The mixed language debate: Theoretical and empirical advances
The study of language contact and its outcomes has expanded greatly over the last few decades, going far beyond its earlier focus on such areas as creole formation and bilingual code switching, to include a much wider range of contact phenomena. Among the latter are the class of contact languages referred to variously as 'bilingual mixed languages', 'intertwined languages', and 'split languages'. This book is devoted exclusively to these outcomes of contact, and reflects the increasing attention that scholars have paid to them in recent times.
A very broad range of such mixed languages is represented here-from the 'classic' and most studied cases like Media Lengua, Michif, Ma'á, and Mednyj Aleut, to less well-known cases such as Para-Romani, Sri Lanka Portuguese, and Sri Lanka Malay, to somewhat controversial cases such as Chamorro and Maltese (Malti). The book consists of ten chapters. The introductory chapter, written by the editors, is an excellent discussion of the main issues addressed in the book, with rich illustrations drawn from several mixed languages. The central concerns of the book reflect those of contact linguistics in general. The most important of these include questions of definition and classification, the linguistic processes that gave rise to mixed languages, the social contexts of their formation, and the relevance of all this to our understanding of contact-induced change in general. This review is organized around these central questions.
The chapters written by SARAH G. THOMASON, PETER BAKKER, CAROL MYERS-SCOTTON, and THOMAS STOLZ are all concerned with the issue of definition and classification. The key problem here is how to delineate the class of contact languages that have been referred to as 'bilingual mixed languages' (BMLs), and distinguish them from other products of language mixture, especially given the fact that all languages are mixed to varying extents. Thomason, appealing to a historical linguist's conception of genetic relationship, offers the classic definition of a BML as 'a language whose grammar and lexical subsystems cannot all be traced back primarily to a single source language' (21). This by and large corresponds to the conventional view of a 'prototypical' BML as 'showing a split between the source language of the grammar and that of the lexicon' (151). But the careful reader will note that hedges such as 'cannot all' and [End Page 223] 'primarily' in Thomason's definition point to the fact that this grammar/lexicon split does not manifest itself in absolute terms in any BML.
Bakker's chapter addresses this problem by positing various subcategories of mixed languages: intertwined languages, converted languages, and lexically mixed languages. Of these, only inter-twined languages such as Ma'a´, Media Lengua, and Anglo-Romani, among others (Bakker claims there are about twenty-five such documented cases) truly qualify as BMLs, or what Bakker calls 'lexicon-grammar (L-G) mixed languages'. The key distinction between BMLs and converted languages is that the former preserve, more or less intact, the morphosyntactic frame (including overt grammatical morphemes) of the grammar language, while converted languages 'use lexical and/or morphological material from one language and map them on to the semantic and grammatical categories from another language' (110). This creates an outcome whose grammar is structurally identical with that of the grammar language, but whose lexical or grammatical morphemes come entirely from the other source language. Among such cases are Sri Lanka Portuguese, Sri Lanka Malay, and presumably also languages such as the Austronesian languages of Papua New Guinea, as described by Ross (1996). Bakker classifies such outcomes as 'form-semantics (F-S) mixed languages'. His third category, lexically mixed languages, have a vocabulary that is 'equally derived from several languages, as a result of heavy lexical borrowing'. These include languages like English, in which the number of words borrowed from Romance languages exceeds that of Germanic words, and Danish, which has borrowed heavily from German. These languages can still be classified in a genetic tree, particularly since most of their grammar, as well as their basic vocabulary, is inherited. Other languages that have experienced heavy lexical borrowing, such as Kormakiti Arabic of Cyprus, Chamorro, and Maltese, also fail to meet the criteria for designation as BMLs.
The chapter by Stolz reaches the same conclusion, on the basis of an in-depth examination of the nature of the grammar/lexicon split in Chamorro and Maltese. The two languages have borrowed extensively from Spanish and Italian respectively, with some incorporation also of function words and limited bound (mostly derivational) morphology. Yet in neither case has the borrowing affected the inherited grammar of the languages, Semitic (Afro-Asiatic) in the case of Maltese, and Austronesian in the case of Chamorro. Stolz concludes that the two languages fail to qualify as BMLs, but seem to fit somewhere in the no-man's land between cases of heavy lexical borrowing and prototypical BMLs. His paper simply reinforces the fact that there are no firm criteria by which we can draw a clear line between the two types of language mixture.
