Signal to Syntax; Bootstrapping from Speech to Grammar in Early Acquisition

In February 1993, a conference was organized at Brown University, Providence, RI, where an interesting mélange of theoretical linguists, computer scientists, acousticians, psycholinguists, and cognitive scientists discussed issues bearing on how children’s perception and representation of the speech stream may contribute to the acquisition of syntax. The contents of the current volume are based on the proceedings of this conference. The basic idea for the conference (and the book) was a good one. Morgan and Demuth had noticed that, in accounts of children’s grammatical development, little attention had been paid to factors involving the perception, representation, or production of speech, despite the large amount of work done in these areas. In bringing together scholars of different plumage but with the same basic research interest, they sought to bridge the disciplinary gaps that might have resulted in the paucity of influence of work on infant perception and representation on work in language acquisition. Both sides could benefit from the exchange of information. On the one hand, theories of grammatical development might become simpler were an appropriately rich input representation assumed, while on the other hand, a theory of grammar is needed to test phonological bootstrapping hypotheses. Another goal of the conference was to encourage further research into the question of whether perceptual analyses yield information about some basic properties of grammar. The conference apparently was a success, and some of the enthusiasm has found its way into the proceedings. The introductory chapter by the editors is very enlightening. It is not only an introduction to the different parts and chapters of the book, but also a clear synopsis of the main results of the research presented in the different chapters. It is worth rereading, after having savored the entire volume, in order to regain a grip on the large amount of information. Apart from the introduction, there are 24 chapters, distributed over five parts, followed by an author index and a subject index. Part I, “The Nature, Perception, and Representation of Input Speech,” contains four introductory chapters in order to familiarize the reader with the data, methodology, and arguments from different disciplines. Eimas presents an account of the research on infant speech perception and representation. Dresher introduces prosodic and metrical theory. Lieberman presents a biological perspective on the study of prosody and draws attention to the difficulty of using appropriate acoustic parameters in studies of prosodic bootstrapping. Price and Ostendorf give an outline of how statistical and linguistic models of prosody can be combined and how this combination can increase the knowledge about the role of prosody in language processing.