Bart de Boer

Self-organization in vowel systems

This paper presents a computer simulation of the emergence of vowel systems in a population of agents. The agents (small computer programs that operate autonomously) are equipped with a realistic articulatory synthesizer, a model of human perception and the ability to imitate and learn sounds they hear. It is shown that due to the interactions between the agents and due to self-organization, realistic vowel repertoires emerge. This happens under a large number of di!erent parameter settings and therefore seems to be a very robust phenomenon. The emerged vowel systems show remarkable similarities with the vowel systems found in human languages. It is argued that self-organization probably plays an important role in determining the vowel inventories of human languages and that innate predispositions are probably not necessary to explain the universal tendencies of human vowel systems. ( 2000 Academic Press

Investigating the role of infant-directed speech with a computer model

A computer model (expectation maximization of a mixture of Gaussians) is used to learn the positions of vowel categories from two sets of recorded words. The number of vowels is known beforehand. The results show that vowel positions learned on the basis of infant-directed (ID) speech correspond better to those in the input than those learned on the basis of adult-directed (AD) speech.

Paying attention to symmetry

Humans are very sensitive to symmetry in visual patterns. Symmetry is detected and recognized very rapidly. While viewing symmetrical patterns eye fixations are concentrated along the axis of symmetry or the symmetrical center of the patterns. This suggests that symmetry is a highly salient feature. Existing computational models of saliency, however, have mainly focused on contrast as a measure of saliency. These models do not take symmetry into account. In this paper, we discuss local symmetry as measure of saliency. We developed a number of symmetry models an performed an eye tracking study with human participants viewing photographic images to test the models. The performance of our symmetry models is compared with the contrast saliency model of Itti et al. [1]. The results show that the symmetry models better match the human data than the contrast model. This indicates that symmetry is a salient structural feature for humans, a finding which can be exploited in computer vision.

The evolution of combinatorial phonology

A fundamental, universal property of human language is that its phonology is combinatorial. That is, one can identify a set of basic, distinct units (phonemes, syllables) that can be productively combined in many different ways. In this paper, we develop a methodological framework based on evolutionary game theory for studying the evolutionary transition from holistic to combinatorial signal systems, and use it to evaluate a number of existing models and theories. We find that in all problematic linguistic assumptions are made or crucial components of evolutionary explanations are omitted. We present a novel model to investigate the hypothesis that combinatorial phonology results from optimizing signal systems for perceptual distinctiveness. Our model differs from previous models in three important respects. First, signals are modeled as trajectories through acoustic space; hence, both holistic and combinatorial signals have a temporal structure. Second, acoustic distinctiveness is defined in terms of the probability of confusion. Third, we show a path of ever increasing fitness from unstructured, holistic signals to structured signals that can be analyzed as combinatorial. On this path, every innovation represents an advantage even if no-one else in a population has yet obtained it.

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.

Emergence of vowel systems through self-organisation

This paper describes a model for explaining the emergence and the universal structural tendencies of vowel systems. Both are considered as the result of self-organisation in a population of language users. The language users try to imitate each other and to learn each other’s vowel systems as well as possible under constraints of production and perception, while at the same time maximising the number of available speech sounds. It is shown through computer simulations that coherent and natural sound systems can indeed emerge in populations of artificial agents. It is also shown that the mechanism that is responsible for the emergence of sound systems can be used for learning existing sound systems as well. Finally, it is argued that the simulation of agents that can only produce isolated vowels is not enough. More complex utterances are needed for other interesting universals of sound systems and for explaining realistic sound change. Work in progress on implementing agents that can produce and perceive complex utterances is reported.

Computer Models of Vocal Tract Evolution: An Overview and Critique

Human speech has been investigated with computer models since the invention of digital computers, and models of the evolution of speech first appeared in the late 1960s and early 1970s. Speech science and computer models have a long shared history because speech is a physical signal and can be modeled accurately. This article gives a brief overview of the use of computer models in the study of the evolution of the vocal tract. We also present a critical case study of one model that has been used to study the vocal abilities of Neanderthals. We argue that this study contains subtle but fatal flaws which invalidate the conclusions drawn from the model, illustrating the dangers of applying computer models outside the area for which they have been developed. Future models need to make use of a broader database of anatomical and physiological data from other animals, especially nonhuman primates, to understand the path leading to modern Homo sapiens.Human speech has been investigated with computer models since the invention of digital computers, and models of the evolution of speech first appeared in the late 1960s and early 1970s. Speech science and computer models have a long shared history because speech is a physical signal and can be modeled accurately. This article gives a brief overview of the use of computer models in the study of the evolution of the vocal tract. We also present a critical case study of one model that has been used to study the vocal abilities of Neanderthals. We argue that this study contains subtle but fatal flaws which invalidate the conclusions drawn from the model, illustrating the dangers of applying computer models outside the area for which they have been developed. Future models need to make use of a broader database of anatomical and physiological data from other animals, especially nonhuman primates, to understand the path leading to modern Homo sapiens.

Multi-Agent Simulations of the Evolution of Combinatorial Phonology

A fundamental characteristic of human speech is that it uses a limited set of basic building blocks (phonemes, syllables), that are put to use in many different combinations to mark differences in meaning. This article investigates the evolution of such combinatorial phonology with a simulated population of agents. We first argue that it is a challenge to explain the transition from holistic to combinatorial phonology, as the first agent that has a mutation for using combinatorial speech does not benefit from this in a population of agents that use a holistic signaling system. We then present a solution for this evolutionary deadlock. We present experiments that show that when a repertoire of holistic signals is optimized for distinctiveness in a population of agents, it converges to a situation in which the signals can be analyzed as combinatorial, even though the agents are not aware of this structure. We argue that in this situation adaptations for productive combinatorial phonology can spread.

