NYU Linguistics

Michael Becker (Reed College)

"The role of markedness in generalizing over lexical exceptions"

Friday, October 10 at 4pm
Silver Center Room 414

Joint work with Lena Fainleib (Tel Aviv University)

Lexical exceptions have been repeatedly shown to influence speakers’ treatment of novel items (Bybee & Moder 1983 and many others since), which was taken to mean that lexical exceptions have an impact on the grammar. With the advent of Optimality Theory (Prince & Smolensky 1993/2004), lexical exceptions have been modeled using markedness constraints (Zuraw 2000, Albright & Hayes 2003, Hayes & Londe 2006, Pater 2006, Becker 2008, among others). These OT accounts share the prediction that speakers will generalize over the output properties of lexical exceptions, since by definition, markedness constraints only assess output forms. I claim that this prediction is correct, and show that Hebrew speakers prefer output-based generalizations to input-based generalizations even in the absence of evidence for it in the source language.

In Hebrew, masculine nouns regularly take the suffix [-im]. Of the irregular nouns that take [-ot], most have [o] in their stem, i.e. the affix agrees with the stem vowel. Since the stem [o] stays unchanged, speakers can’t tell whether [-ot] agrees with the singular stem [o] or the plural stem [o].

In an artificial input-output mapping experiment, 60 Hebrew speakers were assigned to learn one of two artificial languages. In both languages, singulars were the same plausibly native novel nouns. In the plural stems, [o] was switched with [i] (1) and [i] with [o] (2) — changes that are absent from real Hebrew. In the “surface” language, the plural suffixes [-im] and [-ot] were selected to agree with the vowel in the plural stem, whereas in the “deep” language, the plural suffixes were selected to agree with the vowel in the singular stem.

Singular_____“Surface” language plural_____“Deep” language plural
(1) apóz____________apiz-ím___________________apiz-ót
_____agóf_____________agif-ím___________________agif-ót
(2) amíg_____________amog-ót_________________amog-ím
_____axís______________axos-ót__________________axos-ím

When asked to generate plurals for novel items, the “surface” language participants were significantly more successful in applying the required vowel changes and affix selection, demonstrating a universal bias toward output-based generalizations.

I show that the results follow nicely from a model in which speakers assign an Optimality Theoretic grammar to the artificial languages, using the same constraints they use in real Hebrew. Models that make input-based generalizations by learning from input-output mappings fail to capture the results. Interestingly, models that rely on raw phonotactics of the language, without deriving these phonotactics from universal principles, fail as well.

Olga Kagan

"On the Semantics of Genitive Objects"

Friday, October 10 at 1:30pm
Department Conference Room

Abstract:

In this talk, I will investigate the semantics of Intensional Genitive Case in Russian, a phenomenon whereby certain intensional verbs may take genitive objects, as well as accusative ones, as is illustrated in (1):

(1) On ždal čuda / Dimu.
he waited miracle.GEN Dima.ACC
He was waiting for a miracle / for Dima.

Two puzzles arise in connection with the distribution of Intensional Genitive: first, it is licensed only by strong intensional verbs in Farkas' (1985) terminology (e.g. ždat' (wait for), zasluživat' (deserve), trebovat' (demand)) and not by weak intensional verbs (e.g. predstavljat' (sebe) (imagine), predvidet' (foresee), izobražat' (draw, depict)). The second question concerns the choice of Case for objects of strong intensional verbs. As has been noted in the literature (Neidle 1988, Bailyn 2004, Kagan 2005), the choice of Case is dependent on a number of semantic properties. For instance, the genitive tends to be assigned to non-specific, indefinite, narrow scope NPs. At the same time, none of these properties is sufficient to account for the alternation.

I will propose a semantico-pragmatic analysis of Intensional Genitive that is based on the notion of existential commitment, i.e. existential entailment and/or presupposition. It will be argued that Intensional Genitive can only be assigned in the absence of commitment to existence in the reference world, as well as in the set of worlds that is introduced by the intensional predicate (and forms the embedded context set). I will also discuss the relation between existential commitment and individuation, a property that has been argued to affect Case-assignment to objects in a wide range of languages (e.g. Hopper and Thompson 1980, Grimm 2005).

Jennifer Bloomquist

"'Girrrl, my car needs warshed!' Regional dialect accommodation by African American English speakers in the Lower Susquehanna Valley"

Friday, October 17 at 1pm
Department Conference Room

Abstract:

When members of minority groups migrate to majority communities, their rates of assimilation (or non-assimilation) to the majority culture are dependent on several factors including, but not limited to, the following: 1) the degree of physical, social, and economic isolation experienced by the migrants in their new location, 2) the history of the migration and ways in which the newcomers are received by the members of the established community, 3) the construction of a new community identity, and 4) the strength of the connection the migrants maintain to their former community.

