Mixed Languages by Saskia Marina Hewitt UWA linguistics student

Introduction
Language mixing is a phenomenon that has only been discovered relatively recently, and while it is rare, it is not as rare as once thought (Bakker and Muysken, 1995:50). It is probably because of its novelty as an area of study in linguistics that there are a number of differing opinions as to what the definition of a mixed, or split, language should actually be. The examples I will present here do not seem to be under dispute, however.
The exact mechanisms and conditions that lead to formation and crystallisation of a mixed language are still being investigated by linguists, but there are a number of recurrent trends that can explain why and how they start. This essay will discuss those trends and also look at patterns in the structure of mixed languages.

What is a “mixed language”?
The exact definition of a mixed language still seems to be under dispute. A number of linguists define it as a language in which a majority of the grammar comes from a different source to the majority of the lexicon (e.g. Bakker and Muysken, 1995:49) but Thomason and Kaufman (1988) take a slightly less stringent view. They still define a language as mixed even if much of the grammar is from the same language source as the lexicon, as long as the language exhibits “complete grammatical replacement in at least some subsystem” (103). So if the language has grammar from the same language as the lexicon, except that all the noun morphology, for example, is from a different language, this would be a mixed language. They further define a mixed language as being the result of such massive borrowing from one language into another that the transmission of the latter language from generation to generation is obstructed (76).
Linguists also differ in regards to terminology because they like to go about describing and studying these languages differently. For example, Bakker and Muysken prefer to talk about “language intertwining” instead of “mixed languages” because they feel that “the process… can be defined independently from the historical directionality involved” (1995:50), while Thomason (2001) and Thomason and Kaufman (1988) generally define the language through the process of borrowing which they say created it.
Despite slight differences, the broad picture painted by various linguists on the matter is the same: a mixed language, as its name implies, is created by a conjunction of two different language systems. It is significantly different from either of these two languages, and, perhaps most importantly, it cannot be traced back to any one language. And, importantly, language mixing is distinguished from other related phenomena such as language death, pidginization and creolization because a mixed language has all the grammatical and lexical complexity of a normal language (Thomason and Kaufman, 1988:104), is usually formed in bilingual (rather than multilingual) contact situations, and has components which can be traced back to their respective languages easily (Thomason, 2001:197). In the same place, Thomason also says that an important distinguishing factor is that in mixed languages “imperfect learning plays no role” whereas in pidgins and Creoles, “imperfect learning plays a prominent role”. I will give some examples of mixed languages that have been reported in various works. You will see that all these languages fit into the broadest definition of a mixed language, but that some of them may be harder to fit into the more restricted (Bakker and Muysken type) definition. This is of course neither an exhaustive list of known mixed languages in the world nor a comprehensive study of any of them. For further reading see the reference list at the end of this essay.


Michif
Peter Bakker’s work A Language of Our Own is quoted on the Metis Cultural and Heritage Resource Centre (2002). There, Bakker describes Michif as a language spoken in North America by less than a thousand people of Native American and European (mostly French) descent. Its first speakers are thought to have been the children of mothers who spoke the American-Indian language of Cree and French-speaking fathers. The language has both French and Cree components in roughly equal parts, and each component keeps the phonological system of the language it was derived from (Thomason and Kaufman, 1988:229). The basic split in this language is between verbs (Cree) and nouns (French). It is in this way that it differs from the definitions of a mixed language as being split down a grammar/lexicon dividing line. Bakker and Muysken, however, propose that Michif actually does conform to such a split because its “verbs…consist of only bound morphemes and are therefore only in Cree.” It looks as though the split is down a noun/verb line “because of the polysynthetic structure of Cree” (1995:46). This would enable Michif to fit more stringent definitions as well. In any case, the idea that Michif is a mixed language does not seem to be in any dispute.

