Phonetic and lexical processing in a second language

Speech comprehension is more difficult in a second language than in one's native language. This dissertation examines the recognition of speech sounds and the recognition of words in a second language. First, Dutch and English listeners' perception of the British English vowels /ae/-/E/, was investigated, a contrast which does not occur in Dutch. Further, perception of English /z/-/s/, /v/-/f/, /b/-/p/, and /d/-/t/ in word-final position was investigated. These contrasts are similar to Dutch contrasts, but they do not occur word-finally in Dutch. Dutch listeners found the vowel contrast difficult to distinguish. Although they did not use vowel duration as a cue for the word-final voicing contrasts in a native-like manner, they categorized the contrasts as accurately as English listeners did. Thus, phonetic processing does not have to be native-like for nonnative listeners to recognize sounds as accurately as native listeners do. Second, the recognition of words containing these contrasts was investigated. Listeners were presented with 'near-words', differing from real words in one phoneme, e.g., 'daf' ('deaf'), 'lemp' ('lamp') and 'glite' ('glide'). Near-words were taken from an embedding context, e.g., 'DAFfodil' for 'daf', 'eviL EMPire' for 'lemp', and 'biG LIGHT' for 'glite'. Near-words caused more lexical activation for Dutch listeners than for English listeners. For Dutch listeners, near-words often caused as much lexical activation as real words did. Further, the recognition of words with partially overlapping onsets (e.g., 'DAFFOdil'-'DEFIcit'), and of minimal pairs (e.g., 'flash'-'flesh', 'robe'-'rope') was investigated. Hearing one word of these pairs caused more lexical activation of the other member of the pair for Dutch listeners than for English listeners. Thus, lexical processing was less efficient in nonnative listening: there was increased lexical activation for nonnative listeners as compared to native listeners, and this increased activation was not restricted to stimuli that were perceptually ambiguous for the nonnative listeners