The University of Southampton
University of Southampton Institutional Repository
Warning ePrints Soton is experiencing an issue with some file downloads not being available. We are working hard to fix this. Please bear with us.

An investigation into the influence of population structures and dynamics on the emergence of linguistic systems through iterated learning

An investigation into the influence of population structures and dynamics on the emergence of linguistic systems through iterated learning
An investigation into the influence of population structures and dynamics on the emergence of linguistic systems through iterated learning
Human language pervades in a complex and ever-changing social milieu, and although the tendency and ability to learn languages are clearly innate, given the rate at which lexical items change, it is clear that social-cultural factors and ontogenetic development play a significant role in the way in which languages change over time. This has resulted in research concerned with human language evolution being dominated by two, umbrella-like, research questions. First, to what extent is the human language faculty the result of genetic endowment, and to what extent might it result from non-evolutionary factors such as constraints imposed by the fundamental nature of observational learning and social interaction? Second, to what extent are the observed characteristics of human language the result of evolutionary selection on language users, and to what extent are they the result of individuals shaping languages during their usage? This thesis is concerned with both questions, and focuses specifically on the role of social learning in
shaping language.

There is now a growing body of work which indicates that much of the contemporary linguistic form seen in languages around the world is the result of said languages being influenced by the population structure and social dynamics of their language communities. This, combined with emerging evidence that suggests a strong association between the origins of human language and a coincidental, and dramatic, shift in social structure, means that investigating the nature of the relationship between linguistic form and social structure has the potential to offer powerful insights into the nature of human language evolution.

This thesis explores this notion of a relationship between the structure of a language community and the linguistic structures that their language exhibits by modelling language changes as arising within the context of a social-coordination problem. In doing so, it utilises a specific form of expression/induction simulations known as iterated learning models. The key principle of these models is that the training data offered to a language learner is, itself, the result of training and learning on the part of another language user.

Four different models are presented here. The first introduces the concept of iterated learning, and explores how compositional languages emerge in a population of language users. The second adopts the principles of Roth-Erev reinforcement learning to look at the evolution of term-based languages; again, in a population of language users. The third, uses both the iterated learning framework and the principles of Roth-Erev reinforcement learning in order to explore the nature of linguistic change in a situation whereby agents create their own signals and syntactic rules while their population size is in a state of flux. The final model is adapted from the third, and explores the emergence of contact languages that tend to arise when independent language communities interact.

All four models demonstrate that the structure and make-up of a population influences the dynamics of language change over generational time. Specifically, it is shown that, by increasing the number of trainers from which an agent learns, the agent in question tends to learn a more expressive and stable language at a much faster rate, and with less training data. It is also shown that, so long as the number of mature agents is large enough, this finding holds, even if a learner's trainers include other agents that do not yet possess full linguistic competence.

Importantly, the findings presented here demonstrate that it is not population size per se that dictates how long, if at all, a fully expressive and stable linguistic system takes to emerge. Rather, it is how proportionally interconnected a given agent is to other agents in the social group that dictates the success of said population's language.

In addition, the final model, which looks at the nature of pidgin and creole language emergence, presents two key findings. First, and in contrast to the common claim within the pidgin and creole literature, social power need not play a key role in pidgin emergence. Here, the pidginisation process needed to be a bilateral process, with both parties contributing to the subsequent pidgin in order for a successful contact language to exist between the two different populations. Secondly, this model looked at the concept of tertiary hybridisation; the belief that a pidgin will have to be used as the lingua franca between two groups who do not possess a common language, and whose speakers are not native speakers of the original target language. The data from these model runs indicated that, when two groups without any common language come together, tertiary hybridisation is necessary in order for a creole to emerge; otherwise, the resulting language is an entirely new linguistic system.

In summary, the results of these models demonstrate that the evolution of language does indeed have an intimate relationship with population structure and social dynamics. In that linguistic variations and systems become more stable in situations where language users have a higher level of interconnectivity with the rest of the population. The reason for this is shown to be due to the way in which languages themselves evolve in response to individual learner biases so as to become easier to learn. In other words, as language users learn the linguistic system of their particular social group, the language is essentially exposed to a refinement process as it is past on from one generation to the next. Furthermore, although it has been argued that, in order for a language to be learnable, its structure has to adhere to certain constraints placed upon its structure, and that any language that violates such a 'linguistic blueprint' would not exist because it would be unlearnable, the findings presented here demonstrate that this refinement process is highly efficient at producing similar results; even when input is highly variable and inconsistent.
University of Southampton
Brace, Lewys G.
1d2d1e1f-8e0f-4e24-9f6b-359acde75bab
Brace, Lewys G.
1d2d1e1f-8e0f-4e24-9f6b-359acde75bab
Bullock, Seth
2ad576e4-56b8-4f31-84e0-51bd0b7a1cd3

Brace, Lewys G. (2017) An investigation into the influence of population structures and dynamics on the emergence of linguistic systems through iterated learning. University of Southampton, Doctoral Thesis, 197pp.