Acoustic analysis of primate air sacs and their effect on vocalization

This paper presents an analysis of the acoustic impedance of primate air sacs and their interaction with the vocal tract. A lumped element model is derived and it is found that the inertance of the neck and the volume of the air sac are relevant, as well as the mass and stiffness of the walls (depending on the tissue). It is also shown that at low frequencies, radiation from the air sac can be non-negligible, even if the mouth is open. It is furthermore shown that an air sac can add one or two low resonances to the resonances of the oral tract, and that it shifts up the oral tracts resonances below approximately 2000 Hz, and shifts them closer together. The theory was verified by acoustic measurements and applied to the red howler monkey (Alouatta seniculus) and the siamang (Symphalangus syndactylus). The theory describes the physical models and the siamang calls correctly, but appears incomplete for the howler monkey vocalizations. The relation between air sacs and the evolution of speech is discussed briefly, and it is proposed that an air sac would reduce the ability to produce distinctive speech, but would enhance the impression of size of the vocalizer.This paper presents an analysis of the acoustic impedance of primate air sacs and their interaction with the vocal tract. A lumped element model is derived and it is found that the inertance of the neck and the volume of the air sac are relevant, as well as the mass and stiffness of the walls (depending on the tissue). It is also shown that at low frequencies, radiation from the air sac can be non-negligible, even if the mouth is open. It is furthermore shown that an air sac can add one or two low resonances to the resonances of the oral tract, and that it shifts up the oral tracts resonances below approximately 2000 Hz, and shifts them closer together. The theory was verified by acoustic measurements and applied to the red howler monkey (Alouatta seniculus) and the siamang (Symphalangus syndactylus). The theory describes the physical models and the siamang calls correctly, but appears incomplete for the howler monkey vocalizations. The relation between air sacs and the evolution of speech is discussed briefly, and it is proposed that an air sac would reduce the ability to produce distinctive speech, but would enhance the impression of size of the vocalizer.

New perspectives on duality of patterning: Introduction to the special issue

This special issue assembles a number of papers that present recent work on the nature and the emergence of duality of patterning. Duality of patterning (Hockett 1960) is the property of human language that enables combinatorial structure on two distinct levels: meaningless sounds can be combined into meaningful morphemes and words, which themselves could be combined further. We will refer to recombination at the first level as combinatorial structure, while recombination at the second level will be called compositional structure. According to Hockett (1960), duality of patterning is a design feature of human language (meaning that all human languages have it) while it is also unique to human language. He argued that it evolves when a growing number of meanings need to be expressed, so that combinatorial structure helps to keep signals distinct. More recently similar arguments have been made on the basis of mathematical and computational models (e. g. Nowak et al. 1999; Zuidema and de Boer 2009). Although it seems to be uncontroversial that recombination of meaningful elements (i.e. compositional structure) is needed for an unlimited system, the relation between an unlimited set of signals and recombination of meaningless elements (i.e. combinatorial structure) is less clear. 1.1 Is duality of patterning a design feature? On the one hand, simple combinatorial structure has been found in vocalizations of adult male putty-nosed monkeys with a relatively limited set of signals (Arnold and Zuberbuhler 2006; Yip 2006), indicating that duality may not be uniquely human and that it may occur in systems in which it is not needed to keep signals distinct. On the other hand, there is some evidence that duality of patterning is not required for a human language. Combinatorial structure does not appear to have crystallized in a recently emerging sign language: Al-Sayyid Bedouin Sign Language even though it is a fully expressive language (Sandler et al. 2011). In the ~75 years since its emergence, Al Sayyid Bedouin Sign Language (ABSL) has come to serve all the communicative functions usually associated with language and has been shown to have compositional structure at the levels of morphology (Meir et al. 2010) and syntax (Sandler et al. 2005; Padden et al. 2010). However, the community has not yet converged on a conventionalized level of meaningless elements, although the beginnings of phonology can be discerned (Sandler et al. 2011). The ABSL findings show that a language without a clearly phonological level of structure is possible. This contrasts with more established sign languages, whose lexical signs are made up of meaningless contrastive units belonging to the categories of handshape, location, and movement.1 While there is a degree of linear organization in the combination of elements in a sign, certain key formational elements combine with one another simultaneously and, as a result, a likely holistic source is often quite transparent. Because sign languages have a strongly iconic base, so that language users needn’t distinguish large sets of purely arbitrary holistic auditory signals, the ABSL researchers suggest that the development of a large vocabulary before holistic gestures are decomposed into a system of meaningless elements might be easier in signed than in spoken languages. However, as Blevins (2012) shows using data from spoken languages, iconicity is not a necessary condition for isomorphism between the two levels of structure to occur frequently in a language, and explanation of the evolution and predominance of dual patterning remains a challenge. We see then from the case of ABSL that the need to express a large set of signals does not necessarily lead to combinatorial structure, while conversely from the animal systems, it appears that combinatorial structure does not necessarily need a very large set of signals to emerge. As combinatorial structure is the main defining characteristic of duality of patterning, it appears that both the status of duality of patterning as a design feature of language and the evolutionary pathways leading to it need to be rethought.