This study examines the socio-historical acquisition and non-acquisition of the regional dialect by African Americans who are at least second generation residents (i.e., natives) of Pennsylvania's Lower Susquehanna Valley (the area that includes Harrisburg, York, and Lancaster) and investigates the reasons contributing to differences found among these speakers in terms of the degree to which they have acquired the local variety. The linguistic factors that are considered are region-specific elements of lexicon, syntax, and phonology; social and historical factors involve the migrant African Americans' relationships to the European American community including physical location (rural vs. urban, integrated vs. segregated), socio-economic status, rates and types of contact among speakers, and the connections maintained by the relocated members to their home communities.

Findings suggest that while rates of dialect accommodation are somewhat location specific (rural vs. urban), they are also influenced by the ways in which members of each community identify (or resist identifying) both locally and with larger, nearby African American communities. The results also challenge long held assumptions regarding the supra-regionality of African American English and the ways in which regional varieties have influenced the development of
AAE.

Lisa Matthewson (University of British Columbia)

"Moods vs. Modals in St'át'imcets and Beyond" Abstract

Friday, October 24 at 4pm
Silver Center Room 414

Gabriela Alboiu (York University)

"On Silent Categories and Case"

Friday, October 24 at 1:30pm
Syntax/Semantics Lab

Abstract:

By focusing on (c)overt subjects in non-finite domains I provide an account of structural Case reliant on the phase domain (Chomsky 2008) rather than the presence of phi-features (i.e., uφ). I argue that uφ is not needed for either Case checking or for obtaining a Case value. Nonetheless, I show that the presence of uφ (and specifically a Person feature, π) on the Probing domain will ensure a NOM value, while its absence will constrain the DP to an ACC value. I propose that a null expletive checking the EPP feature will trigger NOM subjects whenever (C)-T lacks uφ, specifically, in non-finite domains, such as infinitives and gerunds. Crucially, lexical subjects in non-finite domains are not assigned default Case; rather ACC or NOM are systematic occurrences, dependent on the presence or absence of non-referential pro. In effect, an A-relationship guarantees interaction with specific properties/features of some Probe. This feature complex in turn ensures a certain morphological instantiation of the vocabulary item inserted post-Spell-Out. Under this analysis, the overwhelming cross-linguistic evidence of Case-marked PRO is also readily explained. PRO either has: (i) Case checked against a Probe, so assigned a value (e.g. NOM in Icelandic, ACC in English), or (ii) Case checked at Spell-Out, so default (e.g. NOM in German, Czech), but never 'null Case'. It also explains why default Case will not guarantee lexical subjects (e.g. German has NOM as default but no lexical subjects in non-finite clauses), or why lexical subjects in non-finite domains can have Case values distinct from the default (e.g., see Latin). Crucially, the proposed analysis has the benefit of correctly predicting the various empirical generalizations with respect to DP Case valuation, as well as providing support for divorcing structural Case values from specific functional heads. NOM or ACC are not syntactic primitives or a priori properties of (C)-T and v, respectively, but forms assigned post-Spell-Out based on specific instructions from the computational system. In this sense, then, "syntax has no case features" (Sigurðsson 2007), but A-relations in syntax enable specific Case values.

Ivona Kucerova (UCL)

"A-Scrambling as Grammatical Marking of Prepositions"

Wednesday, October 29 at 5:30pm
726 Broadway, Room 703

Abstract:

Scrambling is usually understood as a feature-triggered operation that is avaialable only under rather strict pragmatic conditions and only in some languages. Even though a lot of attention has been paid to syntactic properties of scrambling, it still remains mysterious why scrambling takes place at all and why it takes place only in some language. Furthermore, there are significant differences between scrambling languages both with respect to type of movement labeled as scrambling and the set of syntactic elements that undergo this movement.

In this talk I will offer an answer to these questions in terms of scrambling as a parasitic syntactic operation which purpose is to Maximize Presuppositons (Heim 1991). I will build on Schwarzschild's (1999) account of givenness, following Rooth (1992), under which givenness is semantically interpreted, but focus is grammatically marked. I will presents a case study of a language in which it is not focus but givenness that that is always grammatically marked (cf. Wagner 2005, 2006 for arguments that even in English givenness must be sometimes grammatically marked). Crucially, movement plays a role in this process.

In Czech, given elements must linearly precede new elements. If this relative ordering cannot be achieved by base generation, movement must take place. I will argue for a natural language operator that marks elements in its scope as given. The operator divides the structure between a given and a new part. The role of Maximize Presupposition is to enforce that every given element is in the scope of the operator. The operator and Maximize Presupposition work in tandem with an economy condition on movement that licenses movement only if it yields otherwise unavailable semantic interpretation (cf. Fox 1995, 2000, Reinhart 1995). The proposal thus provides independent evidence for a structural competition in the grammar and for the role of Maximize Presupposition in the process. Furthermore, understanding scrambling as presupposition marking allows us to understand scrambling in context of other grammatical means for Maximizing Presupposition such as definiteness marking, agreement, or special morphological markers.