Tuuli Morrill Adams, New York University
Listeners use prosodic cues to facilitate lexical access when listening to fluent speech in their native language. This study investigates second language learners' ability to segment words from continuous speech, and the effect of native language prosodic structure on the perception of word boundaries in the second language. In an experiment conducted with natural language stimuli, English speakers learned words and then listened to fluent speech in a previously unfamiliar language (Finnish). After listening to fluent speech, they chose between pairs of correctly segmented real words and incorrectly segmented non-words, to identify possible words of Finnish. Results show that English speakers do exhibit a bias towards identifying words with first-syllable stress as real, likely an effect of a native language segmentation strategy. However, the test group that learned words and then listened to fluent speech performed better than the group that did not listen to fluent speech. This suggests that learning words and hearing them in context aids in second language speech segmentation. More successful speech segmentation, in turn, promotes the learning of other phonological patterns, which makes learners more accurate at identifying possible words in the second language.