|
|
|
|
CSSP 2015
Université Paris Diderot - Paris 7, Amphi Buffon
bât. Buffon, 15 rue Helene Brion, 75013 Paris
Abstracts
- Chris Potts (Stanford, invited talk)
Negotiating lexical uncertainty and speaker expertise with disjunction
- Robin Cooper (Gothenburg University)
Creating lexical meaning on the fly
Cooper (2010; 2012) proposes that the verb rise has slightly different meanings depending on what kind of rising event is being referred to, e.g. a temperature rise or a price rise. The argument is that the events belong to different, though similar, types and that there is no obvious single type to which they all belong. In this paper we will attempt to show that a reformulation will allow us to make a generalization over these types. This raises the question of whether the generalization allows for a single meaning in place of the meanings constructed on the fly. We will suggest that the distinction is related to the contextualist vs. relativist debate and that it is hard to distinguish between the two options (Stojanovic, 2007). We will nevertheless conclude that consideration of a theory of concept learning and the use of underspecified meanings might favour the contextualist view and that we are able to create new meanings on the fly. - Yoad Winter (Utrecht University)
Proto-predicates and the Lexical Origins of Reciprocity
This paper introduces a new generalization about symmetry and lexical reciprocity. Binary predicates are shown to support an equivalence with their reciprocal form if and only if they are logically symmetric. This contrasts with predicates like be in love with, which are non-symmetric and non-equivalent to their reciprocal parallels. The generalization is explained by a new formal theory of lexical reciprocity, which builds on Dowty’s notion of proto-roles and uses it for deriving predicate meanings in reciprocal alternations. - Ariel Cohen (Ben Gurion University)
Non-Temporal Tense
According to Recanati, time, world, location, and other unarticulated constituents are not part of logical form, although they do affect truth conditions.
He makes these points using philosophical arguments; in this paper I argue for them based on linguistic arguments, Specifically, I consider languages where world, location, or person plays the role of tense in languages like English, and I demonstrate that their behavior is analogous.
These findings provide support for Recanati's position, and, moreover, favor an extensional rather than an operator analysis of all unarticulated constituents. - Doreen Georgi (ENS / Institut Jean Nicod) and Martin Salzmann (University of Leipzig)
Complementarity of gaps and resumptives as the result of Case attraction: local modeling under top-down derivation
Deriving the complementarity between gaps and resumptives that holds in some languages presents a challenge to local derivational bottom-up approaches because the choice between the two strategies has to be made at a point where the necessary information (e.g., islands) is not available. A comparison of derivations thus seems unavoidable (cf. e.g., Aoun et al. 2001). We will show that together with the novel proposal that the distribution of gaps vs. resumptives should be reanalyzed in terms of Case attraction, topdown derivation allows for the choice to be made locally without recourse to global economy. - Anne Abeillé, Berthold Crysmann and Aoi Shiraishi (Université Paris-Diderot)
Syntactic mismatches in French peripheral ellipsis
We provide new data showing that the commonly assumed identity constraint on shared material in right node raising, or right peripheral ellipsis, should be relaxed. RNR has always been set apart from other kinds of ellipsis in this respect, and alternative analyses have been proposed: multidominance (McCawley 1982, Bachrach & Katzir 2009) or backward deletion (Kayne 1994, Chaves 2014). The data we provide about determiner, preposition and voice mismatch, put back RNR in the family of elliptical constructions. Since RNR may also involve non constituents, and imposes syncretism on the shared material, we propose an HPSG analysis in terms of phonological identity of meaningful material, allowing for mismatches of grammatical markers. - Oleg Belyaev (Lomonosov Moscow State University), Mary Dalrymple (Oxford University) and John J. Lowe (Oxford University)
Number mismatches in adjective coordination: Feature distributivity and the directionality of agreement
This paper addresses number mismatches in agreement involving attributive adjective coordination. It is argued that an elegant solution for such mismatches can be provided in the LFG framework if, adopting the distinction between CONCORD and INDEX agreement features, one assumes that feature distributivity is constructionspecific and that split readings of adjective coordination involve special functional annotations which make the nominal PRED distribute into the coordinate set created by the adjectives. This leads to the resulting analysis being reminiscent of ellipsisbased accounts without involving any actual deletion. - Vera Demberg (Saarbrücken, Invited Talk)
Towards a structured model of semantic surprisala
- Tohru Seraku (Hankuk University of Foreign Studies) and Akira Ohtani (Osaka Gakuin University)
The Wh-Licensing in Japanese Right Dislocations: An Incremental Grammar View
This paper defends an incremental grammar, a syntactic model which reflects the left-to-right parsing, by exploring Right Dislocations (RDs) in Japanese. We offer new data on the licensing patterns of a wh-word at an RD part and show how the patterns follow from the way in which an RD string is parsed left-to-right. The account is also corroborated by the locality data of RDs. Our grammar is formalised in Dynamic Syntax. - Timm Lichte and Laura Kallmeyer (University of Düsseldorf)
