| |
Abstract:
Evidence that many kinds of linguistic information
(syntactic, lexical, discourse, etc.) can bias parse choices has
led to the development of interactionist models of parsing (see
Tanenhaus & Trueswell, 1995, for review). Among the
interactionist models, one subgroup, connectionist models, make
bottom-up structural organization primary, and thus make an
additional, surprising prediction about the parsing of ambiguous
elements: even grammatically impossible interpretations, if they
find sufficient local motivation, may exert an influence on the
parsing process. The importance of this claim is that it points to
a much richer set of relationships between words and phrases than
current syntactic theories, which focus on well-formedness,
provide. We must ask: where, in the model of grammar and/or
processing, should the ungrammatical structures reside?
Evidence in favor of one simple variety of the "ungrammatical
influences" claim has been known for some time. Tanenhaus, Leiman,
and Seidenberg (1979) found that both the noun and verb
interpretations of ambiguous lexical items (e.g.,
rose)
temporarily (for less than 200 ms after presentation) support
priming even in a syntactic context which rules out one
interpretation (e.g.,
She held the rose; They all rose).
But do such "interference phenomena" ever extend past the one-word
level and the 200 ms time interval? As a first step toward
answering this question, we used a word-by-word self-paced reading
paradigm to compare sentences like (1a), which has an embedded
two-word compound
(waste baskets)
that is irrelevant to the only reasonable parse, with sentences
like (1b) which has no such distractor.
Reading times at words 7, 8, and 9 were significantly higher in
the (1a) cases than the (1b) cases (F1(1,30) = 10.77, p = .003;
F2(1,15) = 4.79, p = .045), thus supporting the hypothesis that
there is interference from the ambiguous, but grammatically
irrelevant, compound. A connectionist simulation based on a Simple
Recurrent Network (Elman, 1990) showed similar interference
effects. We argue that "interaction" is only the tip of an iceberg:
the theory of grammar and processing must also support an
appropriate "intimacy" between the well-formed structures that
compose a language and a ghostly legion of quasi-structures whose
members are sometimes temporarily entertained.
References
Tanenhaus, M. K., Leiman, J. M., & Seidenberg, M. S. (1979).
Evidence for multiple stages in the processing of ambiguous words
in syntactic contexts.
Journal of Verbal Learning and Verbal Behavior,
18, 427-440.
Tanenhaus, M. K., & Trueswell, J. C. (1995). Sentence
comprehension. In J. L. Miller & P. D. Eimas (Eds.),
Handbook of Perception and Cognition: Volume 11,
pp. 217-262. Academic Press, San Diego.
Elman, J. L. (1990). Finding structure in time.
Cognitive Science,
14, 179-211.
|
