 |
 |
 |
 |
 |
 |
Web Resources - past talks
Here you will find information about GSAS past talks in theoretical linguistics
at the Department of Linguistics at Harvard University. The talks were
usually on Wednesdays and Fridays.
If you have any questions concerning the talks, please contact us at the address below:
Adam Szczegielniak
304 Boylston Hall
Department of Linguistics
Harvard University
Cambridge, MA 02138
USA
Tel: (617) 495-4006
Fax: (617) 496-4447
E-mail: szczeg@fas.harvard.edu
Adolfo Ausín
ada96002@uconnvm.uconn.edu
Department of Linguistics, University of Connecticut
Entitled:
Multiple Spell-Out and the locality conditions on A-movement
Day: Wednesday 3 May
Time: 3:30 p.m.
Place: Graduate Lounge, Dept. of Linguistics, Boylston
Hall.
Click here for a longer PDF abstract
ABSTRACT
Uriagereka (1999) and Chomsky (1998, 1999) have argued that the operation of Spell-out can take place more than once as the derivation proceeds. In this talk, I will argue that if we make certain assumptions about A-movement and the operation of Multiple Spell-Out (MSO), a straightforward account of the locality conditions on A-movement is available.
The Department of Linguistics invites you to a GSAS Talk by:
Claire Bowern & Gulsat Aygen-Tosun
bowern@fas.harvard.edu; tosun@fas.harvard.edu
Department of Linguistics, Harvard University
Entitled:
Titan's Tensed Prepositions
Day: Tuesday 25 April
Time: 3:30 p.m.
Place: Graduate Lounge, Dept. of Linguistics, Boylston
Hall.
Click here for a longer PDF abstract
ABSTRACT
Abstract In the Sivisa dialect of Titan (Admiralty Subgroup, Oceanic) there are a number of prepositions that agree in tense/mood with the main verb of the clause. The marker of tense on prepositions is homophonous with the third person subject agreement clitics on matrix verbs. The tense marked prepositions are also homophonous with lexical verbs. Synchronically, however, the deverbal prepositions cannot be analysed as verbs. The facts of Titan's prepositions are challenging for recent work in Minimalism, where Tense feature is claimed to exist on functional heads and subject DPs. These prepositions seem to bear a tense feature that must agree with the tense feature of the subject agreement clitics (hosted by the verb). We argue that the tense feature on this set of prepositions is deleted under the process of PROBE&GOAL agreement. Titan's tense agreement also shows that the feature [tense] is not limited to functional heads like T and C. This implies that the VP is a domain in which tense features can be checked, removing the motivation for any constituent within the VP to move out of it.
Olga Fernandez Soriano
Universita Autonoma de Madrid
Entitled:
Datives in construction with unaccusative "se" in Spanish.
Day: Friday 28 April
Time: 3:30 p.m.
Place: Graduate Lounge, Dept. of Linguistics, Boylston
Hall.
The Department of Linguistics invites you to a GSAS Talk by:
Gulsat Aygen-Tosun
tosun@fas.harvard.edu
Department of Linguistics, Harvard University
Entitled:
How Counterfactuals are marked in Turkish: T-to-C
Day: Friday 21 April
Time: 3:30 p.m.
Place: Graduate Lounge, Dept. of Linguistics, Boylston
Hall.
In this paper, grammatical ingredients of counterfactuals (Cf) in Turkish are investigated as a means to illustrate two claims: (i) it has been argued that Iatridou’s (1998/99) generalizations based on Indo-European languages hold for Turkish, a member of the Altaic/Turkic family. Past tense morphology marks counter-factuality when it ranges over worlds. This interpretation is overtly marked by a shift in the order of morphemes in Turkish, whereas it is overt only in Inverted Conditionals in English. I will argue that T-to-C is what marks counterfactuals in both languages. The Imperfective appears as a "fake aspect" in non-counterfactuals and various environments in which it appears provides evidence for its "possible" default nature in Turkish, too.
(ii) In Turkish, this "shift" in the order of morphemes is in fact part of a larger phenomenon , that is, the multifunctional property of various morphemes. I will argue that the syntactic position where a feature a morpheme bears is checked and interpreted determines its function and meaning. The multifunctional nature of other morphemes such as the question particle can also be accounted in the same line of analysis as noted in Tosun (1996, 1997,1998, 2000b).
