J.
Higginbotham. In F. Pianesi & A.C. Varzi (Eds.) Speaking of events. Oxford/New York:
Oxford University Press, 65-287.
Menéndez-Benito, P. (2001). Aspect and adverbial quantification in Spanish, ms. University of
Massachusetts at Amherst.
Menéndez-Benito, P. (2002). Aspect and adverbial quantification in Spanish, in M. Hirotani (ed.)
Proceedings of the 32nd North Eastern Linguistics Society, GLSA, UMASS, Amherst, MA.
Montrul, S., & Slabakova, R. (2003). Competence similarities between natives and near-native
speakers: An investigation of the Preterit/Imperfect contrast in Spanish. Studies in Second
Language Acquisition, 25, 351-398.
Prévost, P., & White, L. (2000). Missing surface inflection or impairment in second language
acquisition? Second Language Research, 16, 103-1 33.
Schwartz, B. (2003). Child L2 acquisition: Paving the way. In Barbara Beachley et al. (Eds.), BUCLD
27 Proceedings. Somerville: MA: Cascadilla Press, 25-50.
Slabakova, R., & Montrul, S. (2003). Genericity and Aspect in L2 Acquisition. Language Acquisition,
11, 165-196.
Agreement and Case: Impersonal Sentences in Slavic and Accusative and Dative NPs
Luka Szucsich ([email protected])
Humboldt-University, Berlin
The paper discusses two issues in connection with the feature makeup and feature valuing of
functional categories:
(i) the status of impersonal sentences with accusative internal arguments, and
(ii) the status of obliquely marked NPs denoting (potential) causers of events expressed by the
impersonal verbs.
Accusative Impersonal Sentences (AIS) in Slavic languages pose a problem for Burzio-style analyses
of Case licensing having no overt external argument, cf. (1). Some authors (e.g. Lavine & Freidin
(2002), Harves (2006) take the verb in (1a) to lack agreement morphology indicating that T is a (φ-
incomplete head incapable of assigning nominative to a NP. On the other hand, v is considered to be
φ-complete (cf. Lavine & Freidin 2002) valuing the unvalued Case feature of the internal argument (=
accusative). Crucially, in the mentioned analyses, v does not select an external argument (following
Babby 1994).
Apart of a couple of minor shortcomings, the analysis employing T’s alleged defectiveness faces at
least three problems:
(i) This analysis cannot account for the control and binding facts with AIS (cf. below);
(ii) This stipulation leads to a proliferation of derivational dead ends (e.g., T being restricted to the
specification ‘defective’ with AIS due to v’s 4)-completeness);
(iii) T’s morphology is not taken seriously which, although not sharing φ -features with any of the NPs
in the sentence, exhibits markers otherwise capable of spelling out agreement.
Ad (i): AIS in Slavic (with the exception of -no/-to-constructions in Ukrainian) allow for control of the
PRO subject in gerund clauses, cf. (2), the latter requiring control by a prominent argument—in most
cases the matrix subject, but in any case not by an internal theme-argument (cf. Rappaport 1984 for
details concerning Russian). Besides, reflexive AIS with “impersonal subjects” interpreted as
[+animate] allow for binding of anaphoric elements, cf. (3a). Interestingly, there are personal
counterparts of reflexive AIS which do not allow for binding. Binding and control is also attested for -
no/-to-constructions in Polish (cf. Lavine 2005).
In this paper I argue that none of the categories of AIS is defective. In the light of the binding and
control data, I assume that either the category v in AIS selects for a semantically bleached nominal
expression (lacking φ-features) or v is equipped with features capable of establishing the respective
syntactic relations. This also makes v a licenser of unvalued structural Case features of the internal
argument, cf. (4) for the AGREE-relation and the valuation of [uCase]. This is why the case facts in
(3) go hand in hand with the binding facts. In the personal variant of the sentence (= 3b), in contrast to
XVII Colloquium on Generative Grammar – Girona, 13th-15th June 2007
77
(3a), the respective variant of the reflexive marker absorbs the external argument (strips off “external
features” of v) altogether (hence no accusative).
T of AIS has unvalued φ-features which (in the absence of matching φ-features of a goal) have to be
valued as default. The -no/-to-constructions in Polish and nominal impersonal predicates in Russian
provide evidence that the feature valuation of T’s φ-features is rather a consequence of a default
mechanism than due to agreement with a null D.
Polish and Slovenian allow for dative NPs denoting the (potential) causer of the event expressed by
the verb to occur in reflexive AIS, cf. (5). This led several authors to propose either that this NP
occupies the “subject position” or that it discloses an existentially closed null pronoun being
coindexed with the latter (cf. among others Rivero & Milojeviá Sheppard 2003). Analyses in this line
predict that the NPDAT should be capable of binding reflexive possessive pronouns in the same way as
its covert counterpart. (6) shows that neither the personal nor the impersonal variant allows for binding
by NPDAT (unlike quirky subjects in Icelandic). It follows that these NPs are rather introduced in the
complement domain of v (some lower applicative head or V itself) and that they cannot establish
relations to a position which would enable them to bind anaphoric elements. In this, these NPsDAT
sharply contrast with covert elements within the v-projection with AIS without dative NPs.
