Socially-mediated linguistic convergence can drive perception of social proximity

Structural priming – the tendency to re-use syntactic forms after exposure to those forms – fits into a broader pattern of convergence between interlocutors at various linguistic levels. While sentence-level convergence is often explained in terms of cognitive mechanisms like implicit learning, recent work suggests that it can function to manage social distance with an interlocutor, as has been demonstrated for phonetic accommodation. Two experiments are presented that show that structural convergence is mediated by a speaker’s perception of their social proximity to their interlocutor, and that these perceptions can be shifted by convergence that occurs over the course of a conversation.

Evidence from dialogue studies show that interlocutors tend to converge with each other’s language usage patterns at a number of linguistic levels. Phonetic imitation studies have shown that listeners adapt their speech to be more similar to that of speakers they have prior exposure to [1-3]. Interlocutors also converge at on a set of expressions to refer to items in the linguistic environment (lexical entrainment) – a phenomenon often explained in terms of referential pacts formed between interlocutors [4-5]. At the sentential level, interlocutors tend to converge on the same syntactic forms [6-8]. While these varieties of convergence look alike superficially, they have received different kinds of explanations.

While phonetic adaptation has long been conceptualised as a tool for social distance management [9-11], syntactic-level adaptation is typically seen as a form of priming, driven by low-level cognitive mechanisms like implicit learning [12]. However, recent work has shown that structural convergence is sensitive to speakers’ perceptions of interlocutor characteristics, including whether they are a native speaker, and how similar they are judged to be to the speaker on a number of socio-cultural dimensions [13]. We use structural priming as a measure of speakers’ convergence with their interlocutor to address questions raised by prior research: whether native-speaker status and perception of social proximity have independent effects on convergence (Experiment 1), and whether convergence over the course of an experimental session shift speakers’ perceptions about their proximity to their interlocutors (Experiment 2).

Dialogue game. Participants played a picture-matching game involving taking turns with another speaker (a confederate) to describe scenes depicting ditransitive events (e.g. Ann showed Mia the painting) using the verb provided. The participant and confederate communicated by voice only using web-based videoconferencing. The verbs varied by whether they participated in the dative alternation: half were acceptable in either double object (DO) or prepositional dative (PD) form, while the other half were anomalous in DO form (Ann described Mia the monument). Bias toward the PD form (PD-bias) was measured in a norming study. Critically, the confederate always used DO sentences, resulting in many anomalous-sounding descriptions. The outcome measure was the structure produced by the participant: while structural priming predicts increased production of DO sentences, both experiments asked whether the penalty for using a DO form with increasingly PD-biased verbs was mediated by a participant’s perception of their interlocutor’s attributes.

Experiment 1 asked whether individuals’ calculations of their interlocutor’s competence relative to their own predicts convergence. Participants were high-proficiency L2 English speakers (L1 Spanish), and were paired with a confederate who was a native English speaker, another L1-Spanish speaker, or an L1-Slovak speaker, all speaking English. Participants completed a pre-test LEAP-Questionnaire [14], which included self-assessments of their English proficiency (e.g. how often they are identified as a non-native speaker). These scores were included in the analysis, along with post-test judgments about interlocutor attributes: agreement with statements like “The other person was easy to understand” (Easy-to-understand), “The other person sounded similar to me” (Sounds similar), “The other person and I have similar backgrounds (family, education, etc.)” (Similar backgrounds), “The other person and I have similar interests” (Similar interests) and “The other person was a native speaker of English” (Nativeness).

Responses were fitted with mixed-effects regression models predicting DO responses, with PD-bias, Nativeness, Similar interests, Trial type (description, response), Trial order, and two-way interactions included as predictors (post-test variables that would have resulted in collinearity were excluded). Fixed effects were removed from the model using stepwise model comparison, and the maximal random effects structure was included that was supported by the data.

The results revealed an interaction between Nativeness and PD-bias (=0.23, p<0.05): as the interlocutor was perceived as increasingly native-like, the penalty associated with stronger PD-bias weakened. Anomalous DO sentences using PD-biased verbs were therefore judged as less ill-formed when produced by a confederate perceived as more native-like, corroborating prior findings that Nativeness mediates convergence. There was also a main effect of Similar interests (=0.74, p<0.001), with greater convergence for participants who rated their interlocutor as likely to have similar interests to them. This social proximity effect was independent of the native speaker effect: the shared interests and nativeness variables were not correlated in the dataset (t=1.07, p> 0.2), and the fixed effects were not correlated in the regression model (all VIF<1.25).

