Towards automated detection of similarities and differences in bilingual speakers

Bilingualism is the norm rather than the exception in the U.S. Within speech-language pathology this change in linguistic demography has prompted the increased need for information on bilingual speech sound production patterns to reduce misdiagnosis of speech sound disorders in bilingual children. Speakers of Jamaican Creole (JC) and English are an underserved, understudied population, yet they represent the third largest Caribbean-born immigrant group in the U.S. This study employs perceptual-based and acoustic-based measures to model bilingual speech sound production in bilingual children. Eight JC-English bilingual children were recorded speaking eleven words three times in each language. Words were transcribed and similarities/differences analyzed by: (1) Percent Consonants Correct (2) Phonological Mean Length of Utterance; and (3) phonotactic structure. Acoustic landmark detection was used to characterizing similarities and differences in speech production using acoustic-based measures. Analysis revealed that if PCC indicates JC and English words are different, the mismatch between landmark tends to be lower (higher production accuracy); if pMLU indicates word differences, mismatch tends to be higher (lower accuracy); and if phonotactics indicate JC and English words are different, mismatch tends to be slightly higher (lower accuracy). Our findings support traditional linguistic expectations regarding bilingual speakers’ similarities and differences.