Code-switching (CS), ‘the alternating use of two languages in the same stretch of discourse by a bilingual speaker’ (xii), is a flourishing sub-area of the study of bilingualism (Bhatia & Ritchie Reference Bhatia and Ritchie2004), whose findings have far-reaching implications for various neighboring fields, not least linguistic theory. Drawing on several decades of intense scholarship, The Cambridge handbook of linguistic code-switching (LCS) is a rich and rewarding volume. It is to serve ‘a readership ranging from advanced undergraduate students to researchers’ as ‘a guide to the main theoretical and empirical contributions to the [linguistic] study of CS’ (xii). The editors observe that ‘linguists are converging on the view that CS must be examined from structural, social, and psycholinguistic perspectives concurrently’ (xiii), all of which are amply elucidated in the volume. Apart from the editors' overview (‘Themes in the study of code-switching’; 1–17), the bulk of the book consists of eighteen topical chapters organized into five parts. Part I, ‘Conceptual and methodological considerations in code-switching research’ (21–93), contains four articles: ‘Research techniques for the study of code-switching’ (Marianne Gullberg, Peter Indefrey & Pieter Muysken), ‘On the notions of congruence and convergence in code-switching’ (Mark Sebba), ‘Code-switching and transfer: An exploration of similarities and differences’ (Jeanine Treffers-Daller), and ‘Loan translations versus code-switching’ (Ad Backus & Margreet Dorleijn). Part II, ‘Social aspects of code-switching’ (97–159), addresses, beside ‘Sociolinguistic factors in code-switching’ (Penelope Gardner-Chloros), the ‘The Conversation Analytic model of code-switching’ (Joseph Gafaranga), ‘Code-switching and the internet' (Margreet Dorleijn & Jacomine Nortier), and ‘Phonetic accommodation in children's code-switching’ (Ghada Khattab). Part III, ‘The structural implications of code-switching’ (163–237), examines possible ‘Phonetic reflexes of code-switching’ (Barbara Bullock), ‘Code-switching between typologically distinct languages’ (Brian Hok-Shing Chan), ‘Language mixing in bilingual children: Code-switching?’ (Natascha Müller & Katja Francesca Cantone), and ‘Code-switching between sign languages’ (David Quinto-Pozos). Topics of Part IV, ‘Psycholinguistics and code-switching’ (241–306), are ‘Code-switching and language disorders in bilingual children’ (Adele W. Miccio, Carol Scheffner Hammer & Bárbara Rodríguez), ‘Code-switching, imperfect acquisition, and attrition’ (Agnes Bolonyai), ‘Code-switching and the bilingual mental lexicon’ (Longxing Wei), and neurolinguistic aspects of CS (‘Code-switching and the brain’ by Marta Kutas, Eva Moreno & Nicole Wicha). The final part, ‘Formal models of code-switching’ (309–357), presents two specific CS models, portrayed by their respective proponents: the application of the Minimalist framework to CS (‘Generative approaches to code-switching’ by Jeff MacSwan), and the highly influential Matrix Language Frame (MLF) model with the supplementary 4-M model and Uniform Structure Principle (USP) (‘A universal model of code-switching and bilingual language processing and production’ by Carol Myers-Scotton & Janice Jake). Additionally, the book contains miscellaneous frontispiece matter, a comprehensive 58-page-long bibliography (358–415), a subject index (416–420), and an index of languages cited (421f.).

