This article draws attention to three general problems in existing theories and models of contact-induced language change: the problem of autonomous types of change, that of autonomous contact languages, and that of the metaphors used in contact linguistic terminology. Parting from a discussion of these problems and two case studies of contact varieties that heavily challenge existing models of contact-induced change (Chamorro and Zamboangueño-Chabacano), I provide a new and comprehensive model based on the conception of contact-induced change as a continuous space, in which interrelated and interconnected parameters dominate over autonomous types. This model is embedded in an ecological conception of language and language contact, as expressed in Ludwig, Mühlhäusler and Pagel (in press). The relevance of the early years of contact, as seen from the perspective of the presented model, is addressed in the last section and offers one possible prospect to future discussion and research.
1 Introduction and Aims
The categories and types that we isolate from the world of phenomena we do not find there because they stare every observer in the face; on the contrary, the world is presented in a kaleidoscopic flux of impressions which has to be organized by our minds […]. Whorf 1956: 213
The Whorfian view of the relation between “the world of phenomena” and its being organised can be easily adapted to the scientific study of language: linguistics, like any other structured or unstructured attempt to interpret our natural and social environment, is basically a set of categories, types and terms whose meanings, frames and relations are subject to historical and other dynamics (see Gadet and Pagel, in press). While this may seem intuitive with regard to, e.g., the description of languages or its organisation in genetic and typological groups, it is rather challenging if we consider the theoretical framework underlying these approaches, reflected in questions like: What is a language? How does it come into being? What are its paths of evolution? One fundamental issue in this framework concerns the role of language contact in the ontology and evolution of languages. Here, a principal distinction has evolved through the history of linguistics between language evolution on one hand and language contact on the other (see Thomason, 2009). While whole linguistic disciplines are based explicitly or implicitly on either of the two categories (and methodically exclude the other), the precise origin and ontology of this distinction remains a largely unstudied matter.
This volume and the present article are clearly situated in the second of the mentioned categories. The aims of this article can be stated as follows:
to draw attention to some general problems of categorisation, typification and metaphor in existing theories of contact-induced change;
to provide approaches to solving these problems, resorting to, among others, the conception of linguistic ecology (as defined in Ludwig, Mühlhäusler and Pagel, in press),
to present a comprehensive, integrative model of contact-induced change, based on the conception of contact-induced change as a continuous space in which interrelated and interconnected parameters dominate over autonomous types;
to comment, from the perspective of this model, on the question of this volume, whether the early years of contact are of any particular importance for the understanding of contact-induced change.
Aims 1–3 thus refer to the disclosure and discussion of our scientific organisation of the world of language contact and lead to the proposal of a different organisation based more on continuity than on autonomous categories and types. Aim 4 will relate this different organisation to the topic of this volume.
2 Major Problems in Existing Theories of Contact-Induced Change
There have been countless contributions to the theory of language contact and contact-induced language change since the pioneering works of scholars like Hugo Schuchardt, Max Müller, Hermann Paul or William D. Whitney. Among these, the comprehensive theories and models concerning contact-induced change are of primary relevance for the deduction and contextualisation of the ideas I wish to develop here. By “comprehensive” I mean theories and models that claim to cover all major phenomena of contact-induced change in any natural, spoken language at any time, comprising processes and factors at work as well as possible outcomes. Studies concerned only with selected phenomena of contact-induced change like creolisation, code-switching or convergence will not be explicitly contemplated.1
Building on and advancing the frameworks of Haugen (1953/1969), Weinreich (1953/1968), Appel and Muysken (1987) and Thomason and Kaufman (1988), Winford (2003b: 23–24) gives an overview of the “major outcomes of language contact” that can serve as our point of departure here. Despite its title, it also includes types of change, processes and external factors. Its fundamental division is between “language maintenance”, “language shift”, and “language creation (new contact languages)”. Disregarding specific terminological and/or definitional considerations, this tripartite structure is common in contact linguistics, perhaps even intuitive.2 Table 1 gives a summary of Winford’s overview.
A first important notion concerning the structure of this and most other models is that the three basic types of change are arranged side by side, that is they are being understood as autonomous phenomena, showing no, or no overt relation to each other. One implication of this understanding is that speakers in a given contact situation will act according to one, or perhaps more than one of these types, but that we can always tell what this type is/these types are. Another implication is that we can always tell exactly up to which point a code is being maintained, when it is being shifted and when a new one is being created.
Indisputably, we cannot. First of all because that would require a precise definition of “a code” as opposed to another, which is, as is well known, not a matter of linguistic structure alone but depends on a seemingly infinite number of code internal and external factors.3
A second remark on the structure of this and most other models is that they draw on two different strategies of arranging the phenomena in the proposed space: as types, and as transitions – in other words: as non-continuous, and as continuous phenomena. The basic structure in Winford’s overview is one of types. But below that first level, arrangements become rather heterogeneous: the maintenance type (A) divides into two situational types, of which the first is characterised by gradual transitions and a clear correspondence between the degree of contact and the quality of the borrowed elements. The second, in turn, distinguishes between four types of contact that show no overt connection, corresponding to results that display at least some gradual arrangement. The shift type (B) is divided into three subtypes that are at least implicitly gradual, and correspond to fully gradual results. The creation type (C) subdivides into three autonomous language types whose characteristics also do not display any connection.
Perhaps the most striking structural disproportion can be observed in the category of “results”. While this category exists in both type (A) and type (B), and the phenomena therein are arranged in an explicitly or implicitly gradual fashion, it is completely absent in type (C). Instead, results are given here in the form of proper language types, equally autonomous and unconnected as are the main types of contact-induced change (A, B, C). Each of the language types in (C), again, is characterised by certain features, which, although they do not display any relations to those of the other two language types, they do show relations to maintenance in the case of bilingual mixed languages (“akin to cases of maintenance”), and maintenance and shift in the case of creoles (“akin to cases of both maintenance and shift”). There appears to be, then, a fundamental structural difference between contact-induced change in terms of maintenance and shift on the one hand, and contact-induced change in terms of creation on the other. Yet, there also seems to be a covert relation between them.
An important implication to be extracted from the structural properties of the creation type is this: speakers or codes acting in accordance with this type always do so in correspondence with one and only one of the three contact languages it comprises. That is to say, either a bilingual mixed language, or a pidgin, or a creole (or a deviation of these types like e.g. a semi-creole) is formed. This is, of course, problematic in many respects. First of all, it suggests that speakers or their codes acting according to the creation type specify, at a very early phase in contact, the type that will be the outcome of the contact. It is simply inexplicable how and on what grounds such a ‘decision’ could be made. Furthermore, numerous studies have shown in the past how difficult it is to define these language types on synchronic grounds. There appear to be more cases challenging the existing definitions than there are cases affirming them. Of particular interest are those cases combining the characteristics of two or more of the seemingly autonomous types, pointing to at least some inherent relation between them. I will briefly discuss two such cases—Chamorro and Zamboangueño-Chabacano—in section 3.
