Movement in Language: Interactions and Architectures (review)

Since the advent of the minimalist program, one might get the impression that the central focus of research in syntax has shifted to A-movement. Richards’s book, which is a revised version of his 1997 MIT dissertation, however, demonstrates that wh-movement continues to be as interesting and important a topic as before. In this book, R takes up the challenging job of providing a theory of the typology of wh-question formation and succeeds in coming up with ingenious general hypotheses about the way linguistic computation works, which extend beyond the domain of wh-movement itself.

The book consists of six chapters. There are sporadic typographical errors, including those in the arrows that indicate which element in example sentences moves where, though they do not pose difficulties for comprehension. The book does not mention the fact that it is a revision of R’s doctoral thesis and that the core parts of Chs. 3 and 5 are published under the same titles as Richards 1999 and Richards 1998, respectively. For those who have been exposed to R’s dissertation, this book keeps the overall organization of discussion, with modifications limited to details. The bibliography is mostly, but not completely, updated.

Crosslinguistic variations in wh-question formation are most straightforwardly observed in multiple wh-questions. Languages like Bulgarian and Serbo-Croatian front all wh-phrases, while languages like Chinese and Japanese leave them all in situ. The class that includes English and German takes the middle ground, moving only one wh-phrase per clause and leaving the rest in situ. Ch. 1 reviews the four different approaches to this typology proposed in the past literature and announces that the book will argue for the theory according to which covert movement of wh-in-situ takes place after visible displacement of wh-phrases, the so-called T-model of the extended standard theory. The key arguments to this effect from various parts of the book are summarized in Ch. 6.

Ch. 2 proposes that there is a dimension of variation that cuts across the three types mentioned above. R claims that this new typological variation is correlated with the presence or absence of A-scrambling. Those languages that have A-scrambling, such as Serbo-Croatian, Japanese, and German, are characterized as essentially moving multiple wh-phrases to IP (hence the name ‘IP-absorption’), whereas those that lack A-scrambling, such as Bulgarian, Chinese, and English, make use of multiple specifiers of CP (‘CP-absorption’), though only one specifier is available for visible movement in English. A significant consequence of this proposal is that it captures the fact that CP-absorption languages except English fail to display the wh-island effect, in contrast to IP-absorption languages. The reasoning is that the availability of multiple specifiers at CP allows a wh-phrase to be extracted out of a wh-island without violating locality through one of these specifiers. The exceptional behavior of English is therefore not exceptional at all, since multiple specifiers at CP are not yet available in overt syntax in English.

An obvious weakness of the IP/CP-absorption typology, though, is that it is not clear why Ascrambling, but not A-bar scrambling, is correlated. Since Mahajan’s 1990 work, A-bar scrambling has been considered to be closer to wh-movement than A-scrambling. So if there is any correlation, A-bar scrambling is the promising candidate. The question of why movement to IP can count as wh-movement is not addressed in any detail either. The problem is not resolved in this book.

Another point worth mentioning is that R does not discuss adjunct extraction from islands at all. It is known that in contrast to an argument wh-phrase, an adjunct wh cannot escape a wh-island in Chinese. According to R’s theory, there should be nothing syntactically wrong with such extraction, since Chinese is characterized as a CP-absorption language. In this case, one can follow Szabolcsi & Zwart...