Language Evolution and Computation Bibliography

For the complex human brain that enables us to communicate in natural language, we gathered good understandings of principles underlying language acquisition and processing, knowledge about sociocultural conditions, and insights into activity patterns in the brain. However, we were not yet able to understand the behavioural and mechanistic characteristics for natural language and how mechanisms in the brain allow to acquire and process language. In bridging the insights from behavioural psychology and neuroscience, the goal of this paper is to contribute a computational understanding of appropriate characteristics that favour language acquisition. Accordingly, we provide concepts and refinements in cognitive modelling regarding principles and mechanisms in the brain and propose a neurocognitively plausible model for embodied language acquisition from real-world interaction of a humanoid robot with its environment. In particular, the architecture consists of a continuous time recurrent neural network, where parts have different leakage characteristics and thus operate on multiple timescales for every modality and the association of the higher level nodes of all modalities into cell assemblies. The model is capable of learning language production grounded in both, temporal dynamic somatosensation and vision, and features hierarchical concept abstraction, concept decomposition, multi-modal integration, and self-organisation of latent representations. ARTICLE HISTORY Received 25 June 2016 Accepted 1 February 2017

We study the emergence of shared representations in a population of agents engaged in a supervised classification task, using a model called the classification game. We connect languages with tasks by treating the agents' classification hypothesis space as an information channel. We show that by learning through the classification game, agents can implicitly perform complexity regularisation, which improves generalisation. Improved generalisation also means that the languages that emerge are well adapted to the given task. The improved language-task fit springs from the interplay of two opposing forces: the dynamics of collective learning impose a preference for simple representations, while the intricacy of the classification task imposes a pressure towards representations that are more complex. The push-pull of these two forces results in the emergence of a shared representation that is simple but not too simple. Our agents use artificial neural networks to solve the classification tasks they face, and a simple counting algorithm to learn a language as a form-meaning mapping. We present several experiments to demonstrate that both compositional and holistic languages can emerge in our system. We also demonstrate that the agents avoid overfitting on noisy data, and can learn some very difficult tasks through interaction, which they are unable to learn individually. Further, when the agents use simple recurrent networks to solve temporal classification tasks, we see the emergence of a rudimentary grammar, which does not have to be explicitly learned.

This paper proposes a language acquisition framework that includes both intra-generational transmission among children and inter-generational transmission between adults and children. A multi-agent computational model that adopts this framework is designed to evaluate the relative roles of these forms of cultural transmission in language evolution. It is shown that intra-generational transmission helps accelerate the convergence of linguistic knowledge and introduce changes in the communal language, while inter-generational transmission helps preserve an initial language to a certain extent. Due to conventionalisation during transmission, both forms of transmission collectively achieve a dynamic equilibrium of language evolution: On short time-scales, good understandability is maintained among individuals across generations; in the long run, language change is inevitable.

A compositionality-regularity coevolution model is adopted to explore the effect of social structure on language emergence and maintenance. Based on this model, we explore language evolution in three experiments, and discuss the role of a popular agent in language evolution, the relationship between mutual understanding and social hierarchy, and the effect of inter-community communications and that of simple linguistic features on convergence of communal languages in two communities. This work embodies several important interactions during social learning, and introduces a new approach that manipulates individuals' probabilities to participate in social interactions to study the effect of social structure. We hope it will stimulate further theoretical and empirical explorations on language evolution in a social environment.

This study investigates how more advanced joint attentional mechanisms, rather than only shared attention between two agents and an object, can be implemented and how they influence the results of language games played by these agents. We present computer simulations with language games showing that adding constructs that mimic the three stages of joint attention identified in children's early development (checking attention, following attention, and directing attention) substantially increase the performance of agents in these language games. In particular, the rates of improved performance for the individual attentional mechanisms have the same ordering as that of the emergence of these mechanisms in infants' development. These results suggest that language evolution and joint attentional mechanisms have developed in a co-evolutionary way, and that the evolutionary emergence of the individual attentional mechanisms is ordered just like their developmental emergence.

Learning the meanings of words requires coping with referential uncertainty - a learner hearing a novel word cannot be sure which aspects or properties of the referred object or event comprise the meaning of the word. Data from developmental psychology suggest that human learners grasp the important aspects of many novel words after just a few exposures, a phenomenon known as fast mapping. Traditionally, word learning is viewed as a mapping task, in which the learner has to map a set of forms onto a set of pre-existing concepts. We criticise this approach and argue instead for a flexible nature of the coupling between form and meanings as a solution to the problem of referential uncertainty. We implemented and tested the model in populations of humanoid robots that play situated language games about objects in their shared environment. Results show that the model can handle an exponential increase in uncertainty and allows scaling towards very large meaning spaces, while retaining the ability to grasp an operational meaning almost instantly for a great number of words. In addition, the model captures some aspects of the flexibility of form-meaning associations found in human languages. Meanings of words can shift between being very specific (names) and general (e.g. 'small'). We show that this specificity is biased not by the model itself but by the distribution of object properties in the world.

