Language Science Meets Cognitive Science: Categorization and Adaptation

Abstract

Questions of domain-generality—the extent to which multiple cognitive functions are represented and processed in the same manner—are common topics of discussion in cognitive science, particularly within the realm of language. In the present dissertation, I examine the domain-specificity of two processes in speech perception: category learning and rate adaptation. With regard to category learning, I probed the acquisition of categories of German fricatives by English and German native speakers, finding a bias in both groups towards quicker acquisition of non-disjunctive categories than their disjunctive counterparts. However, a study using an analogous continuum of non-speech sounds, in this case spectrally-rotated musical instrument sounds, did not show such a bias, suggesting that at least some attributes of the phonetic category learning process are unique to speech. For rate adaptation, meanwhile, I first report a study examining rate adaptation in Modern Standard Arabic (MSA), where consonant length is a contrastive part of the phonology; that is, where words can be distinguished from one another by the length of the consonants that make them up. I found that changing the rate of the beginning of a sentence can lead a consonant towards the end of the sentence to change in its perceived duration; a short consonant can sound like a long one, and a long consonant can sound like a short one. An analogous experiment examined rate adaptation in event segmentation, where adaptation-like effects had not previously been explored, using recordings of an actor interacting with a touchscreen. I found that the perception of actions can also be affected by the rate of previously-occurring actions. Listeners adapt to the rate at the beginning of a series of actions when deciding what they saw last in that series of actions. This suggests that rate adaptation follows similar lines across both domains. All told, this dissertation leads to a picture of domain-specificity in which both domain-general and domain-specific processes can operate, with domain-specific processes can help scaffold the use of domain-general processing.