nucleation

THEORY-785: Listeners interpret an utterance by inverting a model of a speaker who chooses among alternatives by informativeness, so what a fixed form conveys depends on what else the speaker could have said, as the rational speech act account of Frank, Goodman and coauthors proposes

2 min read

Properties5
StatusProposed
Tagspragmatics, linguistics, cognition, probabilistic-modeling
SourceLIT-821 — Predicting Pragmatic Reasoning in Language Games, LIT-816 — Pragmatic Language Interpretation as Probabilistic Inference, LIT-809 — That's what she (could have) said: How alternative utterances affect language use
ExtendsTHEORY-172 — In Frank and Goodman's one-shot reference games, listeners' mean bets matched a parameter-free listener inverting a speaker who chooses among true utterances by informativeness, with a separately measured salience prior, and not salience alone; and in Bergen, Goodman and Levy's signalling games with novel symbols, listeners read a signal as excluding what the available alternative would have conveyed
AnswersQUESTION-209 — Do listeners in reference and signalling games invert an informative speaker when alternatives and costs are manipulated with the uttered form held fixed, the model's parameters are set in advance, and its predictions are tested on trial-level data against a literal listener and a likelihood-only model?

Graph View

Source

Frank and Goodman (2012, Science), LIT-821, read in NOTE-634; Goodman and Frank (2016), LIT-816, read in NOTE-606; Bergen, Goodman and Levy (2012), LIT-809, read in NOTE-613. The papers present the inversion of an informative speaker as how listeners reason, not only as a model that fits mean bets.

What is asserted

That a listener’s interpretation of an utterance is the result of inverting a model of a speaker who chooses among the available alternatives by informativeness, so that the same form conveys different things when the alternatives differ.

What the record holds toward it

THEORY-172, Active, is the measured part: in Frank and Goodman’s one-shot reference games, mean listener bets matched a parameter-free inversion of a size-principle speaker with a separately measured salience prior, and not salience alone; and in Bergen, Goodman and Levy’s signalling games with novel symbols, listeners read a signal as excluding what its available alternative would have conveyed. THEORY-472 derives the size-principle speaker from informative communication. QUESTION-209 asks for the decisive test: alternatives manipulated with the form fixed, parameters set in advance, trial-level data.

What this does not say

  • Not that comprehension in natural conversation computes the recursion; the games fix the alternatives.
  • Not that speakers are informative; the model is of the listener’s assumption about speakers.

What would confirm or reject it

Confirmed by the promote_when. Rejected if, with alternatives and costs manipulated and the form held fixed, a literal listener with the same prior, or a likelihood-only model, predicts trial-level judgements as well as the inverted speaker.

Backlinks