Shared Context Facilitates Lexical Variation in Sign Language Emergence

Abstract

It has been suggested that social structure affects the degree of lexical variation in sign language emergence. Evidence from signing communities supports this, with smaller, more insular communities typically displaying a higher degree of lexical variation compared to larger, more dispersed and diverse communities. Though several factors have been proposed to affect the degree of variation, here we focus on how shared context, facilitating the use of iconic signs, facilitates the retention of lexical variation in language emergence. As interlocutors with the same background have similar salient features for real world concepts, shared context allows for the successful communication of iconic mappings between form and culturally salient features (i.e., the meaning specific to an individual based on their cultural context). Because in this case the culturally salient features can be retrieved from the form, there is less pressure to converge on a single form for a concept. We operationalize the relationship between lexical variation and iconic affordances using an agent-based model, studying how shared context and also population size affects the degree of lexical variation in a population of agents. Our model provides support for the relationship between shared context, population size and lexical variation, though several extensions would help improve the explanatory power of this model.

How Network Structure Shapes Languages: Disentangling the Factors Driving Variation in Communicative Agents

“…Importantly, the strong influence of the type of scenario suggests that nonstructural factors do play a very important role in shaping language variability, and that network structure cannot alone account for all the diversity present in real-world networks. For example, other extrinsic factors such as the role of a shared context (Mudd, de Vos, & de Boer, 2022) and the openness to strangers (Dale & Lupyan, 2012) were not taken into account here, but they might have additional contributions to the effect of network structure.…”

Iconicity in the evolution of language: computational models and laboratory experiments

“…We will start this final section by describing a recent computational model of the role of iconicity in emerging sign languages (Mudd, Vos, and De Boer 2022). This work is a good illustration of how agent-based models can be used to further explore the gradient and subjective nature of iconicity.…”

“…Both iterated learning and language games are widely used in language evolution experiments, and they have often been deployed to investigate language evolution computationally as well. Only relatively recently were notions of iconicity incorporated and investigated in such models (e.g., Chaabouni, Kharitonov, Lazaric, Dupoux, and Baroni 2019;Mudd, Vos, and De Boer 2022). In order for agents to adapt their behavior according to interactions they have had, they need some sort of algorithm to learn.…”

“…In order for agents to adapt their behavior according to interactions they have had, they need some sort of algorithm to learn. Learning can be implemented in multiple ways, ranging from rule-based (Mudd, Vos, and De Boer 2022) to neural network implementations (Chaabouni, Kharitonov, Lazaric, Dupoux, and Baroni 2019).…”

Distinguishing selection pressures in an evolving communication system: Evidence from color-naming in “cross signing”

“…So, there are many aspects of the shared context that rural signers can draw on, including a shared community history and language and familiarity with each other, while for cross-signers the contextual sharedness is restricted to the immediate setting and the general patterns of human communication (cf. Mudd et al, 2022). Berlin and Kay's (1969) seminal study looked at basic color terms across spoken languages.…”