This page gives details of my research interests. The publications page contains a full list of my publications, including links to download the papers, if they're not here.

The emergence of language is a crucial landmark in the history of humanity, and one of its defining characteristics; the flexibility and expressivity of human language is unparallelled in the world. My research explores the evolution of language by investigating the cognitive and cultural foundations of language and the origin of linguistic complexity. It is grounded in two crucial characteristics of language:

  1. language is fundamentally inferential: the meaning of an utterance is not transferred directly from speaker to hearer, but reconstructed by the hearer;
  2. language is a flexible set of negotiated cultural conventions: the way in which language is acquired, used and transmitted impacts on its structure.
These characteristics of language have important consequences for its evolution, because the continuous reconstruction of meaning in communication both produces variation and innovations in linguistic structure, and hones the viability of constructions as they are used. Constructions whose meanings can be easily and repeatedly reconstructed are preferentially replicated over those whose meanings are difficult to reconstruct.

The cultural evolution of systematicity

Until recently, I was working on the ESRC-funded project The Emergence and Development of Structural Systematicity in Language, with Barbora Skarabela and Mónica Tamariz. We carried out experimental studies exploring the extent to which cumulative cultural evolution is a suitable explanatory mechanism for the emergence of linguistic structure, using a novel paradigm of co-operative language learning, to explore the effect of the context in which language is learnt and used. Our initial results were presented at Evolang (paper in Utrecht book). More details can be found at the project's webpage.

Cross-situational learning

Cross-situational learning is a mechanism for learning the meaning of words despite referential uncertainty (Quine's gavagai problem), by combining information across multiple exposures. Word learning is traditionally explained by socio-pragmatic, representational, interpretational and syntactic heuristics to reduce referential uncertainty (see my chapter in Language Origins for how mutual exclusivity allows successful communication even between individuals with very different conceptual structures). Richard Blythe, Kenny Smith and I have shown in a series of papers that:

  • cross-situational learning alone can lead to the acquisition of large, human-size vocabularies, and that there is no necessary link between the ability to learn individual words rapidly and the capacity to acquire a large lexicon (paper in Cognitive Science);
  • the capacity for cross-situational learning may be more limited than has previously been claimed (paper at the Cognitive Science Conference 2009);
  • people therefore use different cross-situational learning strategies, depending on the difficulty of the task they face (paper in Cognitive Science)

The origins of linguistic complexity and the nature of protolanguage

My paper in Interaction Studies discusses the nature of protolanguge in light of the constraint of semantic reconstructibility. My forthcoming paper in New Directions in Cognitive Linguistics outlines my thoughts on the cognitive origins of linguistic complexity.

Ostensive-inferential communication and grammaticalisation

This Studies in Language paper, written with Stefan Hoefler, describes how both metaphor- and reanalysis-based approaches to grammaticalisation have the same foundations: ostensive-inferential communication and the memorisation of linguistic usage. These mechanisms are not language-specific, but are much more basic, domain-general properties. We also outline how these same properties could also account for the origin of symbolic conventions themselves. My article on grammaticalisation and the evolution of language, for the Handbook of Grammaticalization, will be available here soon.

Meaning inference

My earlier work used computational models to investigate the explanatory power of meaning inference (Adaptive Behavior paper), and how individuals can develop shared communication systems when meaning must be inferred (Artificial Life paper). This ECAL 2003 paper shows how variation in conceptual structure leads to a cycle of innovation and semantic generalisation, and this ECAL 2005 paper shows how individual meaning creation and imperfect learning results in rapid language change across generations, while the language remains communicatively viable within each generation.