Elusive or self-evident? Looking for common ground in approaches to code-switching

The study of code-switching has been undertaken from three distinct perspectives: linguistic, psycholinguistic and sociolinguistic. The linguistic perspective has been concerned with the search for universal grammatical constraints to code-switching. The psycholinguistic perspective has focused on how bilin-gu als’ linguistic systems are stored and accessed in the cognitive system. Finally, the sociolinguistic perspective has paid attention to the social motivations behind code-switching. The methodologies employed in all three cases range from observation in naturalistic settings, more associated with the sociolinguistic perspective, to experimental elicitation, more associated with the psycholinguistic perspective. The linguistic perspective, in turn, does not necessarily identify explicitly with any methodology for data extraction. A need for bringing together these three perspectives has been increasingly felt, giving rise to a few multidisciplinary overviews of code-switching studies, among which the most recent is an edited volume by Isurin et al. (2009). The present volume emphasizes commonalities of approaches to code-switching. Despite the theoretically and methodologically eclectic character of code-switching studies, it seems feasible to bring together various approaches to code-switching spanning all three perspectives as long as they meet the cri-terion of allowing for social explanations. This is the point of view that we took while editing this volume. This volume also seeks to widen the empirical basis of the study of code-switching by including a richly diverse range of contact settings encompassing countries such as Cameroon, Hong Kong, Suriname, Burkina Faso, Dutch Antilles, French Guiana, Netherlands, Papua New Guinea,


Social factors in the study of code-switching
The study of code-switching has been undertaken from three distinct perspectives: linguistic, psycholinguistic and sociolinguistic. The linguistic perspective has been concerned with the search for universal grammatical constraints to code-switching. The psycholinguistic perspective has focused on how bilinguals' linguistic systems are stored and accessed in the cognitive system. Finally, the sociolinguistic perspective has paid attention to the social motivations behind code-switching. The methodologies employed in all three cases range from observation in naturalistic settings, more associated with the sociolinguistic perspective, to experimental elicitation, more associated with the psycholinguistic perspective. The linguistic perspective, in turn, does not necessarily identify explicitly with any methodology for data extraction. A need for bringing together these three perspectives has been increasingly felt, giving rise to a few multidisciplinary overviews of code-switching studies, among which the most recent is an edited volume by Isurin et al. (2009).
The present volume emphasizes commonalities of approaches to codeswitching. Despite the theoretically and methodologically eclectic character of code-switching studies, it seems feasible to bring together various approaches to code-switching spanning all three perspectives as long as they meet the criterion of allowing for social explanations. This is the point of view that we took while editing this volume. This volume also seeks to widen the empirical basis of the study of code-switching by including a richly diverse range of contact settings encompassing countries such as Cameroon, Hong Kong, Suriname, Burkina Faso, Dutch Antilles, French Guiana, Netherlands, Papua New Guinea, United Kingdom, South Africa, Luxembourg and Australia, and using primary data from languages representing a broad variety of linguistic types and affiliations (Niger-Congo, Sinitic, Germanic, Indic, Austronesian, Pama-Nyungan, Celtic, as well as Afro-Caribbean and Pacific English-lexifier Creoles and Mixed Languages).