Record type: Thesis (Doctoral)

Abstract

Human language pervades in a complex and ever-changing social milieu, and although the tendency and ability to learn languages are clearly innate, given the rate at which lexical items change, it is clear that social-cultural factors and ontogenetic development play a significant role in the way in which languages change over time. This has resulted in research concerned with human language evolution being dominated by two, umbrella-like, research questions. First, to what extent is the human language faculty the result of genetic endowment, and to what extent might it result from non-evolutionary factors such as constraints imposed by the fundamental nature of observational learning and social interaction? Second, to what extent are the observed characteristics of human language the result of evolutionary selection on language users, and to what extent are they the result of individuals shaping languages during their usage? This thesis is concerned with both questions, and focuses specifically on the role of social learning in
shaping language.

There is now a growing body of work which indicates that much of the contemporary linguistic form seen in languages around the world is the result of said languages being influenced by the population structure and social dynamics of their language communities. This, combined with emerging evidence that suggests a strong association between the origins of human language and a coincidental, and dramatic, shift in social structure, means that investigating the nature of the relationship between linguistic form and social structure has the potential to offer powerful insights into the nature of human language evolution.

This thesis explores this notion of a relationship between the structure of a language community and the linguistic structures that their language exhibits by modelling language changes as arising within the context of a social-coordination problem. In doing so, it utilises a specific form of expression/induction simulations known as iterated learning models. The key principle of these models is that the training data offered to a language learner is, itself, the result of training and learning on the part of another language user.

Four different models are presented here. The first introduces the concept of iterated learning, and explores how compositional languages emerge in a population of language users. The second adopts the principles of Roth-Erev reinforcement learning to look at the evolution of term-based languages; again, in a population of language users. The third, uses both the iterated learning framework and the principles of Roth-Erev reinforcement learning in order to explore the nature of linguistic change in a situation whereby agents create their own signals and syntactic rules while their population size is in a state of flux. The final model is adapted from the third, and explores the emergence of contact languages that tend to arise when independent language communities interact.

All four models demonstrate that the structure and make-up of a population influences the dynamics of language change over generational time. Specifically, it is shown that, by increasing the number of trainers from which an agent learns, the agent in question tends to learn a more expressive and stable language at a much faster rate, and with less training data. It is also shown that, so long as the number of mature agents is large enough, this finding holds, even if a learner's trainers include other agents that do not yet possess full linguistic competence.

Importantly, the findings presented here demonstrate that it is not population size per se that dictates how long, if at all, a fully expressive and stable linguistic system takes to emerge. Rather, it is how proportionally interconnected a given agent is to other agents in the social group that dictates the success of said population's language.

In addition, the final model, which looks at the nature of pidgin and creole language emergence, presents two key findings. First, and in contrast to the common claim within the pidgin and creole literature, social power need not play a key role in pidgin emergence. Here, the pidginisation process needed to be a bilateral process, with both parties contributing to the subsequent pidgin in order for a successful contact language to exist between the two different populations. Secondly, this model looked at the concept of tertiary hybridisation; the belief that a pidgin will have to be used as the lingua franca between two groups who do not possess a common language, and whose speakers are not native speakers of the original target language. The data from these model runs indicated that, when two groups without any common language come together, tertiary hybridisation is necessary in order for a creole to emerge; otherwise, the resulting language is an entirely new linguistic system.

In summary, the results of these models demonstrate that the evolution of language does indeed have an intimate relationship with population structure and social dynamics. In that linguistic variations and systems become more stable in situations where language users have a higher level of interconnectivity with the rest of the population. The reason for this is shown to be due to the way in which languages themselves evolve in response to individual learner biases so as to become easier to learn. In other words, as language users learn the linguistic system of their particular social group, the language is essentially exposed to a refinement process as it is past on from one generation to the next. Furthermore, although it has been argued that, in order for a language to be learnable, its structure has to adhere to certain constraints placed upon its structure, and that any language that violates such a 'linguistic blueprint' would not exist because it would be unlearnable, the findings presented here demonstrate that this refinement process is highly efficient at producing similar results; even when input is highly variable and inconsistent.

Text
Final_Thesis - Version of Record
Available under License University of Southampton Thesis Licence.
Download (5MB)
Text
Final Thesis signature - Other
Restricted to Repository staff only

More information

Published date: December 2017

Identifiers

Local EPrints ID: 441853
URI: http://eprints.soton.ac.uk/id/eprint/441853
PURE UUID: f19aa13c-8a66-4c1f-91c2-8ac44556ad1d

Catalogue record

Date deposited: 30 Jun 2020 16:31
Last modified: 21 Nov 2021 20:39

Export record

Contributors

Author: Lewys G. Brace
Thesis advisor: Seth Bullock

Download statistics

Downloads from ePrints over the past year. Other digital versions may also be available to download e.g. from the publisher's website.

View more statistics

Atom RSS 1.0 RSS 2.0

Contact ePrints Soton: eprints@soton.ac.uk

ePrints Soton supports OAI 2.0 with a base URL of http://eprints.soton.ac.uk/cgi/oai2

This repository has been built using EPrints software, developed at the University of Southampton, but available to everyone to use.

We use cookies to ensure that we give you the best experience on our website. If you continue without changing your settings, we will assume that you are happy to receive cookies on the University of Southampton website.

×