Semantic Theory and Philosophy of Language.
Judging Generics
In joint work with Ashwini Deo [DM10], we propose a solution to a problem in
semantics: why do human beings tend to agree on their judgements of the truth/falsity
of a wide variety of generic sentences (of form "A are B"), even though the meanings
carried by, say, the subclass judged as true are highly divergent? For example,
"Crows are black" is judged true as is "Mosquitoes carry the West Nile virus", but the
former seems like a simple consequence of the overwhelming majority of crows being black,
whereas the latter is judged true even though only a tiny proportion of mosquitoes might
actually carry the virus. Such problems are well studied in linguistics, and several approaches have been suggested to resolve this problem of generics in particular, notably one by Ariel Cohen using (frequentist) probability. We use a subjective Bayesian approach (thought of as a first-order approximation to human decision-making on sentence soundness) to propose a simpler solution to the problem.
In a team with Ashwini Deo and Maria Piñango, we are also working on other problems in semantics that would benefit from insights from statistics and information. The eventual dream is that these investigations might lead to a better theoretical grasp of how we humans convey "information" through natural language (a much more subtle problem than that studied by classical information theory).
References
| [DM10] |
A. Deo and M. Madiman. Generic sentences and subjective probability.
Abstract presented at the 20th Semantics and Linguistic Theory
conference (SALT 20), Vancouver, Canada, May 2010.
[pdf]
[slides]
[Full paper being prepared]
|
Read the relevant publications for more details,
or go back to research summaries.