Capturing a phenomenon as old as bilingualism itself, the term code-switching was originally coined by Vogt (1954) in his review of Weinreich's seminal work Language in Contact (1953). However, it is only from the 1970s onwards that code-switching began to receive sustained attention as an independent topic of study. As such, it has been the subject of a variety of scholarly perspectives, rooted in different theoretical outlooks, relying on different methodologies, and pursuing different goals. Against this eclectic background, it should be little wonder that code-switching has been given quite a few dis-tinct -not to say contradictory -definitions. To complicate matters, it has been regularly set off against another concept, namely code-mixing -which again has been given very different definitions. For example, whereas in Muysken's view code-switching refers to 'the rapid succession of several languages in a single speech event', code-mixing refers to 'all cases where lexical items and grammatical features from two languages appear in one sentence' (Muysken 2000: 1). By contrast, Bentahila and Davies (1983: 302) defined code-switching as the outcome of 'mixing (...) two codes together to produce some-thing which might itself be called a third code'. The notion of a code defined communicatively rather than structurally underlies conversational perspec-tives on code-switching, which propose a distinctive set of definitions for codeswitching and code-mixing. In the perspective taken by Auer (1998Auer ( , 1999, 'code-mixing' (or 'language mixing') is opposed to 'code-switching' (or 'language alternation') on the ground of the degrees of pragmatic salience of structural switch points. Whereas the pragmatic salience of structural switch points is relatively low in code-mixing/language mixing − to the point that structural-ly distinguishable codes are not communicatively distinguishable, it is con-versely relatively high in code-switching/language alternation, whereby speak-ers show awareness of alternating between two structurally distinguishable codes.
As a whole, this present volume deals with code-switching as conceived of in the broadest terms. Hence we use 'code-switching' as a generic term transcending the occasionally perceived opposition between code-switching and code-mixing, while potentially subsuming all definitions, both grammatical and conversational, that code-switching and code-mixing have been given. This implies that we provisionally subsume in the term 'code-switching' both the communicative and structural definitions of 'code', leaving it to individual authors in this volume to specify which of the two definitions is applicable within the framework of their respective contributions.
Giving centre stage to the social dimension of code-switching perhaps does justice to its historical beginnings as a topic of systematic study. Beyond Weinreich, historical overviews of the study of code-switching usually locate these beginnings in the research undertaken by Blom and Gumperz (1972) on the social motivations behind code-switching, producing the first conversational typology of code-switching in the form of a distinction between 'situational' and 'metaphorical code-switching'. Under the influence of Dell Hymes' Ethnography of Communication (1972) as well as of Conversation Analysis as developed by Sacks et al. (1974), Gumperz (1980Gumperz ( , 1982 subsequently went on to lay the foundations for a descriptive framework of social motivations behind code-switching behaviours observed in naturally occurring conversations. A defining factor in the maturing of Interactional Sociolinguistics -as his framework may be referred to -was its opposition to descriptive frameworks giving centre stage to the assumed social indexicality of code-switching. Myers-Scotton's Markedness Model (MM) in particular (1993b) sought to provide an account of code-switching based on the assumption of one-to-one relationships between specific languages and specific social meanings. The notion of 'language' equating 'code' encapsulated in the concept of code-switching has been questioned increasingly, especially in the face of intensive and seemingly unconscious code-switching in certain settings such as in particular in the postcolonial world (Meeuwis and Blommaert 1998). As a result, the idea that structurally identifiable code-switching patterns can constitute codes in their own right came to be reflected in the terminological apparatus used for describing code-switching. We will see that the distinction proposed by Auer (1998Auer ( , 1999 between code-mixing (or 'language mixing') as a code in its own right and what he calls 'language alternation', in which social indexicality may be a factor alongside interaction-internal factors, is more or less directly relevant to describing phenomena that most contributions in this volume take an interest in.

