Browse Books

Reviewer: Richard L. Frautschi

For several decades, computer modeling and simulations have offered a novel approach to the problems of language evolution. In this book, the authors’ working premise is that artificial reality strategies support a bottom-up, incremental approach to theoretical and pragmatic linguistics. The 24 authors propose that the myriad of extant languages today are autonomous products of complex physiological, psychological, social, and cultural interactions, mutated over time. Given the diversity of chapter subtopics, all of which support aspects of evolutionary linguistics, let me highlight briefly each of the contributions. The editors, Cangelosi and Parisi, launch the volume with a thesis that quantified simulations can promote an efficient examination of language evolution theories, for example, processes of natural selection in a virtual laboratory. Prudently, the limits of simulation are also ventilated. In chapter 2, Turner introduces computational methods used in language evolution: representation of individual agents, learning, and interactive algorithms. In chapter 3, Noble et al. present simulations which augment conventional game theory: trajectories of evolutionary change, such as ecological and Zuhavian handicap feedbacks, and continuous-time simulation of animal contests. De Boer, in chapter 4, focuses on signaling (namely, speech sounds). He describes modeling approaches combining genetic algorithms and language games, challenging the view that speech universals reflect an innate linguistic ability in humans. Livingston, in chapter 5, argues that dialects are not accidents of natural selection. Rather, processes of language learning and transmission explain language diversity. In chapter 6, Kirby and Hurford offer an iterated learning model (ILM), specifically the emergent properties of syntactic universals via neural networks and rule inference systems. Of special interest is the interaction between glossogenesis and phylogeny, where functional pressures can bias grammatical principles. Komarova and Novak, in chapter 7, show how population dynamics also influence grammar learning. A key coherence threshold must be reached before one of many universal grammars can begin to function. Chapter 8, by Christiansen et al., describes a multi-methodological approach to sequential learning as a hierarchically organized structure. Cognitive limitations in artificial language learning, aphasic patients, and primates provide contrastive paradigms. Grounded symbols are the focus in chapters 9 and 10. What mechanisms and models link shared lexical identity with objects in the world__?__ Congelosi et al. turn to cognitive and linguistic models, such as category-as-learning, and symbol grounding processes using symbolic theft and sensorimotor toil. Again, Steels favors a multi-media interface approach to sensorimotor symbol grounding, using robotic groups of talking heads, a video camera, and computer processed categorization, plus devices for audio input and output. He shows how language plays a crucial role as feedback. Also, his evolutionary language games concretize structural couplings between concept formation and grounded symbol acquisition. Arbib, in chapter 11, also proposes that evolution has shaped both the human brain and body to language readiness. In his mirror system, model monkey hand grasping evolves to protospeech, and finally to language. In chapter 12, Parisi and Congelosi suggest a unified computational scenario in order to answer three problems: the origin and history of speaking skills in an individual, the species, and the particular language of the social group of which the individual is a member. The adaptations reveal an ontogenetic acquisition of knowledge via neural network learning, and cultural transmission of linguistic knowledge via cultural learning. Hutchins and Hazelhurst discuss auto-organizational emergence of shared language, for example, axioms and grammars, in chapter 13. Modeling protocols feature expression and induction, form tuning, and embodied guessing games. Their cultural symbol hypothesis argues that public symbols emerge concurrently with internal meaning structures. For Hashimoto (in chapter 14), language is again a dynamic system. He models sense-making activity as the formation of web relationships between words via conversation among individuals. His simulations show consistent closed categories with rigid links, as well as prototype categories with changeable links. A final chapter by Tomasello suggests that simulators seek new directions in their study of language evolution; fresh empirical data is available, for example, in human primate communication, child language acquisition, imitation, and cultural learning, as well as in language change mechanisms, such as pidginization and creolization. To conclude, three common threads dominate the above studies: language emergence, interaction and interdependence between language and non-linguistic abilities in organisms and environments, and the relationships between simulation and empiricism. The present volume, then, presents a rich and refreshing vision. There are two minor caveats. There are copy-editing blemishes, and, regrettably, the constraints of book design have relegated the details of simulative operations to previous publications. Online Computing Reviews Service