The TRACE model of speech perception