Structural factors and constraints in code-switching
The competition between various structural accounts has been particularly influential in the study of code-switching. The search for universal linguistic constraints on code-switching has on a whole been unsuccessful. Poplack (1980) built the first theoretical model of structural constraints on code-switching by positing two specific constraints, namely the Equivalence Constraint (languages tend to be switched at points where the syntactic rules of both lan-guages are similar) and the Free Morpheme Constraint (languages may be switched after any constituent that is not a bound morpheme). The universal validity of Poplack's constraints was soon called into question in a theoretical fray that gave rise to a succession of ever more specified constraints, such as among other things the Government Constraint (Di Sciullo et al. 1986) or the Functional Head Constraint (Belazi et al. 1994). Still particularly influent is Myers-Scotton's Matrix Language Frame (MLF) model (1993a) which postulates an asymmetric relation between the participating languages, among which one (i.e. the matrix language) is expected to set the morphosyntactic frame for the other, and to constrain the available range of function words and bound morphemes in the process (i.e. what Myers-Scotton calls 'system morphemes'). A more recent development in the search for universal structural constraints comes in the form of the Minimalist Programme, derived from the endeavour for a Null Theory of code-switching which avoids formulating a 'third grammar' specific to code-switching (McSwan 1999). However, an 'all-or-none' structural model of code-switching that stands the test of what seems to be an inexhaustible pool of problematic code-switching data has yet to see the light. That such a model might never see the light is what prompted an increasing provision for language-external factors -such as sociolinguistic factors in particular -as additional predictors for code-switching phenomena likely to override universal structural constraints. Muysken (2000) has occupied a prominent position among structural approaches to code-switching in which sociolinguistic factors are provided for. His structural typology comprises three potentially overlapping types, namely insertional (whereby constituents from A are inserted in a morphosyntactic frame dominated by B), alternational (whereby A and B alternate without en-croaching upon their respective morphosyntactic frames), and congruent lexi-calization (where A and B converge in the morphosyntactic frame). 'Congruent lexicalization' accommodates much of the problematic code-switching data that 'all-or-none' structural models stumbled over. The dominance of one given type over the others is -according to Muysken -predictable on structural grounds: typologically close language pairs may privilege the insertional type or congruent lexicalization whereas typologically distant pairs may conversely favour the alternational type. Independently of these structural factors, sociolinguistic factors may privilege one type over the other: the alternational type may dominate in settings marked by puristic ideologies whereas congruent lexicalization may conversely dominate in (often postcolonial) settings where such ideologies are absent. Muysken leaves open to question the extent to which structural factors override sociolinguistic factors or vice versa. While accepting that grammatical regularities occur across code-switching data, we espouse the point of view that the question is indeed unsettled, and that look-ing for social explanations for code-switching patterns should be generally en-couraged.
Stressing the need for social explanations need not imply that code-switch-ing studies are condemned to 'fuzziness', as long as 'social' is clearly defined in relation to language. Language is by nature a social activity, which implies Elusive or self-evident? 5 interactional usage. As such, its functions are not only to convey factual information, but also to project identities, inherited or creatively shaped by the individual, as well as value systems -ideological and/or esthetic. The contribu-tions to this volume intend to clarify the nature of social factors that have an impact on language variation in general, and code-switching in particular. Both diachronic and synchronic aspects are covered in order to shed new light on problematic code-switching data and a range of specific societal factors be-hind code-switching patterns are identified. These include individuals' posi-tions in social networks (e.g. Beyer), language attitudes and language ideolo-gies (e.g. Migge, Parafita Couto, Deuchar and Fusser). Interactional factors be-hind codeswitching patterns are identified and mapped onto societal factors that are consequential for strategies of identity negotiation (e.g. Anchimbe). While the contributions to this book all address the relationship between social factors and code-switching in a general sense, they also explicitly or implicitly problematize 'ungrammatical code-switching' -a controversial notion that pervades structural accounts of code-switching.

'Ungrammatical code-switching' and the rise of mixed languages
Myers-Scotton is perhaps the scholar who has elaborated most on the notion of ungrammatical code-switching, which she loosely defines as those forms of code-switching that do not conform to the two main principles of grammatical well-formedness implied in the Matrix Language Frame model (MLF, 1993(MLF, , 2002, namely the Morpheme Order Principle and the System Morpheme Principle. The Matrix Language Frame model rests upon a vast amount of data that falls under what Myers-Scotton calls 'classic code-switching'. By contrast, those code-switching data that do not conform to the MLF principles are considered to be indicative of 'composite code-switching', i.e. a form of codeswitching that '... shows convergence in regard to the source of some framebuilding procedures, as well as in the features of the abstract grammatical structure in some lexemes' (Myers-Scotton 2002: 8). Focussing on structural factors in her search for constraints to code-switching, Myers-Scotton provides little social background to the occurrence of composite code-switching. All that can be gathered from Myers-Scotton's comments is that it is a form of codeswitching 'more common' than classic code-switching in certain communities, and that it typically occurs in contexts of language attrition or language shift that may or may not be conducive to 'mixed languages' such as Michif or Mednyj Aleut (2002: 8, 22, 105).