On grounds of the observations made so far, two major problems in the existing models of contact-induced change can be formulated:
The autonomous type problem: The three basic types (maintenance, shift and creation) are pictured as autonomous, showing no relation with each other. For a coherent theory, either their edges must be defined sharply, or if thought of as transitions, it must be defined where these transitions lead, i.e. the three must be arranged in an order. The latter solution seems more conceivable than the first.
The contact language problem: While the internal structure of both maintenance and shift is essentially perceived as gradual or continuous, that of creation is not. Instead this type subdivides into further autonomous types with no, or at least unclear relations to each other. Since the three types are of equal status, we would expect a good theory to proceed uniformly in this regard.
Before continuing with these two, a third problem of existing models of contact-induced change will demand our attention. It is metalinguistic instead of structural:
The metaphor problem: A substantial part of the conventional terminology of contact linguistics—reflected in terms like borrowing, interference, diffusion, imposition, mixing, transfer or donor/recipient language—resorts to metaphors that are not in concordance with our theoretical understanding of the concepts and processes denoted by these terms. This is not a trivial matter: metaphors privilege certain perceptions and actions regarding the phenomena they signify (Ortony, 1979; Lakoff and Johnson, 1980). Hence, common misunderstandings about the nature of contact-induced change may be rooted in the metaphors the respective terminology resorts to.
When we speak of transmission or transfer in connection with language, we do not imply that the speaker removes or relocates material from one language (the source or donor language, sl) to another (the recipient language, rl). When borrowing, what the speaker actually does is either imitatesl material in the rl (i.e., copy the sl model) or adaptsl material to the rl (i.e., make similar to the rl). 2000: 49
Meillet thinks that the term “mixture” is impropre; in any case the other, “borrowing”, is even more so, although we do not want to and cannot do without it. It gives the impression, as if something was taken over from a per se foreign possession, and not, as it really is, from the already appropriated possession of something foreign. schuchardt, 1914: 390, my translation4
Although most contact linguists would acknowledge, like Van Coetsem, the problematic implications of the conventional terminology of language contact, the reservation expressed by Schuchardt, according to which linguistics cannot do without this controversial terminology, is also often heard. While this conventionalist or traditionalist argument can be interpreted as paradigmatic behaviour in the sense of Kuhn (1962), conventionalism is at least not a logical element of science, and the most adequate terminology should always be sought.5
Nothing is borrowed in language contact: the “donor language” is not deprived of anything; and – more importantly – the “recipient language” does not take over anything identical with anything in the “donor language”. Terms such as “transfer” pose the same problem, since they also suggest identicality of originals and copies. Non-identicality of originals and copies is a fundamental principle of our framework. Johanson 2002: 288
Johanson introduces code copying (but note the term “copy” in the quote by Van Coetsem above and also a much earlier use in Schuchardt, 1885)6 and code-alternation as the two chief processes in code interaction, and distinguishes between a model code, from which a certain element is copied, and a basic code, into which it is copied. There are several immediate advantages of this approach: first of all, Johanson’s terminology evades the metaphorical pitfalls of the traditional terminology. Second, by speaking uniformly of codes, the problem of defining languages, as opposed to dialects or varieties, is circumvented (2002: 286, 287).7 Third, the approach provides an interesting solution to the notorious problem of distinguishing between borrowing and code-switching: code alternation in Johanson’s framework “means shifting from one code to another, juxtaposing elements belonging to different systems. Many cases of so-called “code-switching” imply alternate choices of codes” (2000: 287). The notion of code copying, in turn, “is construed in a rather wide scope. It includes various phenomena such as borrowing and calquing and deals with them as similar types of interaction within one and the same paradigm” (2000: 286).8 It is important to emphasise that this process is considered to be effective in what is diachronically perceived as maintenance, shift and creation alike, and may therefore already hold out the prospect for a solution to our problems 1 and 2 stated above.
Kriegel, Ludwig and Henri (2009) and Kriegel, Ludwig and Salzmann (in press) develop Johanson’s framework further in several dimensions. Of primary importance for this article is their conception regarding the conventionalisation of copies in language contact. They propose a “conventionalized systemic integration continuum”, ranging from the situational-pragmatic process of “interactional copying” to the systemic stage of (overt or covert) “conventionalised copies”. The “propagation” of a linguistic variant (Croft, 2000) is identified as the mechanism to transform the situational instances of copying (often interpreted as instances of code-switching) from a model code into fully systemic and conventionalised elements of the basic code (borrowings and the like).
We will adopt Johanson’s terminology as well as its expansions when we return to the autonomous type problem (problem 1) and the contact languages problem (problem 2). This will be done in two steps: first, two cases will be illustrated that qualify as borderline cases for existing models of contact-induced change, and hence pose a problem to modelling contact-induced change on the basis of autonomous categories and types (section 3); then, the two problems in question will be theoretically discussed and the elements collected that qualify for a model solving them (section 4). This model will be outlined and explained in the subsequent section (section 5), before in the last section (section 6) the notion of the “early years of language contact” will be addressed.
3 Two Borderline Cases
Chamorro is an isolate in the Malayo-Polynesian branch of the Austronesian family (Dyen, 1965, Greenhill, Blust and Gray, 2008) and the indigenous language of the Mariana Islands, located in the Western Pacific. It has undergone significant contact-induced change since the European discovery of the Pacific in 1521, the major influences coming from Spanish, Philippine languages, and recently English. This overview is concerned only with the contact between Chamorro and Spanish, dating roughly from 1668 (first catholic mission established on Guam) to 1898 (end of the Spanish rule as a result of the lost war against theU.S.). In Pagel (2010: 31–160) an extensive analysis of the Spanish element in Chamorro is provided; following here is a brief summary of its results.9
In quantity, contemporary Chamorro texts (both spoken and written) display an average of between 20 and 40 percent of hispanisms (comprising both lexical and grammatical elements).10 Regarding quality, Chamorran hispanisms can be assigned to three different groups, according to their grammaticality, functionality, and the presence or absence of indigenous (or other) alternatives. A first group (systemically indispensible, because they are highly grammatical, functional, and without alternatives) includes: determiners like the indefinite article Ch. un (Sp. un) or the demonstrative este ‘this’ (Sp. este); prepositions like Ch. sin ‘without’ (Sp. sin) or put ‘about’ (Sp. por) that gave Chamorro’s predominantly agglutinating morphology a more analytical character; the grammaticalised markers for future tense Ch. para and siempre (Sp. para ‘for, in order to’, Sp. siempre ‘always’); or the comparative construction Ch. mas X ki ‘more X/Xer than’ (Sp. más X que).
A second group (systemically important, but not indispensable because non-hispanic alternatives are available) includes discourse markers, competing in modern speech with English ones; quantifiers like Ch. todu ‘all, whole’ (Sp. todo) and kada ‘every, each’ (Sp. cada); and many lexical hispanisms that compete with Austronesian alternatives (e.g. buenu [Sp. bueno] vs. maolek ‘good’, bida [Sp. vida] vs. cho’gue ‘do’, lengguahi [Sp. lenguaje] vs. fino’ language, Salas Palomo and Stolz, 2008: 245).
