Place-Based Accentedness Ratings Do Not Predict Sensitivity to Regional Features

Abstract

Discussions of sociolinguistic awareness are often about how patterns observed in one practice (often linguistic production) appear in others (often person perception or metalinguistic commentary). Models like Labov's indicator/marker/stereotype trichotomy force this complexity into a single dimension, due to presupposing a conscious/unconscious distinction unsupported in current cognitive psychology. A more effective approach takes a theoretical step back, asking basic questions about how analogous sociolinguistic meanings relate across activities. In this article, I do so by asking whether explicit verbal reports and speaker evaluations of accentedness in Ohio correlate in strength across individual language users. Such a correlation would suggest a shared representation and/or a shared learning process. A total of 1106 participants listened to Ohio talkers reading word lists of trap, dress, lot, or goose tokens. Participants rated each talker's accentedness, then the accentedness of seven Ohio places. The expected main effects emerged: southern and rural Ohio were most accented, then northern Ohio, and lastly cities and central Ohio. Likewise, the acoustic features influenced talker ratings. Crucially, however, these two effects largely did not interact: those most likely to describe northern (southern) Ohio as accented were no more or less sensitive to northern (southern) vowel features. These results support the small but growing evidence that indexical relationships are learned and used independently across linguistic practices. They also move us further from a unidimensional model of awareness toward an approach where different systems are treated independently.