What Myers-Scotton calls composite code-switching is reminiscent of Muysken's notion of congruent lexicalization, even if the latter type is more restricted in its grammatical form than composite code-switching since, according to Muysken, it is typical for contact between closely related languages. This raises questions with respect to a language type referred to as '(bilingual) mixed language', 'split language' or 'intertwined language' like Michif -also discussed by Muysken (2000: 269) -a language type that emerges from the normalized and regularized combination of two matrix languages. Whether composite code-switching or congruent lexicalization (in this case between typologically distant languages, i.e. French and Cree) is the dominant factor in the emergence of a mixed language might be a secondary issue since the two phenomena overlap considerably.
A question of theoretical concern, however, is whether community-wide codeswitching is necessarily or overwhelmingly indicative of ongoing language attrition and shift given that the transition zone between code-switching and mixed languages remains largely uncharted territory (cf. Meakins 2011: 142 ff.). The question whether extensive community-wide code-switching is an epiphenomenon of language shift is also asked in this volume, and is implicitly answered in the negative by Blaxter Paliwala when calling for a critical reexamination of the idea that extensive Tok Pisin-English code-switching in Papua New Guinea's urban middle-class be seen as diagnostic of 'decreolization'. The case of Light Warlpiri (O'Shannessy) in fact shows that the rise of mixed languages does not necessarily coincide with language shift or attrition during nativization. The same question could indeed also be put to Amuzu, whose urban Ewe-English data reveals pervasive code-switching as well, while Yakpo appears to lean more towards attrition and shift in his analysis of the Surinamese scenario. In principle, the question can only be answered by deciding whether code-switching, no matter how pervasive, is a transitory phenomenon, with language loss as the ultimate outcome, or whether the resulting lect has reached a state of stabilisation. Unfortunately, only very few contact scenarios offer the kind of diachronic data that is required to make a case for the syn-chronic stabilization of the mixed lect. Where diachronic data is available (al-beit usually of a shallow time-depth, e.g. Gurindji Kriol, cf. Meakins 2011), they bring forward certain structural criteria of code-switching between the languages involved that allow us to assign the status of "mixed language" to a given linguistic system, and specific social conditions provide the backdrop on which these structural factors can take hold.
Firstly, mixed language are characterised by systematic and pervasive mixing, i.e. rather than being optional, the occurrence of mixed structures in specific subsystems is obligatory (cf. e.g. Stolz 2003), while the grammatical sub-systems of the participating languages are left relatively intact (Thomason 1997b;Matras and Bakker 2003). Systematic use of code-switching certainly applies to those cases described in this volume that involve community-wide code-switching (e.g. Burkina Faso, cf. Beyer; Papua New Guinea, cf. Blaxter Paliwala; South Africa, cf. Stell; Suriname, cf. Migge, Yakpo), in which the alternation of two, even three languages in everyday interactions is a sociolinguistic norm for the vast majority of speakers. Such contact scenarios should be seen as distinct from those in this volume that involve bilingual individuals, their family networks and social or ethno-linguistic subcultures in an otherwise rather monolingual (albeit not monolectal) society (e.g. Wales, cf. Parafita Couto et al; Hong Kong, cf. Chen). The conditions for stabilization of a mixed lect are only given in the latter situation if the specific linguistic (sub)culture is socially (and often geographically) distinct, isolated or marginalized, often in combination with a process of ethnogenesis (e.g. Ma'a/Mbugu, cf. Mous 2003;Medial Lengua, cf. Muysken 1997).