A third group (systemically marginal because facultative and/or not productive on Austronesian stems) includes remnants of the Spanish gender system, which is functional only on a handful of copied pairs in Chamorro (mostly involving Ch. bunitu/-a ‘beautiful’ as in i isla bunita ‘the beautiful [fem.] island [fem.]’), but are strictly optional even here; diminutive and augmentative suffixes (e.g. -itu/-ita); or the Spanish plural suffix -(e)s, functional only in a few pairs covering units of measurement like ora/oras ‘hour/-s’ (Sp. hora/-s), metro/metros ‘meter/-s’ (Sp. metro/-s), or simply fossilised in semantically singular forms like Ch. flores ‘flower’ (Sp. flores) or kuentos ‘story’ (Sp. cuentos).
In sum, the impact of Spanish on Chamorro is significant in both lexicon and structure. But the majority of the grammatical categories, including the typological core of the language (predominantly agglutinative morphology, split ergative alignment, vso as the basic word order), has not been affected in a noticeable way, nor has Chamorro undergone significant structural reduction. Contact typological evaluations of the result of the Chamorro-Spanish contact, however, have been very heterogeneous. In its history of being investigated, Modern Chamorro has been described as “reminiscent of a pidgin situation” (Fischer 1961: 262), “a (Hispano-Austronesian) mixed language” (Albalá and Rodríguez-Ponga, 1986: 66, Rodríguez-Ponga, 2009: 42), “a creole” or “semicreole language” (Rodríguez-Ponga, 2009: 42), “an atypical creole” (Munteanu, 1997: 962), “the opposite of an anti-creole” (Couto, 1996: 89), “neither an instance of massive borrowing nor of mixed languages but rather something else” (Stolz, 2003: 291), and a “high copying code […], but undoubtedly a Malayo-Polynesian language without far-reaching typological hispanity” (Pagel, 2010: 147). What this seems to suggest (disregarding the differing methods, foci and language corpora of the cited contributions) is that Modern Chamorro touches a sensitive area of contact-induced change, where different categories, some being defined rather vaguely, overlap.
A second case I want to add to the discussion is Zamboangueño-Chabacano, spoken in South-Western Mindanao and Basilan in the Philippines, and often called a Spanish-based creole. This contact code emerged in the 17th century among the mostly Mexican and Philippine personnel of a military outpost, and is today the primary language of perhaps some 300.000 Filipinos (Quilis and Casado-Fresnillo, 2008: 434). Analyses of this contact code (e.g. Frake, 1971; Grant, 2002; Lipski, 1992, 2001; Pagel, 2010: 346–411; Quilis and Casado-Fresnillo, 2008; Steinkrüger, 2003, 2006, 2008) show that Zamboangueño displays some features typically associated with creoles, such as preverbal tma markers, serial verb constructions, overt subject pronoun, definite article only on specific nps, no formal passive voice, no systematic copula and most vocabulary stemming from the lexifiers Spanish and Portuguese. Many of these features are, however, also characteristics of the Austronesian languages involved in the contact (above all Visayan languages and Tagalog). In addition, many unambiguously Philippine features can be found in Zamboangueño (see the table in Pagel, 2010: 397–399), some being rather recent copies, like the plural pronouns kamé, kitá, kamó and silá, others dating from the time of the formation of the code, such as the plural marker mga, the case marker for proper names as subjects si, or derivational affixes. Several phases of contact must be distinguished in Zamboangueño that correspond to different layers of contact-induced change. Creolisation, in the most general linguistic understanding of the term, may have been an active process in the formation of Zamboangueño, but was probably supplemented by Philippine structures from the earliest years on. In subsequent phases of contact, more Philippine elements were added, obscuring what happened in the very beginning. As a result, Zamboangueño synchronically displays much more complex structures than what is generally assumed for creoles, and also significantly more complex structures than the other (more creole-like and less Philippine) varieties of Chabacano spoken in the Manila region (see Steinkrüger 2006). However, it still shows plenty of structural simplification compared with both the varieties of Spanish and the Philippine languages involved in the contact. The fact that many, if not most, linguistic elements of Zamboangueño can be assigned clearly to either the Indo-European or the Austronesian pool (leaving in fact very few features without a potential direct model) may qualify that contact code also as a bilingual mixed or intertwined code.
Zamboangueño began not as a true creole, but as a natural common intersection of grammatically cognate Philippine languages which had already incorporated a core of Spanish borrowings. Lipski, 1992: 220
The case of Chabacano is still an interesting challenge for contact-linguistics and its structure can only be explained by a multidimensional approach. Steinkrüger, 2006: 10
Two processes, creolization and intertwining, may have been involved in the genesis of Zamboangueño, and were supplemented by extensive copying in different language contacts post genesem. From a synchronic point of view, the combination of all three describes Zamboangueño most accurately. Pagel, 2010: 411, my translation
Just as Chamorro, Zamboangueño seems to point to conceptual problems in the existing theories and typologies of language contact—problems in line with what I have called earlier the autonomous type and the contact language problem. A more continuous conception of contact-induced change is needed, one that allows for outcomes of contact-induced change like Modern Chamorro and Zamboangueño, ‘caught between stools’, but actually situated in the space stretching between prototypical expressions of contact-induced change, such as creole and bilingual mixed language.
4 Assembling the Model: Collection and Discussion
It is the same soil on which both the autonomous type and the contact language problem grow. This soil can be described as a fundamental inconsistency in the existing models of contact-induced change as to the arrangement of phenomena of the same level or within the same category. Phenomena are arranged either non-continuously (like the basic modes maintenance, shift and creation, or the three contact language types in table 1), or gradually/continuously (like the degrees of contact and their correlating outcomes in the category “borrowing situations” in table 1), without transparent conditions for choosing either of the two. This inconsistency is probably the result of a rather successive development and shaping of categories and types and their punctual integration into comprehensive models. It is remarkable in this respect that an explicit call for a continuous conception of contact-induced change already appears in a very early essay on the matter: William D. Whitney’s “On mixture in language” (1881). The ideas expressed herein anticipate many of the accomplishments of later contact linguistics, including also some of the ideas I wish to present here—it is therefore surprising that this text has received (and is still receiving) rather little attention among theoretical contact linguists.
