Printed in the United Kingdom DOI: 10.1017/S0305000903215634 REVIEWS

The authors set three major goals for the book: to increase our understanding of the ‘relationship between lexical semantics and morphology in language acquisition’ (p. 8), to review and evaluate pertinent hypotheses ‘with respect to empirical evidence from crosslinguistic studies of English, Chinese, and Japanese, in both first and second language acquisition’ (p. 8), and to investigate ‘the learner’s remarkable ability to extract input patterns and form linguistic associations’ (p. 9). The authors accomplish these goals, and in doing so, they have written a book that makes a valuable contribution to the issue of language learning.The authors promote a prototype hypothesis that is supported by a connectionistmodel.Their fundamental claim is that ‘children create a semantic representation of tense–aspect morphology which is restricted to the prototype of the morphological category’ (p. 66). This argument is focused on two prototypes. One prototype has the semantic features of dynamic, telic, punctual, and resultative. The morphemes that are connected to this prototype are as follows: the perfective aspectmarker -le inChinese, the past tense form -ta in Japanese, and the simple past -ed /irregular in English. The second prototype has the features dynamic and atelic, and this semantic concept is connected to the following: the progressive marker zai in Chinese, the durative marker -te iin Japanese, and the progressive marker -ing in English. The two prototypes will be referred to as the perfective achievement (PA) and the imperfective activity (IA). What is the nature of the knowledge attributed to the child byprototype theory?The child can construct a semantic feature set, and he/she can form an association. For example, a child learning English might associate the verb to spill with the -ed morpheme, and he/she might connect to crywith -ing. Given the constraints of the theory, what is the nature of the knowledge that has NOT been explicitly attributed to the child? Regarding aspect, we cannot infer that the child uses a morpheme to code external versus internal perspective (or bounded versus unbounded), and regarding tense, there is no basis to infer that the child knows how to specify deictic relations such as event time prior to/subsequent to speech time. In short, we cannot infer that these morphemes represent operators that have independent status within the child’s grammatical system. Obviously, the authors are aware that it is impossible to explain the acquisition of aspect without also considering tense and modality. One of the strengths of the book is that the authors maintain their focus on aspect. J. Child Lang. 30 (2003), 237–251. f 2003 Cambridge University Press Printed in the United Kingdom