References
Bonomi, A. 1997. Aspect,
quantification and when clauses in Italian. Linguistics and Philosophy
20.
Kulwer: Netherlands. pp. 469-514.
Iatridou, S. 1998/99. The Grammatical Ingredients of Counterfactuality. ms. MIT.
Kornfilt, J. 1996.
Kural, Murat. 1993. V-TO(-I-TO)-C in Turkish. UCLA Occasional Papers in Linguistics, v.11.ed.Filippo Beghelli &Murat Kural.
Kuruoglu, 1984. Time Reference in Turkish Conditional Sentences.Turk Dilbilim Konferansi Bildirileri.
Ayhan A. Koc. Eser Taylan (eds). Istanbul: BU Yay. pp.129-43.
Ozturk, Balkiz. 1999.
Turkish as a Non-Pro-Drop Language.
Unpublished MA thesis. Bogazici
University.
Pesetsky, David and Esther Torrego. 1999. Tense to C:Causes and Consequences, ms MIT.
Taylan, E.E. 1997. Turkcede Gorunus,Zaman ve Kiplik liskisi./Tense, Aspect and Modality Relations in Turkish. Proceedings of the 11 National Conference on Turkish Linguistics. Ankara: ODTU.
Taylan, E.E. 1996. The parameter of Aspect in Turkish .Modern Studies in Turkish Linguistics.Proceedings of the VI. International Conference on Turkish Linguistics. Eskisehir:Anadolu Universitesi. pp.153-68.
Tosun, G. (in progress) T-to-C and Scrambling in Turkish. ms. Harvard University.
Tosun G. 2000a. Relative Clauses in Turkish, ms Harvard University.
Tosun, G. 2000b. The Question particle, PL readings and Echo Questions in Turkish. (to appear in) HWPL 2000.
Tosun, G. 1999a. Embedded Clauses in Turkish: Extraction Possiblities. Presented at Ford Foundation Fall Syntax Workshop at Harvard University.
Tosun, G. 1999b. Specificity and Subject /Object Scope Interactions in Turkish. (to appear in) Proceedings of the I International Conference in Turkic Linguistics. Manchester University, April 1999.
Tosun, G. 1999c. Counterfactuals:a case of multifunctional morphemes in Turkish and its syntactic and semantic implications, ms Harvard University.
Tosun G. 1999d. Question Particle in Turkish. ms. Harvard University.
Tosun, G. 1999e. Fusion Functional Heads in Turkish. Ms. Harvard University.
Tosun G. 1998a. Bir Gonderme Birimi olarak Zaman /Tense as a Referential Expression. Proceedings Of the XIIth National Conference in Turkish linguistics, Mersin Universitesi yay.
Tosun. G. 1998b. The Split INFL Hypothesis in Turkish. BA thesis. Bogazici University. Istanbul.
Tosun. G. 1998c. The Adverb Positions and Functional Heads in Turkish. To appear in Proceedings Of the VI International Conference in Turkish Linguistics, Oxford University Aug 1998.
Tosun, G. 1996. A syntactic and Semantic Analysis of Turkish Conditionals. unpublished BA thesis. Bogazici University. Istanbul.
Tosun, G. 1997. Turkce’de Kosul Tumceleri./Conditionals in Turkish. Proceedings of the 11 National Conference on Turkish ,Linguistics. Ankara: ODTU.
Yavas, F. 1982. Future Reference in Turkish. Linguistics 20. Moutune. pp. 411-429.
The Department of Linguistics invites you to a GSAS Talk by:
Kylie Skewes
skewes@fas.harvard.edu
Slavic Department, Harvard University
Entitled:
Small Clauses in Russian, Ukrainian and Icelandic: A Brainstorming Session
Day: Wednesday 19 April
Time: 3:30 p.m.
Place: Graduate Lounge, Dept. of Linguistics, Boylston
Hall.
ABSTRACT
Small Clauses in Russian and Ukrainian are often encoded with the instrumental Case on a predicative adjective. We see this instrumental Casemarking in complement Small Clauses of the type 'I consider him stupid', 'BE' constructions, and in so-called depictive (adjunct) Small Clauses of the type 'I ate the meat raw.' Deviations from this encoding shed light on the interaction between Case checking operations, Theta-theory, and the behavior of Individual- and Stage-level predicates in these languages. Rapoport (1991) claims that only Stage-level predicates (adjectives denoting more temporary properties of individuals) occur in depictive Small Clauses. He also claims that the matrix verb in these constructions must also be a S-level predicate, e.g.,
(1) Roni cut the bread wet.