What kind of motivation drives child language development? This article presents a computational model and a robotic experiment to articulate the hypothesis that children discover communication as a result of exploring and playing with their environment. The considered robotic agent is intrinsically motivated towards situations in which it optimally progresses in learning. To experience optimal learning progress, it must avoid situations already familiar but also situations where nothing can be learnt. The robot is placed in an environment in which both communicating and non-communicating objects are present. As a consequence of its intrinsic motivation, the robot explores this environment in an organized manner focusing first on non-communicative activities and then discovering the learning potential of certain types of interactive behaviour. In this experiment, the agent ends up being interested by communication through vocal interactions without having a specific drive for communication.

Linguistic forms are shaped by forces operating on vastly different time scales. Some of these forces operate directly at the moment of speaking, whereas others accumulate over time in personal and social memory. Our challenge is to understand how forces with very different time scales mesh together in the current moment to determine the emergence of linguistic form.

In this paper, efforts to understand the self-organization and evolution of language from a cognitive modelling point of view are discussed. In particular, the paper focuses on efforts that use connectionist components to synthesize some of the major stages in the emergence of language and possible transitions between stages. New technical results are not introduced, but some dimensions for mapping out the research landscape are discussed.

In this paper, I discuss in which conditions a population of embodied and situated agents that have to solve problems that require co-operation might develop forms of ritualized interaction and communication. After reviewing the most relevant literature, I shall try to identify the main open research problems and the most promising research directions. More specifically, I shall discuss: (a) the type of problems, the agents' characteristics and the environmental/social conditions that might facilitate the emergence of an ability to interact and communicate; and (b) the behavioural and cognitive capabilities that are crucial for the development of forms of communication of different complexity.

Distributed co-ordination is the result of dynamical processes enabling independent agents to co-ordinate their actions without the need of a central co-ordinator. In the past few years, several computational models have illustrated the role played by such dynamics for self-organizing communication systems. In particular, it has been shown that agents could bootstrap shared convention systems based on simple local adaptation rules. Such models have played a pivotal role for our understanding of emergent language processes. However, only few formal or theoretical results have been published about such systems. Deliberately simple computational models are discussed in this paper in order to make progress in understanding the underlying dynamics responsible for distributed co-ordination and the scaling laws of such systems. In particular, the paper focuses on explaining the convergence speed of those models, a largely under-investigated issue. Conjectures obtained through empirical and qualitative studies of these simple models are compared with results of more complex simulations and discussed in relation to theoretical models formalized using Markov chains, game theory and Polya processes.

This article suggests that the parser underlying human syntax may have originally evolved to assist navigation, a claim supported by computational simulations as well as evidence from neuroscience and psychology. We discuss two independent conjectures about the way in which navigation could have supported the emergence of this aspect of the human language faculty: firstly, by promoting the development of a parser; and secondly, by possibly providing a topic of discussion to which this parser could have been applied with minimum effort. The paper summarizes our previously published experiments and provides original results in support of the evolutionary advantages this type of communication can provide, compared with other foraging strategies. Another aspect studied in the experiments is the combination and range of environmental factors that make communication beneficial, focusing on the availability and volatility of resources. We suggest that the parser evolved for navigation might initially have been limited to handling regular languages, and describe a mechanism that may have created selective pressure for a context-free parser.

This research takes grammatical constructions (sentence form-to-meaning mappings) as an alternative to abstract generative grammars in the context of understanding the emergence of language. A model of sentence processing based on this construction grammar approach is presented, and then a series of neuropsychological and neurophysiological studies are reviewed that attempt to validate the model and to establish its neurophysiological underpinnings. The resulting model is demonstrated to provide insight into a developmental and evolutionary passage from unitary idiom-like holophrases to progressively more abstract grammatical constructions. The model is then functionally validated by its insertion into a perceptually grounded system that allows spoken language interaction with a human interlocutor. The potential utility of this emergence approach in understanding language is discussed.

In this paper, we explore various adaptive factors that can influence the emergence of a communication system that benefits the receiver of signals (the hearer) but not the emitter (the speaker). Using computer simulations of a population of interacting agents whose behaviour is determined by a neural network, we show that a stable communication system does not emerge in groups of unrelated individuals because of its altruistic character. None the less, another set of simulations shows that the emergence of a language that confers an advantage only to hearers, not to speakers, is possible under at least three conditions: (1) if the hearer and the speaker tend to share the same genes, as predicted by kin selection theory; (2) if the population is `docile' and the communication system is culturally transmitted together with other adaptive behaviours, as predicted by Simon's docility theory; and (3) if the linguistic system is used not only for social communication, but also for talking to oneself, in particular as an aid to memory.

This paper shows how a society of agents can self-organize a shared vocalization system that is discrete, combinatorial and has a form of primitive phonotactics, starting from holistic inarticulate vocalizations. The originality of the system is that: (1) it does not include any explicit pressure for communication; (2) agents do not possess capabilities of coordinated interactions, in particular they do not play language games; (3) agents possess no specific linguistic capacities; and (4) initially there exists no convention that agents can use. As a consequence, the system shows how a primitive speech code may bootstrap in the absence of a communication system between agents, i.e. before the appearance of language.