Examples
(1) a. Soldat-a rani-l-o pul-ej. soldier.M-SG-ACC wounded.PST.N.SG bullet.F.SG.INST ‘A soldier was wounded by a bullet.’ (Russian) b. T-ę ksiąŜk-ę {czyt-a/czyta-ł-o} się z przyjemnośc-ią. this bookF:SG:ACC read.PRES:3.SG/read.PRT.N.SG REFL with pleasure.F.SG.INST ‘One {reads/read} this book with pleasure.’ (Rivero 2001) (Polish)
(2) Mašin-u zanesl-o na povorot-e PRO razvernuv vopreki šosse. car.F:SG.ACC swerved.PST.SG.N on turn.M.SG.PRE PRO turn.GER against highway.N.SG.DAT ‘At the turn, the car swerved turning against the direction of traffic.’ (Russian)
(3) a. Svoj-e starš-e se ubog-a. REFL.POSS.ACC parents.M.PL.ACC REFL obey.PRES.3.SG ‘One obeys (has to obey) one’s parents.’ (Slovenian) b. * Svoj-i starš-i se ubog-ajo. REFL.POSS.NOM parents.M.PL.NOM REFL obey.PRES:3.PL
(4) a. [TP ... T [vP D v [VP V DPintern ]]] [uφ/NOM] [-φ] [uφ/sel-ft] [iφ/uCase] NON-AGREE AGREE
b. [TP ... T [vP D v [VP V DPintern ]]] [default φ] [-φ] [φ/] [iφ/ACC]
(5) Danes dopoldne se mi je jedlo jagode (Slovenian) today morning REFL me.DAT AUX3.SG eat.PRT.N.SG strawberries.F:PL.ACC ‘This morning, I felt like eating strawberries.’ (Rivero & Milojeviá Sheppard 2003)
(6) a. Janez-ui se jejo *svoj-ii / OKnjegov-ii/j cmok-i.
Janez.M.SG.DAT REFL eat.PRES:3.PL REFL.POSS.NOM / POSS.NOM dumplings.M:PL.NOM ‘Janez feels like eating his dumplings.’ (Slovenian) b. Janez-ui se jé *svoj-ei /
OKnjegov-ei/j cmok-e. Janez.M.SG.DAT REFL eat.PRES:3.SG REFL.POSS.ACC / POSS.ACC dumplings.M:PL.ACC ‘Janez feels like eating his dumplings.’
References
Babby, L. 1994. A theta-theoretic analysis of adversity impersonal sentences in Russian. In: Formal
Approaches to Slavic Linguistics 2, eds. S. Avrutin et al., Ann Arbor, 25-67.
Chomsky, N. 2000. Minimalist Inquiries: the Framework. In: Step by Step: In Honour of Howard
Lasnik, eds. R. Martin et al., Cambridge (MA), 89-155.
Chomsky, N. 2001. Derivation by Phase. In: Ken Hale: A flfe in language, ed. M. Kenstowicz
Cambridge (MA), 1-52.
Harves, St. 2006. Non-agreement, Unaccusativity, and the External Argument Constraint. In: Formal
Approaches to Slavic Linguistics 14, eds. J. Lavine et al., Ann Arbor, 172-188.
Lavine, James. 2005. The morphosyntax of Polish and Ukrainian -no/-to. Journal of Slavic Linguistics
13(1): 75-117.
Lavine, J. and R. Freidin. 2002. The Subject of Defective T(ense) in Slavic. In: Journal of Slavic
Linguistics 10: 253-289.
Rappaport, G. 1984. Grammatical Function and Syntactic Structure: The Adverbial Participle of
Russian. Columbus.
Rivero, M. L. 2001. On Impersonal Reflexives in Romance and Slavic and Semantic Variation. In:
Romance Syntax, Semantics and L2 Acquisition, eds. J. Camps & C. Wiltshire. Amsterdam,
169-195.
Rivero, M. L. & M. Milojevié Sheppard. 2003. Indefinite Reflexive Clitics in Slavic: Polish and
Slovenian. In: Natural Language andLinguistic Theory 21, 89-155.
XVII Colloquium on Generative Grammar – Girona, 13th-15th June 2007
79
Subjunctive Dependents
Juan Uriagereka ([email protected]) & Angel J. Gallego ([email protected])
University of Maryland & Universitat Autònoma de Barcelona
This paper proposes that subjunctive clauses in Romance (e.g., Juan desea que Maria venga - Juan
wishes that Mary come) be analyzed as ECMs (e.g., Juan wants Mary to come): in particular,
assuming that [1, 2, 3]’s “defectivity” is not restricted to φ-features alone and can thus be applied to
Case/Tense features, it is claimed that C-TSubj though φ-complete, is Case/Tense- defective. The
proposal plausibly explains why bona fide subjunctive dependents lack wh-movement (see [14, 16]),
have severely restricted peripheral fronting (see [13]), and yield long-distance obviation (see [4, 9])
without resorting to ill-understood truncation processes (see [10, 11]). Crucially, an account like this
can only be pursued in a system like [7, 8]’s, where Case and Agreement are differentiated: we assume
so (contra [1, 2, 3]), and argue that the embedded subject, while agreeing with the embedded verb,
receives Case from the matrix v*-TO complex.
