For Preschoolers, Lexical Access is Purely Lexical: Neighborhood Density Effects in Child Speech Perception and the Emergent Phoneme Hypothesis

Both infants and adults are sensitive to phonotactic probability, the statistical distribution of sequences of sound below the level of the whole word. A now classic study by Jusczyk, Luce, and Charles-Luce (1994) demonstrated that 9-month-olds (but not 6-month-olds) prefer to listen to sounds patterns that occur more frequently in their native language, suggesting that sensitivity to sublexical sound patterns develops rather rapidly during the second half of the first year of life. At the other end of the spectrum, work with adults has shown that adults process speech at both the sublexical and whole-word levels. Vitevitch and Luce (1998, 1999) presented adult speakers of American English with both words and nonwords of either high or low phonotactic probability/neighborhood density (these factors were always correlated, such that no stimuli had high phonotactic probability but low neighborhood density, or vice versa). In several different perceptually-oriented tasks, adults were found to process words from dense lexical neighborhoods more slowly than words from sparse lexical neighborhoods, while nonwords with high phonotactic probability were processed more quickly than nonwords with low phonotactic probability. Vitevitch and Luce argued that the frequency of sound patterns in the stimuli was either inhibitive or facilitatory depending on whether adults were processing at the lexical or sublexical level. Real words with frequent sound patterns were responded to more slowly due to a lexical competition effect, while nonwords with frequent sound patterns were responded to more quickly due to a phonotactic facilitation effect. Lexical competition and phonotactic facilitation were argued to be operative forces at all times, with the “winning” level of processing driving the response time; in essence, if stimuli were processed more quickly as words, a lexical competition effect was observed, whereas if they were processed more quickly via their component sounds, a phonotactic facilitation effect emerged. The picture is less clear regarding the effects of phonotactic probability on speech processing in young children. Given that both infants and adults are sensitive to phonotactic probability, one might UC Berkeley Phonology Lab Annual Report (2010)