Report

Teaser, summary, work performed and final results

Periodic Reporting for period 2 - ELC (The evolution of linguistic complexity)

Teaser

Human language is unique among the communication systems of the natural world, providing our species with an incredibly flexible and powerful open-ended system of communication. The expressive potential of language undoubtedly contributes to the human capacity to build on the...

Summary

Human language is unique among the communication systems of the natural world, providing our species with an incredibly flexible and powerful open-ended system of communication. The expressive potential of language undoubtedly contributes to the human capacity to build on the achievements of our peers and ancestors, producing the capacity for cumulative culture that defines our species, and which has enabled us to spread across the planet and beyond. The communication systems of non-human animals are far more limited, both in their expressive potential and in the structural devices they deploy to facilitate communication; while there are important homologies and analogies between the human capacity for complex language and the communication systems of non-humans, human language seems to be qualitatively different. How did this state of affairs come to pass?

Humans learn the language of our speech community based on observing their utterances and the way in which they are used, inferring the underlying system of rules which govern how linguistic forms are constructed, the meanings those complex utterances convey, and how form and meaning are related. Consequently, language undergoes cultural evolution: it changes over time as a result of pressures applied during learning and use. On this grant we are exploring how these processes of linguistic evolution might drive the evolution of complexity in humans languages – under what conditions is complexity favoured by the cycle of learning and use by which languages persist?

We are adopting two main approaches to this problem. Firstly, the right kind of complexity in the right place in a linguistic system might facilitate language learning: if so, complexity would be predicted to emerge as a result of the transmission of language from learner to learner over long timescales. Secondly, complex social environments might drive the evolution of expressive power and complexity in language – communication in between cognitively sophisticated individuals who reason about their interlocutors’ linguistic knowledge, world knowledge, and social status, and adjust their linguistic behaviour accordingly, might provide a rich environment in which linguistic complexity can accrue.

These ideas are of far-reaching importance for our understanding of the evolution of language and linguistic complexity, but have not yet been subjected to a concerted empirical examination: our aim in this project is to provide such an examination.

Work performed

We are doing so using two main methodologies to answer these questions. Firstly, we simulate the processes of language learning and interaction using computational models of populations of learning, interacting, communicating individuals. Secondly, we study the same processes in the lab, studying how human participants learn languages which differ in complexity and how they reason about the linguistic knowledge of their interlocutors.

Looking at the impact of complexity on language learning, we have found that factors which are widely claimed to effect the learnability of linguistic systems (including the predictive complexity of morphological paradigms and the extent to which multiple elements of semantics are loaded onto single morphemes) have at most subtle effects on learning in controlled experiments and simulation models, but adults and children show surprising differences in their propensity to use redundant cues of different types. Looking at the impact of social complexity on linguistic complexity, we have conducted a raft of experimental and modelling work showing how human participants adapt to the (inferred) linguistic knowledge of the people they speak to, and used computational models to show how the consequences of those moment-to-moment interactions filter out to reshape the language of whole populations.

Final results

In the next stage of the project we will look at increasing social complexity, studying language emergence in more complex laboratory microsocieties and investigating the emergence of more complex linguistic structures.