Levels of processing in speech production.

Abstract

When subjects repeated short alphabetic sequences at experimenter-controlled rates while retaining consonant trigrams, subsequent recall was unrelated to speech rate if single letters or a long sequence was repeated, but recall was inversely related to speech rate for sequences three or five letters in length. The latter effect was predicted by a model in which sequences were organized as single response units in a central phonetic processor but independently executed by means of mechanisms not requiring the services of the central processor. Quantitative predictions were generated from this model by estimating organization time for the sequences and assuming that the probability of forgetting would be equal to the organization time/execution time ratio. All obtained recall probabilities were as predicted for sequences three of five letters in length with both experimenter- and subject-controlled speech rates. Repetition of a nine-letter sequence from the end of the alphabet produced forgetting patterns clearly out of accord with any organization time/execution time ratio, but a similar sequence from the beginning of the alphabet produced evidence of being a marginal case with unstable organization.