LCS treats or raises a vast range of linguistic, psycholinguistic and sociolinguistic issues in the investigation of CS. Since a brief review cannot do justice to all of these, let us merely touch on a few selected points. As for basic concepts, Bullock notes ‘a good bit of terminological confusion surrounding the term “code-switching”’ (164), and Treffers-Daller remarks that ‘the term CS has become so polysemous and unclear’ (66). Indeed, descriptions or definitions of what CS is vary markedly in LCS. While the editors by way of introduction provide what is a reasonably standard definition of CS (xii, cited above), they later describe CS as ‘the ability on the part of bilinguals to alternate effortlessly between their two languages’ (1, emphasis added; similarly, Miccio et al. 241f., citing Meisel Reference Meisel1994), i.e., as a latent skill rather than actualized linguistic behavior. Nevertheless, the traditional meaning as well as the face value of the term ‘code-switching’ suggests an overtly manifested activity (‘the use of overt material (from single morphemes to entire sentences) from Language B in Language A discourse’; Backus & Dorleijn 76, emphasis added), making it preferable to utilize another expression such as ‘CS competence’ about the underlying ability. Apart from this, it is questionable whether it inheres in the notion of CS that the switching is carried out without difficulty. In their introduction, in turn, the editors describe CS as the alternation ‘between languages in an unchanged setting, often within the same utterance’ (2, emphasis added), thereby including in CS the switching between utterances, whereas MacSwan (309) states that ‘CS involves the mixing of phonologically distinctive elements into a single utterance’ (emphasis added; cf. also Chan 182). Khattab for her part employs CS as ‘an umbrella term for discourse phenomena which [in her investigation] also included borrowing, mixing, loan words, alternation, code-shifting, and [the momentary return to a monolingual base]’ (149). The terminological difficulty is augmented by the existence of many related designations. Describing language decline in children, Bolonyai notes that ‘L2 influence on the unstable L1 may manifest itself in various forms of language mixing, such as transfer, interference, transference, structural borrowing, selective copying, calquing, convergence, restructuring, creative innovation, incorporation of L1 into L2, covert CS, composite CS, or “third-system” innovation, among others’ (263). Elsewhere, LCS records still other terms like code-mixing, code-changing, language switching, language shifting, style shifting, etc. The readable chapters on congruence/convergence, CS/transfer, and loan translation clear up much, though not all, of this terminological diversity.

The editors indicate that their ‘broader objective is to help dispel the myths and misperceptions that surround the bilingual practice of CS’ (xii). One target is ‘the popular misperceptions of CS as an uncontrolled speech form’ (13), another the frequent social stigmatization of CS (13). Hence, ‘CS is not an indicator of deficiency in either or both language(s)’ (Chan 182), but – as opposed to code-mixing, which does not obey the constraints of normal CS – is done ‘with intent or as a matter of choice’ (Miccio et al. 243), ‘reflects the skillful manipulation of two language systems for various communicative functions’ (Bullock & Toribio 4), and constitutes ‘a measure of proficiency’ (Miccio et al. 244). Thus, as Bolonyai (262) notes, ‘[f]rom a grammatical perspective, CS has been characterized in the literature as a sophisticated, structurally coherent, rule-governed behavior that requires a great deal of bilingual competence’. That CS demonstrates considerable linguistic skill and creativity is undeniable. On the other hand, not so little CS takes place out of sheer necessity, simply because speakers lack, or do not have easy access to, the right word, expression, or construction in the language they momentarily use (cf. Bolonyai 260f., 263 on bilingual children). From this angle, albeit highly skilled, CS serves not infrequently as an emergency expedient, when the speaker's resources in one of his/her languages do not fully suffice.

Of special interest to general linguists are the discussions in LCS that broach theoretical linguistic issues. The matter of formal model construction is tackled prominently in MacSwan's chapter, which, like his earlier work (e.g. MacSwan Reference MacSwan2000), advocates a Minimalist approach (Chomsky Reference Chomsky1995) to CS. Discussing German–Italian CS, Müller & Cantone (214) represent MacSwan's assumptions about the set-up of the bilingual language faculty as shown in Figure 1.

One of the features of the model in Figure 1 is, as MacSwan states in his contribution to LCS, that ‘[e]very syntactic head must be phonologically parsed at Spell Out’ (331). Following the tenets of Optimality Theory, he claims that (a) the phonology of a language is a set of constraints (or, more precisely, constraint dominance hierarchies), ordered on a language-specific basis, and (b) phonological representations are converted into surface forms in a single step (330, 331). From these starting points, he arrives at the conclusion that (c) ‘bilinguals have a separately encapsulated phonological system for each language … in order to avoid ranking paradoxes’ (331, stated twice; emphasis added). Thus, ‘[w]hile both phonological systems are involved in deriving PF, each is compartmentalized and restricted to terms of its respective lexicon’ (MacSwan Reference MacSwan2000: 52). Yet, even though the postulates that phonology consists of constraints exclusive of rules and that derivations are completed in one sweep are highly controversial (see, for example, several papers in Vaux & Nevins Reference Vaux and Nevins2008), these assumptions are introduced with no warning to the unsuspecting student reader. Moreover, the claim that the bilinguals' desire to evade Optimality-Theoretic ranking paradoxes is the reason for keeping their phonologies distinct is not self-evident and needs to be justified empirically. Finally, from the angle of second-language acquisition at least, learners frequently transfer native productive phonological rules from one language to another, as when Russians palatalize, Icelanders preaspirate and Swedes/Norwegians postalveolarize in miscellaneous L2s. Such cases suggest that phonologies are not necessarily kept absolutely separate, requiring some addition to, or emendation of, the model if it is to be valid for sequential as well as simultaneous bilingualism (cf. also MacSwan Reference MacSwan2000: 52).

