Automatic And Social Effects on Accommodation in Monolingual and Bilingual Speech

This dissertation investigates the automatic and social mechanisms underlying accommodation, and how these mechanisms influence the time-course of accommodation.

In particular, I examine whether accommodation occurs for reasons related to social factors (e.g., affiliation) or whether accommodation occurs automatically (e.g. recency, novelty)—and how these automatic and social factors influence accommodation at various points both within an interaction and after.

The social dimensions of accommodation are addressed by examining accommodation in monolingual and bilingual speech. Specifically, I test whether monolingual and/or bilingual participants converge more with either a monolingual or bilingual model talker, and whether a participant’s speech community influences whether they accommodate to a model talker.

To investigate these questions, participants completed a referential communication task with two pre-recorded model talkers: an English monolingual model talker and a Spanish-English bilingual model talker. The participants themselves were either English monolinguals or Spanish-English bilinguals, from either a majority monolingual community (Ithaca, NY: 7.1% Hispanic/Latinx) or a majority bilingual community (Miami, FL: 68.6% Hispanic/Latinx) (U.S. Census Bureau, 2017b, 2017a).

Thus, there were four participant groups: English monolinguals from Ithaca (M-Ithaca), Spanish English bilinguals from Ithaca (B-Ithaca), English monolinguals from Miami (M-Miami), and Spanish-English bilinguals from Miami (B-Miami). In order to address the time-course of accommodation, each participant interacted with each model talker one-at-a-time, and those interactions were divided into four blocks. Changes in accommodation were then examined by block.

During the experimental task, model talkers asked participants about words on a game board. The words on the boards appeared in pairs, containing both a prime word and a target word. The target words included the dependent variable, Voice Onset Time (VOT) in voiceless stops in English. VOT was selected because it differs in English and Spanish: English has long-lag VOTs at the beginning of a stressed syllable, where Spanish has short-lag VOTs.  Half of the prime words contained the dependent variable, creating a priming condition, and half of the prime words did not, creating a non-priming condition.

Using these methods, two experiments were conducted.

Experiment 1 addressed the following two questions: (1) Is accommodation automatic and/or socially-modulated? (2) Does accommodation increase during a priming condition? If so, does priming interact with automatic and social factors?

Experiment 2 addressed the following two questions: (1) What is the time-course of accommodation? Specifically, how quickly do participants accommodate to a model talker, and how long does accommodation last within an interaction? (2) Will the most-recent, previous interaction influence accommodation during the following interaction?

The results of Experiment 1 provide evidence that accommodation is both automatic and socially-modulated. All participant groups produced longer VOTs with the monolingual model talker and shorter VOTs with the bilingual model talker, indicating that all participants automatically adjusted their speech to accommodate to both model talkers. Additionally, participants accommodated to model talkers when primed by speech that is not common in their speech community. For example, participants from the monolingual community (B-Ithaca, M-Ithaca) produced shorter VOTs when primed by the bilingual model talker.

Thus, participants automatically accommodated to speech that is novel to them. Accommodation was also influenced by social factors. Specifically, bilinguals from the monolingual community (B-Ithaca) produced the longest VOTs with the monolingual model talker, and monolinguals from the bilingual community (M-Miami) produced the shortest VOTs with the bilingual model talker.

In both cases, the model talker did not share the same linguistic background as the participant group but did share the same linguistic background as the majority population in their speech community. Thus, a need to affiliate with their speech community, as linguistic outsiders in their speech community, led B-Ithaca and M-Miami to converge more with the monolingual and bilingual model talkers, respectively.

Experiment 2 examines the time-course of social and automatic accommodation. The results provide evidence that socially-motivated accommodation is more persistent (i.e., longer-lasting) than accommodation that occurs due to automatic causes.

Specifically, automatic effects found in Experiment 1 occurred in earlier blocks of an interaction, while social effects occurred in later blocks. Also, social factors (e.g., affiliation) related to the most-recent, previous interaction with a model talker influenced participants’ accommodation to a different model talker in the following interaction. For example, participants from the bilingual community (M-Miami, B-Miami) who interacted with the bilingual model talker first produced shorter VOTs when interacting with the monolingual model talker afterward.

Thus, the time-course for socially-motivated accommodation is longer than the time-course for automatic accommodation. As such, socially-motivated accommodation is more likely to lead to long-term accommodation and, ultimately, language change.