From natural to conventional word order: iconicity, simplicity and the mechanisms of linguistic evolution

Lead Research Organisation: University of Edinburgh
Department Name: Sch of Philosophy Psychology & Language


All languages have systematic structure, which allows us to communicate our thoughts. Where does this linguistic structure come from? Do the origins lie in the way humans think, or are the structures we find in languages the result of cultural processes like interaction and learning? There is increasing recognition that both individual and cultural processes play a role in the evolution of language, but it is not clear how these interact. We will provide new insights into these central questions of language evolution by conducting a series of laboratory experiments looking at the emergence of miniature artificial languages in the visual-gestural modality (artificial sign languages), comparing our results to data from naturally occurring sign languages, and extending our investigation to two other modalities: the graphical (using graphical icons) and the vocal (using iconic sounds).

In the first strand, we start from improvisation: how do people convey information when they cannot use the languages they know? In experiments where participants convey information using only their body and no speech, we have found that they do not use the word order of their native language. Instead the order depends on the meaning being conveyed. We call this meaning-dependent pattern 'natural order': it represents preferences in individuals when they improvise one-off solutions to communication problems. Real languages are not one-off solutions to communication problems but instead have been used in interaction, and transmitted through learning over many generations. We will investigate what happens to natural word order (the result of improvisation) when it is learned by new individuals, and used in interaction. Natural word order can vary from meaning to meaning, however languages typically use a conventional order systematically across meanings. We hypothesise that this is due to a learning bias favouring simple systems. Accordingly, we will conduct artificial sign language learning experiments, to investigate whether word order in a system becomes more consistent when it is learned. Moreover, it is known that when individuals interact, structural priming occurs: individuals are likely to repeat structures that have just occurred in conversation. In an improvised gesture study that implements communicative interaction, we will investigate how word order variability is affected by priming. These experiments together will uncover how learning and interaction contribute to the process that takes us from improvised utterances to a linguistic system.

Conducting laboratory experiments allows us to study learning and interaction in detail in a controlled setting. However, pairing laboratory work with an investigation of the same phenomena in a naturalistic context would broaden the empirical basis of our findings. Using the gestural modality in the lab allows us to build a bridge between experimental work and the study of existing sign languages. In the second strand of the project we will examine word order in British and German Sign Language in order to see how preferences for naturalness and simplicity are shaped by learning and interaction, in large populations, over longer periods of time.

Lastly, in a suite of innovative experiments, we will compare improvised gesture to the graphical modality (where participants manipulate iconic signs) and the vocal modality (where participants produce sounds, but not words, to convey information). Looking at improvisation in these different modalities will allow us to test the generality of our previous results from gesture experiments, and discover the affordances of different modalities with respect to iconicity (resemblance-based mappings between form and meaning). Together, our multi-modal experiments and sign language studies will provide a new way to look at the mechanisms that create structured languages.

Planned Impact

Language is unique to humans, and questions about the origins of language have always been a topic of general public interest. It is not unusual to find that members of the public have a set of intuitive beliefs about why we have language, and how it may have arisen. This represents a great opportunity to open up a conversation about the possibility of studying language origins scientifically, and perhaps test some of these beliefs.

In addition, there is a growing awareness of the societal importance of sign language, but public understanding of the scientific status of sign language is low. Our impact plans are set up in a way to address these topics, and to reach various audiences. We plan to set up a range of activities: a science fair exhibit (year 1), a visual art performance (year 2), a public talk and discussion (year 3) and a project YouTube channel that provides video diaries and explanations of the project in BSL and with subtitles. With these activities we aim to reach the general public (both deaf and hearing), and stimulate discussion about both the origins of language and the nature of sign language.

We will create a science festival exhibit in which the public takes part in a demonstration version of a silent gesture experiment. Having their own gestures recorded, and watching other participants' recordings will actively engage participants in questions about the differences between improvised gesture and conventional sign. Further, using the existing extensive contacts between the Centre of Language Evolution and the art world, we will collaborate with a visual artist to create a work of art aimed at a broad audience that illustrates mechanisms in the emergence of manual-gestural languages. Visual art is a powerful way to illustrate, for example, the difference between improvised manual movements, and manual signs that are part of a set of conventions. This will trigger a public discussion about the origin of language, as well as properties of sign languages. In the final year of the project we will organise a public talk and conversation with our project advisor Carol Padden, to present the main results of our project.

Throughout the project, we will update a YouTube channel with video diaries, presented in BSL and/or with subtitles, to maximise our potential audience to members of the Deaf community, and make the general public more familiar with fundamental properties of sign languages, and show them how improvised pantomime-like gestures differ from the signs in a sign language.


10 25 50
Title Technique for measuring similarity between gestures using 3d motion capture and dynamic time warping. 
Description We used a consumer grade 3d motion capture camera to measure movement of joints and 3d dynamic time warping in order to compare these gestures to those produced by a model. This allowed us to measure how well participants reproduced gestures that they saw. 
Type Of Material Physiological assessment or outcome measure 
Year Produced 2019 
Provided To Others? Yes  
Impact Published as journal paper in Language and Cognition. 
Description Blog post for "A year of conversation" website. 
Form Of Engagement Activity Engagement focused website, blog or social media channel
Part Of Official Scheme? No
Geographic Reach International
Primary Audience Public/other audiences
Results and Impact Blog post relating research to general issues of communication targeted at interested members of the public.
Year(s) Of Engagement Activity 2019
Description Presentation at interdisciplinary conference on collaboration between arts, humanities and science. 
Form Of Engagement Activity A talk or presentation
Part Of Official Scheme? No
Geographic Reach National
Primary Audience Professional Practitioners
Results and Impact A wide variety of artists and scientists attended the event, which opened up new avenues for collaboration between disciplines.
Year(s) Of Engagement Activity 2019