EVIDENCE FOR MULTIPLE STAGES IN THE PROCESSING OF AMBIGUOUS WORDS IN SYNTACTIC CONTEXTS

A variable time delay naming latency paradigm was used to investigate the processing of noun-verb lexical ambiguities (e.g., watch) in syntactic contexts which either biased the noun or verb reading (e.g., I bought the watch; I will watch). Target words related to either the noun or verb reading were presented at 0, 200, and 600 msec following the sentence-final ambiguous word. At 0 msec, naming latencies related to either reading were facilitated regardless of the biasing context. By 200 msec, facilitation obtained only for targets related to the reading of the ambiguous word biased by the context. The results support a two-stage model in which all readings of ambiguous words are initially accessed and then the inappropriate readings are rapidly suppressed. © 1979 Academic Press, Inc.