What enables an organism to perform behaviour we would call cognitive and adaptive, like language? Here, it is argued that an essential prerequisite is the ability to build up mental representations of external situations to uncouple the behaviour from direct environmental control. Such representations can be realized by building up cell assemblies. The recurrent neural network presented to cope with this task has been used for generation of action but can also be utilized as a basis for mental representations due to its attractor characteristics. In this context, a new learning algorithm (Dynamic Delta Rule) is proposed, which leads to a self-organized weight distribution yielding stable states on the one hand and which, on the other hand, only activates subpopulations of larger networks that code for the respective situation. In a second step, ways are shown of how the static information of these internal models can be transformed into time-dependent behavioural sequences.

A description is given of a `back-tracking' approach that could be used to model neural language development and language evolution. This approach aims to develop a neural model of human language capacity that incorporates important constraints on language structure, language comprehension and performance, and brain structure. The model is then to be used as a target for development or evolution. To this end, the target model would have to be simplified, so that the target model can be derived from the simplified version and learning algorithms (or structural changes). The benefit of this approach is that the development of important constraints such as the combinatorial productivity of human language is ensured. The paper illustrates the importance of including this constraint in models of language development and evolution. It then describes a neural model in which this constraint is satisfied. Finally, the paper describes how such a model could be used to investigate language development and/or evolution.

Language is about symbols, and those symbols must be grounded in the physical world. Children learn to associate language with sensorimotor experiences during their development. In light of this, we first provide a computational account of how words are mapped to their perceptually grounded meanings. Moreover, the main part of this work proposes and implements a computational model of how word learning influences the formation of object categories to which those words refer. This model simulates the bi-directional relationship between word and object category learning: (1) object categorization provides mental representations of meanings that are mapped to words to form lexical items; (2) linguistic labels help object categorization by providing additional teaching signals; and (3) these two learning processes interplay with each other and form a developmental feedback loop. Compared with the method that performs these two tasks separately, our model shows promising improvements in both word-to-world mapping and perceptual categorization, suggesting a unified view of lexical and category learning in an integrative framework. Most importantly, this work provides a cognitively plausible explanation of the mechanistic nature of early word learning and object learning from co-occurring multisensory data.

Human language is learned, symbolic and exhibits syntactic structure, a set of properties which make it unique among naturally-occurring communication systems. How did human language come to be as it is? Language is culturally transmitted and cultural processes may have played a role in shaping language. However, it has been suggested that the cultural transmission of language is constrained by some language-specific innate endowment. The primary objective of the research outlined in this paper is to investigate how such an endowment would influence the acquisition of language and the dynamics of the repeated cultural transmission of language. To this end, a new connectionist model of the cultural evolution of communication is presented. In this model an individual's innate endowment is considered to be a learning rule with an associated learning bias. The model allows manipulations to be made to this learning apparatus and the impact of such manipulations on the processes of language acquisition and language evolution to be explored. These investigations reveal that an innate endowment consisting of an ability to read the communicative intentions of others and a bias towards acquiring one-to-one mappings between meanings and signals results in the emergence, through purely cultural processes, of optimal communication. It has previously been suggested that humans possess just such an innate endowment. Properties of human language may therefore best be explained in terms of cultural evolution on an innate substrate.

Neural network models of categorical perception (compression of withincategory similarity and dilation of between-category differences) are applied to the symbol-grounding problem (of how to connect symbols with meanings) by connecting analogue sensorimotor projections to arbitrary symbolic representations via learned category-invariance detectors in a hybrid symbolic/non-symbolic system. Our nets are trained to categorize and name 50 2 50 pixel images (e.g. circles, ellipses, squares and rectangles) projected on to the receptive field of a 7 2 7 retina. They first learn to do prototype matching and then entry-level naming for the four kinds of stimuli, grounding their names directly in the input patterns via hidden-unit representations ('sensorimotor toil'). We show that a higher-level categorization (e.g. 'symmetric' versus 'asymmetric') can be learned in two very different ways: either (1) directly from the input, just as with the entry-level categories (i.e. by toil); or (2) indirectly, from Boolean combinations of the grounded category names in the form of propositions describing the higher-order category ('symbolic theft'). We analyse the architectures and input conditions that allow grounding (in the form of compression/ separation in internal similarity space) to be 'transferred' in this second way from directly grounded entry-level category names to higher-order category names. Such hybrid models have implications for the evolution and learning of language.

The evolution of language implies the parallel evolution of an ability to respond appropriately to signals (language understanding) and an ability to produce the appropriate signals in the appropriate circumstances (language production). When linguistic signals are produced to inform other individuals, individuals that respond appropriately to these signals may increase their reproductive chances but it is less clear what the reproductive advantage is for the language producers. We present simulations in which populations of neural networks living in an environment evolve a simple language with an informative function. Signals are produced to help other individuals categorize edible and poisonous mushrooms, in order to decide whether to approach or avoid encountered mushrooms. Language production, while not under direct evolutionary pressure, evolves as a byproduct of the independently evolving perceptual ability to categorize mushrooms.