last changed 31 Jan 2006
This paper was published in UCL Working Papers in Linguistics (8:283-314, 1996), but never published or (as far as I can remember) submitted for publication beyond that; I can't remember why not. It developed the ideas in "Measuring syntactic difficulty", an unpublished draft from 1995.
Why is sentence (1) impossible to process?
(1) The rat the cat the dog chased ate died.
The standard explanations all focus on its syntactic structure, but the present paper offers an alternative explanation in terms of semantic structure. The syntactic account cannot explain why some sentences which are syntactically similar to (1) are much easier to process. The difficult sentences seem to be those in which an embedded clause which modifies a noun has its own subject modified by another clause whose subject is a common noun (not a pronoun). The explanation offered here is that the senses of the modified nouns are not sufficiently distinct to be maintained as separate concepts in short-term memory, so the various concepts stored `interfere' with one another.