Phonetic effects of the ambient language in early speech: Comparisons of monolingual- and bilingual-learning children

Date of Award




Degree Name

Doctor of Philosophy (Ph.D.)



First Committee Member

D. Kimbrough Oller - Committee Chair


To evaluate the phonetic effect of the ambient language on young children already speaking in short sentences, the present study assessed early utterances of thirty 24--30 month-old English-, Spanish-, and English/Spanish-learning children. The utterances were digitized, stripped of all potentially biasing context, presented in random order, one utterance at a time, and judged for lexical content and language spoken by ten English-Spanish bilingual adults. This study eliminates potential effects of context that have been present in many prior studies of phonological adaptation.All of the children's utterances selected as stimuli for the present work were intelligible to parents and/or examiners at the time they were produced and to the author when she listened to and transcribed the recorded play sessions. Two-thirds of the stimuli were single words and the remainder were short phrases (e.g. "There's a duckie in there.")Fewer than 25% of those same utterances were given correct lexical identifications by the experimentally blinded listeners who participated in the study. Of the remaining utterances, namely those with unidentified lexical targets, about 60% were assigned to the right language; that is, listeners were able reliably to discern the language of utterances that were unintelligible to them. Although the language identification was above chance levels across groups, the results included a surprisingly large proportion of errors. Correct identifications were made slightly more often in judgments of the monolingual children. For unintelligible utterances (those whose targets were not identified), the presence of a single language-specific segment (a segment that occurs in well-formed Spanish or English but not in both) was sufficient to act as a cue for language identification above chance levels, but language-specific elements were inconsistently present in children's utterances, whether intelligible or unintelligible. About one fourth of all utterances had no language-specific elements present. This was true even for some utterances whose targets were identified by a majority of listeners. The best cue for English was the presence of non-Spanish word-final consonants and for Spanish was the presence of final point vowels in isolation.These findings suggest that the majority of what listeners understand from children---perhaps as much as 75%---is contributed by context, since fewer than 25% of utterances that were intelligible in context could be lexically identified when contextual cues were eliminated by means of random presentation of individual utterances. In the context-free situation, only one child in this study had over 50% target intelligibility (compared to 93% for the adults in the same protocol). Only slightly more than one half of these typically-developing children gave evidence at 26 months of phonological adaptation to the ambient language sufficient for blind listeners reliably to identify the language in which utterances were spoken. The tendency for children's utterances to be correctly identified as to the language in which they were spoken occurred for the bilingual children as well as for the monolinguals, but 6 of the 7 bilinguals whose data yielded reliable language-identification showed the effect in only one of the two languages spoken. Only one child showed reliable language identifiability in both languages. (Abstract shortened by UMI.)


Language, Linguistics; Psychology, Developmental

Link to Full Text


Link to Full Text