The plethora of possible outcomes of language mixing, including the emergence of mixed languages per se invites us to rethink code-switching as a linguistic phenomenon embedded in a social context. Essential to a new perspective on what structural accounts of code-switching regard as ungrammatical code-switching is to observe code-switching -not just as part of decontextualized example sentences -but as part of usage in the holistic sociolinguistic and cognitive sense developed for it by Croft (2000) among other scholars. Taking as a point of departure the example of the distinction between loanwords and codeswitches made in the literature on code-switching, Backus (this volume) argues that the integration of code-switching into a given base language is more directly identifiable on the basis of frequencies of occurrence and speakers' acceptability judgments than on the basis of whether it complies with matrix language rules. What counts as ungrammatical code-switching in many grammatical models of code-switching can thus be accounted for in terms of 'altered replications' (whereby a new linguistic pattern is used) competing with -or sometimes superseding -'normal replications' (whereby the established linguistic pattern is reiterated). In other words, grammatical 'irregularities' in code-switching -such as the use of morphosyntactic patterns foreign to the host language -may as well just form an instance of an altered replication, which may or may not be read as part of an ongoing process of language change leading to that altered replication gradually superseding the normal replication. Altered replications which -given their general sociolinguistic environment -need not be considered indicative of language attrition or shift, are exemplified by most of the Ewe-English and Akan-English data presented by Amuzu (this volume), and some of the Tok Pisin-English data 8 Gerald Stell, Kofi Yakpo presented by Blaxter Paliwala (this volume), which count as composite codeswitching in the perspective of the MLF. By contrast, the case of Light Warlpiri -discussed by O'Shannessy (this volume) -could serve as an example of composite code-switching, or -in the terms of Backus -of altered replications eventually turning into a norm to the point of laying the foundations to a conventionalized 'mixed code' that has cristallized into a new language. The codeswitching data from Suriname (Migge; Yakpo, this volume), in turn, may be seen as intermediary between these two poles in that code-switching, rather than monolingual discourse, is the communicative norm in these societies, and may or may not be indicative of language attrition or shift.

Attitudes, ideologies, style and social networks
What causes code-switching to become conventionalized or not amounts to identifying which specific social factors cause code-switching in the first place. This book tackles the question of the relationships between code-switching patterns and socio-structural characteristics as they manifest themselves in attitudes, ideologies, and social networks. Yakpo, Migge and Blaxter Paliwala (this volume) relate unexpected structural features of code-switching in the form of congruent lexicalization involving typologically distant languages to multilingual practices in postcolonial environments. Parafita Couto et al. (this volume) experimentally test the correlation between self-reported speakers' attitudes to code-switching forms, and find a positive correlation between disapproval of code-switching and the use or non-use of code-switching at syntactic sites where it could violate the Equivalence Constraint. Attitudes to codeswitching as reported on by Parafita Couto et al. are a function of broader ideological environments, which Ehrhart (this volume) discusses from both a western perspective -in which the concept of language (and grammaticality) goes hand in hand with that of standardization and boundedness -and a nonwestern one -in which the notion of language is not assigned clear-cut boundaries and may thus favour code-switching, as in particular code-switching that could be counted as ungrammatical in the view of structural models of codeswitching. Irrespective of attitudes or language ideologies, Beyer (this volume) demonstrates the merit of assessing the likelihood of code-switching as a function of speakers' position within social networks, showing peripheral and geographically more mobile speakers in multilingual environments to be more likely to use code-switching.
Patterns of language variation need not be directly linked to societal factors, such as those subsumed in language ideologies and expressed in atti-tudes. That fact has been established -also in specific relation to code-switching -by Conversation Analysis (CA), as elaborated on by Gumperz (1982), and later by Auer (1998), as well as by psycholinguistic research. Both Stell and Amuzu (this volume) look at structural patterns of code-switching and examine the degree to which they reflect conversational patterns of code-switching in the same terms as they reflect societal factors. We are reminded by Kootstra's (this volume) psycholinguistic approach that code-switching can be triggered by factors other than social. Detailing the principle of automatic alignment across pair parts as a cognitive mechanism, Kootstra comes to the conclusion that certain patterns of triggering across pair parts, including triggering causing 'ungrammatical' code-switching, can be the outcome of alignment strategies deployed by speakers, as such requiring an account in which interactional factors are given as much consideration as they are in orthodox versions of CA (see further Auer 1998). Stell, Amuzu and Kootstra generally illustrate a specific approach to code-switching whereby the causes of code-switching can be sought within the interaction itself. Another approach can be taken, however, whereby the causes of code-switching are sought simultaneously within the interaction and in the surrounding societal environment by focusing on the role of social identity in code-switching.