Myers-Scotton's paper also addresses the issue of classification of mixed languages, or in her terms, 'split languages'. She defines a split language as one in which 'some of the abstract grammatical structure underlying the morphosyntactic frame comes from a source other than the major source of its lexicon' (73). She argues that there are three types of split languages. In type A, 'the actual surface late system morphemes come from the less dominant language in one or more constituent types and function as they would in that language' (92). Note that 'less dominant' here refers to social or political dominance. Late system morphemes include case marking, agreement, and other devices that signal syntactic relations. These are the kinds of morpheme that make up the core morphosyntactic frame of a language. She includes Ma'a´ in this category, arguing that all of its late system morphemes seem to be Bantu, while only 'early' Cushitic morphemes (e.g. derivational affixes) are present. It is odd that Myers-Scotton views the Bantu language as 'less dominant' in this situation. She also includes Mednyj Aleut in type A, arguing that Aleut, as the dominant language, provides the major portion of the morphosyntactic frame, while Russian supplies such grammatical features as a complete present-tense paradigm, expression of person/number for past tense, and an infinitival suffix. Placing Ma'á and Mednyj Aleut together in the same category is curious, to say the least, given the stark differences between the two in the composition of their grammars. Moreover, Myers-Scotton's view of Mednyj Aleut contrasts sharply with that of Yaron Matras in this volume.
Myers-Scotton's second category, type B, consists of languages in which the less dominant language 'supplies abstract grammatical structure underlying surface-level late system morphemes [End Page 224] in one or more constituents of the dominant language'. She cites Gangou Chinese and Chaupi Lengua as examples. In the former, the socially dominant and subordinate languages are Chinese and Mongolian varieties (Minhe Munguor), respectively; in the latter, they are Quechua and Spanish, respectively. In both cases, the important point is that part of the abstract grammatical structure comes from the subordinate language. Myers-Scotton's type B seems to correspond to Bakker's converted languages. Finally, Myers-Scotton's third category, type C, is least clearly defined, and only one example is suggested: Michif.
All of the attempts to classify BMLs offered in this book remain inconclusive in one way or another, though in my view, Bakker's is the most satisfactory of all. The basic distinction I would propose is one between BMLs that are the result of what van Coetsem (1988, 2000) refers to as 'recipient language (RL) agentivity', and those that result from 'source language agentivity'. The former mechanism is associated with borrowing, while the latter is associated with transfer in second language acquisition, or what van Coetsem calls 'imposition'. The prototypical cases included among Bakker's intertwined languages all seem to be of the former type, though several other cases-for example, Michif and Mednyj Aleut-remain problematic, as Bakker himself acknowledges. By contrast, 'converted languages' seem to arise in situations where a community shifts to another dominant language, but transfers the abstract grammatical structure of its ancestral language to its version of the language shifted to. Note that in these cases, the subordinate language is linguistically, though not politically, dominant. Like other classifications, this one is by no means exhaustive, and there are many outcomes of contact akin in various ways to intertwined and converted languages, which elude precise classification.
Perhaps the most problematic of these is Mednyj Aleut, which everyone agrees is a BML, but whose grammatical composition appears unique among these contact languages. Mednyj Aleut challenges the lexicon/grammar split of the prototype in many ways, to the point where scholars in fact disagree about the precise make-up of the grammar. YARON MATRAS's paper provides an insightful new perspective on this question. He points out that some languages, for example, Mednyj Aleut and Michif, simply do not adhere to the structural prototype, since their grammatical frame comes from two sources. Matras argues, however, that only one of the languages is the source of the grammatical structures involved in anchoring the predication, as well as the word order and clause-combining devices. He refers to this language as the INFL language, and argues that Russian must be regarded as the INFL or 'matrix' language in Mednyj Aleut, despite the strong grammatical input from Aleut. This is in keeping with Golovko's (1996) view, and opposed to that of Myers-Scotton, as described earlier. Matras also makes the very interesting suggestion that the predominance of Russian-derived syntax in contemporary Mednyj Aleut is the result of a shift of the INFL language from Aleut to Russian over time. This is a very reasonable suggestion, since the history of the language community suggests a slow shift in dominance from Aleut to Russian. In general, too little attention has been paid to historical developments in BMLs from their creation through to their present make-up, so Matras's perspective is a welcome one. He offers a very compelling account of the stages of development that different BMLs undergo over time. For this, he appeals to the notion that different 'layers' of contact phenomena can be introduced to a BML over time. This, he argues, can lead to different types of BML, based on the degree to which grammatical elements from another source are incorporated into the grammatical frame of the INFL language. In the prototypical cases, the first layer consists of wholesale incorporation of lexicon from the external source language, as in the case of Media Lengua and Anglo-Romani. In other cases, a second layer consists of grammatical elements that are readily borrowable, which pattern with the lexifier language. Such elements may include negators, personal pronouns, deictics, lower numerals, and the like. Michif is a case in point, as well as, to a lesser extent, Ma'a´ and Media Lengua. BMLs that manifest the features found in layers 1 and 2 constitute the more prototypical cases. In other BMLs, athird layer is added, consisting of grammatical features such as case markers, adpositions, agreement markers, and the like. Such features are not readily transferred across languages, and their presence in BMLs such as Mednyj Aleut is what distinguishes such cases from the prototypical ones. Matras's account, apart from providing a rationale for different classifications of BMLs, also provides insight into the continuum of structural changes that link degrees of structural [End Page 225] diffusion in BMLs with degrees of structural diffusion in other situations of bilingual language contact.
