The Weekly Newsletter of MIT Linguistics

LFRG 4/19 - Ben George

Speaker: Ben George
Date/Time: 4/19 (Thu) 10 am
Title: Question Embedding: Varieties of Reducibility and Non-Reducibility


In this talk is an effort to collect some of my recent concerns about question embedding under predicates, like ‘know’, that have both a propositional (1) and a question-oriented (2) use.

1. Anne knows that Maggie destroyed the records.

2. Anne knows who destroyed the records.

A usual assumption, sometimes implicit and sometimes explicit, is that the truth of sentences like (2) is to be evaluated in terms of the facts reported by sentences like (1). That is, most theories of question embedding employ some version of the reducibility assumption in (3).

3. The truth of a sentence like (2) is determined entirely by which answers to the embedded question (‘who destroyed the records?’) are known by the subject (Anne).

In this talk, I briefly introduce some of the (quasi-)formalizations and weakenings of (3) that I and others have suggested, and then introduce three groups of descriptive claims that challenge some or all formulations of (3), with an emphasis on moving beyond the ‘mention some’ cases that I considered before, and engaging with ‘exhaustive’ examples (4, after an example from Lahiri and an observation of Kratzer) and examples involving ‘agree’ (5, after a claim from Beck and Rullmann).

4. The witnesses know which conspirators were present at the secret meeting.
(Truth depends not only on witnesses’ knowledge, but on whether they have any relevant false beliefs.)

5. Robin and Rupert agree on which of their colleagues are spies.
(Truth depends not only on which answers Robin and Rupert agree to, but on which answers they are opinionated about.)

I do not reach any firm conclusions, but attempt to get a sense of the scope of the problem, and suggest a preliminary taxonomy of problem cases.