The interaction of perceptual processes and ambiguous sentences

Abstract

(;cnnai pruuiplc-, of speech perception resolve <cvcral expcrimc ntal conflict-, about whether listeners interpret one or all meanings of an ambiguous sentl'nc<' We argue that during an amhiguou x clause, both meanings are processed. but immediately after the clause is over. it is rccodcd with only one meaning retained. This model resolves the apparently conflicting results of previous experiments: it also predicts that underlying structure ambiguity in incomplete clauses increases comprehension time. In complete clauses, ambiguity docs not increase relative comprehension time; it may reduce COin prehension time for ;lInbiguities whose interpretations arc perceptually distinct in those tasks where either meaning is appropriate. Two IlL'''· experiments offer preliminary confirmation of these predictions.Many sentences could mean more than one thing, but in a normal context we usually notice only one meaning. This fact prcscnts a puzzle to the studcnt of language behavior: do listeners unconsciously decide between the interprctations of an arnhiguous sentence or do they analyze the first meaning that comes to mind without analyzing the other') Some recent experimental studies have presented evidence that listeners unconsciously analyze both meanings of an ambiguous sentence, even though they arc aware of only one meaning. Other studies have support cd the opposite claim. that listeners ordinarily analyze only one meaning of an ambiguous sentence. In this paper. we review a theory of the normal mechanisms of speech processing, usc this theory to resolve the previous conflicting results, and present two experimental results which support our view of the percept ion of ambiguous sequences.Recent research has suggested three principal features of speech percept ion: (I) the clause is a primary perceptual unit. (~) within each clause direct mapping rules assign underlying structure relations among lexical items. and (3) after each clause is processed it is receded into a rclatively abst ruct form, thereby leaving immediate storage available for processing the next clause.

Discourse context and sentence perception∗

“…A great deal of research in sentence processing has demonstrated that clauses are important units (Bever, Garrett, & Hurtig, 1973;Hurtig, 1978;Townsend & Bever, 1978). The results of Experiments I and 2 suggest that clausal processing strategies can be affected by discourse context.…”

“…Bever, Garrett, and Hurtig (1973) presented subjects with complete and incomplete clauses that were structurally ambiguous (e.g., Although the solution was clear . .…”

“…As a clause is heard, the listener uses perceptual mapping rules or strategies (Bever, 1970) to develop potential representations of the clause by mapping each word onto its deep structure role. Once the clause ends, it is recoded into a more abstract form which frees limited capacity resources to process subsequent input (Bever, Garrett, & Hurtig, 1973;Fodor, Bever, & Garrett, 1974).…”

Lexical ambiguity of words used in english text

“…All but one found effects for both lexical and grammatical ambiguities (Bever et al, 1973;cf. Hogoboam & Perfetti, 1975).…”

Strategies of Discourse Comprehension

“…One suggestion that we would like to offer is that clause boundaries serve as decision points: Alternatives are explored in parallel within a clause, but when a major boundary is reached, a selection is made and the nonselected alternatives are discarded. Bever, Garrett, and Hurtig (1973) After having crossed the clause boundary, people are apparently left with only one reading of (18). point out that the reason clause boundaries are so important may simply be due to that fact that most strategies deal with constituents no larger than the clause.…”