(2) *Roni cut the bread white.
Russian and Ukrainian depictive Small Clauses, however, may have both Stage-level and Individual-level (adjectives denoting more permanent properties of individuals) properties; these properties depend on whether the predicative adjective agrees with its NP-controller in Case or not, e.g., Russian (Nominative subject-control):
(3) Stepan (NOM), mesjac provaljavshis' v bol'nice, vernulsja zdorovyj (NOM) /zdorovym (INSTR)
Steven, having disappeared for a month in hospital, returned healthy.
(example from Timberlake 1986, tested on my native informants).
The predicative adjective with nominative Case in this example has an I-level interpretation, while the adjective with the instrumental Case has a S-level interpretation.
Russian (accusative object-control):
(4) Ja(NOM) voz'mu ego(ACC) zhivogo (ACC).
I'll take him alive.
(5) (Vzjat' ego(ACC)) zhivym(INSTR) ili m'ortvym(INSTR).
(Take him) alive or dead.
(Russian equivalent of 'Wanted Dead or Alive')
Ukrainian (accusative object-control):
(6) Borys znajshov Sashu(ACC) holoho(ACC)/ holym(INSTR)
Boris found Sasha naked.
We see the same contrasts in meaning between agreement and the instrumental Case in these examples with object control--agreement imparts a I-level interpretation, the instrumental Case a S-level interpretation. These oppositions disappear in constructions with long-distance control (ie., the instrumental is obligatory once there is a non-finite verb or an overt complementizer), and in constructions in which the predicative adjective is controlled by an NP with lexical Case.
Unlike Russian and Ukrainian, Icelandic predicative adjectives agree in case with their controlling NP in both simple clauses and in constructions with long-distance control. Interestingly, in non-finite constructions the predicative adjective agrees with the Case that an overt NP-controller would have received in a finite construction (note: in the following example the verb "to drift" takes an accusative argument in Icelandic):
(7) Hana(ACC) rak a land eina(ACC).
She drifted ashore alone.
(8) Hun vonast til [PRO ad reka a land eina ]
She(NOM) hopes to drift ashore alone(ACC).
(data from Andrews 1990)
To muddy the waters even more, Russian (and Ukrainian) have a construction that resembles the Icelandic long-distance control constructions--the so-called "second datives" (Comrie 1974) 'odin' and 'sam' 'alone'. Unlike predicative adjectives, 'odin' and 'sam' always agree in case with their NP-controller in simple clauses. In certain clauses with long-distance control, however, the Small Clause either agrees in case with its controller, or occurs in the dative case (the Case of subjects of non-finite verbs in Russian):
(9) Ja(NOM) poprosyla Ivana(ACC) [PRO prijti odnogo(ACC) / odnomu(DAT)]
I asked Ivan to come alone.
For some speakers these two constructions have a different semanticinterpretation, for others only one or the other is possible. These constructions differ in both structure and interpretation from all other predicative adjective structures in Russian, and resemble the Icelandic state of affairs more closely.
The Russian, Ukrainian and Icelandic data suggest that Theta-theory, Case Theory and the semantic interpretation of Small Clauses are closely connected. The exact nature of this connection will be open for a brainstorming session.
The Department of Linguistics invites you to a GSAS Talk by:
Javier Martín-González
jmartin@fas.harvard.edu
Department of Linguistics Harvard University
Entitled:
Sentential ‘no’
Day: Wednesday 12 April
Time: 3:30 p.m.
Place: Graduate Lounge, Dept. of Linguistics, Boylston
Hall.
Click here for
a PDF abstract.
ABSTRACT
It is generally claimed that the overt sentential negator ‘no’ in Spanish negative sentences is incompatible with preverbal n-phrases (Laka 1990), its occurrence being obligatory otherwise (Bosque 1980, 1994, Laka 1990, Zanuttini 1991):
(1) a. *(No) vino nadie.
not came nobody
‘Nobody came.’
b. Nadie (*no) vino.
nobody not came
‘Nobody came.’
However, sentential ‘no’ seems to be optional when the preverbal n-phrase is a Topic in a Clitic-Left Dislocated construction:
(2) A ninguna de estas personas, (no) las vi en la
fiesta.