Stylistically-oriented studies of code-switching have already shown that the use of code-switching, including ungrammatical forms of code-switching in the view of structural accounts of code-switching, can be read as strategies of identity negotiation deployed by individuals in reaction to both their inter-locutors and societal environments they find themselves in (see e.g. Auer 2007). Anchimbe (this volume) presents us with trilingual code-switching data from Cameroon, involving English, Cameroon Pidgin English (CPE) and French, providing examples bordering on congruent lexicalization between the former two of which the function is to simultaneously signal identification with Cameroonian English-speakers and distance from Cameroonian French-speak-ers. In the same line, Migge (this volume) presents us with code-switching data from French Guyana which generally defy notions of grammaticality as pro-posed by structural models of code-switching. In that case, what could count as ungrammaticality in the perspective of those models seems to be more or less consciously dictated by the deployment of a general strategy of neutrality pervaded with a self-presentational concern for dis-identification from any salient social stereotype that the exclusive use of specific varieties might index. Avoiding negative social indexicalities could, therefore, constitute a potent factor in composite code-switching or congruent lexicalization. Both of the abovenamed case studies refer to code-switching between languages that are structurally related, i.e. English-CPE in the former case, and Eastern Maroon-Sranan Tongo in the latter case, which can be seen as potentially favouring congruent lexicalization. The role of identity negotiation in the emergence of congruent lexicalization between typologically distant languages needs to be investigated further, as in the case of the patterns of Cantonese-English code-switching in Hong Kong, which Chen (this volume) describes as a style deployed by Hong Kong residents with a background of bilingual education for signalling dis-tance from their more monolingually educated peers.
Ungrammatical code-switching is an ill-defined notion, and it seems clear at this stage that none of the grammatical models so far devised can lay a claim to universality. This implies that there may as well be as many sorts of grammatical code-switching as there are sociolinguistic settings. The perspec-tives presented in this book are favourable towards steering away from the search for grammatical universals in favour of a search for language-external regularities in code-switching practices, while providing a range of code-switching data that do not fit neatly into available grammatical models. As such, this volume will hopefully be a stepping stone on the way to a richly detailed social typology of code-switching practices to which future structural and psycholinguistic frameworks will refer in their hypotheses on code-switch-ing.

Authors' contributions
The contributions in this volume are grouped into three thematic blocks that reflect the level of analysis of the authors and together demonstrate the possibilities of integrating structural and socio-linguistic approaches to code-switching.
The contributions in Part 1 by Backus, Kootstra, Parafita Couto, Deuchar and Fusser, and Amuzu have in common that they dedicate (parts of ) their analysis to the interaction of cognitive and socio-pragmatic factors in the creation of code-switched utterances by the individual mind. The articles constitute departures from classical structural approaches to code-switching by taking micro-sociological factors into account. The individual-level cognitive processes of frequency, entrenchment and replication (Backus) usher into communitylevel contact-induced language change only if they are cumulative. The importance of interlocutor alignment for the grammatical structure of code-switched utterances is shown (Kootstra) and evidence is provided for the effects of individual attitudes on the likelihood to use particular types of bilingual structures (Parafita Couto, Deuchar and Fusser). Equally, structural differences are identified between socio-pragmatically oriented, conscious 'marked' choices of codeswitching versus 'umarked' code-switching by default (Amuzu).