The standard assumption about ECMs (and raising) has a truncation-like flavor: these
structures are taken to be introduced by a TP which, due to the lack of the CP layer, is defective (see
[1, 2, 5, 6, 12]), as (1) shows. Secondly, ECMs display the possibility for the embedded subject to
raise to the Case checking position of objects in the matrix clause, which explains both why the
himself (not him) is licensed in (2a) and why objects can bind into matrix adjuncts in (2b) (see [3, 5,
6]). Here we argue that the analysis in (1) should be as in (3), where we introduce a defective CP layer
to capture the subordination dependency that holds between matrix and embedded domains (mediated
through C, we assume) and the idea that T is present if C is (sensu [3]).
Let’s now turn to Romance subjunctives —which, as advanced above, we want to analyze just
like (3). As noted in the literature (see [9, 10, 12, 13]), Romance languages such as Spanish and
Catalan lack true ECM structures (putting aside an ECM-like analysis of perception verbs of the see
and hear sort): see (4). We want to qualify this by claiming that Romance subjunctives are the natural
counterparts of English ECMs, although for that to be feasible we need to make some assumptions, the
first one having to do with [1, 2, 3]’s “defectivity”. In this respect, we argue, following [8], that
defectivity can be understood not as ‘lack of (some) feature’ (typically, [number]), but as ‘lack of
value’, and that it is not a prerogative of φ-features alone —hence, it can apply to other features, like
Case/Tense (see [7, 8]). With these assumptions in mind, we propose that subjunctive clauses have a
C-T structure that is φ-complete but Case/Tense-defective, as roughly indicated in (5). We believe (5)
is able to explain various things: first, given that the embedded subject María shows full agreement
with the verb venga (Eng. come-SUBJ), the former is “frozen in place” and cannot raise (differently
put, A-freezing is related to total checking of φ-features), nor can it plausibly bind into matrix
adjuncts; second, since Case is dependent on the matrix v*-TO complex, it comes as no surprise that
long-distance obviation (see (6)) follows from the same reason local-obviation does: because subject
and object receive a distinct Case value (we therefore adopt [15]’s idea that formal distinctness —
Nominative vs. Accusative— entails interpretive distinctness). Our proposal makes the next
prediction: if the embedded subject cannot receive (abstract) structural Case from matrix v*-TO, then
correference with the matrix subject should in principle be possible: (7), which contains the quirky
subject a él (Eng. to him), shows that the prediction is borne out.
Finally, consider the well-known fact, noted by [15], that consecutio temporum effects only
arise with subjunctives. The asymmetry in (8) can be accommodated into our proposal if, just like
Case, Tense values (i.e., [present], [past], and [future]) are valued from the matrix clause.
Synthesizing, in the preceding lines we have argued that Romance subjunctives are the
counterpart of English ECMs. In order to do so, we have extended [1, 2, 3]’s “defectivity” to the
Case/Tense realms, a possibility that allows us to provide a rationale not only to long-distance
obviation, but also to the strong connectivity of subjunctives. A promising way of looking at this
connectivity ensues if subjunctives do not qualify as (strong) phases: that is not only empirically
tenable, but also conceptually sound, for it would bring more symmetry to the system, both v and C*
having ‘weak’ counterparts: v (passive/unaccusative) and C (subjunctive).
Examples
(1) [Ccomp [ John Tcomp believes [ Mary T-todef work hard ] ] ] ECM (standard analysis)
(2) a. [ Johni believed [ {*himi/himselfi} [ ti to be immoral ] ] ] b. [ The DA proved [ the defendantsi [ ti to be guilty during each otheri’s trials ] ] ]
(3) [Ccomp [ John Tcomp believes [ Cdef [ Mary T-todef work hard ] ] ] ECM (final version)
(4) a. *Vull la Maria tenir sort. (Catalan) Want.l.SG the Maria have luck ‘I want Maria to be lucky.’ b. *Creo a Maria ser afortunada. (Spanish) believe.L.SG to Maria be lucky ‘I believe Maria to be lucky.’
(5) [ C [ Yo T quiero [ v*-TO [ C-que Maria T venga] ] ] ] (Spanish) [1.SG] [1.SG] [1.SG] [3.SG] [3.SG] [3.SG] [3.SG] [Case:Nom] [Case:Nom] [Case:Nom] [Case:Acc] [Case: ] [Case: ] [Case:: ] I want.1SG that María come.SUBJ.3SG ‘I want for Mary to come.’
(6) [ C [ Juan T quiere [ v*-TO [ C-que pro{*j/k} T venga] ] ] ] (Spanish) [3.SG] [3.SG] [3.SG] [3.SG] [3.SG] [3.SG] [3.SG] [Case :Nom] [Case:Nom] [Case :Nom] [Case:Acc] [Case:Acc] [Case Acc] [Case Ace] Juan want.3SG that he come.SUBJ.3.SG ‘Juan wants him to come.’
