Tessa Verhoef

Emergence of combinatorial structure and economy through iterated learning with continuous acoustic signals

Abstract Human speech has combinatorial structure, but it is still unclear how this type of organization emerged in the course of language evolution. There are two positions in the debate about the evolution of combinatorial structure: one stresses the importance of distinctiveness, while the other stresses economy and efficient reuse of building blocks. Different sources of evidence can be used to investigate the origins of combinatorial structure, such as emerging sign languages, animal communication systems, analysis of modern language and computer simulations but each source has its problems. In this article it is demonstrated that a novel empirical method from the field of language evolution can help to gain insight into the emergence of phonological combinatorial organization. This method, experimental iterated learning, allows investigating cultural evolution and the development of structure over time with human participants. We present data from an experiment in which combinatorial structure emerges in artificial whistled languages. We show that our experiment can give insight into the role of distinctiveness and reuse of building blocks and how they interact. We argue that experimental iterated learning offers a valuable new tool for investigating questions on evolutionary phonology and phonetics.

The origins of duality of patterning in artificial whistled languages

Abstract In human speech, a finite set of basic sounds is combined into a (potentially) unlimited set of well-formed morphemes. Hockett (1960) placed this phenomenon under the term ‘duality of patterning’ and included it as one of the basic design features of human language. Of the thirteen basic design features Hockett proposed, duality of patterning is the least studied and it is still unclear how it evolved in language. Recent work shedding light on this is summarized in this paper and experimental data is presented. This data shows that combinatorial structure can emerge in an artificial whistled language through cultural transmission as an adaptation to human cognitive biases and learning. In this work the method of experimental iterated learning (Kirby et al. 2008) is used, in which a participant is trained on the reproductions of the utterances the previous participant learned. Participants learn and recall a system of sounds that are produced with a slide whistle. Transmission from participant to participant causes the whistle systems to change and become more learnable and more structured. These findings follow from qualitative observations, quantitative measures and a follow-up experiment that tests how well participants can learn the emerged whistled languages by generalizing from a few examples.

Language dynamics in structured form and meaning spaces

This paper reviews how the structure of form and meaning spaces influences the nature and the dynamics of the form-meaning mappings in language. In general, in a structured form or meaning space, not all forms and meanings are equivalent: some forms and some meanings are more easily confused with each other than with other forms or meanings. We first give a formalization of this idea, and explore how it influences robust form-meaning mappings. It is shown that some fundamental properties of human language, such as discreteness and combinatorial structure as well as universals of sound systems of human languages follow from optimal communication in structured form and meaning spaces. We also argue that some properties of human language follow less from these fundamental issues, and more from cognitive constraints.We then show that it is possible to experimentally investigate the relative contribution of functional constraints and of cognitive constraints. We illustrate this with an example of one of our own experiments, in which experimental participants have to learn a set of complex form-meaning mappings that have been produced by a previous generation of participants. Theoretically predicted properties appear in the sets of signals that emerge in this iterated learning experiment.

Holistic or synthetic protolanguage: Evidence from iterated learning of whistled signals

Many arguments have been proposed in favor of and against the idea of protolanguage as a set of holistic utterances that were later segmented into words. This paper presents data from a human iterated learning experiment which corroborates arguments in favor of holistic protolanguage and forms a counterexample to some of the arguments that were proposed against it. This experiment involves iterated learning and recall of a set of (initially) holistic whistles produced with a slide whistle. Over generations, the whistle sets become easier to learn and discrete, combinatorial structure emerges.

Bio-sensing for Emotional Characterization without Word Labels

In this article, we address some of the issues concerning emotion recognition from processing physiological signals captured by bio-sensors. We discuss some of our preliminary results, and propose future directions for emotion recognition based on our lessons learned.

Iconicity and the Emergence of Combinatorial Structure in Language

In language, recombination of a discrete set of meaningless building blocks forms an unlimited set of possible utterances. How such combinatorial structure emerged in the evolution of human language is increasingly being studied. It has been shown that it can emerge when languages culturally evolve and adapt to human cognitive biases. How the emergence of combinatorial structure interacts with the existence of holistic iconic form-meaning mappings in a language is still unknown. The experiment presented in this paper studies the role of iconicity and human cognitive learning biases in the emergence of combinatorial structure in artificial whistled languages. Participants learned and reproduced whistled words for novel objects with the use of a slide whistle. Their reproductions were used as input for the next participant, to create transmission chains and simulate cultural transmission. Two conditions were studied: one in which the persistence of iconic form-meaning mappings was possible and one in which this was experimentally made impossible. In both conditions, cultural transmission caused the whistled languages to become more learnable and more structured, but this process was slightly delayed in the first condition. Our findings help to gain insight into when and how words may lose their iconic origins when they become part of an organized linguistic system.

