Japanese quail categorize /d/ across allophonic variation

Abstract

Studies using animal subjects provide a unique opportunity to assess the nature of processes for perceiving speech. Demonstrations that animals have the ability to categorize speech sounds in ways similar to humans, serve to undermine claims that some levels of speech processing either are specific to humans or carry substantial cognitive requirements. In our study, Japanese quail were taught to discriminate natural CV + /s/ syllables beginning with /d/ from those beginning with /b/ or /g/. The phoneme /d/ was chosen because of its well‐documented context‐conditioned variability. Birds received food reinforcement for pecking a lighted key during repeated presentation of /d/ syllables, but were required to refrain from pecking in order for the presentation of /b/ and /g/ syllables to be terminated. The quail were first trained to discriminate /dæs/, /das/, /dus/, and /dis/ from tokens with the same vowels following /b/ and /g/. After reaching asymptotic performance for the /æ/, /a/, /u/, and /i/ contexts, birds were tested on novel tokens with the vowels /ε/, /Λ/, /v/, and /ɪ/. They received no reinforcement for pecking to novel tokens. Quail pecked substantially more to novel /d/ tokens than to novel /b/ or /g/ tokens, suggesting that they treated the novel stimuli as similar to their non‐novel counterparts. Apparently, there exist significant acoustic/auditory commonalities among the class of /d/ allophones that may serve as the basis of category formation for both humans and animals. [Work supported by NICHD.]