Figure 1 A Minimalist view of the organization of the bilingual language faculty (adapted from Müller & Cantone 214).

A long-standing issue in CS research is whether there exist special structural restrictions on CS (14) and, if so, whether these are valid for all languages. Discarding CS-specific principles (320), MacSwan rejects Shana Poplack's Equivalence Constraint, but sympathizes with the factual contents of her Free-Morpheme Constraint, which prohibits word-internal CS (313). Since syntactic heads are parsed phonologically at Spell Out in Figure 1, ‘the boundary between heads (words) represents the minimal opportunity for CS’ (331), a situation that results in ‘the prohibition against switching head-internally’ (330). Seeming counterexamples are explained as (nonce) borrowings. However, researchers have turned up data from several languages that cannot easily be dismissed as merely apparent violations of that prohibition (cf. 313). One case stems from Maori–English CS, which documents linguistically mixed passive forms of the type turnhia ‘be turned’, helptia ‘be helped’, changedngia ‘was changed’, etc. (Eliasson Reference Eliasson1991: 22; Reference Eliasson1995: 48). The English stems of such formations include all the forty-odd phonemes of New Zealand English as against the merely fifteen basic phonemes of native Maori words and may exhibit any of the complex English phonotactic structures CCVC, CVCC, CCCVCC, CCVCCC, and so on, as opposed to the simple (C)V(V) syllables of Maori, which have no consonant clusters or syllable-final consonants. With respect to (a) sound inventory and (b) phonotactic structure, then, putative loans of the type turn, help deviate strikingly from established English loans in Maori. Moreover, not infrequently, they (c) duplicate phonologically adapted counterparts, as when trucktia ‘be trucked’ is sometimes used in spite of the existence of tarakatia with a fully nativized stem. Given these facts, Thomason (Reference Thomason, Eliasson and Jahr1997: 191) concludes: ‘Maori has not (so far) undergone major phonological restructuring through borrowing from English, and unassimilated words, or parts of words, are therefore easily identified as code-switched elements’. Evidently, word-internal switching does exist, and the model in Figure 1 would need to be modified accordingly.

Compared to the object of monolingual investigations, the subject matter of CS research is doubly complex, involving, in Sebba's words, ‘[p]laying two games at once’ (48). Generalizing from findings is not always easy. Backus & Dorleijn note in their contribution that ‘we have raised more questions than we have answered’ (92), an observation that holds for several other discussions in LCS as well. The volume makes it clear that the study of CS is a young field that is still to an extent searching for firm foundations. But it is to the book's credit that its authors do not shun away from the open issues. Its topical coverage, moreover, is truly impressive. Even areas like sign language CS, with thus far only a modest amount of literature (228), and the possible phonetic reflexes of CS – ‘an open field of inquiry’ (180) – find their place in the book. Most fundamental concepts and many core principles of CS research are presented or touched on in some way or other. As a whole, LCS judiciously abstains from advocating a particular theory or model of CS. Two specific frameworks, those of MacSwan and Myers-Scotton, are, however, accorded their own chapters. The proposals made by these scholars point in rather disparate directions, though (cf. the dispute between MacSwan and advocates of the MLF model in the journal Bilingualism: Language and Cognition 5.1 (2002), 8.1, 8.3 (2005)), for which reason a further article critically comparing the strengths and weaknesses of both models would have been of interest. Coherent full-length presentations of other structural approaches such as those of Shana Poplack and Pieter Muysken are missing as well as a separate review of the history of CS research (but see, for example, Chan 182–188, MacSwan 310–319). Chapters on the implications of CS for linguistic theory, CS and language universals, and the role of CS in language change (for the latter, see, for example, Thomason Reference Thomason, Eliasson and Jahr1997: 189–195) might have been further desiderata for inclusion. As for the real-life execution of CS research, students using LCS could have benefited from a brief chapter on the purely practical aspects of recording, transcribing, organizing, storing and analyzing CS data, matters that are not quite within the purview of Gullberg et al.'s excellent review of research techniques.

In sum, in the recent flood of linguistic handbooks, some of which partly duplicate each other, The Cambridge handbook of linguistic code-switching is a particularly welcome addition, meeting a long-felt need. The volume surveys a wealth of CS research from the past few decades, scrutinizes insightfully major issues and many proposed principles, and opens up new paths in the investigation and understanding of CS. Being the first handbook of its kind and unique in its depth and breadth of coverage, LCS will not only serve as a useful reference work, but is also likely to stimulate new CS research in the next few years.