Ma’a (Mbugu)
Ma’a, also called Mbugu, is probably one of the most widely reported mixed languages in the literature, possibly because it was discovered early on, and also because it seems to be of a prototypical form. According to Mous (1994) it is spoken in Tanzania in the Usambara Mountains (175) by a group that is thought to have come originally from Kenya (177). The group is surrounded by the Bantu people, who speak Shambaa, which is the dominant language in the area (175). Thomason and Kaufman (1988) say in their case study of the language that its grammar is primarily of Bantu origin, while its lexicon is mainly of Cushitic origin (223). They say that this is a result of massive borrowing of grammatical structures from Bantu into the original Cushitic language, in a situation where that language was being maintained (rather than in a situation of shift) (227).

Shelta
Anthony Grant describes this language in his case study (1994). It is spoken by the Irish Travellers, a nomadic group of people who share “a number of cultural traits, such as common patterns of livelihood, [certain] customs…and adherence to Roman Catholicism” (123). It is basically composed of an Irish lexicon and an English grammar, in which respect it can be analogised to another mixed language, Anglo-Romani, which features a Rom (gypsy) lexicon and an English grammar (Thomason and Kaufman, 1988:103). The two are also similar in that they function as secret languages. (ibid; Grant, 1994:123). One major difference is that Shelta is acquired as a first language, while Anglo-Romani is learnt in adolescence (Grant, 1994:140). The speakers of Shelta number about ten to twenty thousand (the figure is uncertain), living mainly in Ireland but also in Great Britain, the USA, Canada, Australia and South Africa (Grant, 1994:123).

Mednyj Aleut
Mednyj Aleut, also called Copper Island Aleut, is a case somewhat similar to Michif in that it exhibits nominal and verbal morphologies that can be traced back to different languages; the noun morphology, with, indeed, the main part of the grammar, are from Aleut, and the verb morphology is Russian (Thomason and Kaufman, 1988:234). The verb morphology of Aleut, a polysynthetic language, is extremely complicated (ibid.) and this may provide an explanation for why it was replaced with the simpler Russian system. This complicated verbal system is another point in case for the similarity of Mednyj Aleut with Michif; recall that Cree’s verbal morphology is also complex. Further, Cree is also a polysynthetic language (Golovko, 1994:120). However, Thomason and Kaufman (1988:238) make the following important distinction between the two languages: that, “Cree speakers borrowed French lexical morphemes, while Aleut speakers borrowed only the grammatical morphemes.”
Mednyj Aleut is spoken by a small number of people on Mednyj (Copper) Island, one of the islands in the Aleutian chain, near Russia. It began to be spoken by the children of Russian men and Aleut women, who had a special status in the community on the island, somewhere between the Aleuts and the Russians (Golovko, 1994:113).

Kormakiti Arabic
Thomason and Kaufman (1988), in their discussion of mixed languages, outline three broad types existing in the world today. The first type is exemplified above by Ma’a and Shelta. These languages have lexicons which are predominantly from one source language, but grammars which are almost entirely from the other source language (103-104). The second type is that of the Michif and Mednyj Aleut examples, which feature “borrowed grammatical structures [which] are confined to particular grammatical subsystems” (105). As an example of the third type, they give Kormakiti Arabic, a hybrid between Arabic and Greek spoken on Cyprus (105). They quote Newton, who studied the language, on page 106: “Words of Arabic… origin retain the full morphological apparatus of Arabic, while those of Cypriot-Greek… origin appear exactly as they do [in that language].” Newton found that 38% of lexical items were of Greek origin, including some basic vocabulary such as body parts and numerals. So in this case, the mixture is diffused evenly across grammatical structures and lexical items (Thomason, 2001:202).