PREP none of these people not them saw-I in the party
‘I didn’t see any of these people at the party.’
(Lit. ‘None of these people, I didn’t see them at the party.’)
Furthermore, there are two cases in which sentential ‘no’ is obligatory in spite of the fact that the sentence contains a preverbal n-phrase. One such case takes place when the Topic n-phrase appears to have been displaced from an ambedded wh-clause:
(3) A ninguna de estas personas, dime por qué
*(no) las invitaste.
PREP none of these people tell-me why not them invited-you
‘Tell me why you didn’t invite any of these people.’
(Lit. ‘None of the se people, tell me why you didn’t invite them.’)
The second case is instanced by sentences where the Topic n-phrase occurs between a non-wh complementizer and its doubled counterpart (i.e. ‘que’ + n-phrase +’que’):
(4) Me dijeron que, a ninguno de ellos, que *(no)
los invitaste.
me told-they that PREP none of them that not them
invited-you
‘They told me that you didn’t invite any of them.’
(Lit. ‘They told me that, none of them, that you didn’t invite them.’)
Except for those cases like (3) involving embedded wh-clauses, analysed in Suñer 1992, no proposals have been made for cases like (2) and (4).
In the present proposal, Move and Merge (Chomsky 1995) are adopted as the two distinct derivational processes that account for the preverbal position of n-phrases. Secondly, it is argued that Topic negative phrases, unlike non-Topic negative phrases- contain a clitic (overt or covert) as head and the negative phrase as Specifier (along the lines of Uriagereka 1995). It is also argued that Topic n-phrases may occur preverbally due to application of either Move or Merge, whereas non-Topic n-phrases -including non-arguments (Escobar Alvarez 1995)- can only appear in preverbal position due to application of Move.
In this paper it is also claimed that NegP in Spanish contains an uninterpretable feature that must be checked off before Spell-Out. Checking may be achieved by having overt ‘no’ as head of NegP, or by having an n-phrase move to Spec,NegP. This correctly predicts that sentences without preverbal n-phrases (e.g. 1a) will require an overt sentential ‘no’. Furthermore, considering that preverbal non-Topics must have been moved to that position, it also makes the correct predictions about the obligatory absence of sentential ‘no’ in sentences with preverbal non-Topic n-phrases (e.g. 1b). Likewise, considering that Topics may occur preverbally due to Move or Merge, the optionality of ‘no’ in examples containing preverbal Topic n-phrases in examples like (2) is successfully predicted as well.
Finally, an attempt is made to account for the obligatory presence of ‘no’ in cases like those exemplified in (3) and (4). First of all, Rizzi’s (1997) enriched left-periphery structure is adopted, together with his claim that fronted wh-phrases sit in Spec,FocP. Then, it is shown that the doubled form of the non-wh complementizer occupies Foc0 and it is claimed that an assertive Operator fills the Spec,FocP position in those cases. Furthermore, drawing from similarities between negative and interrogative sentences (Klima 1964, Haegeman 1995, Rizzi 1990), it is argued that the Focus head can also check the negative feature. Thus, any movement past the checking domain of Focus0 will constitute a violation of "shortest move" (especially under Fergusson’s (1996) Shortest Move Requirement) and will result in a nonconvergent derivation. Therefore, the only possibility for a n-phrase to occur in some position to the left of FocP is by application of Merge, which makes the presence of ‘no’ obligatory since the uninterpretable negative feature of NegP must be checked. This makes the correct predictions about the obligatoriness of sentential ‘no’ in cases of preverbal n-phrases to the left of wh-complementizers (ex.3), and between non-wh complementizers and their doubled form (ex.4).
Thus, by making use of claims and postulates that are independently motivated, non-language and non-construction specific, the proposed analysis offers an elegant account for all the "new" cases (ex. 2-4), plus the familiar instances of (non-)occurrence of sentential ‘no’ in Spanish negative sentences (ex.1).
References
Bosque, Ignacio. 1980. Sobre la negación. Madrid: Cátedra.
Bosque, Ignacio. 1994. ‘La negación y el PCV’, in V. Demonte (ed.) Gramática del. español. Publicaciones de la Nueva Revista de Filología Hispánica. El Colegio de México
Chomsky, Noam. 1995. The Minimalist Program. Cambridge, Mass.: MIT Press.
Escobar Alvarez, M. de los Angeles. 1995. Lefthand Satellites in Spanish. Utrecht: OTS.