(7) [ C [ Juan quiere-T [ C-que a él{k/j} le guste Charlie Mingus ] ] ] Juan want.3SG that to him CL-to.him like.SUBJ.3.SG Charlie Mingus ‘Juan wants for him to like Charlie Mingus.’ (Spanish)
(8) a. Platón dice [ que Aristóteles {lee/leía/leerá} a Sócrates] Ind. Plato say.3.SG that Aristotle read{PRES/PST/FUT}.IND.3.SG to Socrates ‘Plato says that Aristotle {reads/read/will read} Socrates.’ (Spanish)
XVII Colloquium on Generative Grammar – Girona, 13th-15th June 2007
81
b. Platón quiere [ que Aristóteles {lea/*leyera/*leyere} a Sócrates] Subj. Plato want.3sg that Aristotle read{PRES/PST/FUT}.SUBJ.3.SG to Socrates ‘Plato wants Aristotle to read Socrates.’ (Spanish)
References
[l] Chomsky, N. (2000): Minirnalist inquiries: the framework” in R. Martin, D. Michaels & J.
Uriagereka, Step by step. Essays on minimalist syntax in honor of Howard Lasnik, Cambridge
(Mass.): MIT Press, 89-155.
[2] Chomsky, N. (2001): Derivation by Phase”, in M. Kenstowicz (ed.), Ken Hale: A Life in
Language, Cambridge (Mass.): MIT Press, 1-52.
[3] Chomsky, N. (2005): On Phases”. Ms., MIT.
[4] Kempchinsky, P. (1987): Romance subjunctive Clauses and Logical Form, Ph.D. Dissertation,
University of California.
[5] Lasnik, H. & M. Saito (1999): “The Suiject of Infinitives”, in H. Lasnik, Minimalist Analysis,
Maiden (Mass.): Blackwell, 7-24.
[6] Lasnik, H. (2003): Levels of Representation and the elements of Anaphora”, in H. Lasnik.
Minimalist Investigations in Linguistic Theory, New York: Routledge, 42-54.
[7] Pesetsky, D. & E. Torrego (2004a): Tense. Case, and the Nature of Syntactic Categories”, in J.
Guéron & J. Lecarme (eds.). The Syntax of Time, Cambridge (Mass.): MIT Press, 495-537.
[8] Pesetsky, D. & E. Torrego (2004b): The Syntax of Valuation and the Interpretability of Features”,
Ms., MIT/UMass.
[9] Picallo, C. (1985): Opaque Domains, Ph.D. Dissertation, CUNY.
[10] Rizzi, L. (1994): Early null subjects and root null subjects”, in T. Hoekstra & B. Schwartz (eds.),
Language acquisition studies in generative grammar, Amsterdam: John Benjamins, 15 1-177.
[11] Rizzi, L. (1997): The Fine Structure of The Left Periphery”, in L. Haegeman (ed). Elements of
Grammar. Handbook in Generative Syntax, Dordrecht: Kluwer: 28 1-337.
[12] Torrego, E. (1998): The Dependencies of Objects, Cambridge (Mass.): MIT Press.
[13] Torrego, E. & J. Uriagereka (1992): Indicative Dependents”, Ms.. U.Mass. Boston/UMD.
[14] Uriagereka, J. (1995): An F Position in Western Romance”, in K. Kiss (ed). Discourse
configurational languages. Oxford (NY): OUP, 153- 175.
[15] Uriagereka, J. (1997): Formal and Substantive Elegance in the Minimalist Program (On the
Emergence of Some Linguistic Forms)”, in M. Bierwisch et al. (eds.), The role of economy
principles in linguistic theory, Berlin: Akademie Verlag, 170-204.
[l6] Uriagereka, J. (1999): Minimal Restrictions on Basque Movements”, NLLT 17:403-444.
The Syntax of Little Things
Vidal Valmala Elguea ([email protected])
University of the Basque Country (UPV-EHU)
In all natural languages, phrases of different types are often uttered which are (superficially) not
contained in full sentences. These appear both as short answers to questions (1B) and in discourse
initial position (2). In the generative tradition, two hypotheses have been put forward to deal with
these constructions. Some scholars (Hankamer (1971,1979), Morgan (1973), and Merchant (2004),
among others) argue that they are the result of applying deletion/ellipsis to parts of full sentences and
thus tend to call them fragments. For others (Yanofsky (1978), Barton (1990), Valmala (1999), and
Stainton (2005), among others), they are non-sentential maximal projections at all levels of syntactic
representation and thus tend to call them non-sentential constituents.
To my knowledge, all previous analyses of non-sententials have concentrated exclusively on
declarative utterances. In this talk I will argue in favour of a non-sentential approach on the basis,
among other things, of non-sentential questions (3) and imperatives (4) which have gone unnoticed so
far in the literature. This talk is structured as follows.
In sections 1 and 2 I introduce the phenomenon and discuss the basic properties of Hankamer
(1971,1979) and Morgan’s (1973) “traditional” deletion analysis and the main features of the non-
sentential approach.
