ATTENTION AND THE ROLE OF DUAL CODES IN PHONEME MONITORING

In a series of 16 experiments, the reaction time to decide which of two stop consonants, /b/ or /p/, for example, was the initial phoneme of a single target-bearing item was measured. In half of the experiments word frequency was varied and in the other half, lexical status (words vs. nonwords) was varied. When listeners had only to monitor the target-bearing items for the initial consonant, nonsignificant effects of word frequency and lexical status were found. However, with the addition of a secondary task that focused attention on the consequences of lexical processing by requiring listeners to indicate whether the targetbearing item was a noun or verb or a word or nonword, highly reliable lexical effects were obtained: phoneme monitoring times were faster for high than for low frequency words and faster for words than for nonwords. A secondary task that required a judgment regarding item duration did not yield reliable effects of frequency or lexical status. Finally, degrading the targets with pink noise also resulted in reliable lexical effects. An explanation of the results was offered that rested on the following assumptions: (1) competition (i.e., a race) normally exists between which representation, the pre- or postlexical, will be the basis for phoneme decisions; with the prelexical code the usual winner, but that (2) attention to postlexical representations may at times override the consequences of the competition between representations. © 1990.