Cultural transmission of melodic and rhythmic universals: Four experiments and a model

Corresponding Author: andrea.ravignani@gmail.com Corresponding Author: tessauva@gmail.com Department of Cognitive Science, University of California, San Diego, La Jolla, USA Artificial Intelligence Lab, Vrije Universiteit Brussel, Brussels, Belgium Research Department, Sealcentre Pieterburen, Pieterburen, Netherlands Language and Cognition Department, Max Planck Institute for Psycholinguistics, Nijmegen, Netherlands Leiden Institute of Advanced Computer Science (LIACS), Leiden University, Leiden, The Netherlands Center for Research in Language, University of California, San Diego, La Jolla, USA Centre for Language Evolution, School of Philosophy, Psychology and Language Sciences, University of Edinburgh, Edinburgh, UK.

The Evolution of Language: Proceedings of the 11th International Conference (EVOLANG11)

In this work we compare, on the well explored domain of Indo-European languages, the phylogenetic outputs of three different sets of linguistic characters: traditional etymological judgments, a system for phonetic alignment of lists of cognates, and a set of values for generative syntactic parameters. The correlation and relative informativeness of distances and phylogenies generated by the three types of characters can thus be for the first time accurately evaluated, and the degree of success of the last two, innovative, alternatives to the classical comparative method can be so assessed.

ITERATED LEARNING OF SOUND SYSTEMS AND THE EMERGENCE OF TONE CATEGORIES

In this abstract we investigate whether emerging sets of signals in cultural evolution experiments show similar structural tendencies as the world’s tone languages. An iterated learning experiment (Kirby, Cornish, & Smith, 2008) was conducted with human participants. The universal phonetic tendencies of tones in human language are relatively well-understood. Following Maddieson (1978), linguistic tones can be level (i.e. only one pitch level), simple contours (two pitch levels) or complex (more than two pitch levels). For instance Mandarin Chinese of Beijing has a high tone, a rising tone, a falling tone and a falling-rising tone and therefore has all three types of tones. Level tones tend to be much more prevalent than contour tones, while there is a simple implicational hierarchy for the different types of tones. If a language has complex contour tones it also has simple contour tones. If it has simple contour tones it also has level contour tones. In short, there seems to be a tendency favoring less complex contours. The experiment was conducted as part of Science Livea, in museum Nemo in Amsterdam. Participants learned and reproduced sets of sounds, similar as in Verhoef, Kirby, and Padden (2011), but the signals were produced with the use of a mouse in a virtual synthesizer-interface on the computer. By pressing the mouse inside a horizontal bar and moving the cursor left and right, the pitch was manipulated. The sound stopped when the mouse was released, therefore a signal could not contain any silent parts, as opposed to the whistled signals used by Verhoef et al. (2011). The output produced by one participant was used as the input for the next, following the design of Kirby et al. (2008). Twelve chains were created with twelve participants in each. The signal sets consisted of 8 signals (two different initial sets were used, each used for 6 chains)

The critical period and preservation of emerged vowel systems

The critical period for language acquisition is often assumed to be nothing more than a by-product of development. However, evolutionary computer simulations show that it can be explained as a result of biological evolution (Hurford, 1991). In the present study the aim is not to explain how and why this age sensitivity evolved but to investigate the consequences of this individual-level disadvantage on a culturally evolving vowel system as a whole. Using two different agent-based computer models it will be argued that a difference in learning ability between children and adults can improve the stabilization and preservation of complexity of vowel systems in a changing population. The first model is a re-implementation of the one described by de Boer and Vogt (1999), which consists of a population of agents that interact through imitation games using realistic mechanisms for production and perception of vowels. The agents have a vowel memory in which they store learned prototypes of vowels and in response to their interactions with other agents they update their memory and learn new sounds. Analogous to the results of de Boer and Vogt (1999) the model shows that a population in which new members are born and old members die, a critical period stabilizes vowel systems over the generations. In this case the adults provide the learners with a stable target facilitating the acquisition process. Figure 1 shows the difference in the changes of the vowel system after transmission in a population with or without age structure. The second model is a variation on the first which integrates the linguistic paradigm of Optimality Theory (OT). In this version of the model, the agents imitate each other using their own bidirectional stochastic OT grammar (Boersma & Hamann, 2008) consisting of a ranked set of articulatory and cue constraints. To produce or perceive a speech signal, a set of possible candidate forms is evaluated by the grammar. The candidate that violates the fewest highly ranked constraints is selected. In response to their interactions with other agents they learn by adjusting (a) No age structure (b) Age structure Figure 1. Emerged vowel systems in the first model. Initial vowel system in grey. the ranking values in their grammar. This new approach replicates the results of the stabilizing effect on the emerged vowel systems. The results suggest that the critical period might be more than just an unfortunate consequence of development since its influence …