Homework 4 more comments

This is pretty late in the game here, but I’ve been continuing to get questions and comments about parts 2.3 and 2.4. As I said in class too, these are… pretty terrible questions. I am aware of that.

Let me say a couple of things about them, in case someone is still reading this, even if it is only something like 10 hours before class. At least it might be reassuring to read.

In the first part, you are designing a grammar to account for the facts given in (7) and (8) and I gave an example of how you would define the determiner no so that it would occur only with nouns that were not proper names, by specifying it as follows:

no, Det, [+ __ N[-proper] ]

By putting this kind of subcategorization frame on the determiners, you can get the facts in (7-8), and since they vary both in terms of number and in terms of whether they are a proper noun or not, you’d need to use both kinds of features. (Like [-plural] or [+proper].)

Part 3 then shows you a new set of facts in (9). It should be clear enough from looking at them that they have no pronounced determiner (they are bare nouns), and they differ in number. The plural ones are ok when there is no pronounced determiner, but the singular ones are ungrammatical without a pronounced determiner.

Now, if you designed the grammar the way I anticipated you would, by putting subcategorization frames on the determiners, then it is not clear how to distinguish (9a) from (9b) – if there is no determiner there upon which to put a subcategorization frame. Right? So when I asked whether you can do it “by specifying complements in some way” what I meant was whether you can do it by having a subcategorization frame like [+ __ N[+plural]]. And the reason it seems unlikely is that there is probably no determiner to put that subcategorization frame on.

Really, all you need to do with this problem is observe the pattern and the potential difficulty you would have in incorporating these facts into the grammar you already have.

Now, one presupposition here is that the subcategorization frame always goes on the determiner, not on the noun. If you allow boy to have a subcategorization frame like [+ Det __ ], that will solve the problem too, if you put that frame on every singular noun. It feels a little bit clunky, but it’s not really ruled out by anything we’ve talked about. So this possibility was not one that I had anticipated, but if you do allow for this, it kind of undercuts the question, since there is a way to solve it after all, just by specifying subcategorization frames on lexical items.

However, the flow of the problem was supposed to be to observe the facts in (9) and point out the difficulty in using subcategorization frames to constrain what nouns can appear bare if there is no determiner to put the subcategorization frame on. Leaving this as kind of an open issue, then part 2.4 attempts to resolve the problem.

Now, in section 2.4, the solution proposed is to say that NPs always have a determiner. That there is no NP -> N rule, only a NP -> Det N rule. And which Dets and Ns you can use together depends on satisfying the subcategorization constraints. If we suppose that there is always a determiner, but that at least one of them has no pronounciation, then you can suppose that a silent indefinite determiner SOME (the plural counterpart of the singular a(n)) goes with plural nouns to make an indefinite plural like boys or babies.

I say something a little bit confusing at the end there about how you need another silent determiner to finish the whole data set. What I mean by that is that we need to be able to account for the appearance of proper names (like Bart) without a determiner. The problem I was alluding to is that SOME is indefinite (again, the plural counterpart of the singular a(n)), and an NP consisting of a proper name is definite, not indefinite. So, what I meant was that you’d want a silent definite determiner (a counterpart of the but that goes only with proper names) in addition to the SOME determiner. But, if you didn’t consider indefiniteness to be a fundamental part of SOME, then this issue doesn’t come up and you could just use SOME with a proper name or a plural noun.

The actual question here (“Does this help…?”) is really asking for just a little explanation of how having SOME there can allow you to attach the subcategorization frames that it seemed like we needed in the prior part.

So, as I say, these questions as they are stated are pretty confusing. By the time I’ve gotten through explaining what I had in mind, I’ve already kind of given you the answers I was looking for, but if you get it and write something that kind of makes these points, it’s fine.