UGGS meetings are characterised by comfort, style, overpronation, and audience participation. With inquiries and suggestions, please contact Karen.
Link 1983 noted that predicates built from many transitive verbs can be interpreted as collective as well as distributive on their subjects, whereas predicates built from most intransitive verbs are inherently subject-distributive. For example, Ana and Bob built a raft can mean that they each built separate rafts (distributive: the predicate it separately true of each member of the group Ana and Bob), or that they worked together to build one (collective: the predicate is not separately true of each member of the group, but only true of the group as a whole). In contrast, Ana and Bob died means that they each died (distributive), not that they died jointly without doing so separately (collective). This project's data contribution is to refine and confirm Link's observation.
We show experimentally that many predicates built from transitive verbs can be interpreted as either distributive or collective on their subjects. But, we add, there are also many such predicates that can only be subject-distributive: Ana and Bob like Cara means Ana likes her and Bob likes her, not that they like her jointly but not separately. In contrast, as Link suggested, many predicates built from intransitive verbs can only be interpreted as distributive. Moreover, we build on Link by showing that transitive verbs verbs are almost always interpreted as distributive on their objects: Ana hugged Bob and Cara, even if there was only one group hug, conveys that she hugged Bob and that she hugged Cara.
The theoretical contribution is to isolate the crucial property that allows arguments to be interpreted collectively, and to explain why this property tends to hold of subjects of VPs built from transitive verbs, but not subjects of VPs built from intransitive verbs, or of the objects of transitive verbs.
The connective "unless" is often suggested as a challenge to compositionality, on the grounds that it combines differently with positive and negative quantifiers. In particular, it has been argued that "unless" is biconditional in positive contexts but unidirectional in negative ones. An attractive compositional solution is the exceptive approach of [von Fintel 1992,Leslie 2008]. We report on an experiment (joint work with Dan Lassiter) which demonstrates that unless is not biconditional under "every," and in which a majority of participants endorse sentences predicted false by both exceptive theories. The results additionally show three new theoretical puzzles. I propose a revised semantics and pragmatic s which accounts for the first two,consider some directions for explaining the third, gradient pattern, and discuss the consequences of this proposal for other exceptive constructions.
In a truth conditional semantics, the meaning of a proposition is taken to be the conditions that must obtain in order for it to be true. In an information exchange semantics, the meaning of a proposition is taken to be the change that a context will undergo when updated with it. Generally, information exchange meanings are derivable from truth conditional meanings: a proposition p in an information exchange semantics denotes the set of worlds in which p is true—that is to say, the set of worlds that satisfy the truth conditions of p. I will argue that this method does not work for statements of the form might-p. Instead, I will propose that epistemic modality can be operationalized in an information exchange semantics as the imposition of conditions on information states, and as the highlighting of certain possibilities within those information states.
In this talk, I will look at the non-future interpretations of will. I show that will and must differ in the inferences they allow. For example, the situations where a speaker can felicitously utter "My neighbors will be barbecuing" are distinct from those in which the speaker can utter "That will be my neighbors barbecuing." However, the counterparts of these sentences with must do not show similar behavior. I claim that while must can be accounted for by quantifying over epistemic states, an account of will needs to make reference to the speaker's source of evidence.
A concept that has much advanced our understanding of discourse in recent years is the concept of a question under discussion (Ginzburg, 1995a; Ginzburg, 1995b; Roberts, 1996). The idea is that conversation does not unfold randomly, but is guided by discourse topics or decision problems (van Rooij, 2003). When formulated as a question, such a problem is called a question under discussion — QUD for short. Among other things, QUDs have advanced our understanding of focus (Büring, 2003), anaphora resolution (Roberts, 2003; Schoubye, 2009), speech acts (Roberts, 2004), and scope resolution (Zondervan et al., 2008).
As theoretical primitives, QUDs are useful because they can be made formally precise. Each individual QUD can be given a precise structure via question semantics, and we can also make exact statements about the relationship between different QUDs (Groenendijk, 1999a). For these reasons, QUDs have formed the basis for exciting new theories of discourse strategies (Roberts, 1996; Groenendijk, 1999a; Büring, 2003) and invigorated the study of discourse particles (McCready, 2006; Eckardt, 2007; Davis, 2009).
I will propose a particular QUD-formalism, called S-trees, and then illustrate how it can help us understand the German particle überhaupt in a unified way. The German particle überhaupt exhibits a variety of uses with seemingly unrelated meanings. Correspondingly, only partial and non-unified theoretical accounts have been proposed. I show how the various intuitions and ostensibly different meanings can be derived from a unified characterization of überhaupt as a move to a higher-level question under discussion.
This paper brings together two strands of research, work on bias in high negation polar questions (HNPQs) and work on evidentiality, to reach a novel approach to discourse bias as relevant to HNPQs and related questions. The paper formalizes bias as defective discourse evidentiality, with participants' discourse commitments and other contextually-rooted sources acting as the evidential base. The paper presents novel data that builds on Sudo (to appear) and Roelofsen et al. (2012) to disentangle different types of contextual evidence, and what it means for an evidential base to be pro-, anti-, or neutral with regard to a given proposition.
The availability of quantifiers inside ellipsis sites is often used to test for internal structure in anaphors. The ability of quantifiers to introduce new referential dependencies and to participate in inverse scope are particularly important. The talk deals with two problems for using the "availability" of quantifiers as tests for internal structure. The first problem is a set of "overactive quantifier" puzzles for the missing antecedent test; in this case, quantificational dependencies seem to be introduced where our theory predicts they shouldn't be. The second is a set of "underactive quantifier" puzzles for the inverse scope test, which deals with instances where inverse scope should be possible according to our theory---but isn't. (Note that this will be a very informal workshop---much of the data is quite raw.)
The premiere of UGGS will have as its theme modified numerals, and, rumour has it, there will be rugelach (the committee had to Google this) and a frisbee!
Karl DeVries will be talking about the interaction between modified (and maybe some unmodified) numerals and aspectual modifiers.
Matthijs Westera will be talking about the pragmatics of modified numerals, based on inquisitive semantics.
The Audience will be talking about related topics.