The recognition of words after their acoustic offset: Evidence and implications

It was found that more than half of the monosyllabic words were isolated after their offset, and that all these words, with one exception, received a perfect confidence rating during the next word or words.

Abstract

The aim of this study is to show that continuous spoken word recognition cannot always be a strictly sequential, left-to-right, word-by-word process. The gating paradigm was used to present monosyllabic and polysyllabic words in context and to follow the responses given to them through the next three words of the sentence. The isolation points and total acceptance points were calculated with respect to the acoustic offsets of the words. It was found that more than half of the monosyllabic words were isolatedafter their offset, and thatall these words, with one exception, received a perfect confidence rating during the next word or words. Reasons for this delay in recognition are examined, and the implications for sequential, left-to-right models of word recognition are discussed.