a pure scientific induction from the observed facts of mixed languages, dependent for its authority and its extensibility to further cases, on the one hand, upon the number and variety of the cases already observed, and, on the other hand, upon the degree of success with which the facts they present have been reasoned out and put in connection with the fundamental principles of language-using and language-making. That, in either of these essential respects, the subject has been fully worked up, no one would be justified in asserting; yet there is a considerable body of knowledge respecting it, enough to establish among students of language a prevalent doctrine, held with a fair degree of confidence, though also held open to modification by further evidence, or by the bringing-in of examples radically different from those thus far taken into account. Whitney, 1881: 9
[t]here is no definable limit to the amount of accessions that may be brought […] into a language; but they can hardly fail to leave untouched its forms, and the central kernel of its vocabulary, its words of commonest use. Whitney, 1881: 11
The blood of a people may, for example, become prevailingly different from what it was, by a process of gradual mixture, such as is now bringing a never ending current of immigration to our American shores, with only a minimal effect on the original speech; and, on the other hand, the great bulk of a community may give up its old tongue for that of a small intruded element, as in the case of the countries of southern Europe which were Romanized and in consequence Latinized: and between these two extremes lie numberless intermediates. Whitney, 1881: 11–12, emphasis mine
According to Whitney (1881: 12), the possible linguistic outcomes of the “mixture of populations” fall into three kinds, the first corresponding to what in later models is often called language maintenance (accompanied by borrowing), the second to language shift (“almost without mixture”, p. 12), and the third to language emergence or creation. As regards the latter, “there arises a notably mixed language, containing abundant elements derived from both the one and the other of the tongues whose speakers were brought together to form the community” (p. 12). It is above all in the third group, that Müller’s alleged “axiom” breaks down, “for grammatical mixture […] actually does take place, and its effects are clearly to be seen in English” (p. 16).12
A basically continuous conception of contact-induced change also seems to have been on Schuchardt’s mind, when stating three years after Whitney that “[t]he possibility of language mixture has no limit on either side; it stretches to the maximum and also to the minimum of language diversity” (Schuchardt, 1884: 6, translation mine).13 It is therefore surprising that for the next century to come non-continuous conceptions came to dominate the discourse.
Recently, however, there appears to be a new tendency towards more coherence, or at least more permeability in contact linguistic theory. Putting into a simple schema the ideas expressed in Johanson’s code copying framework we discussed in section 3, the different phenomena of code interaction can be set in relation according to their proximity or distance to either the basic or the model code, and be modelled as potentially overlapping. The framework lacks, however, a proper representation for code creation, hence many outcomes of language contact would be difficult to identify (see figure 1).
Muysken’s (2000) “typology of code-mixing” is perhaps the first to (again) move beyond a rigorous separation of the three basic modes of contact-induced change, if rather implicitly (see figure 2). A central idea in this framework is the combination of the model of Thomason and Kaufman (1988), which is essentially diachronic, i.e. focusing on the outcomes of language contact, with a synchronic model for bilingual behaviour (2000: 264). In the respective figure the three possible scenarios of language contact—“shift”, “code-mixing” and “borrowing”—are displayed as merging into one another, while each is assigned to a dominant orientation of the speaker community (“L1-oriented”, “L2-oriented”, “bilingual”).
Muysken’s book, however, mainly focuses on code-mixing as a kind of code interaction in bilingual speakers (i.e. the middle section in figure 2). Non-bilingual phenomena like “lexical borrowing”, “second language learning” or “language genesis” (pidgins and creoles) are addressed briefly, but not (or only in part) explicitly included in the typology. The emphasis on the “simultaneous access to (modules) of two different languages” may therefore be favourable for the understanding of “many of the phenomena in contact-induced language change” (2000: 274), but it does not result in a comprehensive model of contact-induced change. Furthermore, a distinction is made on the level of contact languages between those that emerge through “relexification” or “intertwining” on the one hand (e.g. Media Lengua), and pidgins and creoles (“language genesis”) on the other, and is perhaps motivated, too, in the bilingual perspective of Muysken’s approach.
The relation between bilingual mixed/intertwined codes like Media Lengua and so-called creoles like Sranan or Papiamentu is subject to an ongoing debate. Being a proponent of the synchronic validity of the creole type, McWhorter (2005: 247–259) challenges the idea “that intertwined languages are a discrete phenomenon, taxonomically and ontogenetically distinct from creole languages” (2005: 258).14 He departs from two socio-historical conditions that are typically mentioned as being characteristic for the emergence of bilingual mixed codes:
The creators must be proficient in the lexifier.
The languages are used as in-group codes, reflecting a pointedly intermediate bicultural identity. 2005: 249
In contexts conditioning a culturally intermediate identity, of which plantations are one of several, an intertwined language results when there is only one substrate language, and a creole when there are two or more. 2005: 258
In other words, intertwined languages retain substrate morphology not because of sociological circumstances unique to them, but simply because when there is only one substrate language, it is feasible for such morphology to be retained by substrate populations. 2005: 258
one manifestation of a general process of language mixture. This process has usually resulted in creoles, but only because historical realities have brought multiple substrate languages to most contexts in which the process took place. 2005: 258
Terminologically, then, intertwined languages are creoles. 2005: 258
[C]reoles and intertwined languages are simply variations on a general theme, their differences determined by clinical aspects of language contact. 2005: 248
The problem is that pidgins and creoles are typically taxonomized in ways that discourage awareness that pidginization and creolization are clinal phenomena. First, pidgins and creoles are typically classified separately from other cases of language contact, partly because of a certain taxonomic unity suggested by most of them having arisen amid plantation slavery during four centuries under a few Western European powers.
Those seeking what is often termed a “metric” or “yardstick” for what a creole is will be disappointed – a necessary disappointment, as language contact processes are inherently scalar. The epistemological difference between quantum physics and language contact theory is vast. It is no more theoretically or empirically incoherent to judge one creole as more “prototypical” than another than it is to judge rabbits as more akin to mice than rhinoceroses (which they are, genetically) – the nature of the thing is a matter of degree.
There are several conclusions to be drawn from these statements. First of all, there is no reason to treat the different outcomes of code creation as separate categories. Second, the “general process of language mixture” can produce other outcomes, or, there can be other variations on the general theme described by McWhorter than those leading in the discourse on contact codes, i.e. pidgins, creoles, bilingual mixed/intertwined codes. More precisely, an infinite number of possible manifestations of this process is conceivable, the complexity of which is covered neither by these dominant types nor by their manifold derivations like “semi-creole” (Holm, 2004), “anti-creole” (Couto, 2002), or “mixed-creole” (Matras and Bakker, 2003). Third, in the genesis of a contact code, elements of more than one of these types or scenarios can be involved. And fourth, contact-induced change following the immediate genesis phase (i.e. the early or earliest years) can take yet other directions. Combining these conclusions, it is all but surprising that some codes in contact, or contact codes, synchronically display elements of several of the prototypical outcomes of code contact, as has been shown in section 3 for Chamorro and Zamboangueño.