In section 3 I review Merchant’s (2004) theory of non-sententials, probably the best articulated
and most elegant ellipsis approach to the phenomenon, which solves the non-constituent-deletion
problem of “traditional” ellipsis analyses on the basis of the idea that non-sententials are the result of
two operations: a movement operation by which the non-sentential moves to the specifier of a left-
peripheral head F plus ellipsis of the complement of the head F (5b).
In section 4 I provide arguments against Merchant’s proposal on the basis of different anti-
connectivity effects of a) sequences which are not possible as non-sententials but can undergo left-
peripheral movement in full sentences (6B,b), and b) sequences which cannot be moved in full
sentences but can unexpectedly appear as non-sententials (7B’,c) and (8B’,b) in declarative,
interrogative and imperative contexts. I will also take the existence of island-insensitive non-
sententials (9) and the impossibility of VP-ellipsis fed by non-sententials (10) as evidence against a
movement plus ellipsis analysis.
In section 5, I provide additional evidence supporting my claim that even non-sentential
answers to questions cannot be the result of a movement plus deletion strategy on the basis of the
existence of individuals who, as a result of an impaired theory of language or mind, do not have
movement to the left periphery and ellipsis as part of their linguistic competence but can nevertheless
produce and understand non-sentential constituents.
XVII Colloquium on Generative Grammar – Girona, 13th-15th June 2007
83
Section 6 discusses the consequences of assuming a non-ellipsis approach to non-sententials
for our conception of different aspects of the grammar of natural languages.
Examples
(1) a. A: What are you reading? B: A mystery novel. (DP) b. A: What did you do yesterday? B: Clean the carpet. (VP)
(2) Two pints of lager and a packet of crisps. (A customer in a pub)
(3) a. A: He’s gone out for lunch. B: Where? b. A: That can’t be true.
B: Why not? c. A: There must have been a car. B: Any tyre tracks?
(4) Home! Straight away!
(5) a. A: What are you reading? B: A mystery novel. b. [FP [DP a mystery novel]i [F’ F [TP I am reading ti ]]]
(6) a. A: They told me that John got sacked. What did he do? B: *Often break the rules. b. She said that John would often break the rules, and [often break the rules]i he did ti.
(7) a. A: What have you done? B: Fix the car. B’: Fixed the car. b. John said that he would fix the car, and fix the car he has.
c. *John said that he would fix the car, and fixed the car he has.
(8) a. A: The rope is from a factory in Denmark. B: Denmark?
b. *DENMARK, the rope is from a factory in?
(9) a. A: Ann left because she didn’t want to meet who!? (echo question) B: Bill.
b. *Bill Ann left because she didn’t want to meet.
(10) a. A: Which novel have you been asked to read? B: Disgrace. A’: *Will you?
b. A: Which novel have you been asked to read? B: I have been asked to read Disgrace. A’: Will you?
References
Barton. E. (1990) Nonsentential Constituents. Amsterdam: John Benjamins.
Hankamer, J. (1971) Constraints on Deletion in Syntax. Ph.D. Dissertation, Yale University.
Hankamer, J. (1979) Deletion in Coordinate Structures, New York: Garland.
Merchant, J., (2004), ‘Fragments and Ellipsis’, Linguistics and Philosophy 27: 661-738.
Morgan, J.L. (1973) ‘Sentence Fragments and the Notion “Sentence”’, in B. Kachru et al (eds.), Issues
in Linguistics: Papers in Honor of Henry and Renée Kahane, pp. 719-715, University of
Illinois Press, Urbana.
Stainton, R. (2005) In Defense of Non-Sentential Assertion, in Z. Szabo (ed.), Semantics versus
Pragmatics, Oxford University Press.
Valmala, V. (1999), VP Fragments and the Pro-drop Parameter, in S.J. Billings et al. (eds.) CLS 35.1:
Papers from the Main Session, pp. 323-337.
Yanofsky, N. (1978) NP Utterances, in Donka Farkas et al (eds.) CLS 14: 491-502, Chicago Linguistic
Society, Chicago.
XVII Colloquium on Generative Grammar – Girona, 13th-15th June 2007
85
Consonant Clusters and phonotactic constraints. A generative perspective
Bakhta Abdelhay & Ouahmiche Ghania
University of Mostaganem
The present paper uses the transformational generative apparatus and notational conventions and
applies them to the data of Mostaganem Spoken Arabic. Since the structure of consonant clusters
exhibits by and large regular patterns, they were chosen as the main object of analysis.
Some observations concerning various aspects of the structure of consonant clusters are made ,and
rules are suggested to account for the language processes that are most productive. In the course of this
study, a number of issues are raised and discussed in accordance with the data on which it is based.
For instance, it is demonstrated that there are some phonotactic constraints prevent some consonant
clusters to occur in the language.