Experiment 2 asked whether social proximity effects on convergence emerge in the absence of differences in native speaker status, and whether convergence over the course of a conversation can shift speakers’ judgements about their proximity with an interlocutor. Native British English speaker participants were paired with a confederate speaking one of two regional varieties of British English (confederates were from Cork, Ireland and Southeast England). Before playing the game, participants heard a pre-recorded clip of the confederate they were paired with, and provided ratings about the speaker’s attributes (same as in Experiment 1 post-test) based on the recorded speech. Participants’ perceptions of their geographical proximity to the speaker were measured by asking them to indicate their own hometown and where they thought the speaker’s hometown was on a map (Map distance).

These measures of inter-personal proximity were included in the analysis as predictors of convergence (excluding those that would have resulted in collinearity), with PD-bias, Trial type, Trial order, and two-way interactions except interactions between pre-test variables and Trial order. There was an interaction between Map distance and PD-bias (=-0.93, p<0.001): as a participant’s perception of their geographical proximity with their interlocutor decreased, the penalty associated with stronger PD-bias weakened. There was a similar, marginal interaction between PD-bias and Similar backgrounds (=0.33, p=0.05): as perception of shared background increased, anomalous DO sentences using strongly PD-biased verbs were less penalised.

To assess whether convergence during the session shifted participants’ judgments about their interlocutors, the pre-test questionnaire (excluding Map distance) was readministered to collect post-test judgments. Overall, agreement scores increased from pre- to post-test for Easy-to-understand (t=-2.22, p<0.05), Sounds similar (t=-2.98, p<0.01) and Similar backgrounds (t=-3.05, p<.005). The data was divided into halves by the extent of convergence observed by participant, measured by comparing DO productions in the first and last thirds of trials. High-convergence participants showed reliable increase in agreement scores for the Easy-to-understand (t=-2.43, p<0.05), Sounds similar (t=-2.53, p<.05) and Similar backgrounds (t=-2.03, p<0.05) statements. By comparison, low-convergence participants showed an increase for only Similar backgrounds (t=-2.27, p<0.05; Easy-to-understand: t=-0.96, p=0.34; Sounds similar: t=-1.64, p=0.11).

Together, these experiments demonstrate that structural convergence is mediated by a speaker’s perception of their social proximity to their interlocutor – independently of effects of the interlocutor’s native-speaker status, and that these perceptions can be shifted by convergence that occurs over the course of a conversation.

References

[1] Goldinger, S. (1998). Echoes of echoes? Shadowing words and non-words in an episodic lexicon. Psychological Review, 105, 251-279.

[2] Namy, L. L., Nygaard, L. C., & Sauerteig, D. (2002). Gender differences in vocal accommodation: The role of perception. Journal of Language and Social Psychology, 21, 422–432.

[3] Kim, M., Horton, W. S., & Bradlowm A. R. (2011). Phonetic convergence in spontaneous conversations as a function of interlocutor language distance. Laboratory Phonology, 2(1), 125-156.

[4] Brennan, S. (1996). Lexical entrainment in spontaneous dialog. Proceedings of the 1996 International Symposium on Spoken Dialogue, 96: 41–44.

[5] Brennan, S., & Clark, H. H. (1996). Conceptual Pacts and Lexical Choice in Conversation. Journal of Experimental Psychology: Learning, Memory, and Cognition, 22(6): 1482–1493.

[6] Bock, J. K. (1986). Syntactic persistence in language production. Cognitive Psychology, 18(3), 355–387.

[7] Pickering, M. J., & Branigan, H. P. (1998). The Representation of Verbs: Evidence from Syntactic Priming in Language Production. Journal of Memory and Language, 39, 633-651.

[8] Bock, J. K., & Griffin, Z. M. (2000). The persistence of structural priming: transient activation or implicit learning. Journal of Experimental Psychology: General, 129(2), 177-192.

[9] Bourhis, R. Y., & Giles, H. (1977). The language of intergroup distinctiveness. In Giles, H. (ed.), Language, Ethnicity and Intergroup Relations. Academic: London, pp. 119-135.

[10] Babel, M. (2010). Dialect divergence and convergence in New Zealand English. Language in Society, 39(4), 437-456.

[11] Babel, M. McGuire, G., Walters, S., & Nicolls, A. (2014). Novelty and social preference in phonetic accommodation. Laboratory Phonology, 5(1), 123-150.

[12] Chang, F., Dell, G. S., Bock, K., & Griffin, Z. M. (2000). Structural priming as implicit learning: A comparison of models of sentence production. Journal of Psycholinguistic Research, 29(2), 217-229.

[13] Kim, C. S., & Chamorro, G. (2021). Nativeness, social distance and structural convergence in dialogue. Language, Cognition and Neuroscience, 36(8), 984–1000.

[14] Marian, V., Blumenfeld, H. K., & Kaushanskaya, M. (2007). The Language Experience and Proficiency Questionnaire (LEAP-Q): Assessing language profiles in bilinguals and multilinguals. Journal of Speech, Language, and Hearing Research, 50(4), 940-967.