Returning to our problems 1 and 2, we can further conclude that contact linguistics is approaching a certain symmetry regarding the internal structure of the three basic modes of contact-induced change: maintenance, shift, and creation. This is not surprising: Why would we not expect them to be essentially structured the same way? Or put differently: What makes autonomous types like pidgin, creole and bilingual mixed codes so dominant in the code-creation discourse, whereas both maintenance and shift are modelled rather consensually as internally continuous (ranging from none over merely lexical to more structural copying in the first, and from shift without copying to the massive copying from the basic code into the model code in the second case, see table 1)? The idea of a continuous structure of the code creation complex, however, has been expressed only occasionally (see Thomason, 1997, Mufwene, 1986, 2000).15
In a first step towards a model of contact-induced change that provides solutions to our problems 1 and 2, we will assume with McWhorter a general process or space of code creation that shares the feature of internal continuity with those of maintenance and shift.
First, pidgins and creoles are indeed a phenomenon worthy of address as a class, but only in being extreme ends of a continuum of structural reduction, upon which various languages could be placed at each point in an unbroken line. p. 253
The basic claim is that in language contact, mixture and reduction operate on complementary axes. After contact, some languages become sharply reduced with little or no mixture beyond the inevitable lexical borrowings. In contrast, some languages become highly hybridized not only in lexicon but also in all levels of grammar, but without being structurally reduced to any significant degree. Meanwhile, in most cases, the result of language contact exhibits both reduction and mixture to varying degrees.
If this claim is true, then we expect that there will be examples exhibiting all possible degrees of this mixture, including those where either reduction or structural mixture is absent. p. 254
An important premise of McWhorter’s claim is that the basic structure of the contact-induced change universe is one of continuity instead of discrete types and categories. In a second step towards our model we will adopt both this premise and the claim. The typology of language contact McWhorter developed on grounds of this claim, however, is not quite complete from the perspective of this article.
McWhorter’s claim is that differences in the outcomes of language contact can be translated into different ratios of “mixture” and “reduction”. Consequently his typology is situated between two respective axes, each subdivided into three sections corresponding to different degrees. While language contact is modelled here essentially as an open space, the division into altogether nine blocks enables McWhorter to locate specific cases in this space and associate them with a certain ratio of mixture and reduction.17
Could these two parameters also be the requisite ones to interconnect our three modes of contact-induced change: maintenance, shift and creation? The answer, I think, must be yes and no. First of all, McWhorter’s typology makes no explicit reference to these modes, nor does it include the processes by means of which codes in contact will become mixed and/or structurally reduced. It basically extends into two dimensions: linguistic structure and diachronic change. Its scope is therefore restricted to statements regarding structural properties of the outcomes of contact, relative to those of the codes involved in it. Certainly the concepts of structural reduction and mixture are valuable in this respect, which is why—in a third step towards our model—we will adopt them, if with a slight alteration (see next section).
What these concepts do not provide, however, are answers to questions like: Why does contact-induced change in a given code and within a given time span prefer one mode of change to another (say maintenance to shift)? How do factors like linguistic competences and attitudes, (perceived) power relations, or the awareness of structural similarities and dissimilarities between the codes in contact shape contact-induced change? Or put more generally: How do social, political, historical, geographical, psychological, cultural, interactional and variational-linguistic factors relate to the restructuring processes in language contact? Assuming that all of the mentioned categories have at least potentially an impact on contact-induced change (which is, I believe, a safe assumption), there can be no simple answer to these questions, no single category or factor be held responsible for what happens in contact-induced change.18 Rather, a different approach to language and language contact is needed that incorporates the immense spectrum of parameters covered in these categories.
[W]herever two tongues come in contact, each is liable to borrow something from the other; and more or less, according to wholly indeterminable circumstances: the measure and nature of the intercourse, the resources of the respective tongues, their degree of facilitating kinship or structural accordance, and so forth. Whitney, 1881: 10
5 A Continuous Model of Contact-Induced Change: Depiction, Description and Application
On the grounds of the elements and terms discussed so far, we can now outline our model. In a first draft (figure 3) it covers the participating codes, dimensions and processes of contact-induced change as well as the ecological parameters affecting them. The (prototypical) outcomes of contact-induced change will be left out for the moment (see figure 4).
At the core of this model are three overlapping triangles, each representing one of the three modes of contact-induced change: code maintenance, code creation, and code shift. These three modes make up the very structure of what was earlier called the “contact-induced change universe”, i.e. any phenomenon of contact-induced change can be located herein.
There are three participants in our idealised contact situation:20 codes A and B in contact, and a new code C as one potential result of the contact. In our model each participant corresponds to one of the three modes and is displayed as a pivot in that mode: code A corresponds to maintenance (speakers of A maintain A as their primary code), code B to shift (speakers of A replace A with B as their primary code), and code C to creation (speakers of A and B create a new code C that becomes their primary code).
The three modes are primarily diachronic constituents and must be set apart from the processes of copying and alternation, which represent synchronic strategies of the speakers in contact, and in the form of conventionalised copies also synchronic results of copying, propagation and conventionalisation at a later stage of contact. Alternation is a strategy speakers may choose at basically any stage and time of the contact. However, the likelihood of this strategy being chosen is probably highest in the lower central and lower right sections of the model where good conditions for elaborated bilingualism can be found.
Copying is further specified as to its direction by the terms in brackets and, from a different perspective, also in the phrase in italics below (the terms being mainly Johanson’s): adoption refers to copying from a model code to a basic code (from B to A by speakers of A) and is the direction found in maintenance, where thus B moves A; imposition indicates the reverse, i.e. copying from a basic code to a model code (from A to speakers of A’s variety of B) and is the direction found in shift, where A moves B. Finally, both adoption and imposition are possible directions of copying within the dimension of creation, where A and B merge.
Due to the wide applicability of this term, copying has a core function in this model: it spans and interconnects the three basic modes. Each mode covers a wide range of possible processes and outcomes, essentially defined by the quality and quantity of copying. Maintenance spans from zero copying (A remains A) via few and occasional (above all lexical copying) to the massive copying and subsequent conventionalisation of material from all linguistic domains (whereby A becomes A’).21Shift spans from the perfect taking over of the model code as the new primary code of speakers of A, involving no significant copying (A is replaced by B), to the mere approximation of the model code by the shifting speakers, involving massive copying from the basic code (A is replaced by B’). Creation, in turn, comprises all kinds of codes that emerge in contact situations but are difficult to assign genetically and/or typologically to either A or B (traditionally labelled as jargons, pidgins, creoles, bilingual mixed languages, etc.).
Between the three modes are zones of transition, indicating that the phenomena located in these zones can be interpreted from both sides. The many controversies regarding situations of significant contact-induced change demonstrate that it is often impossible to decide whether a code A’ or B’ must still be considered a variant of A or B, respectively, or a new and independent code C. Theoretically, however, the edges of the modes can be defined as follows: zero copying from B into A in terms of adoption marks the point of origin of the maintenance space, which is identical with A (i.e. maintenance of A equals A here). Massive copying from B into A by means of adoption defines the points farthest from A in the maintenance space and thus its edge. Zero copying from A to B in terms of imposition marks the point of origin of the shift space, which is identical with B (here shift to B equals B). Massive copying from A into B in terms of imposition defines the points farthest from B and thus its edge. In between these points farthest from A but only with adoption, and farthest from B but only with imposition, is extended the creation space, which is characterised by copying processes that do not correspond clearly to either the adoption or the imposition direction, but rather to both. The fundamental implication of this arrangement is that a new code emerges when (and only when) either of the two modes—maintenance or shift—is executed but fails, or both are executed at the same time and equally fail because they compensate each other.22
If we connect all points farthest from A or B in their respective triangle, they form the inner side (relative to the model) of this triangle, overlapping with the left and right side of the creation mode triangle respectively. Except for the bottom sides of maintenance and shift (which meet halfway with the tip of the creation triangle), all sides stretch into two dimensions. The vertical dimension is defined as structural reduction and is, in essence, adopted from McWhorter’s (2007) framework. The horizontal dimension, although similar to McWhorter’s complementary dimension “mixing”, is defined here as the structural congruence between A and B and has been adopted from Kriegel, Ludwig and Pfänder’s (forthcoming) convergence framework. One advantage of the congruence conception is that in depicting the structural proximity or distance of or between the codes in contact, it already limits the possibilities of both mixture and reduction.
Statements like the following can be made from the two dimensions: first of all, codes A and B (or their combined structural properties) are the entities by which any structural reduction is to be measured. Copying in all three modes will result in alterations in one or both dimensions. While a certain amount of copying will always lead to some approximation in structural congruence—the basic code, into which the copying was made, then being closer to the model code (hence the triangular shape of the modes)—it can result in a largely unaltered structural complexity (as in Media Lengua) as well as in a highly reduced structure (as in jargons and pidgins) and in anything in-between.
Since the vertical dimension indicates the level of structural reduction, the possibility of structural complexification is explicitly excluded from the model. This needs further explication. As mentioned above, the (theoretical) entities to measure any reduction with are the original codes in contact A and B. Therefore it is important to emphasise that it is not only the one code copying and changing that constitutes the measuring instance, but both codes that constitute a pool of structural properties by which measuring is to be made. It is this pool that, in the logic of this model, cannot be complexified but only reduced in contact-induced change. Hence the claim is not that the copying of a certain structural feature from, say, B to A could not possibly result in a complexification of the structure of A (vis-à-vis its pre-contact stage)—this is by all means possible. The claim is merely that contact-induced change can never result in a complexification of the combined structural properties of the codes involved in the contact. Here, the theoretical possibilities are restricted to either the maintenance of the complexity level represented in this pool or its reduction.
Finally, at the very top of the model, affecting synchronic strategies, intensity, direction and outcomes of contact-induced change are ecological parameters, a dynamic set of active and dominant parameters from many different domains, including e.g. the pressure the speakers of B exert on the speakers of A, the availability of the linguistic apparatus of B for speakers of A, or the functionality of a certain linguistic feature from one code in the other.
On the basis of the observations made so far, a second version of our model gives a proposal how to integrate and set into relation some prototypical outcomes of language contact. Note that any such positioning can only be made relative to the original codes involved in the contact. Jargons, pidgins, creoles and bilingual mixed or intertwined codes are positioned in the creation complex, arranged in descending order along the axis of structural reduction. High-copying codes display a significant number of fully conventionalised copies. The relations between these prototypes are defined strictly as transitions, according to the fundamental principles of the model. Any point between and beyond the prototypes is also a possible manifestation of contact-induced change. In the respective context (defined by ecological parameters) manifestations do not need to correspond to types in a prototypical way in order to be interpreted as representatives of that type: correspondence can be merely one of Gestalt or family resemblance. That is why cases like Modern Chamorro and Zamboangueño have been interpreted in terms of, e.g., the creole, the semi-creole and the bilingual mixed type, because they share at least some characteristics with the respective prototypes. My own analyses of these outcomes of language contact, however, indicate that the contact with Spanish has brought Chamorro closer to the high-copying code than to any other scenario, and that contemporary Zamboangueño is the result of a process sharing characteristics with both prototypical creolisation and intertwining, to which substantial copying from Philippine languages was added at a later stage. An approximate location of these two cases (abbreviated Ch and Zm) is also given in figure 4.23
I hope to have shown that contact linguistics gains much from looking beyond the established categories, types and prototypes, and, by doing so, can tie in with thoughts expressed quite early by people like Whitney or Schuchardt. Reducing empirical findings to these templates is not only problematic, but also unnecessary, because there are other ways of organising contact linguistic phenomena. This does not mean, however, that contact linguistics should do entirely without (these) categories, types and prototypes. They constitute valuable landmarks on the map of contact-induced change, but only if treated with the necessary caution.
6 Some Remarks Regarding the Early Years of Contact
There is an essential limit to our model, and perhaps to any attempt of modelling language contact: it can be either restricted to one particular episode of a given contact, or can be very generalising about processes which in fact are more complex. The evolution of any natural language comprises many contact episodes and contact situations, both consecutive and simultaneous, in which different codes are in contact with the one in question, and different processes and ecological parameters operate together.
The case of Zamboangueño is again enlightening in this respect. In order to fairly understand the synchronic state of this code, at least four different (and already generalised) episodes must be distinguished.24 In the first of these episodes, varieties of Spanish and Philippine languages provide the basic constellation, which is supplemented by at least some input from existing regional pidgins and/or creoles, foremost of all the Chabacano varieties of the Manila region.25 It is here that Zamboangueño evolves as a contact code, sufficiently different from both its Spanish and Philippine parents, as well as from the regional pidgin and creole varieties involved. This episode covers in a sense what is called in this volume “the early years”, but extends over a period of time that comprises several generations of speakers (approximately from the mid-18th to the mid-19th century, see Lipski, 2001: 148). The dominant process in this episode shares characteristics with both intertwining and creolisation, and should therefore be settled in the code creation mode of our model.26 In a second episode, the contact with Spanish is continued, but under entirely different circumstances. Most importantly, as a result of the first episode the basic constellation for this episode has changed: Spanish is now in contact with the created code Zamboangueño. Furthermore, the influence of Spanish increased in Zamboanga City during the 19th century, the timespan of this episode, due to the increasing presence of civilian native speakers (Lipski, 2001: 148). Not creation, but maintenance is the principal modus operandi in this episode, and is characterised by copying in terms of adoption from Spanish. A third episode partly overlaps with the second, as it begins in the early 19th century and continues to the present. Here, Zamboangueño is in contact with several cognate Philippine languages, above all Hiligaynon and Cebuano. The guiding process is again copying in terms of adoption, and must therefore be settled in the maintenance mode of our model. Finally, the same can be stated for the fourth contact episode of Zamboangueño, which started approximately in the 1930s and involves varieties of English: maintenance and copying in terms of adoption from English are the dominant mode and process.
What general conclusions can be drawn from this example with regard to the questions of the present volume? It is conceivable that the early or earliest years of contact dictate the general direction of contact-induced change. One way to interpret this direction is in terms of the three modes that provide the basic structure of our model: maintenance, creation and shift. While the intensity of the processes active in the chosen mode may vary in later years of contact, speakers will most likely pursue the same mode, as long as the ecological setting does not change radically, e.g. in terms of the basic contact constellation, demographic parameters, etc. If it does, it seems practical for various reasons to speak of a new episode of contact. In this new episode, all options are open again.
But our claim can obviously not be true for the creation mode. Once the processes in this mode have become effective, and a contact code has emerged, and reached a certain degree of stability, this mode cannot be pursued any longer. The main function of the creation mode is to answer to a communicative crisis. Speakers are (or feel they are) unable, for whatever reason (limited access, identity, attitudes), to speak either A or B. The increase of linguistic congruence is a logical by-product of finding an answer to this crisis. Once a linguistic solution to this crisis has emerged and stabilised, once the congruence has increased, a new contact episode begins, in which the created code itself represents one of the parties in contact. Thus, even if the ecological setting remains the same, the predominant direction or mode of contact-induced change will be a different and a less radical one, i.e. maintenance or shift, perhaps even both. Compared to maintenance and shift, then, creation is a marked mode, and in contacts in which it becomes effective, the conditions of the early and earliest years are likely to be of particular relevance.
One important question, however, remains: How do we set the early years of a particular language contact anyway? Is it realistic to assume a starting point for contact? Can we isolate a particular contact from its synchronic and diachronic contact environment? There may be contact situations for which we can, e.g. creole genesis on previously uninhabited islands, or perhaps even cases like Media Lengua, for which the time, place and circumstances of genesis can be estimated quite fairly (see Muysken, 1997). But the majority of the world’s language contacts are embedded profoundly in other contacts, both synchronic and diachronic. Again Zamboangueño offers a nuanced picture in this respect. Earlier I have called the formative phase of Zamboangueño “the first episode” of this contact, and from this point of view the early years are significant, because they gave the contact code its basic contours which later on were altered comparably little. But from a holistic, or ecological viewpoint my claim is actually illegitimate. The Philippine languages involved in the creation of Zamboangueño had a history of contact with Spanish for almost two centuries when Zamboangueño came into existence. In addition, the oldest of the Manila-Bay varieties of Chabacano had also emerged a century before. Whose early years, one could ask, should we take into account when trying to explain Zamboangueño? Most certainly the early years of a Magellan, who was the first European to reach the Philippine islands in 1521, or a Legazpi, who claimed these islands for the Spanish crown in 1565, were not of particular relevance for the Zamboangueño case. But does that mean they were insignificant at all? A tentative solution to this methodological problem may be in the distinction between different contact episodes of a certain contact situation, as I have done in this section. For each episode, then, a period of early years can be established. This may shift the topical questions of this volume to a whole different level.
FrakeCharles O.1971. Lexical and semantic structure in Philippine Creole Spanish. In HymesDell H. (ed.) Pidginization and Creolization of Language: Proceedings of a Conference held at the University of the West Indies Mona Jamaica April 1968223–242. Cambridge: Cambridge University Press.
KriegelSybilleLudwigRalph and HenriFabiola. 2009. Les rapports entre créole et bhojpuri à Maurice: Contact de langues et actes identitaires. In HookoomsingVinesh Y.LudwigRalph and SchnepelBurkhard (eds.) Multiple Identities in Action: Mauritius and some Antillean Parallelisms203–252. Frankfurt am Main: Peter Lang.
KriegelSybilleLudwigRalph and SalzmannTabea. in press. Reflections on discourse ecology and language contact: The crucial role of some scalar terms. In LudwigRalphMühlhäuslerPeter and PagelSteve (eds.) Linguistic Ecology and Language Contact. Cambridge: Cambridge University Press.
KriegelSybilleLudwigRalph and PfänderStefan. forthcoming. Convergence in language contact: On the interaction of structures speakers’ perceptions and change.
MuyskenPieter. 1997. Media Lengua. In ThomasonSarah G. (ed.) Contact Languages: A Wider Perspective365–426. Amsterdam/Philadelphia: John Benjamins.
PagelSteve. 2008. The old, the new and the in-between: Comparative aspects of Hispanisation on the Marianas and Easter Island (Rapa Nui). In StolzThomasBakkerDik and Salas PalomoRosa (eds.) Hispanisation: The Impact of Spanish on the Lexicon and Grammar of the Indigenous Languages of Austronesia and the Americas167–203. Berlin/New York: Mouton de Gruyter.
PoplackShana and LeveyStephen. 2010. Contact-induced grammatical change: A cautionary tale. In AuerPeter and Erich SchmidtJürgen (eds.) Language and Space – An International Handbook of Linguistic Variation: Volume 1 – Theories and Methods391–419. Berlin/New York: Mouton de Gruyter.
Salas PalomoRosa and StolzThomas. 2008. Pro or contra hispanisms: Attitudes of native speakers of modern Chamoru. In StolzThomasBakkerDik and Salas PalomoRosa (eds.) Hispanisation: The Impact of Spanish on the Lexicon and Grammar of the Indigenous Languages of Austronesia and the Americas237–267. Berlin/New York: Mouton de Gruyter.
SteinkrügerPatrick O.2006. The puzzling case of Chabacano: Creolization, substrate mixing and secondary contact. Paper presented at the Tenth International Conference on Austronesian Linguistics January 17–20 2006 Palawan Philippines. http://www.sil.org/asia/philippines/ical/papers/Steinkrueger-The_Puzzling_Case_of_Chabacano.pdf.
SteinkrügerPatrick O.2008. Hispanisation processes in the Philippines. In StolzThomasBakkerDik and Salas PalomoRosa (eds.) Hispanisation: The Impact of Spanish on the Lexicon and Grammar of the Indigenous Languages of Austronesia and the Americas203–236. Berlin/New York: Mouton de Gruyter.
ThomasonSarah G.2008. Social and linguistic factors as predictors of contact-induced change. In NicolaïRobert and ComrieBernard (eds.) Language Contact and the Dynamics of Language/Theory and Implications. Journal of Language Contact THEMA 2. http://www.jlc-journal.org/: 42–56. http://www.jlc-journal.org.
1 It should be noted that studies on convergence enjoy a privileged position in language contact theory over those on, e.g., code-switching, because they are mainly concerned with structural change. In typological evaluations of contact-induced change priority is mostly given to changes in linguistic structure (as opposed to those in vocabulary, for instance), as can be observed from early works by, e.g., Müller (1862), Whitney (1881), or Schuchardt (1884 et alibi).
2 Essentially similar structures can be found e.g. in the frameworks of Van Coetsem (2000) (“recipient language agentivity”, “source language agentivity”, “neutralization”) and Muysken (2000) (“L1-oriented”, “L2-oriented”, “bilingual”). Rather exceptional in this respect is Johanson’s (2002) framework because it lacks a proper category for “creation” or “mixing”.
3 The ecological approach to language and language contact (see Ludwig, Mühlhäusler and Pagel, in press) perhaps meets best this complexity and dynamics. It will therefore be resorted to later in this article.
