All representations are not equal: Differential activation of words and phonological variants across accents

Abstract

Given the massive variation in natural speech, how listeners recognize words is a central issue for linguistic theory. Listener sensitivity to acoustic fluctuations in speech has provided us with one piece of the puzzle: detailed, specific representations. We provide another piece: differential activation of these representations. We conducted a semantic priming with General American listeners of positively viewed non-rhotic British English accents and negatively viewed non-rhotic New York City accents (slender'THIN; slend-UH(BE)'THIN; slend-UH(NYC)'THIN). Controlling participant experience and self-reported familiarity with both accents, only the words produced in a British accent primed semantically related targets. There is no a priori reason to expect this pattern, as the phonological variant and lexical items are controlled. We suggest that two variants equally experienced in number (raw frequency) are perceived differentially because positive and negative attitudes influence the activation lexical representations. Additionally, by examining particular participant groups from non-rhotic regions, we show an oscillation between effects of raw frequency and listener attitudes. This work (1) provides a broader understanding of factors influencing activation, (2) increases our understanding of how frequency effects are modulated by other factors (subjective listener attitudes), and (3) illuminates the interactive nature of linguistic and social factors in speech perception.