Causal approaches to investigating language evolution

Research context

Human language is the most complex communication system on earth. Why did it evolve only in humans? Other animals share some of our capacities: birdsong has complex syntax, vervet monkeys have 'semantic' alarm calls, and bees communicate the location of nectar, but none come close to human language. While genetics and cognition are part of the story, perspectives from anthropology and archaeology make it increasingly clear that our ancestors found themselves in critical social, economic and ecological situations which provided the selective pressures for the evolution of complex language. Some primates can learn symbolic communication systems via intense training in captivity, but their natural habitats do not provide the right selective pressures for these latent abilities to develop. Theories explaining language evolution are scattered across biology, linguistics, psychology, anthropology and archaeology. What is missing is an attempt to pull these theories together and test them systematically against each other.

Solving the mystery of language evolution is important. It changes the way we see ourselves as a species: not as a predestined pinnacle of creation, but as evolved animals shaped by our history and environment, and on a continuum with other species. Now more than ever it is important to demonstrate that researchers can work together across fields to resolve debates and tell meaningful stories that capture people's imaginations.

Aims and objectives

This project aims to develop the Causal Hypotheses in Evolutionary Linguistics Database (CHIELD, https://chield.excd.org). This uses formal tools from the field of Causal Inference to represent hypotheses as a series of causal connections (a causal graph). This clarifies theories and allows researchers to spot connections between them. Causal inference has revolutionised fields like epidemiology, and the time is ripe for its application to language evolution and the social sciences more generally. The aim is to provide a model for scientific theory building that can be applied to many other questions.

The second aim is to develop and use a "common task framework" to test competing theories of language evolution. In a pilot study, the project team tested the proposal that the use of symbolic signals emerged to help early humans build structures together (Irvine & Roberts, 2016). We ran an experiment where human participants had to build a shelter together in a virtual world (Minecraft) without using natural language. Unexpectedly, participants chose to rely on simple pointing rather than innovate a symbolic communication system. We concluded that collaborative construction would not provide a selective pressure strong enough for symbolic communication to emerge. Several theories suggest alternative scenarios such as the division of labour or collaborative hunting. A common task framework would allow us to manipulate the number of participants or the resources available while keeping everything else constant.

Objective 1: Collect and formally describe theories of how a particular linguistic ability evolved as a response to particular social, economic or ecological factors. The CHIELD database will be public and open access.
Objective 2: Develop open-source tools for a common task framework.
Objective 3: Test three case-studies from the set of identified theories through experiments using the common task framework.
Objective 4: Communicate the results to researchers.
Objective 5: Communicate the results to the general public.

Applications and benefits:
This project will provide a new way of solving one of humanity's greatest mysteries: why we evolved a capacity for language. It will create impact by engaging science fiction authors. More generally, this project will provide a model for how to develop theory in interdisciplinary research. Source code will be freely available for other fields to utilise.

1) Engaging science fiction authors

Science fiction authors are becoming increasingly interested in ideas about how complex communication systems evolved in humans and how they might evolve in other species. In particular, hard science fiction authors need to design realistic worlds and find points of scientific speculation around which to build their plots. Reciprocally, the project needs to convert its findings into an engaging story for the general public. The PI will host a free workshop for science fiction writers in Bristol and across the UK, communicating the cutting-edge of research and what remains to be answered.

The project will also host a short story competition to promote stories that involve language evolution, judged by award winning author Mary Doria Russell. The winning story will be published in a literary magazine, helping to promote the writer and the ideas behind the story.

The workshop and short story competition have been designed in collaboration with Bristol-based company Write Club, who specialise in training for writers. This collaboration will be a pilot for a potential wider collaboration between the University of Bristol and Write Club. It is hoped that this will lay the groundwork for other researchers and departments to offer similar workshops that engage writers with their research.

This project will create impact in two ways. First, it will change the way writers research their stories and represent language and evolution in their industry. Second, it will function as a pilot for how researchers in other fields can engage fiction authors in a similar way. This has the potential to change how authors do research for writing and encourage them to utilise academics in their process.

In return, the authors will produce "imaginaries": alternative ways of looking at our questions and results that may provide new insights for the researchers. They will also provide new, engaging ways of communicating the research to the public.


2) Public engagement:

People from all walks of life are fascinated by questions about what makes humans unique. Language is one of our defining abilities, and there is a huge public interest in how it evolved, as evidenced by the Royal Institution choosing it as a topic for the most recent Christmas Lectures. The experimental approaches are fun and engaging, particularly because participants get to play Minecraft (one of the most popular video games of all time with 144 million copies sold to date). There is an opportunity to communicate an important message to the public: how languages serve our communicative needs and cognitive capacities.

The PI will bring the experimental task to the Cheltenham Science Festival. The PI will utilise existing contacts to further publicise the research, including blogs, science journalists and Bristol-based YouTube channel The Yogscast, who create videos about computer games such as Minecraft.