Computation of Conditional Probability Statistics by 8-Month-Old Infants

TLDR

Infants can segment continuous speech into wordlike units after brief exposure, indicating early statistical learning. The study investigates whether 8‑month‑olds compute transitional probabilities to segment words. Eight‑month‑olds listened to a 3‑minute stream of trisyllabic nonsense words and were later tested on their ability to distinguish whole words from part‑words that differed only in transitional probabilities. Infants reliably distinguished words from part‑words, demonstrating rapid segmentation of continuous speech based on transitional probabilities.

Abstract

A recent report demonstrated that 8-month-olds can segment a continuous stream of speech syllables, containing no acoustic or prosodic cues to word boundaries, into wordlike units after only 2 min of listening experience (Saffran, Aslin, & Newport, 1996). Thus, a powerful learning mechanism capable of extracting statistical information from fluent speech is available early in development. The present study extends these results by documenting the particular type of statistical computation—transitional (conditional) probability—used by infants to solve this word-segmentation task. An artificial language corpus, consisting of a continuous stream of trisyllabic nonsense words, was presented to 8-month-olds for 3 min. A postfamiliarization test compared the infants' responses to words versus part-words (trisyllabic sequences spanning word boundaries). The corpus was constructed so that test words and part-words were matched in frequency, but differed in their transitional probabilities. Infants showed reliable discrimination of words from part-words, thereby demonstrating rapid segmentation of continuous speech into words on the basis of transitional probabilities of syllable pairs.