The articles in Part 2 by Stell, Anchimbe, Chen, Migge, and Paliwala give centre stage to multilingual interaction in their enquiry. These authors link their analyses of individual code-switching patterns to broader socio-structural entities that speakers index and negotiate as they interact, i.e. ethnicity (Stell, Migge), local, regional and national identity (Anchimbe, Migge) as well as social class (Paliwala) and geographical mobility (Chen). The contributions in Part 3 by Beyer, O'Shannessy, Yakpo, and Ehrhart show varying degrees of a macrosociological perspective in their analysis of code-switching. They cover the rise and stabilization of mixed grammars and lexicons via code-switching, by integrating diachronic and synchronic data, while incorporating the sociohistorical context (O'Shannessy, Yakpo), making links with language policy and 'national' identity issues (Ehrhart), and underlining the role of local and regional social networks in the spread and consolidation of specific code-switching and contact features (Beyer).

Part 1: Code-switching between cognition and socio-pragmatics
Backus' contribution is rooted in usage-based approaches to language varia-tion in which diachronic variation features as a central issue. The author shows that the use of the concepts of normal replication and altered replication can be extended to code-switching data, as a result of which the levels of entrench-ment of foreign lexemes and morphosyntactic structures in a given host lan-guage can be determined. The author argues that a usage-based approach to code-switching is in a position to tackle the issue of the distinction between loanwords and codeswitches, a regular thorn in the side of code-switching studies.
On the basis of experimental techniques that simulate aspects of language use in dialogue (confederate-scripting technique), Kootstra investigates sociopragmatic, structural, and lexical factors that may influence code-switching from a psycholinguistic perspective (most notably the interactive alignment model). The main conclusion from this research is that the choice to code-switch or not and syntactic choices during code-switching are driven by a dy-namic interplay between socio-interactional mechanisms, and cross-language lexical and syntactic overlap.
Parafita Couto, Deuchar and Fusser's contribution aims to test the validity of two distinct structural models of code-switching, namely Myers-Scotton's MLF and McSwan's Minimalist Program (MP) against Welsh-English data, while simultaneously looking for correlations between these data and gram-maticality judgments and self-reported attitudes to code-switching. The au-thors generally find on the basis of specific switch sites where the respective morphosyntactic frames of Welsh and English do not match (i.e. determiner phrases with adjectives) more support for the MLF than for the MP while also identifying a strong correlation between self-reported attitudes to code-switching and code-switching patterns.
Amuzu's contribution presents us with Ewe-English code-switching data which the author places in the perspective of Myers-Scotton's Matrix Language Frame (MLF) model and Markedness Model (MM). The author uses the data to first illustrate the distinction made by Myers-Scotton between 'classic codeswitching' and composite code-switching on the one hand, and between marked and unmarked code-switching on the other. He then goes on to estab-lish correlations between phenomena that the two models tend to account for separately, thus laying the foundation for a joint grammatical and conversational typology of code-switching compliant with Myers-Scotton's terminology.

Part 2: Multilingual interaction and social identity
Stell uses as a point of departure Muysken's grammatical typology of codeswitching (in which a distinction is made between insertional code-switching, alternational code-switching, and congruent lexicalization) and Auer's conversational typology of code-switching (in which a distinction is made between language alternation and language mixing). On the basis of South African codeswitching data extracted from different sociolinguistic settings and in-volving different language pairs, the author shows regularities in patterns of cooccurrence of grammatical and conversational types of code-switching. He eventually proposes a joint grammatical and conversational typology of codeswitching governed by sociolinguistic factors.
Detailing in-depth some of the identity-related functions of code-switching is the purpose of Anchimbe's contribution, in which the geographical focus is placed on Cameroon. Investigating patterns of trilingual code-switching (in-volving English, Cameroon Pidgin English (CPE) and French) in chatrooms, the author establishes links between the use of code-switching made by English-speaking Cameroonians and strategies of group profiling versus the French-speaking Other. Such strategies -which tend to imply the denigration of Cameroonian French-speakers -result in derogatory switching into French, as well as intensive code-switching between English and CPE -invested with strong ingroup values associated with Cameroonian English-speakers.
