Summary: | This is a study about how one-year-old Swedish-learning infants presumably use probabilistic information, such as prosody and phonotactic regularity, in segmentation of speech. The variables studied were the Swedish tonal word accents I & II and the distributional regularities of within-word and between-word consonant clusters in Swedish infant-directed speech. The results – which were not as clear-cut as the results obtained in earlier experiments on English-learning infants – suggest that 12-month old Swedish infants might be sensitive to prosodic cues to word boundaries: in experiment 1, altering the phonotactics of the stimuli reversed the infants’ preference for word accent types. However this was not confirmed in experiment 2, instead there was a general preference for listening at the accent II words. The results also suggest that 12-month old Swedish infants might not use phonotactic cues to word boundaries to the extent as expected: in experiment 1 and 2, altering the word accent types did not reverse the infants’ preference for phonotactics. Instead, both in experiment 1 and 2, there was a general preference for listening at the within-word stimuli. When compared with earlier research these findings indicate that infants, besides being able to integrate multiple statistical cues to word boundaries, might early in life be assisted by pattern-recognition in speech segmentation. === <p>Eeva Klintfors är född Koponen.</p>
|