Social Factors in Language Mixing
How and why does language mixing occur? We have already looked at a few different factors in the definitions above – they occur where there is language contact between two different cultures, and where there is widespread bilingualism (so that the structures of the other language can be borrowed intact). This bilingualism is usually just one-way (Thomason, 2001:197). This factor is important, because it means that the two cultures do not need a third language in order to bridge any sort of communication gap, which is the sort of situation in which you would get a creole or a pidgin. Instead, a mixed language arises because of a desire for an in-group language (ibid:198). Sometimes the mixed language is never or rarely revealed to outsiders – for example Shelta and Anglo-Romani, as discussed above, and also the mixed language Media Lengua, which is a hybrid between Quechua and Spanish. Pieter Muysken spent three months studying Quechua before discovering that “his hosts spoke a different language… when they thought he could not overhear them” (Bakker and Muysken, 1995:50). Every work on this matter that I have come across attributes this desire for an in-group language to one of two causes: either so that conversations can be kept secret from outsiders, or as a symbol of identity, or both. For example, recall from the discussion of Mednyj Aleut that its speakers had a special status. They were more privileged that the Aleuts, but not as privileged as the Russians. They affectively had a new culture which arose rapidly and had no particular ties to any other culture. For this reason, the language of the Copper Island Aleuts now “is the only factor for their self identification” (Golovko, 1994:117).
In what socio-historical settings might we find this sort of thing happening? Myers-Scotton (2002:253) identifies a few conditions that seem to recur:
(i) …subjugation of an indigenous group… by a foreign power; (ii) migration and a new life that requires regular use of a dominant L2; (iii) long term employment in an alien culture, using the L2 of that culture; and (iv) indigenous, but minority, status under totalitarian rule.
She also identifies one other factor, which is the deciding one – that “the speakers of the besieged language must perceive their language as an essential part of their ethnic identity.” If the speakers are hostile towards the dominant language and culture, then they will try to hold onto their own culture through their language. But if pressure to change is intense and prolonged, a mixed language may result (as in Thomason and Kaufman, 1988:50). This scenario is applicable to languages like Ma’a and Anglo-Romani, but in the cases of Mednyj Aleut and Michif there was no pressure to shift to the other language (ibid:108). In these two languages, also, the speakers were of mixed blood. In cases of that type, there surely can be no great attachment to either culture on the part of the speakers. Instead, the speakers would see themselves as belonging to a third, separate group (Bakker and Muysken, 1995:51). It would make sense, then, to use a third, separate language.
So it seems that there are two broad socio-historic settings in which a mixed language comes about. The first is when a minority language is in intense contact with a dominant one for a long time. The second is when two cultures with different languages come together and have children, especially when these children feel for some reason separate from the cultures of their parents. Note that these two settings do not correspond to the reasons given above as to why a mixed language is desired.

How will a Mixed Language Look?
Most mixed languages follow similar patterns, and a short typology is given in Golovko (1994:119ff), where he uses Bakker (1992). He says that in a mixed language derived from languages A and B, “bound morphemes (always of a grammatical nature)… [and] syntax” will come from language A, while “free lexical morphemes are in language B” and “free grammatical morphemes can be in either language”. He notes, though, that Michif and Mednyj Aleut are exceptions to this rule, although this may be because of the polysynthetic natures of Cree and Aleut. He says further that “if language B were an isolating language (the only type not represented among source languages)”, then the resulting mixed language may be different again. In other words, the morphological type of the source languages will affect what the mixed language will look like in terms of which components come from which language.
Bakker and Muysken (1995) also note that in a language where the speakers of a mixed language are descended from mothers of one culture and fathers of another (e.g. Michif), the grammar will be that of the mother’s language. Alternatively, if the language is the result of one language trying to hold out against another (like Ma’a), the grammar will be that of the intruding language (50).

Conclusion
The idea of mixed languages even existing is a fairly new one, and for this reason study of the phenomenon still is in early stages. There is some disagreement about what a mixed language actually is, but even so most languages that have been identified as mixed are not under any dispute. Despite this, linguists have been able to thoroughly investigate a number of different cases and this has helped them to discover something of what goes on when a mixed language is formed. There is general agreement about the social conditions in which a mixed language can arise and the reason for its development, but there is still much more work to be done. We do not yet have a complete picture of the phenomenon and there are gaps to be filled and questions to be asked before we can fully understand everything about it.