Abstract
Subjects were presented with single consonants, consonant-consonant pairs and consonant-vowel pairs, and asked to produce words beginning with these letters. Three presentation methods were used: visual tachistoscopic, acoustic with the letters " spelled ", and acoustic with the letters pronounced as they would be in a word. Latency measures revealed that the acoustic " spelling " condition took longer than the other two, and consonant-vowel took longer than consonant-consonant. These results are interpreted as supporting a conception of verbal memory in which initial consonant clusters are stored as integral units.

This publication has 5 references indexed in Scilit: