A meeting of tongues

This is an important book, written predominantly for Carol Myers-Scotton's fellow linguists. It is at a level of technicality such that few others are likely to get through it unless they need to incorporate linguistic insights into discussion of adjacent disciplines.

Neither would it be an easy book for non-linguists to dip into.

Like many such works, the book exhibits a tension between its identity as a "textbook" presenting a historical review and a state-of-the-art survey of its field and as a vehicle for the promotion and defence of Myers-Scotton's own theoretical stance. The author is one of the most eminent world figures in the field of language contact, and her position is well supported, but - as she acknowledges at a very early stage - it is nevertheless controversial, and this book will not settle all the arguments. For this reason, even students of linguistics (especially undergraduates) would need guidance in using the book. (Myers-Scotton does help further by giving an account of her own intellectual development, allowing assessment of the influences on her thinking.) The controversy here is an instance of the theoretical uncertainty that pertains in all subjects but is rather salient in linguistics. Many linguistic theories that are fashionable at any given time - even in some apparently quite strictly empirical sub-fields such as the one covered in this book - fail at numerous points, making predictions that are not borne out or avoiding failure only through philosophically dubious strategies.

Often, there are competing, sometimes simplistic-sounding theories, all of them threatened by large amounts of anomalous data.

However, in this case Myers-Scotton claims to have reduced the number of troublesome anomalies greatly by progressively refining her own theory, and claims to have explained bodies of hitherto intractable data. This has not been demonstrated to the satisfaction of all her colleagues, but it is not obvious that she is wrong. If her theory came to be generally accepted, this would be a very significant gain (not least because this is an area of linguistics with major educational ramifications).

Although this is a work of theoretical rather than descriptive linguistics, Myers-Scotton works heavily from data and uses data from many languages. Of course, the book does not deal with all aspects of contact between languages. It focuses on formal linguistic aspects of contact rather than social (but without denying the importance of the latter), on the outcomes of contact and on grammatical phenomena rather than, say, phonological. But the scope and depth are still impressive.

Like many who have studied multilingualism, Myers-Scotton argues that linguistics has been distorted and weakened in various respects through an excessive focus on single languages in isolation and monolingual speakers.

Historical linguists, for instance, have (overall) worked much more on the divergence of genetically related languages than on the convergence of (often-unrelated) languages that have come into contact. If the theoretical study of multilingualism and language contact is further strengthened, this situation will surely improve.

Briefly, Myers-Scotton holds that in all contact phenomena, and most clearly in cases of "classic code-switching", the participating languages are unequal, in that one of them can be identified as the "matrix language", from which the core grammar is drawn; the other language is the "embedded language". (Not all scholars would accept this opposition as universally valid, at least as Myers-Scotton frames it, but she argues that objections to it involve misunderstanding.) The matrix language can be replaced ("matrix language turnover") but this is a slow and complex process that can be recognised from data. The "matrix language principle" is one of a set of four connected principles at the heart of Myers-Scotton's view of contact. One of these interacts with a classification of morphemes into four types, three of which are different types of "system morpheme" (grammatical). A further key principle is that all system morphemes that have relations external to their own phrasal head will be found to come from the matrix language (the "system morpheme principle", or SMP).

Critics of Myers-Scotton (as of others who propose widely applicable theories in this area) are liable to question the empirical status of some of the analytical concepts used, including the degree to which they are sufficiently well defined. This could readily be said, for instance, of "classic code-switching", and it might even be said of the notion of "system morpheme" as employed in the SMP. If the SMP were thereby invalidated, the consequences would be serious, because the matrix language/embedded language opposition and the SMP form a nexus at the core of Myers-Scotton's theory. She believes that surface syntactic relationships are not important here; it is thus inevitable that some of the definitions of these notions involve abstract syntactic analyses - and other more general theoretical positions about the mental organisation of language - which themselves may be seriously disputed. But there are also data-based criticisms of the SMP that are grounded within Myers-Scotton's own analytical framework. She herself again attributes these criticisms to misunderstanding, but it is not clear that this is the whole story.

(Tension between existing theoretical notions and the interpretation of "facts" that arguably disconfirm them is a recurrent issue in linguistic theory.) One very obvious and important focus of contact studies involves pidgin and creole languages. Myers-Scotton includes a programmatic discussion of this area. But the main focus of the chapter in question is on another intriguing language-type: "split" languages that display forms drawn substantially from two genetically defined language families after a long period of convergence. The author attributes such languages to incomplete matrix language turnover. Without doubt, anyone with a serious interest in contact linguistics will read this book with profit.

Mark Newbrook is honorary research associate in linguistics, Monash University, Melbourne, Australia.