Leveraging Social Network Data to Ground Multilingual Background Measures: The Case of General and Socially Based Language Entropy

Abstract

Recent research on multilingualism highlights the role of language diversity in modulating the cognitive capacities of communication and suggests a gap in available measures for quantifying socially realistic language experience. One questionnaire-based measure that potentially fills this gap is Language Entropy (e.g., Gullifer & Titone, 2018, 2020), which quantifies the balance between compartmentalised and integrated language use. However, an open question is whether questionnaire-based Language Entropy is a valid reflection of socially realistic language behaviours. To address this question, we grounded questionnaire-based Language Entropy using personal social network data for a linguistically diverse sample of speakers of French and English in the city of Montréal (n = 95). Specifically, we used exploratory factor analysis to characterise the factor structures resulting from questionnaire-based and social network-based Entropy. In addition, we examined the generalisability and stability of the relationship between both entropies across three bilingual groups with different social network compositions: simultaneous, English-dominant, and French-dominant. Our findings indicated that both questionnaire-based and social network-based entropies loaded onto the same factors and that the relationship between them was not affected by group differences in social network composition or by context. This suggests that questionnaire-based Language Entropy aligns well with social network-based Entropy and that this relationship is stable across different sociolinguistic realities, validating Language Entropy as a useful tool for quantifying language diversity.