Fergusson, K. Scott. 1996. ‘Shortest Move and Object Case Checking’, in W. Abraham et alii. (eds.) Minimal Ideas. Syntactic Studies in the Minimalist Framework. Philadelphia: John Benjamins Publishing Company.
Haegeman, Liliane. 1995. The Syntax of Negation. Cambridge: C.U.P.
Klima, Edward. 1964. ‘Negation in English’, in J. Fodor
and J. Katz (eds.) The Structure
of Language. New Jersey: Prentice-Hall.
Laka, Itziar. 1990. ‘Negation in Syntax: on the Nature
of Functional Categories and
Projections’. PhD dissertation. MIT.
Rizzi, Luigi. 1990. Relativized Minimality. Cambridge, Mass.: MIT Press.
Rizzi, Luigi. 1997. ‘The Fine Structure of the Left Periphery’, in L. Haegeman (ed.) Elements of Grammar. Handbook of Generative Syntax. Boston: Kluwer Academic Publishers.
Suñer, Margarita. 1993. ‘Negative Elements, Island Effects, and Resumptive No’. ms.Cornell University.
Uriagereka, Juan. 1995. ‘Aspects of the Syntax of Clitic
Placement in Western
Romance’. Linguistic Inquiry 26,1.79. 79-123.
Zanuttini, Raffaella. 1991. ‘Syntactic Properties of Sentential Negation.
A Comparative Study of Romance Languages’. PhD dissertation. University
of Pennsylvania.
jmartin@fas.harvard.edu
Juan
Uriagereka
juan@wam.umd.edu
Department
of Linguistics, University of Maryland
Entitled:
Reprojection: An Argument for Derivations
Day: Friday 24 March
Time: 3:30 p.m.
Place: Fong Auditorium, Boylston
Hall (same building as the Department).
A reception will follow in the Ticknor Lounge, Boylston Hall.
Abstract:
(download here a PDF version)
It is often hard to decide whether a given proposal must be expressed in derivational or representational terms, or rather either expression is possible. Arguably the best kind of argument for a derivational system is one involving representations that are destroyed as the derivation unfolds, and do not make it to the final representation in any direct form. This 'loss of information' situation is hard to obtain in models -such as the classical Principles and Parameters system- which make use of highly enriched representational codings (including traces, indices, and similar elements). Intuitively, whatever information may have been lost in the course of the derivation can be reconstructed through some abstract coding. Interestingly, the Minimalist Program drastically reduces this notational artifact, for intriguing methodological and even ontological reasons relating to grammatical 'good design'. The present study can be seen in this light, as a minimalist exercise on whether one particular case study that we present is an instance of a 'loss of information' situation. We argue that it is, and rather than being of negative consequence, the lost information has good semantic and syntactic properties. So far as we can see our results can be replicated in representational terms only at the cost of much added, otherwise unnecessary notational machinery. The basic idea concerns the labeling mechanism in (1): (1) X <=> Y When X merges to Y, obtaining the construct {X, Y}, what is the label of the expression? We must emphasize that by 'label' we simply understand the type of the expression in question, assuming this is important. Following Chomsky (1995) we may code that label as in: (2) { L {X, Y}} Where L is the label. This is the only substantive issue: is L identical to X or to Y? (Chomsky (1995) argues that among the simplest, set-theoretic relations between X and Y that one could consider to determine L's properties, identity is the only one that does not create formal problems.) The options are thus: (3) a. X (Formally: {X, {X, Y}} b. Y (Formally: {Y, {X, Y}}) / \ / \ X Y X Y Since Chomsky (1998) it is assumed that what determines the choice in (3) are the properties of X and Y. For instance, if X is a verb and Y is a DP, then X will project. If in contrast X is a DP and Y is a v', then v will project. Collins (1999) has suggested that, if appropriately extended, this actually deduces the labeling operation proper. Suppose this is true in some form; labels are in no sense primitive objects, although of course they may still be real to the system (just as, say, molecules are not primitive, yet they are real at some level). The question that intrigues us is this: could it be that during the course of the derivation we actually change from an object like (3a) to an object like (3b)? Call this a reprojection, which can be characterized as in (4): (4) If X projects upon taking Y as a dependent (i.e., whenever X is a head), and in the course of the derivation Y projects (thus 'turning' X to a maximal projection and making X a head) we say that Y has been 'reprojected'. If this is possible, and (3b) is the only object that reaches the interpretive component that LF bleeds into, then we would have a nice argument for a derivational system, since the only way a representational system could reproduce the fact that (3b) was 'once' (3a) is in terms of some notational trick that enriches the system in an undesirable way. Chomsky (1995) presupposes two reasons that prevent a class of reprojections involving heads and their specifier (the only type we will be concerned with here). Assuming specifiers are obtained by movement, then reprojected specifiers will involve either improper or non-uniform chains: (5) a. Y b. XPi / \ / \ XPi Y' ====REPROJECTION====> X'i YP / \ / \ Y ... ti ... Y ... ti ... Clearly, the chain (XPi, ti) in (5a) is tampered with after the reprojection in (5b). If the intended chain is still (XPi, ti), then it will be improper (the first link doesn't c-command the last); if the chain in question is now (X'i, ti), then it will be non-uniform (the first link is not a maximal projection, while the last one is). At the same time, a