Ever since Thomason & Kaufman 1988 reawakened linguists to the role of external contacts in language change, and the origin and development of creoles became a major research topic (largely after Bickerton 1981), the field of contact linguistics has burgeoned. Sociolinguists have always been interested in the social side of language contact, but today the grammatical side of contact also is attracting researchers. Books published just since 2000 give some idea of the research activity. Three monographs exemplify the wide interest in grammatical aspects of codeswitching: Hlavac 2003 deals with codeswitching among second-generation Croatian speakers in Australia; and Kovács 2001 treats Australian Finnish and Hungarian in contact with English; and Canut & Caubet 2001 is a collection of articles on codeswitching involving French in various francophonie communities.

Other types of contact also are attracting a good deal of attention. For example, generalizations about the structure of mixed languages are debated in an edited volume, Matras & Bakker 2003. Renewed interest in grammatical aspects of first language attrition is also evident (e.g., Schmid 2002).

What Michael Clyne thinks of as codeswitching (but under a new label, “transference”) is a main focus in the book under review. Drawing on his extensive experience in studying the language use patterns of immigrants to Australia, Clyne has produced a study with an impressive database. The ways in which immigrants to Australia, especially German and Dutch ones, incorporate English elements into their speech is the main subject.

Because of its wide scope, Clyne's volume is best grouped with a number of recent overviews of language contact (e.g., Muysken 2000, Myers-Scotton 2002, Thomason 2001, Winford 2003). Unfortunately, except for Muysken's volume, these were all too new for Clyne to consider their approaches in his own analyses. However, of them all, Clyne's volume may be the most theoretically ambitious because of the many topics that he stakes out as his territory. In some ways, this breadth is the real virtue of the volume; it certainly demonstrates the extent of Clyne's expertise. But breadth can also mean that when many topics are introduced and copiously exemplified, there is less emphasis on analysis and fewer attempts to explain the data. It's not that I necessarily have serious criticisms of Clyne's analyses; rather, it is just that sometimes I find them too sketchy to evaluate and sometimes, even fully to understand. Knowing that Clyne has much knowledge to draw on, I wish he had been less reluctant to offer explanations.

The main thrust of the book is to illustrate the types of semantic and syntactic inroads made by English into the first languages of immigrants to Australia. English, of course, is the dominant language spoken in Australia, even though that society is one of the most multicultural in the world today. Clyne proposes the term “transference” as an umbrella term for most changes that are motivated by external contacts. As is clear in chap. 5, Clyne is most interested in what he has called “trigger” elements, which he argues facilitate a variety of structural changes. Let me outline the organization of the volume.

Chap. 1 is intended as an introduction, giving details about the corpus and the Australian immigrant situation. Clyne indicates that what he calls his “core data” were collected over nearly four decades from L1 speakers of German or Dutch, either bilinguals or trilinguals. The German corpus includes “data from different generations, vintages, and bilingual situations” (p. 6). For example, some data from the 1970s come from German-English bilingual children attending Saturday schools (privately organized classes intended to familiarize the descendants of immigrants with their ancestral culture). Some more recent corpora consist of conversations self-recorded by the bilinguals with their friends, and trilingual data come from interviews, with different segments conducted in different languages. In addition to bilingual data from German and Dutch L1 speakers, examples come from L1 speakers of Croatian and of Vietnamese. The trilingual data seem largely to come from Dutch-German-English, Hungarian-German-English, and Italian-Spanish-English speakers.

Few researchers can claim to have strictly comparable data (from similar consultants in similar situations) or data from situations that result in the truly “natural” combining of languages. But the diversity in Clyne's database (when and how speakers were recorded) is a potential problem when speech examples are cited to support critical points. Specifics about the source of a specific example are not always clear. For example, who is this speaker, and how representative is he or she?

Chap. 2 is a background chapter, dealing entirely with macro level features of language maintenance and shift in the Australian setting. It includes many statistics on such matters as which immigrant groups show the lowest percentages of shift by the second generation. Clyne, who has written a good deal about language maintenance and shift in Australia, is very much “at home” in this chapter, and he writes with more authority here than elsewhere in the volume. He does a fine job of synthesizing facts about what has happened in Australia. The chapter includes a lengthy section on predictive frameworks proposed by researchers elsewhere which take account of sociopsychological factors in regard to potential shift.

