The Weekly Newsletter of MIT Linguistics

Colloquium 5/13 - Roni Katzir

Speaker: Roni Katzir (Tel Aviv University)
Title: On the roles of anaphoricity and relevance in focus
Date: Friday, May 13rd
Time: 3:30-5:00 PM
Place: 32-141

The placement of accent on elements in sentences interacts both with the felicity of sentences in their conversational context — so-called free focus (FF)—and, in the presence of certain operators, with the truth conditions and presuppositions of sentences—so-called association with focus (AF). For example, John DRINKS tea is acceptable as a response to John sells tea but not to John drinks coffee (FF); and John only DRINKS tea, with the AF operator ‘only’, can entail that it is false that John sells tea but not that it is false that he drinks coffee. It is commonly assumed that focus-sensitivity in both FF and AF is related to focus alternatives, sets of sentences that are identical to the original modulo focus-marked constituents (e.g., {John drinks tea, John buys tea, John sells tea, …} for John DRINKS tea). Moreover, this connection is often taken to be anaphoric: in FF, the focus alternatives of a sentence are required to have a contextually salient element or subset (Jackendoff 1972, Rooth 1992, Schwarzschild 1999); and in AF, focus alternatives are matched against an anaphoric element that determines the domain restriction of an operator like ‘only’ (Rooth 1992, von Fintel 1994).

My goal in this talk is to argue that the role of anaphoricity in focus sensitivity is more limited than commonly thought and that the main factor, both in FF and in AF, is relevance to a question (in the sense of Groenendijk & Stokhof 1984). I start by reviewing several empirical puzzles for Rooth 1992 and Schwarzschild 1999. These puzzles suggest a central role for questions in focus sensitivity, though they do not help choose between relevance and anaphoricity to a question. I proceed to present the details of a relevance-based account of focus sensitivity, building on Fox 2007’s architecture in which the grammar, enriched with a silent exhaustivity operator, is responsible for all specialized alternative-sensitive computations, while the pragmatic component does not perform such computations and is mostly limited to disambiguating between parses and determining possible values for contextual variables. Finally, I use an extension of Wagner 2005’s ‘convertible’ paradigm to argue that the dependence of focus sensitivity on questions in both FF and AF must be one of relevance rather than anaphoricity. The argument relies on a crucial difference between the two mechanisms: anaphoricity can pick up arbitrary sets of alternatives, while relevance, due to contradiction avoidance, is sometimes incapable of making selections that would lead to arbitrary alternative-based inferences.