LING419: Fitting Syntax and Semantics Together
Spring 2010

When a human being knows a language, he/she knows that certain sounds are paired with certain meanings. In your syntax classes, you've basically learnt that certain sounds are paired with certain tree structures. The aim of this course is to complete the picture by looking at how those tree structures relate to meanings, and look at why the syntax-centric approach that you have been exposed to has become popular.

Course syllabus

Homeworks

• Homework 1: Categorial Grammar Basics and Lambda Notation, due Mon 15 February [partial solutions]
• Homework 2: Categorial Grammar, due Wed 10 March
• Homework 3: Trees and Movement with Semantics, due Mon 5 April
• Homework 4: Event Semantics and Phrase Structure, due Mon 19 April

Final paper guidelines

• You should identify some facts about a language (perhaps English, perhaps another) that seem to pose a puzzle for the theories we have looked at in the class. By "facts about a language", I mean a small collection of judgements that certain sound-meaning pairings are grammatical and that other pairings are not.
• You need to show why these facts are a puzzle for the theories we have looked at, by trying to account for the facts using the existing theory and showing what goes wrong. Imagine that you're responding to someone who looks at the facts you've identified and says "OK, so what? I bet our current theories get that right. Why should I care about these facts?"
• Once you have shown why your facts are problematic for an existing theory that we have looked at, try to suggest some minor tweaks that we might make to our theory, in order to allow your facts to fit in. Showing exactly why the existing theory can't handle your facts, as described above, will help you towards thinking of some suggested adjustments.
• These suggested tweaks still might not work. That's OK! By working out exactly where an adjusted theory hits problems, you will learn more about the nature of the puzzle, and maybe have ideas for some further adjustments which will take you in the right direction. Don't fall into the trap of thinking that an idea must be perfect in order to be worth writing down carefully!
• Do not worry if you do not arrive at perfect explanation for the facts you have chosen to investigate. But be sure to include a description of why you feel that your ideas don't get things perfectly right. I'm much more interested in seeing that you have approached the problem carefully and systematically, and that you have explained your thoughts clearly, than I am in how good an explanation you actually end up with.
• You don't necessarily need to discuss any "real semantics" explicitly (i.e. you don't need to include anything that involves writing logic or lambdas), but your approach definitely should make it clear that the facts to be explained concern the existence of certain sound-meaning pairings and not others. Using lambdas and logic would be one way to do this, but careful plain English can also do the job.
• One way to try to generate some ideas is to "mix and match" pieces of different theories: take a phenomenon which you know how to deal with in one theory, and see if you can tell an equally good story in a different theory. You might take a phenomenon which we saw worked out neatly in categorial grammar (eg. "John bought and Mary ate apples"), and think about how to account for that in tree-based/transformational grammars. Or you might do it the other way round: take a phenomenon that you learnt about in LING311 which has a nice explanation in transformational grammar (eg. passives, raising, control, multiple-auxiliary sentences, binding constraints) and see if you can find a way to account for the same facts in categorial grammar, or in the "formal transformational grammar" framework that we looked at (part A of homework 3 was an example of this). If you do this kind of thing, your first goal should be to see if you can assign categories to words such that the rules we looked at in class give the right result; or maybe that won't work, and you need to change/add some rules, in which case you can ask whether the things that worked correctly in class will still work correctly with your new changes/additions.
• To repeat: Don't fall into the trap of thinking that an idea must be perfect in order to be worth writing down carefully! If an idea is half-good, and you write it out carefully and try some derivations and see what goes wrong, you'll often find that a better version of the idea pops into your head — this is a perfectly normal way to make progress. If an idea is half-good, and you just sit around feeling unsatisfied with it, it's much less likely to lead you anywhere better.

Schedule

This is subject to change, as stated on the syllabus.

Controversy about islands

• Hofmeister and Sag argue that island effects result from processing difficulty
• Sprouse et al. look for a correlation between working memory capacity and island effects

Required background from Syntax I

• Most importantly, you'll definitely need to be familiar with X-bar theory, which is covered in chapters 6 and 7. The only prerequisite for reading those chapters should be a basic understanding of phrase-structure trees — look in chapters 2 and 3 if necessary for this but it's really just LING240 stuff.
• Ideally you should also be familiar with the basic idea of wh-movement, as it's covered in chapter 11 up to the end of section 1 (p.333). If you're feeling unsure about the general concept of movement (eg. if you look at chapter 11 and you find yourself wondering how on earth we ended up with so many arrows on our trees), you should review chapters 9 and 10.
• We will not be talking about binding theory at all, so you can safely ignore chapter 5.
• You can definitely ignore everything after chapter 11.

Back up to Tim's web page