Same syntax, different semantics: A compositional approach to idiomaticity in multi-word expressions
Idiomatic MWEs are commonly analyzed as phrasal units in syntax, in addition to their literal counterparts, and accordingly introduce syntactic rather than semantic ambiguity. However, an analysis of idiomaticity based on syntactic ambiguity is disadvantageous, because it neglects recent psycholinguistic findings about the processing of idiomatic MWEs, and it furthermore obscures the possible connection between their literal and idiomatic meaning. In this contribution we sketch two alternatives, employing the framework of LTAG, where idiomaticity is not subject to syntactic ambiguity, but emerges in the semantics. - Glyn Morrill and Oriol Valentín (Universitat Politècnica de Catalunya)
Computational Coverage of TLG: The Montague Test
This paper reports on the empirical coverage of Type Logical Grammar (TLG) and on how it has been computer implemented. We analyse the Montague fragment computationally and we prior this task as a challenge to computational grammar: the Montague Test. - Jean-Pierre Koenig and Karin Michelson (SUNY Buffalo)
Fractured domain-dependent grammars: The grammar of possession in Oneida
It is quite common for some semantic or pragmatic domains to behave differently grammatically. In this paper, we look at a complex case, that of possession in Oneida, a Northern Iroquoian language. We show that the grammar of possession in Oneida works differently from that of other domains and, crucially, that its exceptional behavior pertains to distinct areas of the language’s grammar (exponence rules, derivational morphology, linking, syntax of quantification). We suggest that to model this kind of exceptional behavior, a global ‘tweak’ of the overall architecture of the morphosyntax of languages is unlikely to be successful. What is needed is a more localized model of these exceptions: Each relevant rule/constraint must provide special dispensation for how to handle possession. We suggest that constraintbased lexicalist theories provide the right architecture for seamlessly modeling this kind of fractured domain-dependent grammar. - Staffan Larsson (Gothenburg University) and Alex Berman (Talkamatic AB)
Domain-specific and general syntax and semantics in the Talkamatic Dialogue Manager
We have presented a design philosophy for dialogue system development, where domain-specific domain knowledge is clearly separated from the logic for generic dialogue capabilities. We hope that this provide a useful illustration of how one may approach the division of labour between general and domainspecific syntax, semantics and pragmatics. - Ray Jackendoff (Tufts, invited talk)
Morphology in the parallel architecture
- Ira Noveck (L2C2, Lyon, invited talk)
When mindreading is not taboo: Discerning intention in pragmatic processing
- Laura Kallmeyer (University of Duesseldorf), Rainer Osswald (University of Duesseldorf) and Sylvain Pogodalla (LORIA/INRIA Lorraine)
Progression and Iteration in Event Semantics LTAG Analysis Using Hybrid Logic and Frame Semantics
In this paper, we propose to use Hybrid Logic (HL) as a means to combine frame-based lexical semantics with quantification. We integrate this into an LTAG syntax-semantics interface and show that this architecture allows a fine-grained description of event structures by quantifying for instance over subevents. As a case study we provide an analysis of iteration and progression in combination with foradverbials. With the HL approach and with standard techniques of underspecification we can account for the behaviour of these adverbials without the assumption of an additional iteration operator on events. This is due to the fact that frame semantics allows to express general constraints on event types that require an event to be an iteration in certain contexts. - Eva Csipak (Universität Göttingen) and Sarah Zobel (Universität Tübingen)
Discourse particle "denn" in conditionals
The talk analyzes the meaning and discourse function of the German discourse particle denn in the antecedent of a conditional. Using denn signals that the antecedent challenges background assumptions in the preceding discourse. Our considerations are supported by a preliminary corpus study. - M. Ryan Bochnak (UC, Berkeley)
Optional past tense and the nature of presuppositional alternatives
I analyze the verbal suffix -uNil in Washo as an optional past tense. It is optional in the sense that it is not part of a paradigm of tenses, and morphologically tenseless clauses are also compatible with past time reference. I also address an apparent problem that this state of affairs presents for the pragmatic principle Maximize Presupposition, and propose a solution. - Miriam Nussbaum (MIT)
Tense and Scope in Superlatives
This paper concerns a puzzle in the interpretation of superlative constructions that contain relative clauses: the lack of sequence of tense (present under past, in this case) forces the superlative to receive an absolute reading. I discuss this in terms of the relationship between the binding of the relative clause’s tense and the distribution of weak and strong NPs in these relative clauses. - Bernhard Schwarz (Mcgill) and Alexandra Simonenko (Labex EFL)
Two pragmatic accounts of factive islands
We explicate and compare two semanticpragmatic approaches to so-called factive island effects: the contradiction analysis (Abrus´an, 2011; Abrus´an, 2014), which excludes factive island questions by virtue of assigning them contradictory presuppositions; and the triviality account (Oshima, 2007; Simonenko, to appear), under which factive island cases are bad by virtue of lacking informative semantic answers relative to any context where they are otherwise felicitous. We present new evidence to argue that the triviality account is superior to the contradiction account.