On-line word processing in infants and toddlers: The role of age, vocabulary and perceptual degradation

The effects of acoustic degradation on lexical processing were investigated in children ranging from 12 to 31 months of age. Using a visual fixation technique, familiar target words (in infant-directed speech) were presented either acoustically unaltered, time compressed (50%) or low-pass filtered (1.5kHz). The children's ability to correctly identify the respective target was assessed by examining (1) accuracy, (2) peak latency (= response time) and (3) peak duration (as a measure of visual engagement). All three factors were sensitive to the acoustic manipulations; the severity of the effect was dependent on the nature of the acoustic distortion, the child's age and— more importantly—her vocabulary level. Adults tested using identical and more severe perceptual degradations mirrored the results obtained in children in their overall pattern but differed in the magnitude of the effects. Low-pass filtering had minimal effects on word recognition in adults, but the same manipulation had devastating effects on infants, with accurate looking observed only in the more sophisticated groups (vocabularies >100 words). In contrast, infants had relatively little difficulty recognizing temporally compressed stimuli, although they did show enhanced stimulus engagement for compressed compared with unaltered words (i.e. " staring longer "). Reaction time data indicate a sharp, non-linear increase in the efficiency of word recognition that coincides with the well-known acceleration in expressive vocabulary known as the " vocabulary burst ". Introduction Comprehending spoken language is a complex cognitive task that works at high speed and requires precise timing in the integration of multiple sources of information from the incoming speech stream. Adult listeners process language incrementally, rapidly merging linguistic and contextual information from the speech signal (Marslen-Wilson, 1984, 1987, 1993). A skilled listener can handle around 10-15 phonemes per second, cope easily with 140 to 180 words arriving per minute, overcome coarticulation conflicts, adjust to variabilities in talkers, continuously update the interpretation of the running speech signal, and extract meaning before the actual offset of the acoustic event Correct identification of words in a sentence context is performed around 200 ms post-stimulus onset; word recognition in isolation is performed, on average, within an additional 130 ms (Grosjean 1980, Marslen-Wilson, 1984, Tyler 1984). The astonishing efficiency and optimal use of speech processing strategies in adults may sometimes conceal the enormous precision and " teamwork " that is required to decode spoken language correctly. In fact, more taxing processing climates affecting signal clari-ty—such as background noise, simultaneous presentation of …