In the middle of my “oh my God, I have how many days to complete my Speech Prosody project?” panic, I started thinking about relative clauses again. It’s a nice comfortable place to rest my head sometimes. Anyway, here are some new thoughts on that.

Ok. Assume any theory of grammar. Whether you take a movement-based, variable-full, direct compositional, variable-free, whatever–any theory which takes the Det-Nom view of relative clauses is going to have to get intersection somehow. I should probably go back to Heim & Kratzer and check the standard theory, but to the best of my memory, the rule intersecting the lambda-abstracted relative clause with the noun has got to be pretty stipulative. (Can you do predicate modification there? Hm. I guess that’s not as bad as a whole separate rule. But when you have a theory which can take care of adjectives without predicate modification, it returns to being pretty stipulative.) Anyway, the categorial grammar version of Det-Nom that I’ve seen just builds that into the relativizer (remember we don’t need lambda abstraction here because we have function composition!). Which all seems nice and great except that this doesn’t really seem to capture the fact that relativizers can be optional sometimes. For that matter, neither does the standard theory.

So, let’s think about our N->N/RC rule again. Its semantics is λP[P∩N], where N is the set denoted by the noun, of course. This simple rule is something that the Det-Nom view could also use. And, whatever theory you pick, when you have this rule, you have a way to account for how to intersect the relative clause when you don’t have a relativizer.

You may be asking now then what’s a relativizer? A pronoun, of course! I think this makes a lot of sense. Take the distribution of relativizers. From what I can tell, they are not optional when the relativizer is occupying the subject position of the relative clause sentence. That is Every woman who loves John requires the relativizer whereas Every woman who John loves does not. I’m not quite sure why this should be (some syntactic constraint? without the relativizer there, there’s no reason to not just hear that NP as instead a full S–one reason to be skeptical of any approach that says that when there is no relativizer, there’s some silent, deleted thing there, I think), but I have this feeling that this will show the way to why relativizers should be thought of as pronouns.

I obviously haven’t been thinking about this long enough, but I really sort of like the observation about the distribution of relativizers and how the N->N/RC rule seems to be something that everyone (both Det-Nom’ers and NP-S’ers) will need.

emma :: May.07.2007 :: misc, syntax, semantics :: 2 Comments »