Language down the garden path: The cognitive and biological basis for linguistic structures ed. by Montserrat Sanz, Itziar Laka, Michael K. Tanenhaus (review)

Language down the garden path leads the reader into the meanders of more than forty years of research since the publication of Bever’s 1970 article ‘The cognitive basis for linguistic structure’ (CBLS). The book mirrors the nature of CBLS in compiling an eclectic set of chapters addressing a variety of key theoretical questions, hypotheses, and observations about language and its relation to other domains of cognition. The book also mirrors the evolution of the cognitive science of language, which has spread in a wide array of directions. It encompasses different theoretical frameworks, diverse objects of study, and various levels of key phenomena explored through multiple research methods. While the diverse ideas of CBLS were developed in a single paper by a single researcher with the aim of reaching a unified view of the study of language learning and processing, a significant part of the field today is parceled, with sometimes little interpermeability of people and ideas. Notably, then, this book puts people and ideas together again, with the challenging aim of reunifying the different research pathways around the major question of the relation between competence and performance. The last decade has indeed witnessed a new interest in drawing links between the various pockets of knowledge accumulated in isolation, with models linking production, comprehension, and learning (see chapters by Thomas G. Bever; Gary S. Dell and Audrey K. Kittredge; Chien-Jer Charles Lin; Maryellen C. McDonald; Colin Phillips; and David J. Townsend) or research programs linking behavioral or neurophysiological data to the fine theoretical tools of linguistics (see chapters by Ina Bornkessel-Schlesewsky and Matthias Schlesewsky; Ewan Dunbar, Brian Dillon, and William J. Idsardi; Janet Dean Fodor; Yosef Grodzinsky; Simona Mancini, Nicola Molinar, and Manuel Carreiras; Jacques Mehler; Massimo Piattelli-Palmarini; colin phillips; Montserrat Sanz; and Virginia Valian).

The book starts with a reprint of CBLS, whose core question is the relation between internal linguistic structures and external input sequences. On Bever’s view, the key link between the two lies in the ‘analysis-by-synthesis’ approach, according to which language processing takes place in two steps. During the first step, ‘perceptual strategies’ ensure a direct mapping between the surface structure of the sentence and thematic roles, generating a rough, plausible meaning for the sentence. This early analysis is then ‘synthesized’ during a derivational step responsible for building a phrase structure for the sentence and checking that it converges with the output of the initial quick-and-dirty analysis. CBLS introduced the idea developed later that ‘we understand everything twice’ (Townsend & Bever 2001). It is interesting that this approach seeded the idea of a two-stage decomposition of the comprehension process, one of the major tenets of the modular approach, although the modular model flipped the temporal order of the two components, arguing for an early, syntax-based process followed by a late integration of other types of information.

Bever’s hypothesis that the form of grammar is determined by general constraints from cognition constituted a radical shift from the generative approach, dominant at that time. It questioned its overall relevance as a model of language processing, but also its methodology. Indeed, if the grammaticality judgments used to develop the model of grammar are themselves the output of the performance system, Bever asks what the science of linguistics is a science of. As Tanenhaus writes in his afterword, CBLS was an invitation to ‘thinking out of the box’, and this invitation helped spawn an immensely varied and fruitful research program exploring the cognitive bases of language structure. But CBLS may...