Homework 4 comments

I talked through a number of things in class about the upcoming homework 4.

One thing I didn’t mention, but was asked about, concern the meaning of the # in problem 3, and the do so replacement.

The # in problem 3 is supposed to be analogous to the * or ? or ?* marks on a sentence; it is an indication of a judgment. In particular, the # is supposed to indicate a sort of semantically ill-formed sentence, even if it is syntactically grammatical.

So, the * in *Bart pizza the ate represents a syntactically ill-formed sentence. The # in #Homer worded the message carefully, but Bart did so sloppily indicates that this sounds kind of like a contradiction. The reason it sounds like a contradiction is that Bart did so really stands for Bart worded the message carefully, so the whole meaning is something more like “Bart worded the message carefully sloppily” and it feels somewhat contradictory to be both careful and sloppy.

The “do so” replacement is a form of proform replacement, it’s replacing the VP with “do so”. So, like Bart read the magazine, and Homer did so too (do so replaces read the magazine).

Among the things I did mention, but will reiterate here:

In problem 1, the task generally involves deciding whether some of the things that you see in the data are obligatory or optional. In all cases, there is a position in the sentence where something can either be or not be. So, we talked about Bart behaved and Bart behaved poorly. The question is whether the adverb poorly is an argument (required) or an adjunct (optional). It is clear that you can say Bart behaved with no overt adverb. The question is, if you say that, does the semantics still imply that a silent adverb is there. It kind of seems like it does. If you say Bart behaved you mean Bart behaved well. If so, then the adverb is not really optional, it is required—but it can be left unpronounced. There’s a difference between being not there at all and being there but silent. And that difference generally arises from the semantics, whether it is possible to behave without a manner. Since it seems like a manner is required (even if it is essentially implicit, unpronounced), then it should be considered an argument, and the frame should look like [+ __ AdvP]. (Or, possibly Adv, given the data.)

In problem 2, pay attention to the note that the NP -> N rule will be discarded later. What that means is that eventually, NP will require a determiner. But, as with problem 1, it is conceivable that a required element like a determiner might be unpronounced. The hypothesis in part D is that there is a determiner which will name SOME (and note that this is different from the pronounced determiner some, although both share an indefinite semantics) and which is not pronounced. Once you hypothesize a silent determiner, you can assign to it a subcategorization frame that can require of it that it be in a constituent with a particular sort of N.