Intelligibility of Excerpts from Fluent Speech

Résumé du livre

The intelligibility of samples excised from the fluent stream of speech improves as the duration of the sample is increased by adding succeeding words. Beforehand, the listeners were told each word, except the initial word, of an excised speech sample. The initial word was to be written by the listeners. The listeners then heard samples consisting of the initial word alone, the initial word plus its subsequent word, the initial word plus its subsequent two words, etc. If the subsequent words simply serve to provide structural context, intelligibility should not improve with the addition of subsequent words to the heard sample, since the subsequent words were known beforehand. However, continued improvements are observed with the longer samples, even with known structural context. The advantage of longer samples was assumed, therefore, to result from the contribution of auditory context. This result is interpreted to favor a theory of speech perception based upon a set of contextual factors operating over communication units longer than a single word. (Author).