This brings us to the second major issue explored in this book, the question of which linguistic processes lie behind the formation of BMLs. There is general agreement that these languages can arise in two ways: from processes of borrowing and processes of convergence. But beyond this there is still disagreement or uncertainty about the nature of each process. The outcomes attributed to borrowing include a broad range of BMLs, from Media Lengua to Mednyj Aleut. Outcomes like the former, as well as Anglo-Romani and Ma'á, involve heavy incorporation of lexical items (or lexemes) from the lexifier into the grammar language, and are clearly cases of lexical borrowing taken to an extreme. Yet the different authors in this book propose very different scenarios for how this came about. WILLIAM CROFT argues that Ma'á and Para-Romani varieties like Anglo-Romani are cases of 'death by borrowing' (53), a view with which Matras concurs, while Media Lengua is the result of 'semi-shift' (55). The former view accords with Thomason and Kaufman's (1988) claim that these languages arose via massive borrowing of grammatical structure from an external source language (a Bantu language in the case of Ma'á, English in the case of Anglo-Romani). Bakker, however, presents compelling arguments against this point of view, noting that 'it is impossible to follow the pathways of borrowing outlined in Thomason & Kaufman and end up with a language like Anglo-Romani or Ma'a´' (137). In addition, it seems counterintuitive to suggest that languages like Media Lengua could have arisen through different processes from those that gave rise to languages like Ma'á, given the striking similarity in the lexicon/grammar split in all of these intertwined languages. Such similarity would suggest that they arose in similar ways, via the mechanism of extreme lexical borrowing under recipient-language agentivity. The only difference would be that for languages like Media Lengua, the recipient language would have been the community's ancestral language, while for languages like Ma'a´ and Anglo-Romani, it was the language of the host community to which the migrant groups had already shifted.
Another related issue explored in this book relates to the role of 'lexical manipulation' and code switching in the creation of BMLs. MAARTEN MOUS, EVGENIJ V. GOLOVKO, and Matras all argue that lexical manipulation is a key process in their formation. Mous defines the process as involving 'the "conscious" creation of lexical forms that are parallel in semantic and morphosyntactic properties to an existing lexical item in the language' (209). Matras goes further, distinguishing two types of process: lexical reorientation (insertion from an L2 into an L1) and selective replication (preservation of the lexicon of an ancestral language with shift to a new INFL language). It is clear that the different types of lexical insertion that these authors describe all involve the same mechanism-heavy lexical borrowing under recipient-language agentivity. Notions such as lexical reorientation and selective replication simply describe the results of this mechanism.
The contributors to this volume seem to agree, more or less, that code switching (CS) was not one of the 'mechanisms' involved in the emergence of BMLs. Based on his investigation of the kinds of language mixture found in contact between Russian on the one hand, and Karelian and Yakut on the other, Golovko claims that these mixtures, which he calls 'cases of ethnic and linguistic mélange', hardly fit any known model of CS. This leads him to conclude that 'the idea of code switching as an initial start for the emergence of mixed languages should be rejected' (196). Bakker also rejects the idea that 'insertional code switching' was a path toward intertwined languages (129), arguing that the quantity of embedded lexicon in BMLs is far greater than in ordinary CS and no transitory stage between the initial CS behavior and the resulting mixed language has been documented. Thomason, for her part, seems undecided on this issue, acknowledging that CS was an important 'mechanism' in the creation of Michif (27), but that the linguistic mix found in Media Lengua does not fit most CS patterns all that closely, since the latter do not involve a complete replacement of L1 vocabulary. This is odd, given her earlier suggestion that the differences between the two are a matter of degree rather than of kind. What all of these arguments seem to overlook is the similarity of the underlying process in both insertional or 'classic' CS, and the emergence of intertwined languages. There may be differences in degree, but the mechanism is essentially the same, viz. borrowing under recipient-language agentivity. [End Page 226]
By contrast, it appears clear that the other major pattern of CS, alternational CS, has never led to the creation of a stable conventionalized BML, as the contribution by AD BACKUS persuasively demonstrates. As he argues, a putative BML emerging from alternational CS would require 'fixed constraints on what language contributes which clauses and sentences. But such total compartmentalization is in principle impossible, because the range of semantic content either language must be able to contribute is essentially open-ended' (263).
