Makoto Kanazawa on NL semantics
- Zahlangabeheft
- 2406
- 0
- 0
- 0
Now back in Japan. I had a great time at LACL 2011 in Montpellier!
2011-07-03 03:23:36In linguistics, syntax : semantics = grammaticality judgments : truth value judgments.
2011-07-03 03:30:47Truth is of course relative to given situation/information, not absolute.
2011-07-03 03:33:22Entailment, unlike truth in a model/structure, is a theoretical construct that does not correspond to any reliable data.
2011-07-03 03:35:40"What is meaning?" is not an important question in NL semantics any more than "What is a linguistic expression?" is in NL syntax.
2011-07-03 03:39:48Model-theoretic semantics in linguistics is what John Etchemendy once called "representational semantics" (as opps'd to "interpretational").
2011-07-03 03:45:45But I don't think a model is a "representation" of a concrete situation, any more than a cardinal number is of a concrete set …
2011-07-03 03:51:30If a model is a representation of a concrete situation, then a certain finite group would be a representation of Rubik's Cube …
2011-07-03 03:54:29Maybe the last one is not a very good analogy. Anyway, this is just a terminological point.
2011-07-03 04:15:02@makotokanazawa Can you elaborate on entailment? Entailment relation can be inspected wrt the t-value judgements of the two sentences.
2011-07-03 11:47:21@wataruu How? To see whether A entails B, you need to have truth value judgments against all situations.
2011-07-03 11:57:01@wataruu One rarely has reliable intuitions on the question "Does A entail B?", …
2011-07-03 12:00:56@wataruu … whereas, given sufficient information to decide the truth value of B, one can answer the question "Is B true or false?"
2011-07-03 12:03:34@wataruu That's how truth value judgment tasks work in psycholinguistic experiments.
2011-07-03 12:05:30@makotokanazawa @wataruu strictly speaking wouldn't be tv judgments theoretical constructs as well? Only assenting behaviors are present.
2011-07-03 12:07:27@wataruu Psycholinguists working on acquisition rightly don't ask a 10-year old a question of the form "Does A entail B?"
2011-07-03 12:07:32@yuizumi @wataruu That's why I said truth value judgments are like grammaticality judgments. Important data, but not naturally occurring.
2011-07-03 12:09:51@makotokanazawa @wataruu True/false,right/wrong,approving speaker's attitudes etc seem all theoretical. Though we need to start fr somewhere
2011-07-03 12:13:20I totally agree. “@makotokanazawa: That's why I said truth value judgments are like grammaticality judgments."
2011-07-03 12:16:29@ytb_at_twt @wataruu I admit sometimes "sufficient information to decide the truth value" is hard to provide.
2011-07-03 12:19:33@ytb_at_twt @wataruu In the case of a counterfactual, linguists typically give a rather elaborate context agnst which it is to be evaluated.
2011-07-03 12:21:47@makotokanazawa @wataruu It might be difficult to give a context of "If I were PM of Japan ..." or "If the sun rose in the west", anyway.
2011-07-03 12:33:04@ytb_at_twt @makotokanazawa @wataruu theory of imagination in cog sci seems very interesting in that regard. how do we judge anyt. unreal?
2011-07-03 12:42:42