This study looks for support for the suggestion in the Speech Learning Model that the phonetic category established in childhood for a first language (L1) sound may evolve gradually if it is linked perceptually to a second language (L2) sound. Assessed here is voice onset time (VOT) in two groups of five bilingual (English and Japanese) speakers, aged 35-60, one group whose L1 is English and the other whose is Japanese. All reported no significant L2 exposure before age 15, and they had been working in the L2 regularly for 15 to 35 years. Both groups read the same list of English and Japanese words containing word initial /p/, /t/, or /k/. In all 880 tokens (10 speakers multiplied by 2 languages multiplied by 44 tokens per language) were examined. Results found that, for both groups, the L2 VOT mean values were positioned between the normal L1 Japanese and L1 English VOT values. No support was found for the suggestion that speakers' L1 VOT values evolve over time. The L1 values appear to have stayed the same. (Contains 2 tables and 21 references.) (KFT)