Assignment 1: Verb Learning

Posted Thursday September 9th, due Thursday September 16th.

This written assignment is based on articles by Lila Gleitman and Steven Pinker on how children might exploit connections between verb meanings and verb syntax in order to learn aspects of their native language. You will have read and written about these articles by the time we discuss them in class, so you will be especially well-prepared to participate in the discussion.

1. Summarize what a child can (and cannot) learn about verb meanings using syntactic information, as discussed by Gleitman (this is sometimes known as the "Syntactic Bootstrapping" approach, although Gleitman objects to this terminology). Does this approach make it possible to learn the fine details of verb meanings, or only very general properties. Provide examples where appropriate.
2. Is there anything more that a child could learn using Pinker's "Semantic Boostrapping" approach? Does this approach disagree with Gleitman with respect to what information the learner is able to extract from his/her environment? Provide examples where appropriate.
3. Are these two approaches incompatible with one another, or could they be usefully combined? Justify your answer.

Reading guidelines. There are three readings relevant to this assignment. Lila Gleitman's The Structural Sources of Verb Meanings is at this point a classic article in language acquisition. The first few pages of the article set out the problem of learning verb meanings extremely well. Pages 23 and onwards are particularly useful to the questions above. Steven Pinker's article What could a child learn about verb semantics using verb syntax? is a characteristically insightful analysis of what can and cannot be learned about verbs using syntactic or semantic cues; pp.383-388 and pp.395-403 are particularly useful. These two articles should be your primary sources. The review article by Jean Gilette and Lila Gleitman provides a more up-to-date summary of Gleitman's views and experimental findings.

Writing guidelines. A good answer will contain carefully presented arguments, together with justification for these arguments in the form of examples/evidence etc. Vague claims are always extremely hard to evaluate, and are scientifically not particularly useful. Explicit discussion of how learning mechanisms could be applied to specific verbs are likely to be very useful. You are strongly encouraged to use a word processor to write your answer - this makes it much easier to review and edit your answers, which you should always plan to do.