REVIEW OF Michael Tomasello and Elizabeth Bates (eds) Language Development: the essential readings

Michael Tomasello and Elizabeth Bates (Eds),
Language Development: the essential readings
Blackwell publishers, Oxford, 2001.

James R Hurford,
Language Evolution and Computation Research Unit,
School of Philosophy, Psychology and Language Sciences,
University of Edinburgh.

(This review is to appear, perhaps in a shortened form, in Quarterly Review of Biology.)

Bates and Tomasello are both productive and highly respected researchers in the acquisition of language. This collection of `essential readings' brings together 19 articles from the scholarly literature between 1990 and 2001, with the average being very recent. The stance that informed the selection is summarized thus:
``In the field of language development, which has more than its share of controversy, the one-of-everything strategy can lead to a collection that is ... like a jungle, with the warring factions of nativism and empiricism all given equal time to fight today's battles. A book of that kind might leave our readers well informed about those battles, but it is not clear that it would prepare them for tomorrow's science. The papers in Parts I-III present an array of solid facts that must be explained regardless of ones theoretical framework. But by focussing on papers that emphasize learning and change, we have also made a carefully judgment about the future of the field.'' (p.292)
The collection is definitely from the empiricist tradition in language development. Authors from the rival `formalist, nativist' point of view (e.g. Chomsky, Pinker, Hyams, Atkinson, Radford) are occasionally mentioned and generally disagreed with. The papers in this collection are empirically based, citing experiments with children and data gleaned from children. A general message is that `` ... children must do much more work to learn their language than they would have to do if Chomsky's account were true.'' (p.8)

The book is structured into three main parts: I, Speech Perception; II Word Learning; III, Grammatical Development. Highlights culled, essentially verbatim, from each of these parts are presented below.

SPEECH PERCEPTION. P.Jusczyk shows that, in English learners, the ability to detect familiar words in fluent speech appears to develop between 6 and 7.5 months of age. Infants use multiple cues, such as stressed syllables, allophonic regularities and phonotactic constraints in picking out words from the stream of speech. At least initially, infants' representations of words may be stored exemplars of previously heard words rather than abstract prototypes. J.Werker and R.Desjardins review the kinds of initial abilities infants bring to the task of phoneme perception, how these sensitivities are influenced by experience in a particular linguistic community, and whether events that ocur during the prelinguistic period help prepare the child for language acquisition. F.Ramus and co-authors, on the basis of experiments with tamarins, show that aspects of human speech perception may have built upon pre-existing sensitivities of the primate auditory system; the mouth evolved to meet the ears, and not vice-versa. R.Gomez and L.Gerken report studies with an experimental technique based on subjects learning experimentally controlled miniature artificial languages. Their study, in the vein of the whole collection, reconsiders the view of the constrained language learner which has dominated the field for the last 30 years. A.Fernald and co-authors quantify the gains in speed and efficiency with which infants in their 2nd year understand familiar words in continuous speech.

WORD LEARNING. H.Shwe's and E.Markman's results suggest that young children are aware that their communicative signals have an impact on the mental state of their listeners. Children persisted in clarifying their signals more when their listener expressed misunderstanding than when she conveyed understanding. M.Caselli and co-authors show that Italian and English provide strong evidence for universal constraints on the development and composition of vocabulary, although subtle cross-language variations in content are observed as well. M.Tomasello's experiments show that the major social cognitive skill that underlies children's ability to learn words is their understanding of the intentional actions of others. Children's ability to reproduce these intentional communicative actions via some form of cultural or imitative learning involves a role reversal. L.Markson and P.Bloom demonstrate that children learn words relatively quickly with learning skills that do not seem to differ in kind from those used to acquire other pieces of information about the world. E.Bates and J.Goodman conclude that the distinction between the lexicon and grammar has been exaggerated. Many lexical items serve grammatical functions in larger syntactic constructions, and many syntactic constructions represent symbolic patterns that must be learned in the same basic way as words. The dependence of early grammar on vocabulary size is so strong and the nonlinear shape of this function is so regular that it approaches the status of a psychological law.

GRAMMATICAL DEVELOPMENT. M.Tomasello documents the fact that young children's early multi-word productions are highly concrete, being based on particular words and phrases, not on innate and abstract linguistic categories; gradual and word-specific grammatical development is widespread. N.Akhtar's data show that young children up to beyond three years treat new verbs idiosyncratically, differently from other verbs in their vocabulary. She argues that these results cannot be explained by a parameter-setting approach in which innate and abstract verb-general structures are simply triggered by the environment. Children learn the appropriate use of grammatical relations (e.g. word order) on a verb-by-verb basis. The overall claim is that learning of syntactic structure is data-driven. K.Köpcke, on the basis of plural marking in English and German, argues for a model which eliminates the strict division between the grammar on the one hand, which covers regular productive morphology (e.g. how most English verbs form their past tense), and the lexicon on the other hand, in which all exceptions to regular processes are stored. L.Bloom and co-authors reinforce this message with their conclusion that the acquisition of the syntax of complementation was lexically specific. Rather than learning a general rule for complementation per se, or even separate rules for Wh-complements, S-complements, to-complements, if-complements, and so forth, the children's grammatical knowledge was specific to the matrix verbs. D.Slobin shows that there is too much cross-linguistic variability for the human species to have any preconceived ideas about what kinds of notions will and will not be indicated by grammatical rather than by lexical items.