Research Interests


My primary research interests are in the field of formal semantics and pragmatics.
  1. Generics

    Our knowledge about the world is often expressed by generics, sentences such as Birds fly or The tiger has stripes. Yet it is far from clear exactly what such sentences mean. Generics turn out to exhibit a surprising variety of interpretations, and a number of unique and puzzling properties, and have therefore attracted the attention of many researchers. Previous approaches to the semantics of this construction have not been successful, because they focused on only one or two of the aspects of their interpretation, thus missing the big picture and failing to account for other aspects. By looking at the entire range of phenomena exhibited by generics, I have been able to come up with a novel theory that accounts uniformly for the full richness of the meaning of generics. Specifically, I claim that generics express probability judgments, and that, if a suitable interpretation of probability is used, all their puzzling properties are accounted for.

  2. Other quantifiers

    My work on generics has prompted me to consider similar constructions, such as frequency adverbs, and attempt to account for the similarities and differences between them. It turns out that the study of generics can even given insight into puzzles of seemingly unrelated phenomena, such as the quantifiers many and few, and I have proposed a unified account of these phenomena.

  3. Focus

    One of the ways in which generics are similar to many other constructions is that their interpretation is sensitive to focus, often conveyed by intonation. Focus is a much discussed topic of current research, and its contribution is generally held to be in providing a set of alternatives, with respect to which the sentence is evaluated. However, the mechanism by which these alternatives are induced has been unsatisfactorily discussed in current literature. I have proposed that focus does not introduce alternatives directly, is in providing a presupposition, and it is this presupposition that induces alternatives. Thus, by applying tools developed for the analysis of presupposition, I have been able to provide specific rules detailing how the alternatives to a given sentence are computed.

  4. Presupposition

    My interest in presupposition extends beyond its role of providing alternatives. I have also investigated the more philosophical aspects of presupposition; in particular, I claim that, contrary to common opinion, in certain cases, sentences whose presuppositions are unsatisfied are true.

  5. Bare plurals

    Bare plural nouns often express generics, e.g. Firemen are intelligent. But bare plural nouns may also be used non-generically, as in Firemen are available. I have therefore been interested in the semantics of bare nouns, and in the rules that determine which interpretation, generic or non-generic, is the preferred one. In joint work with Nomi Erteschik-Shir we have determined that here, too, focus plays a crucial rule.

  6. Default reasoning

    We often reason using default, rather than absolute, rules. Such rules are often expressed, informally, using generic sentences. I have proposed taking this informal characterization, and defining a default rule to be sound just in case the corresponding generic sentence is true. I have been able to prove some desirable properties of default reasoning systems that are compatible with this proposal.

  7. Anaphora resolution

    The computational problem of identifying the referent of a pronoun, definite description and other anaphoric triggers requires the interaction of many factors. I am interested in systems that make all such factors explicit, in terms of prioritized defaults. In particular, there is one misleadingly simple constraint, namely that antecedents to triggers must be looked for, that is usually not formalized, but is left to the inner workings of the algorithm. Formalizing this principle on a par with other factors, as a prioritized default, results in a system that is fully explicit, hence its behavior can be dependably predicted in any given situation, and theorems about it can be proved.


Return to Home Page