It seems like you need to settle the question of whether concepts are holistic before you can get an natural language understanding (NLU) system off the ground and understanding some text. Holism is, I bet, one of the most important foundaitonal issues in NLU.
Wednesday, March 15, 2006
Sunday, March 12, 2006
To what extent does interpretation factor into the meaning of a sentence? Does the speaker have the complete and/or final say in what she means? Part of me wants to say, yes, she does. The speaker is the one who determines what she means because she produces and has better access to her intentions. Part of me wants to say, no, she does not. Communication is a two-way street and the interpreter places some restrictions on what the meaning of an utterance is. Maybe the problem here is that meaning and communication don't necessarily have to go together. Do they? I started thinking about this last night when I was asked what produces semantics (the question was: where does semantics come from?). The standard answer, I suppose, is the conventions of language use within a population. There is something unsatisfying about this answer, but I'm not yet sure what.
I was surprised to find another philosopher besides Davidson who denies that there is such a thing as languages, as philosophers construe them. Peter Ludlow makes this claim. I haven't read his essay in detail yet, but I am quite curious.
Thursday, March 09, 2006
I think that philosophers of language have overlooked issues in semantics that they really should not have overlooked. My main thing is plurals. For example, Cappelen and Lepore stick a bunch of plurals in their basic set by claiming that the singular indexicals (I,he,you,etc) and 'all their various forms' (approximate quote) are as well. This overlooks the ways in which the plurals are different from the singular. For example, in the first-person case, it looks like 'we' is not automatic while 'I' is. Nunberg has some interesting examples of ways in which 'these' and 'those' differ from 'this' and 'that'.
Wednesday, March 08, 2006
It occurred to me that one of the reasons that philosophers have not tackled the problem of similarity of meaning may be this. Quine pretty thoroughly refuted one notion of synonymy. He proposed another, stimulus synonymy. I don't think that ever caught on. But, since the previous kind of synonymy was considered vanquished, it may have made philosophers shy away from looking at similarity of meaning. Similarity seems harder than identity. Would this even be a transitive relation? My inital answer is no.
On a different topic, there is an argument in one passage of Husserl's Ideas that seemed odd to me. He argues that the vey idea of possible worlds that are causally and spatio-temporally isolated from ours is incoherent due to considerations of perception. I will have to review the passage at a later date in order to do it full justice. It is Husserl offering phenomenological objections to David Lewis-style possible worlds metaphysics. It sesems like an original idea. I don't think David Lewis was fazed by this objection though. Dagfinn Follesdal said that he taught Lewis phenomenology but not much seemed to have stuck.
Monday, March 06, 2006
What are word meanings? One characterization is that they are functions of various kinds. This may work as a way of formalizing a theory of meaning, but I'm not sure how to reconcile it with the following phenomenological observation. When I understand what a word/sentence/etc. means, it doesn't seem like I'm grasping a function or a composition of several functions. There is certainly something going on when I understand, say, a proof in recursion theory that uses functions, but that seems somewhat differnet than when I understand what a new word means.
On a similar topic, are there any theories about similarity of word meaning? I don't mean identity of word meaning, synonymy. The meaning of 'black' is similar to the meaning of 'white', but is it more similar to the meaning of 'white' than to the meaning of 'blue' or 'gray'? VSM, LSA and their derivatives have measures of word meaning similiarity, but there is not a way (yet?) to connect their meaning vectors to a truth-conditional semantic theory. This is certainly a project to pursue.
Thursday, March 02, 2006
Cappelen and Lepore claim that intuitively the following is false: There are true utterances of 'George knows that he has hands' even though George doesn't know he has hands. They need the ancillary premise that the George in the quote and the George after the quote are the same George. Otherwise, surely it is true.
Wednesday, March 01, 2006
Question for the day: does the semantics/pragmatics distinction carry any implications for the application of logic to conversation or discourse? When will we start talking about disambiguation as an important activity in understanding rather than jumping straight into saturation,etc. in interpretation?