Word-frequency and phonemic similarity in a word avoidance task ☆

Abstract An experiment is reported where subjects were required to describe picture postcards while trying to avoid words beginning with one of the consonants /ptkbd or g/. There are differences in the strategies used by different subjects, some subjects making more errors (failures to avoid forbidden words) with high frequency words, while other subjects make more errors with low frequency words. Systematic changes were observed in the frequency of classes of words other than the class of words being avoided, and these changes were consistent with the hypothesis that words beginning with a consonant that was similar to the forbidden consonant were used relatively less often than words beginning with a dissimilar consonant. However, the similarity between consonants was different for high and low frequency words. Low-frequency words showed a pattern of similarities similar to ‘acoustic’ confusions in perception, and it is suggested that such words are encoded phonemically by the subject; on the other hand high-frequency words showed a pattern of similarities which had a large articulatory component, and it was suggested that such words are encoded as articulatory events in terms of units larger than a phoneme.