second reason why the reprojection in (5) is undesirable is that it modifies checking configurations: whereas XP in (5a) is in the checking domain of Y, in (5b) YP is in the checking domain of X. Those concerns are representational. It could be that the checking domain of Y (vis-a-vis XP) is used only at some derivational time Dt, while at some later derivational time Dt', instead, the checking domain of X (vis-a-vis YP) is what is relevant. Our work argues that if this is indeed the case, some peculiar syntactic and semantic properties of binary quantifiers can be naturally explained, thus arguing for reprojections and with them for a derivational system. The gist of the idea comes from an observation in Larson and Segal (1995) that binary quantifiers like 'most' behave much like transitive verbs in that they take ordered arguments (thematic arguments in the verbal case, and a restriction and a scope in the case of a quantifer). If this is so, and provided that verbs 'take' their arguments in familiar syntactic ways (within verbal projections), it is reasonable to ask whether quantifiers too 'take' their arguments in familiar syntactic ways. The problem is easily seen in a sentence like (6): (6) Most people love children. Whereas the argument 'people' is in a standard syntactic relation with regards to the quantifier 'most' (as its complement), the argument 'love children' is in no serious sense a dependent of this quantifier; quite the opposite is the case: syntactically, 'most children' is the dependent (specifier) of the relevant IP, within customary assumptions. But now suppose reprojections are possible. Then nothing prevents the quantifier 'most' from handling its argumental requirements after the Case and agreement requirements of IP have been met. At that stage in the derivation, (7b) is what we would have obtained from (7a): (7) a. IP b. QP / \ / \ QPi I' ====REPROJECTION===> Q'i IP / \ / \ / \ / \ most ... I ... ti ... most ... I ... ti ... Aside from directly addressing a serious question about the syntax/semantics of binary quantification, the present proposal has a consequence that is really worth exploring. To put it basically: what was a 'right branch' in (7a) suddenly becomes a 'left branch' in (7b). This may relate to a very complex set of facts involving so-called 'quantifier-induced' islands, known at least since Linebarger (1980) (see Den Diken and Szabolcsi (1999) for a recent summary of the literature). Basically put, whenever we try to establish an LF relation of grammar between positions X and Y in (8) across a binary QP, an ungrammaticality ensues, whereas the same type of relation of grammar is possible if it takes place in overt syntax: (9) ... [X ... [QP ... [Y...]...]...] ... Z One can always code this fact as some kind of requirement on the relevant relations, but needless to say the issue is why such a requirement should hold. From the point of view of reprojections the expectation is that the process in question should in effect produce an island which was not there to begin with. It is hard to see how this could be replicated in representational terms. The actual mechanism that we propose to induce the quantifier induced island has to do with the fact that the chain of QP in (7) must be uniform, and as we saw chains whose first link reproject are no longer uniform. In a system making use of levels of representation this is a reason to either reject uniformity or reprojection. However, in a radically derivational system of the sort argued for by Epstein and Seely (1999) or Uriagereka (1999), all this means is that chains must be 'cashed out' right before they become non-uniform. Suppose we do that, then, prior to reprojection. How does this affect the resulting structure? It does rather drastically, assuming a chain is 'cashed out' by the system in an optimal fashion: by sending to the interpretive component the minimal amount of structure that contains all links of that chain. In the instances that are to reproject, that means that very structure. For instance, IP in (7a). The issue then is what happens to the constituent parts of IP. Inasmuch as IP is essentially gone from computation, those constituents should be inaccessible for further computation from higher positions. For instance, if IP is Z in (9), then a relation of grammar between X and Y would become impossible; the same type of relation would have been fine prior to the chain of QP being 'cashed out' prior to reprojection. As we implied, only binary quantifiers induce these LF islands. Nothing in the logic of our proposal forces a unary quantifier to reproject, hence its chain is never at risk of not being seen by the system as uniform, thus is in no need of an early 'cashing out'. If so no islands should emerge for them, as is the case. The paper discusses other instances of possible reprojections, including those involving negation, and others that plausibly prevent them (because 'segments' of an adjunction are involved in the phrase-marker that undergoes reprojection). The latter provides an interesting account of seemingly unrelated facts: why binary quantifiers never incorporate and why there is a 'definiteness effect' in expletive-associate pairs (why associates cannot be binary quantifiers). Reprojection is not an operation, just a property of derivations strictly considered as having different demands (theta and Case/agreement relations, quantificational relations) as the derivation unfolds. Although labels are real in reprojection instances (something that changes exists), they are crucially not primitive. It remains to be seen whether the analysis is replicable in representational terms, but prima facie this is very hard to imagine.
