In this talk, I will discuss two lines of research in sentence processing. In the first part of the talk, I will present theories and evidence addressing the question of how people resolve syntactic category ambiguities (e.g., "duck" as a noun or a verb; "that" as a complementizer or a determiner).  It is proposed that people combine (a) context-dependent syntactic expectations (top-down statistical information) and (b) context-independent lexical-category frequencies of words (bottom-up statistical information) in order to resolve ambiguities in the lexical categories of words.

In the second part of the talk, I will present experimental evidence relevant to the possible existence of a syntactic constraint in English that I will refer to as "phrasal adjacency" (cf. Hudson, 1990).  Two constituents Yi and Yj are said to be phrasally adjacent if they are generated by a head-dependent context-free rule of the form X -> Y1 ... Yn.  Consistent with the hypothesis that English grammar is constrained by phrasal adjacency, self-paced reading studies indicate that extraposed structures like (1a) and (2a) are more difficult to understand than controlled non-extraposed structures like (1b) and (2b).

(1) a. A musician [ who was hired for the wedding ] arrived.
   b. The reporter interviewed the star of the movie [ who was married to the famous model.

(2) a. A musician arrived [ who was hired for the wedding ] .
   b. The reporter interviewed the star about the movie [ who was married to the famous model.