The short vowel in this system is examined.It is shown that due to a very productive process of vowel
centralisation, the short vowels often surface as a shwa in the context of __ C (C) C except when these
vowels occur in what is reffered to her as protective environments
Movement and agreement in Italian past participles and spell-out domains
Roberta D’Alessandro ([email protected]) & Ian Roberts ([email protected])
University of Cambridge
In Standard Italian past-participle (pp-) agreement is found with promoted internal arguments
(passives, unaccusatives, medio-passive si), reflexive constructions, and with preposed direct object
clitics (see (1)). Kayne (1989) proposed that the agreement was triggered by moving the internal
argument through SpecAgrOP. This triggers Spec-head agreement between the DP and AgrO (see(2)).
Kayne accounts for three things:
(i) only internal arguments trigger pp agreement;
(ii) agreement always and only takes place in a Spec-head relation;
(iii) object agreement is connected to movement.
However, more recent versions of minimalism have effectively abandoned Spec-head agreement
(Chomsky (2004:109)). We propose an analysis of pp agreement using Agree combined with a PIC-
related restriction on spell-out domains.
The obvious account of pp-agreement involves the idea that v probes φ-features of the internal
argument. This approach, however, has the problem of being unable to distinguish moved direct
objects, which trigger pp-agreement (e.g. (3a)), from unmoved ones, which do not (see (3b)). We
propose a different, phase-based account of the relation between Italian pp agreement and the
argument structure of the verb. We assume the structure in (4), involving an iterated vP, for
periphrastic tenses in Romance. The external argument is merged in SpecvPrtP. vPrt θ-marks the
external argument and Agrees with the direct object in φ-features, thereby valuing its Case feature. As
such, vPrt heads a strong phase.
We propose that the reason vPrt does not overtly show agreement with an unmoved direct
object (see (3b,c)) can be found in a phase-based approach to derivations, along with assumptions
regarding the mapping to PF. We assume, following Cinque (1999), that, at the point of the Spell Out,
the participle occupies vPrt. Since vPrt heads a strong phase, its complement VP is sent to PF on a
distinct cycle. Let us now suppose the following condition on the morphophonological realisation of
agreement:
A. Given an Agree relation A between Probe P and Goal G, morpho-phonological agreement
between P and G is realised iff P and G are contained in the complement of the minimal phase-
head H.
B. XP is the complement of a minimal phase head H iff there is no distinct phase head H’
contained in XP whose complement YP contains P and G.
XVII Colloquium on Generative Grammar – Girona, 13th-15th June 2007
87
(A) and (B) state that morphophonological agreement takes place within the complement to a
phase head, i.e. the substructure which is transferred to PF as a single unit.
Consider now the unaccusative in (1a). Here we assume the same structure for the periphrastic
tense as in (4). The difference with (3b,c), however, is that vPrt is not the head of a strong phase: there
is no external argument and vPrt is unable to Case-license the object DP. Therefore, given (A), even if
the participle raises to vPrt, the participle and the object are contained in the complement of the same
minimal phase head (in this case the TP dominating the higher vP, the complement of C). Because of
this, overt morphophonological agreement between the participle and the direct object is required, as a
reflex of the Agree relation. Passive examples like (1b) and structures containing mediopassive si like
(1c) are essentially the same as unaccusatives. We follow Kayne (1988) in taking sentences with
reflexive si also to involve a “medio-passive” like structure, thereby accounting for (1d). In both of
these cases, presumably owing to the presence of si, vPrt cannot be a phase head. Finally, in (1e) and
(3a), the relation between clitic-movement and agreement is very direct in our terms: the clitic moves
to the higher v and is thus in the complement of the same phase head, C, as the raised participle at
Spell Out. Hence, by (A), overt morphophonological agreement is predicted.
Examples
(1) a. Le ragazze sono arrivate. the girls-F.PL. are arrived-F.PL. “The girls have arrived.”
b. Le ragazze sono state arrestate. the girls-F.PL. are been-F.PL. arrested-F.PL. “The girls have been arrested.”
c. Si sono viste le ragazze. SI are seen-F.PL. the girls-F.PL. “We have seen the girls / the girls have been seen.”
d. Le ragazze si sono guardate allo specchio. the girls selves are seen-F.PL. in-the mirror “The girls have seen themselves in the mirror.”
e. Le abbiamo salutate. them-F. we-have greeted-F.PL. “We have greeted them.”
(2) DP … [AgrOP (DP) AgrO [VP V (DP) ] ] …
(3) a. Le ho salutate. them-F. I-have greeted-F.PL. “I have greeted them.”
b. *Ho salutate le ragazze. c. Ho salutato le ragazze.