Ju-Eun Lee
jueunlee@fas.harvard.edu
Department of Linguistics Harvard University
Entitled:
A note on Verb Movement, Scrambling, and Scope Interaction in Korean Negation
Day: Wednesday 22 March
Time: 3:30 p.m.
Place: Graduate Lounge, Dept. of Linguistics, Boylston
Hall.
Abstract:
A Note on Verb Movement, Scrambling, and Scope Interaction in Korean Negation There is scope rigidity in quantifier-quantifier scope interaction in Korean. When clause-internal scrambling is involved, however, the rigidity effect is nullified and we get ambiguity. The same fact holds in Korean short form negation such that we get scopal ambiguity when one quantifier is scrambled over the other. The interesting phenomenon is we do not get this scopal ambiguity in long form negation. In this talk, I will show that this scope phenomenon is well explained in terms of Miyagawa (1999)'s theory on scrambling, which argues that A-scrambling is triggered by EPP feature on T. It will be argued that scrambling to a clause initial position in long form negation (where verb movement to T is blocked) is an instance of A'-scrambling rather than an A-scrambling. If we assume that A-chains do not reconstruct but A'-chains do for scope, then we can explain the contrast between the two forms of negation construction in Korean. I will show a couple of more evidence arguing for this analysis. This analysis leads us to ask two questions: First, why is obligatory reconstruction required for scope in A'-chain? Second, scope and binding seem to have opposite requirement such that scope requires reconstruction and condition C facts seem to require no reconstruction for A'-chains. As for these problems, relation between case checking and interpretation of nominals might give us a clue.
Ken Nakatani
knakatan@fas.harvard.edu
Department of Linguistics, Harvard
Entitled:
The processing complexity of nested structures in Japanese
Day: Wednesday 8 March
Time: 3:30 p.m.
Place: Boylston Hall,
3rd floor Grad Student lounge.
Albert Costa,
acosta@wjh.harvard.edu
Department of Psychology, Harvard
Entitled:
The cognate facilitation effect: Implications for models
of lexical access.
Download a background paper:
PDF format
ps format (zipped)
Day: Friday 25 February
Time: 4:00 p.m.
Place: Boylston Hall,
3rd floor Grad Student lounge.
Abstract:
In two picture naming experiments we tested whether non-selected lexical nodes activate their phonological information. Catalan-Spanish bilinguals were asked to name two sets of pictures: a) pictures whose names are cognates in the two languages (words that are phonologically similar in the two languages), and b) pictures whose names are non-cognates in the two languages. We argue that if non-selected lexical nodes are phonologically encoded, naming latencies should be shorter for cognate words. Furthermore, since the cognate status of words is only meaningful for bilingual speakers, this difference should disappear when testing monolingual speakers. The results of Experiment 1 fully support these predictions. In Experiment 2, bilingual speakers were asked to name the same types of pictures in their dominant and in their non-dominant language. The difference between cognate and non-cognate words was larger when naming in the non-dominant language than when naming in the dominant language. The results of the two experiments are interpreted as providing support to cascaded-activation models of lexical access.
LAST UPDATE: 05/15/00