From Idiosyncrasy to Generalization: The Role of Sociolinguistic Experience on Adaptation to Novel Talkers

Read full text Summarise with VIRAT
Authors: Nicholas B. Aoki
Cited by: 0
References: 204

Summary

No summary yet. VIRAT can read this paper and write a plain-language summary of its aim, methods, key findings and limitations.

Summarise with VIRAT Free to try with your Science Catalogs account.

Abstract

Cognitive scientists have long been interested in cross-talker generalization. When listeners are exposed to phonetic variation, when do they apply what they have learned to novel talkers? A debate about generalization has emerged between two ostensibly opposing theories. Some studies find that exposure to multiple talkers is necessary for generalization (a "numerosity account"); others claim that acoustic similarity between the exposure and novel talkers is what drives generalization, not multiple-talker exposure (a "sufficient similarity account"). But are numerosity and sufficient similarity as different as they seem? This dissertation makes the following insight – multiple-talker exposure only ever affects generalization when the exposure and novel talkers are acoustically similar. This suggests that, rather than being in competition, numerosity and sufficient similarity only differ in the amount of evidence required for generalization. Numerosity relies on two types of evidence (acoustic similarity and multiple-talker exposure), while sufficient similarity relies on just one (acoustic similarity). This observation leads to a novel "linguistic experience hypothesis" – when listeners possess less experience with variation, more evidence is needed to prove that such variation is truly generalizable and not just an idiosyncrasy (i.e., a numerosity account should be observed in situations of lower listener familiarity, and vice versa for sufficient similarity). To investigate the linguistic experience hypothesis, this dissertation presents two experiments as a proof of concept. Experiment 1 focused on a test case of relatively high listener familiarity (L1-English listeners were exposed to L2-accented English). Experiment 2 then presented a test case that was specifically designed to be unattested in L1-English (a gendered /p/ to [b] phonetic shift produced by text-to-speech voices). The results were consistent with the linguistic experience hypothesis – less evidence was required for generalization in Experiment 1 (just acoustic similarity); meanwhile, more evidence was needed for generalization in Experiment 2 (both acoustic similarity and exposure to multiple talkers). The linguistic experience hypothesis makes specific, testable predictions about how social context mediates adaptation to novel talkers, with implications for typology, theories of sound change, and more. In a broader sense, this dissertation highlights the theoretical and practical value of integrating a deeper sociolinguistic perspective into cognitive science.

Fields of Study & Subjects

Phonetics and Phonology Research Linguistic Variation and Morphology Language Development and Disorders

Publication details

Journal / source
eScholarship (California Digital Library)
Publication date
1 January 2026
Language
EN
Document type
Thesis