Elusive or self-evident? 13 Code-switching between typologically distant languages is documented by Chen from the perspective of identity negotiation and style in the form of Can-tonese-English code-switching in Hong Kong. To that end, the author focuses on the stylistic repertoire of one local individual with a US background and contrasts that stylistic repertoire with mainstream local stylistic repertoires. While that individual's stylistic repertoire involves multidirectional insertional and alternational code-switching, mainstream local stylistic repertoires only involve unidirectional English insertional code-switching. The author finally uses these contrasts, as well as the perceptions of these contrasts, as the basis for a reflection on practices of social profiling on the grounds of code-switching usage.
Migge takes us to French Guyana's Maroon communities, among which codeswitching practices involve the joint use of Maroon creole, Sranan Tongo, French, Dutch and English. Pointing out that her code-switching data chal-lenge the notion of a single matrix language, she analyzes them as a 'neutral' frame made up of material shared by typologically close Surinamese Creoles into which speakers variably insert 'marked' features from different linguistic sources. She concludes that the selection of 'distinctive' versus 'neutral' linguistic items appears to be governed by speakers' interactional and self-presentational concerns, possibility for lexical variation or distinction and/or the makeup of speakers' linguistic repertoires.
Paliwala's analysis of code-switching between a creole (Tok Pisin) and its lexifier (English) leads us to Papua New Guinea, where an urban mixed code has evolved in the past decades. The authors shows that in spite of how com-mon this code now is, its concentrated presence in particular communicative contexts and its use by particular social groups (the urban middle class) still makes the code-switching highly indexical of specific social identities. Paliwa-la also contributes to the growing evidence that code-switching is central to the relationship between the lexifier and the creole much more than 'decreolisation' or the 'post-creole continuum'.

Part 3: Code-switching and social structure
Beyer employs the toolkit of social network analysis in the Burkina Faso and Mali border region to account for variation in the intensity of multilingual codeswitching (involving the base language Pana and the contact languages Jula and French) and the degree of proliferation of contact-induced phonologi-cal change. He concludes that linguistic change is driven by socially less inte-grated actors, who code-switch more and are phonologically more innovative. In this linguistically pluralistic environment where normalising pressures are comparatively low, such changes appear to percolate through the linguistic community relatively fast inspite of the peripheral position of the innovating speakers.
Yakpo describes the complex patterns of code-switching observable in Suriname, a postcolonial society marked by an extraordinary ethnolinguistic diversity. The widespread practice of code-switching in Suriname may involve languages with varying degrees of typological proximity/distance, yet the author finds grammatical forms of code-switching across language pairs to be surprisingly regular. This may imply that the occurrence of complex patterns of "fusional mixing" -characterized by complex calquing, congruent lexicalization, innovative approximations of donor structures and constant changes of the matrix language -is possibly more a function of the linguistic enactment of 'Surinamese-ness' than of typological factors.
O'Shannessy's contribution focuses on Light Warlpiri, a new mixed language in Northern Australia, which combines elements of Warlpiri (Pama-Nyungan) and varieties of English and Kriol (an English-lexified creole), and emerged from code-switching between these languages. The author describes unusual characteristics of Light Warlpiri, which involve morphosyntactic innovations that bear little resemblance with any of the co-founding languages. One of those innovations concerns the verbal auxiliary system, in which Eng-lish/Kriol tense, aspect and mood forms are re-analyzed and overlayed on Warlpiri clitic structure, resulting in a formal future-nonfuture distinction in the auxiliary which is not made in the input languages.
Ehrhart takes an ecolinguistic approach to code-switching, emphasizing the constructive role it can play in the educational realm. The author uses case studies from Europe and the South Pacific to illustrate the variety of possibilities in the management of classroom ecologies. In the process, she situates multilingual strategies at school on a continuous scale from more implicit to more explicit language policies. She sheds light on the formation of certain creole languages -which go by the name of 'school creoles' -by presenting them as a linguistic outcome of the processes of learning and acquisition of a new language in a multilingual contact situation.