From chap. 3 on, the footing changes, and the subject matter for the rest of the volume becomes the grammatical and then the psycholinguistic side of language contact. Chap. 3 deals with many subjects, but most prominently with terminology and Clyne's proposal to replace existing terms, especially to replace “codeswitching” and “code mixing” with “transference” as the label for the contact process, and “transfer” as the label for the product of this process. He goes on to speak of transference at different levels of language (e.g., syntactic transference, morphological transference, semantic transference). The term “convergence” is also introduced; Clyne says that he will use it to mean “making languages more similar to each other (including through transference)” (79). I think it more accurate, however, to view convergence as largely a one-way street, but he does not consider this.

A nagging problem for me starting here, but continuing throughout the volume, is this: What is the status of the term “transference”? Is it intended as a cover term for all grammatical processes that happen during language contact, or is it meant just to replace “codeswitching” and “code mixing”?

In this chapter, Clyne also tries to survey the constraints proposed in the codeswitching/code mixing literature. He overviews the recent literature extensively, but in particular he discusses three models, those of Muysken 2000, Myers-Scotton 1997 (with Myers-Scotton & Jake 2000 also mentioned), and Poplack 1980 along with Poplack & Meechan 1988. Each model is discussed very briefly, and this seems to result in misinterpretations or enigmatic statements. For example, here are four points on which I disagree with Clyne's interpretation of Myers-Scotton 1997. (i) The MLF model was never intended as a model for all instances of elements from two languages in the same clause, so it is not relevant to much of the data Clyne exemplifies. It is a model only for classic codeswitching, meaning a model for data in which only one language is the source of the morphosyntactic frame in the relevant clause or CP (projection of complementizer, which through an editing error is stated in this volume as “complementizer of projection.” (ii) The Matrix Language frame is a frame and no more; all the elements within the frame do not have to come from the participating language that is identified as the ML, only the elements that the model states identify the source of the frame (morpheme order and the critical system morphemes). (iii) The definition Clyne gives of these critical morphemes is “missing words” and therefore does not make sense. It should be “those system morphemes which have grammatical relations external to their head constituent.” Clyne's definition is “those system morphemes external to their head constituent.” (81). (iv) The 4-M model (introduced in Myers-Scotton & Jake 2000) is not a revised version of the MLF model, but a new model on its own; yes, it has applications to codeswitching, but also to other phenomena.

Clyne seems to favor Muysken's model because it goes beyond what I have just called “classic codeswitching.” He is particularly interested in one of Muysken's categories of contact phenomena, congruent lexicalization. This phenomenon occurs when there is a good deal of structural equivalence between the participating languages, which results in a composite structure. Clyne notes approvingly that Muysken's model also takes triggering into account in its recognition that cross-linguistic similarities promote certain outcomes. Clyne has long written about triggers (from Clyne 1967 on). Triggering is a process that promotes a switch in languages within a clause. Such a switch occurs when the speaker uses a word that is a homophone with a word in the other language salient in the conversation, or a word that calls up the other language's saliency in some other way.

Clyne recognizes that there is controversy in the literature over whether the same model should be able to provide constraints on switches of singly occurring words as well as of full phrases when two languages occur in the same clause or sentence. He interprets Poplack as holding that codeswitching operates on the grammatical constraints of both languages, while borrowing is constrained grammatically by the recipient language (84). Myers-Scotton 1997 holds that in a given data set, the same language, which is the recipient language for established borrowings and the Matrix Language in codeswitching, constrains the form of singly occurring words in both codeswitching and borrowing. The Matrix Language in codeswitching also constrains the positioning of Embedded Language phrases within the larger clause; its abstract grammatical features also determine whether the elements meeting a speaker's semantic and pragmatic intentions must occur as Embedded Language phrases or whether these can occur as singly occurring words (cf. Myers-Scotton & Jake 2001, Myers-Scotton 2002). However, Clyne is able to skirt the codeswitching vs. borrowing issue entirely by referring to singly occurring words only as a type of transference. But does this solve anything?

Finally, in this chapter Clyne introduces syntactic transference and convergence. He argues that there are many known examples of grammatical transference (i.e., what are called “outsider late system morphemes” under the 4-M model). In his discussion of transference, Clyne largely follows the five-stage classification of lexical to structural borrowing found in Thomason & Kaufman 1988, finding examples for all five stages. I dispute Clyne's claim that there are structural borrowings that are outsider late system morphemes (i.e., more or less what occurs at the fifth stage), but space does not permit the full argument here (cf. Myers-Scotton 2002).