Another important issue addressed in the book is the extent to which structural borrowing plays a role in the creation of BMLs. We have already ruled that out in the case of the more prototypical intertwined languages. Languages further from the prototype, such as Michif and Mednyj Aleut, however, pose a ticklish problem, since so much structure has transferred from each of the languages in contact. The contributions by Bakker and Stolz provide sound arguments as to why these BMLs should be distinguished from languages like Kormakiti Arabic and Chamorro, in which some degree of structural borrowing accompanied heavy lexical borrowing. It could be argued that Michif was somewhat similar, since its creation involved the incorporation of French nouns as well as French NP grammar into the morphosyntactic frame of Plains Cree. But the process in this case is far more similar to the incorporation of NP (and other) islands that is typical of classic CS. The case of Mednyj Aleut, however, remains somewhat of a mystery, since so many grammatical elements from both Russian and Aleut make up the morphosyntactic frame of this language. The pattern of mixture here has parallels in other cases of structural diffusion accompanied by heavy lexical diffusion, such as Asia Minor Greek.
One of the most difficult questions regarding BMLs is the role that social factors played not only in their creation, but in determining the differences in their overall makeup. The papers by Thomason, Golovko, and Croft all address this issue, and agree that BMLs are, first and foremost, the result of a conscious choice on the part of a community that wishes to signal its own unique identity. Thomason argues that BMLs are the result of 'change by deliberate decision, [which] is a quintessentially social factor' (35). Golovko's paper expands on this, arguing that 'folk linguistic engineering' lies behind the emergence of BMLs, and in particular the strategies of lexical manipulation they involve. All three authors also appeal to the notion that BMLs result from conscious 'acts of identity'. Croft offers a very detailed argument for this claim, appealing to parallels between biological evolution and language evolution. In the former case, he argues, 'the environment is the ecosystem and other organisms of the same species'. In language evolution, 'the environment is the experience to be communicated and other interlocutors' (45). Croft also tries to make connections between types of social environments and the types of linguistic elements that are incorporated into BMLs. He argues for three different kinds of social context, each leading to a different result. First, 'death by borrowing' leads to outcomes like Ma'á and Para-Romani; second, 'semi-shift' leads to outcomes like Media Lengua; finally, 'mixed marriage languages' include Mednyj Aleut and Michif, which symbolize speakers' identification with a new community.
Unfortunately, it is not clear whether the social contexts Croft describes tell us anything about the nature of the social settings or environments themselves. Croft's first two contexts really seem to parallel his conception of the different historical processes of change that, in his view, led to the emergence of BMLs. We argued against this view earlier. In the third case, there is no compelling evidence that it was mixed marriages, per se, that led to the peculiar mixtures found in the BMLs in question. Thomason also notes that social factors determine whether the community's decision to combine two languages will lead to a stable BML, but she offers no exploration of precisely how different social settings lead to particular outcomes. While these papers all make interesting suggestions about the role of social factors, their failure to explain that role merely emphasizes the need for in-depth study of the social ecologies in which BMLs emerged. Similar research on the social contexts of creole formation could supply a model.
On the whole, this volume offers a great deal of food for thought by exploring a number of key issues in the study of BMLs in particular, and language contact in general. It provides rich information on diverse cases of bilingual language creation and the linguistic and social processes that underlie it. It also points to the need for further research to address the three most important [End Page 227] challenges facing the field of contact linguistics today. First, we should heed Bakker's call for a comprehensive taxonomy of BMLs and other contact languages, since, as he argues, this is 'a necessary step in an advancement of a theoretical discussion of mixed languages' (142). Second, we need to build on previous research that has explored the linguistic processes involved in the creation of BMLs, and investigate further how they parallel those involved in other cases of contact-induced change, or for that matter, in change due to internal motivation. Finally, there is pressing need for further insight into the social forces and processes that shaped the creation and course of development of BMLs. This book certainly offers a sound foundation for such future exploration.
Department of Linguistics
The Ohio State University
222 Oxley Hall
Columbus, OH 43210