(4) vP
vAux vPrtP
DP vPrt
vPrt VP
V DP
XVII Colloquium on Generative Grammar – Girona, 13th-15th June 2007
89
Reflexive se and Agreement in Brazilian Portuguese
Dorothy Bezerra Silva de Brito ([email protected])
Federal University of Alagoas/PPGLL
The aim of this poster is to present a discussion on some syntactic contexts of Brazilian Portuguese
(henceforth BP) in which the reflexive clitic SE is the dependent part, as we can see in the following
sentences: (a) Ele se lava ‘He washes’ and (b) O João e a Maria se amam ‘João and Maria love each
other’. These predicates show different syntactic patterns: (a) is a reflexive predicate, (b) is a
reciprocal one and this makes the two occurrences of the clitic SE functionally distinct. For instance,
in (a) we can have the clitic doubling Ele se lava a si mesmo/ele mesmo ‘He SE washes himself’, in
which the co-indexing relation is established between the clitic SE and the object si mesmo and the co-
reference relation is stated with the subject Ele in its local domain. As for (b), the symmetrical verb
amar allows a commutation pattern O João ama a Maria ‘João loves Maria’ and A Maria ama o João
‘Maria loves João’, in which there is no co-indexing with the object position, once the clitic SE, in this
context, is an affix detransitivising the verb. Since in some BP dialects we can have sentences like
(c)Eu se lavo ‘I SE wash’ and (b)Nós se ama ‘We love each other’, in which there is no matching
between the grammatical features of the clitic and its antecedent, I argue if the functionality of the
clitic SE in (c) and (d) can be described as in (a) and (b). Several syntactic contexts show up this lack
of morphologically overt agreement between the clitic and its antecedent, not only those ones in which
the clitic SE is reflexive. Given the data presented above, I suggest an analysis in which this
phenomenon is related to the generalized weakening of the morphologically overt agreement in BP,
according to other researches on Agr in BP and its influence on the verbal and the pronominal
paradigms in this language. To achieve this aim, I intend to go through native speakers’ judgments of
grammaticality, with the objective to detect similarities and/or differences with relation to the
manifestation of agreement into these syntactic contexts.
Main References
BURZIO, L (1986). Italian Syntax. Dordrecht, Holland: D. Reidel Publishing Company.
BURZIO, L (1992). The Role of the Antecedent in Anaphoric Relations. Ms, Jonh Hopkins University.
GALVES, C (2001). Ensaios sobre as Gramáticas do Português. Campinas, SP: UNICAMP.
GRIMSHAW, J. B (1990). Argument Structure. Linguistic Inquiry, n. 18.
MARELJ, M (2004). Middles and Argument Structure across Languages. Utrecht, Netherlands: LOT.
REINHART, T. & SILONI, T (1999). Against the Unaccusative Analysis of Reflexives (to appear in
Studies on Unaccusativity: the Syntax-Lexicon Interface, CUP).
Agreement weakening in Brazilian Portuguese: the complementizer se
Danniel Carvalho ([email protected])
Federal University of Alagoas/PPGLL
Some researches developed about Brazilian Portuguese (BP) lead to a gradual loss of the
morphological agreement verb mark in this language (cf. DUARTE, 1995; COELHO, 2000;
GALVES, 2001; CAVALCANTE, 2001; TAVARES SILVA, 2001, 2004; among others). Several
syntactic contexts show up this loss, for instance, overt pronominal subject in finite sentences.
However, other contexts, considered by some as "intact" for their structural dependence, show this
morphological loss, as we can see in finite embedded clauses. In some dialects in BP, this type of
clause started losing its agreement mark which was believed as something necessarily required doing a
differentiation between these structures and the infinitive ones. On the other hand, in the same dialect,
one can find sentences in which the embedded verb does not show any agreement mark even with the
presence of the complementizer se, behaving like an infinitive sentence. But, if one takes the DP out
from the embedded sentences, they will have an obligatory control interpretation, being, therefore,
different from the former. Thus, from the considerations above, this paper analyses the correlation
between the loss of the morphological agreement mark in embedded finite verbs (subjunctive
sentences) to the complementizer se in some dialects in BP. This analysis is based on Carvalho (2005;
2006a,b) who proposes a detachment of abstract Case and agreement mark from feature interpretation
in LF in such language. Thus, I suggest that this loss is due to the failure of interpretation of T in C in
embedded finite clauses in those dialects, being the overt DP responsible for establishing the
agreement relations.
Examples
(1) a. (Inflected Infinitive) É agradável para os ricos os pobres fazerem o trabalho árduo. It is pleasant for the rich the poor to do(Infl.3pes.pl.) the hard work "It is pleasant for the rich for the poor to do the hard work" b. (Inflected Infinitive with covert morphological mark) É agradável para os ricos os pobres fazer o trabalho árduo It is pleasant for the rich the poor to do(InflØ) the hard work "It is pleasant for the rich for the poor to do the hard work" c. (Subjunctive with QUE)
É agradável para os ricos que os pobres façam o trabalho árduo. It is pleasant for the rich that the poor do(3pes.pl.) the hard work "It is pleasant for the rich for the poor to do the hard work"
d. (Subjunctive with SE) É agradável para os ricos se os pobres fizessem o trabalho árduo.
It is pleasant for the rich if the poor do(3pes.pl.) the hard work "It will be pleasant for the rich if the poor does the hard work"
XVII Colloquium on Generative Grammar – Girona, 13th-15th June 2007
91
e. (Subjunctive with SE with covert morphological mark) É agradável para os ricos se os pobres fazer o trabalho árduo It is pleasant for the rich if the poor do(InflØ) the hard work "It will be pleasant for the rich if the poor does the hard work"
(2) a. É agradável para os ricos fazerem o trabalho árduo. It is pleasant for the rich PRO to do(Infl.3pes.pl.) the hard work
b. É agradável para os ricos fazer o trabalho árduo It is pleasant for the rich PRO to do(Infl.3pes.pl.) the hard work c. É agradável para os ricos que façam o trabalho árduo.