Chap. 4 is a key chapter. It deals with how the language systems of some plurilinguals (speakers of three or more languages) converge and how this is resisted by some speakers, as well as with examples of different types of transference. Largely, the chapter consists of many interesting examples of different types of convergence and then transference. It finally seems to become clear how Clyne differentiates convergence and transfer. He seems to see convergence as the creation of a new form in a language, but under the influence of another language; in contrast, he sees transference as the actual surface transfer into a language of an actual form from another language.

Clyne emphasizes two points about convergence. First, bilingual commonalities are a major source of instances of convergence. For example, in speaking Italian, rather than using the Italian word pronuncia, the speaker makes the form closer to English pronunciation. The fact that Spanish has the form pronunciación is cited as a factor, too. The result is that the speaker says pronunciazone. Second, trilinguals show convergence in often extending to a third language a feature shared by their other two languages

The bulk of the chapter consists of examples of different types of transference. Many of the examples of lexical transference may be established borrowings in the speech of the immigrant community; that is, it is not clear whether one would even consider some or all of them to be examples of codeswitching. If they are still examples of transference, they are of a different order than elements or structures that may appear only once or twice. Clyne does not consider this issue. For example, if a speaker from a German enclave says “eine kleine car,” is this an example of transference or an established borrowing, or both?

Many of the examples are of changes in word order (syntactic transference). I would refer to such examples as cases of convergence; that is, even though all the surface elements come from language X, the abstract features controlling word order come from another language. To me, this is convergence, an abstract process affecting the morphosyntactic frame. This is a change of a different order from the act of inserting a lexical element from another language, resulting in a clause with surface forms from both languages, but not any change in the abstract features of the frame.

When Clyne gets to his discussion of morphological features, it seems to me that it is hard to maintain a line between transference and convergence. For example, in certain cases (e.g., loss of accusative case marking in Hungarian) one could argue that this is a matter of convergence to English, which does not overtly mark accusative case. Otherwise, Clyne reports a number of interesting findings, too numerous to detail. One of them is that the case system is still intact in his urban German data in the first generation, an argument against rapid attrition of this type of morpheme.

Here as elsewhere, I wish that Clyne had spent more time elaborating on the point of interest that an example illustrates; he tends to present examples and leave it to the reader to understand their significance. Again, this makes me wonder if Clyne tries to cover too much territory in one book.

Chap. 5 is an important chapter for Clyne because it is here that he argues at length about the role of trigger words in facilitating change from one language to another within a clause. When a bilingual homophone comes up in the clause, the speaker may then switch to the other language. Although the idea that trigger words facilitate switching languages is attractive, and although Clyne reports quantitative evidence that such switches do occur after trigger words, the problem is that he does not report on all the possible opportunities when a trigger word does not result in a switch.

Chap. 6 is an especially ambitious chapter because Clyne attempts to survey a number of psycholinguistic models of language production and to relate them to his plurilingual data and the types of convergence and transference that occur.

Chapter 7 introduces a new topic, the ways in which certain aspects of language contact reflect cultural values and identity change across generations. That is, Clyne considers the pragmatic interpretations that using one form rather than another can convey. He touches upon pronouns of address, but also modal particles that may indicate the speaker's attitude toward a proposition, especially in German or Dutch. He also considers the pragmatic force of discourse markers in English in contrast with those in the immigrant languages discussed.

In a final chapter, Clyne summarizes his approach and the results he thinks he has achieved. He makes two points very clear. First, the degree of typological distance between two languages makes a difference in the type and extent of convergence and transference that occurs. Second, he argues it is more useful to speak of facilitation than of constraints (“e.g. syntactic overlap facilitates transversion, not: different syntactic structures constrain switching,” 241). That is, he takes the position that when two languages are in contact, certain changes are facilitated by certain conditions, such as typological or genetic similarities; but he does not seem to see the researcher predicting outcomes in any absolute or even near-absolute way.

In sum, this book is admirable for its ambition, with many examples of contact phenomena. The chapter of most interest to anthropological linguists and sociolinguists is chap. 2, on language maintenance and shift. The core chapters have definite value, but they are for a more specialized audience of researchers who study the same phenomena and therefore are willing to slog through the many examples. In recompense for the slogging, the book has many virtues (e.g., not just data from bilinguals, but also from trilinguals).