Socially Indexed Phonological Variation and Lexical Competition in Spoken Word Recognition

Abstract

Speech perception requires listeners to map variable acoustic input onto stable words in real time. Because speech also carries socially structured variation, listeners must interpret words under conditions of uncertainty. This dissertation investigates how dialect-specific experience shapes lexical competition during spoken-word recognition, focusing on the pin–pen merger, in which /ɪ/ and /ɛ/ are neutralized before nasal consonants in many varieties of American English. Across four studies, the findings show that dialect-specific alternatives can remain active even when the signal clearly supports the intended referent. This competition was not explained by general phonological similarity alone, but instead reflected structured experience with dialect variation. More broadly, the results suggest that listeners are sensitive to the patterned variability of speech in their linguistic environment and carry those expectations into real-time recognition. Dialect-specific competition also influenced downstream decision dynamics and incidental memory, revealing that alternatives considered during recognition can leave measurable traces in later encoding. These effects varied with listeners’ own merger use and dialect background. Overall, the dissertation argues that spoken-word recognition can sustain parallel activation of socially indexed phonological mappings. These findings show that sociophonetic experience is not peripheral to recognition, but part of the process that structures lexical competition as speech unfolds.