Co‐Occurrence, Extension, and Social Salience: The Emergence of Indexicality in an Artificial Language

Abstract

We investigated the emergence of sociolinguistic indexicality using an artificial‐language‐learning paradigm. Sociolinguistic indexicality involves the association of linguistic variants with nonlinguistic social or contextual features. Any linguistic variant can acquire “constellations” of such indexical meanings, though they also exhibit an ordering, with first‐order indices associated with particular speaker groups and higher‐order indices targeting stereotypical attributes of those speakers. Much natural‐language research has been conducted on this phenomenon, but little experimental work has focused on how indexicality emerges. Here, we present three miniature artificial‐language experiments designed to break ground on this question. Results show ready formation of first‐order indexicality based on co‐occurrence alone, with higher‐order indexicality emerging as a result of extension to new speaker groups, modulated by the perceived practical importance of the indexed social feature.

Lexical and social effects on the learning and integration of inflectional morphology

“…This is consonant with findings from the phonetic convergence literature showing that people are good imitators (German et al, 2013) and their social attitudes play into what and when they imitate (Pardo et al, 2012). It also aligns with findings from artificial language learning paradigms which indicate that social salience influences the success of picking up and transmitting a linguistic pattern (Li and Roberts, 2023). It remains true that, at least on some level, processing shows very clear top-down influences (Niedzielski, 1999;Hurring et al, 2022).…”

“…One, but not the other, has clear higher-order indices in the language community. The task was similar to Li and Roberts (2023) in exposing participants to patterns with and without higher-order indexicality, to Baer-Henney et al (2015) in using nonce word distributions generated from existing lexical patterns in the participants' native language, and to Lindsay et al (2012) in testing for long-term accommodation. Overall, it was closest to Rácz et al (2020), except that it used two large-scale, productive inflection patterns (instead of the English past tense, which is fragmented and has limited productivity), tested for the role of higher-order indexicality (or sociolinguistic salience) in learning, and looked at long-term accommodation.…”

“…In a series of papers (Lai et al, 2020;Li and Roberts, 2023;Roberts, 2008;Sneller and Roberts, 2018), Roberts and colleagues used the artificial language learning paradigm (Kirby et al, 2008) to investigate the role of higher-order indexicality and the resulting, subjective social salience, in language learning in a social setting. They found that people can readily learn or even create higher-order indexicality in an artificial language context, that variants with higher-order indices are picked up and generalised more readily, that transposing a variant to a new grammatical or non-linguistic context increases its subjective salience, and that the abstraction or generalisation of the non-linguistic context itself makes it more likely that a linguistic variant acquires higher-order indexicality.…”

Large Language Models: A Historical and Sociocultural Perspective

“…Interestingly, the above advancements in contemporary linguistic anthropology and sociolinguistics have progressed more or less in parallel with notable developments in psychology and neuroscience over the past two decades, in which the situational, sociocultural, and reflective facets of human cognition have increasingly become a research focus in experimental cognitive research (Christiansen & Chater, 2015, 2022; Decety & Wheatley, 2015; Forgas et al., 2001; Kanske, 2018; Saxe & Baron‐Cohen, 2016; Tomasello, 2003). More recent work in cognitive science has also begun to utilize and evaluate theories about indexicality in particular in cognitive and behavioral experiments (Li & Roberts, 2023). Together, an interdisciplinary collaboration encompassing sociolinguistics, linguistic anthropology, psychology, neuroscience, linguistics, and other allied fields within the cognitive and social sciences spectrum is essential for offering critical theoretical and empirical insights for the current work in language‐related AI modeling.…”