It is pleasant for the rich that PRO to do(Infl.3pes.pl.) the hard work d. É agradável para os ricos se fizessem o trabalho árduo.
It is pleasant for the rich if PRO to do(Infl.3pes.pl.) the hard work e. É agradável para os ricos se fazer o trabalho árduo It is pleasant for the rich if PRO to do(Infl.3pes.pl.) the hard work
Main References
CARVALHO, Danniel (2005). PARA vs. QUE em orações encaixadas no PB. Ms.
CARVALHO, Danniel (2006). Caso: morfologia vs. sintaxe em PB. Ms.
CHOMSKY, Noam (2001). Derivation by phase. In: Ken Hale: A life in language.
Michael Kenstowicz (ed.). Cambridge, Mass.: MIT Press.
HORNSTEIN, N., J. NUNES and G. KLEANTHES (2006). Understanding Minimalism. Cambridge:
Cambridge University Press.
MARKMAN, Vita G.(2005). Syntax of Case and Agreement: its Relationship to Morphology and
Argument Structure. PhD dissertation. The State University of New Jersey.
RAPOSO, E. (1987). Case theory and Infl-to-Comp: the inflected infinitive in European Portuguese.
Linguistic Inquiry, 18(1), p.85-109.
SIGURDSSON, Halldór Ármann (2003). Case: abstract vs. morphological. In: New Perspectives in
Case Theory. Ellen Brandner and Heike Zinsmeister (eds.). CSLI Publications.
Subject-verb agreement in Brazilian and European Portuguese: the case of the partitive
constructions and complex DPs
Mirian Santos de Cerqueira ([email protected])
New University of Lisbon/ Federal University of Alagoas
The main aim of this talk is to analyse the establishment or not of the overt morphologically
agreement between the DP subcategorized by the preposition (P) which is head of an PP included into
the subject DP and the verbal inflection in declarative structures with partitive expressions (see
examples in (1) ). Besides, one intends to do a distinction between this type of agreement and another
that is established in constructions with complex DPs (see examples in (2)). As it is known, in the
literature on the verbal agreement, the Brazilian Portuguese Grammar (henceforth BP) and European
Portuguese Grammar (henceforth EP) can license two types of number feature marking
(singular/plural) for partitive constructions (Peres & Móia, 1995). This marking takes the inflection to
agree either with the head of the subject or with the DP subcategorized by the P which is head of an
included PP into the subject DP. In this paper, concerning to complex DPs, one defends the following
hypothesis: there exist morphological and/or semantic constraints for the establishment of the
agreement in BP, but not in EP because this type of agreement would be ungrammatical. Therefore,
this talk aims to compare, initially, finite declarative sentences in BP and EP which are results of
native speakers’ judgment of grammaticality, with the objective of detecting similarities and/or
differences with relation to the manifestation of the verbal agreement into the grammars of the two
languages. To develop this research, the analysis bases itself on Minimalist Program framework
(Chomsky, 1999) taking into account more specifically the operation Agree. In this analysis, one puts
in question the notions of Probe and Goal and discusses the validity or not of a research that only
takes a strictly syntactic approach. Furthermore, one starts comparing the results of Rodrigues (2006),
concerning to the processing of the agreement in the two types of structures mentioned in BP, with the
results concerning to EP, to verify similarities, differences and implications for a more general study
of agreement about the grammars of the two languages in question.
Examples
Overt morphologically agreement with partitive expressions: (1) a. A maioria dos alunos fizeram o trabalho.
the most of students do-PAST.3PL. the homework. ‘The most of students did the homework’.
b. Uma parte dos professores compareceram à reunião. a part of the teachers attend-PAST.3PL. the meeting ‘A part of the teachers attended the meeting’.
XVII Colloquium on Generative Grammar – Girona, 13th-15th June 2007
93
Overt morphologically agreement with complex DPs: (2) a. A vida das pessoas são importantes.
the life of people be-PRES.3PL. important-PL. ‘The life of people is important’.
b. A troca de conhecimentos acontecem em grupo. the exchange of knowledges happen-PRES.3PL. in group. ‘The exchange of knowledges happens in group’.
References
CHOMSKY, N. Derivation by phase. MIT Working Papers in Linguistics, n. 18, Cambridge, Mass.:
The MIT Press, 1999.
PERES, J. A.; MÓIA, T. Áreas críticas da Língua Portuguesa. 2. ed. Lisboa: Editorial Caminho,
1995.
PEREIRA, S. M. B. Gramática comparada de “a gente”: variação no português europeu. Dissertação
(Mestrado em Gramática Comparada) – Universidade de Lisboa. Lisboa, 2003.
RODRIGUES, E. S. O processamento da concordância de número entre sujeito e verbo na produção
de sentenças. (Tese de Doutoramento). Pontifícia Universidade Católica do Rio de Janeiro. Rio
de Janeiro, 2006.