4 In the German original: “Meillet meint, der Ausdruck “Mischung” sei impropre; jedenfalls ist es der andere, “Entlehnung” noch viel mehr, obwohl wir auf ihn nicht verzichten können und wollen. Er erweckt die Vorstellung, als ob etwas aus einem an sich fremden Besitze herübergenommen würde, nicht, wie es in Wirklichkeit ist, aus dem schon angeeigneten Besitz von etwas Fremdem”.
5 That such a modification is easier in some cases and almost impossible in others can be demonstrated e.g. by the pair nebula (Lat. ‘cloud’) as used in astronomy and atom (from Greek átomos ‘the indivisible’) in physics. While earlier in the history of astronomy all luminous and non-punctual objects in the sky (including e.g. galaxies) were labelled nebula, today the term is pretty much reserved for interstellar clouds of gas and matter. The term atom, on the other hand, was coined as a concept in philosophy, later adopted in physics, and continues to be used there, although it is known that the basic elements of matter it designates are further divisible.
6 “Wenn endlich, wie sich ja historisch belegen lässt, irgend eine Lauteigenthümlichkeit einer wirklich tonangebenden Persönlichkeit, eines Fürsten, Höflings, Schauspielers in deren Kreis freiwillig copirt oder die eines Lehrers von diesem seinen Schülern aufgezwungen wird, so lässt sich auch die Möglichkeit nicht bestreiten dass der Ursprung eines Lautwandels ein willkürlicher sei. [“If, in the end, as can be attested historically, any particular peculiarity of pronunciation of a really style-setting personality, a prince, a courtier, an actor is copied voluntarily in his own circle, or, if a teacher imposes his own pronunciation upon his pupils, it cannot be disputed that the impulse for sound change is one of personal choice”.] (Schuchardt, 1885: 15–16, English translation by Theo Vennemann and Terence H. Wilbur, p. 51)
7 See the heterogeneous pattern in the traditional terminology, e.g. in language maintenance, language shift, language contact, but code-switching.
8 See also Johanson (2008: 62): “The term copying […] covers what is otherwise called ‘borrowing’, ‘diffusion’, ‘transfer’, ‘interference’, ‘replication’, etc.”.
10 Significantly higher estimations, between 50 and 60 per cent, have been given by Albalá and Rodríguez-Ponga (1986) and Rodríguez-Ponga (1995). See Pagel (in press) for a discussion.
11 As is apparently Schuchardt’s a few years later: “[Die Sprachmischung] ist nicht sowohl Ausnahme als Regel. Mit mehr Recht als Max Müller gesagt hat: “es gibt keine Mischsprache”, werden wir sagen können: “es gibt keine völlig ungemischte Sprache”. [“[Linguistic mixture] is not so much exception as rule. With greater right than Max Müller has said, ‘There is no mixed language’, shall we be able to say, ‘There is no fully unmixed language’”.] (Schuchardt, 1884: 5, English translation taken from Bailey, 1979: 144). Müller does by no means deny the possibility of mixture in general and, in fact, explicitly rejects the idea of a “fully unmixed language”: “There is hardly a language which in one sense may not be called a mixed language” (Müller, 1862: 63 ).
12 In view of this, Poplack and Levey’s (2010: 392) impression, according to which Whitney’s opinion in the essay discussed here displays “an outright denial of the possibility of a mixed language”, is quite baffling.
13 In the German original: “Die Möglichkeit der Sprachmischung hat nach keiner Seite hin eine Grenze; sie geht bis zum Maximum wie bis zum Minimum der Sprachverschiedenheit”.
14 For a five volume compilation of contributions to the study of contact languages, comprising the major discourses on pidginisation, creolisation, intertwining, and respective typologies, see Holm and Michaelis (2009).
15 “[A]s a group creoles form a prototype category rather than a classical one” while “they differ among themselves on the Wittgensteinian family resemblance model” (Mufwene, 2000: 65).
16 See also Mufwene (2000: 66): “there are no particular linguistic evolutionary processes likely to yield (prototypical) creoles; these vernaculars are produced by the same restructuring process that bring about change in any language”.
17 Some examples from McWhorter’s (2007: 254) schema are: No simplification/ Lexicon only [mixed]: German in Finnish; No simplification/ Lexicon and syntax: Romanian and other Balkan Sprachbund languages; No simplification/ Lexicon, syntax, morphology and phonology: Media Lengua; Extreme simplification on all levels/ Lexicon only: Hawai’i Creole; Extreme simplification on all levels/ Lexicon and syntax: most creoles, Extreme simplification on all levels/ Lexicon, syntax, morphology and phonology: Saramaccan. See also McWhorter (2011: 15–16).
18 See Thomason (2008: 54): “We can make rough predictions about types of changes to be expected under different social conditions […] and we can often predict the direction of change; but we can’t even be sure that contact-induced change will occur in any given contact situation, much less predict what form it will take if it does occur”.
19 Similarly Haugen states in a paper that became seminal for ecolinguistics: “Language exists only in the minds of its users, and it only functions in relating these users to one another and to nature, i.e. their social and natural environment. Part of its ecology is therefore psychological: its interaction with other languages in the minds of bi- and multilingual speakers. Another part of its ecology is sociological: its interaction with the society in which it functions as a medium of communication”. (Haugen, 1972: 325)
20 Most contact situations involve more than two codes. For modeling contact-induced change, however, it seems legitimate to reduce the number of possible participants.
21 I cannot go further into the hierarchy of copying here, but there is probably a general consensus about lexical elements being unmarked and structural elements marked copies in language maintenance, while in shift the opposite is true: structural copies are considered unmarked and lexical ones marked. For more details see e.g. Thomason and Kaufman (1988) or Van Coetsem (2000).
22 As any reasoning in this section, this must be understood as purely theoretical. I do not claim that speakers consciously decide for either of the three modes in contact situations, nor that they actually feel they speak language A, B or C (let alone during the process of copying), nor that these codes exist as empirical entities.
23 In this application of our model for the Chamorro scenario, pre-contact Chamorro would be code A, and Spanish code B. In the Zamboangueño case A would be a rather heterogeneous group of Philippine (and perhaps other) languages, and B Spanish. Note that in the latter case the perspective can be turned around easily, then A being Spanish and B the other languages involved. Seen from this perspective, Zm would be located somewhere left of its current position, closer to the maintenance mode.
24 The generalising aspect must not be underestimated here. See Lipski’s (2001) scenario, consisting of six “tentatively proposed” stages (p. 148).
25 Although a genetic relation between Zamboangueño and the Manila Bay varieties of Chabacano is undeniable, the latter are significantly older and display less elaborated and less “Philippine” structures. Their status as “creole” is therefore less disputed. For more details see the works cited in section 3.2.
26 See Lipski (2001: 147–148): “Zamboangueño began not as a true creole, but as a natural common intersection of grammatically cognate Philippine languages which had already incorporated a lexical core of Spanish borrowings”.