Contact Linguistics. Chap

Exercise: Thomason (1983a:218) offers a more detailed account of the similarities and differences between Ma'a, Bantu and Cushitic. Make your own list of these and decide to what extent the features peculiar to Ma’a might be attributed to L1 influence under shift as opposed to borrowing. How far do they provide evidence for Thomason’s claim that Ma’a arose through a process of “massive grammatical replacement”?


6.3. The genesis of Ma'a.

The various non-Bantu structural elements described here have been the focus of a continuing debate over Ma'a origins. As noted earlier, some scholars (e.g. Thomason 1997a.b) explain its genesis as the result of gradual but massive structural "borrowing" or replacement (similar to Asia Minor Greek), with maintenance only of core Cushitic vocabulary. I suggest below (section 7.3) that Copper Island Aleut is a better point of comparison with Asia Minor Greek. Others (e.g. Bakker 1994:24) argue that Ma’a arose as a result of the intertwining of a Bantu grammatical system and Cushitic vocabulary. This suggests that the language originated in a way similar to languages like Media Lengua, except that the Matrix Language in this case was a newly acquired L2, rather than the group’s ancestral language.

Thomason (1997b:481-3) uses the evidence of Cushitic structural features and core vocabulary as well as what appear to be chronologically ordered changes in Ma'a to support her view that the language arose via gradual Bantuization of a previously Cushitic grammar. She finds further support for this scenario in Whiteley's (1960:96) observation that the Bantu affix system is better established among the younger people than the older generation. But neither piece of evidence seems sufficient. The Cushitic structural features could just as well be explained as L1 retentions which accompanied the shift to Bantu. In other words, the shift may have produced an imperfectly learnt second language variety of Bantu with some degree of substratum influence from Cushitic. If we assume that it was this variety of Bantu (Pare) that served as the grammar-source language for the initial creation of Ma'a, the Cushitic retentions can be explained (as well as perhaps the intergenerational differences in the use of the Bantu affix system). Younger people's greater use of these affixes would then be due to continuing contact with Bantu and more successful learning of its structure.
In short, all of the non-Bantu structural features in Ma'a are consistent with what might be expected in a case of language shift. Some (like the loss of agreement in demonstratives and possessives) are simplifications, while others (like the actual forms of demonstratives and pronouns as well as the non-Bantu phonemes) appear to be retentions from Cushitic. The interweaving of Cushitic vocabulary with Bantu grammar, on the other hand, seems to have been the result of a process of content morpheme insertion ("relexification") similar to that found in other bilingual mixtures like Media Lengua and Anglo-Romani. This appears to offer the most feasible explanation for the genesis of Ma'a. In other words, the language was created at the point where the shift to a Bantu language (Pare) was (almost) complete. The second-language variety of Pare, with substratum influence from Cushitic, provided the morpho-syntactic frame for Ma’a. This is similar to the position taken by scholars such as Brenzinger (1987), Sasse (1992a) and Bakker (1997).
The shift scenario has two advantages over alternative explanations. First, it allows us to group together a number of contact languages that arose under similar historical circumstances. These would include, apart from Ma’a, various Romani-lexicon contact languages such as Anglo-Romani, Armenian Romani and Caló. Thomason (1995:19), in fact accepts the shift scenario in the case of Anglo-Romani and Armenian Romani, but rejects it for Ma’a and Caló, as noted earlier.
Consigning these intertwined languages to distinct categories seems to be somewhat unmotivated, given their basic similarity in history as well as linguistic make-up. The shift scenario at least provides a unified treatment of the processes involved, while allowing for the retention of structural and lexical features that distinguish Ma’a and Caló from other intertwined languages in the group. Perhaps the difference between Asia Minor Greek on the one hand, and Ma’a and Caló on the other, is merely a matter of the degree to which the shift took place. But the creation of the latter seems to have followed the shift, rather than being products of shift-induced changes. As Bakker (1997:205) has pointed out, we need to distinguish the historical process of shift from the actual linguistic processes that created these languages.
The second advantage of the shift scenario is that it allows us to identify similarities in such processes across bilingual mixed languages with different histories. It seems likely that the intertwined languages that originated after language shift were created by processes quite similar to those that produced languages like Media Lengua and Michif. In both cases, the dominant language of bilinguals provided the Matrix Language (ML) frame into which elements from the other language were incorporated. The only difference would be that, in the case of Ma’a, Anglo-Romani, etc., the ML was the newly acquired language, while for Media Lengua, Michif, etc, it was the ancestral language. In either case, as Bakker (1994;24) notes, “it is the language that is best known that provides the grammatical system.” In short then, while their sociohistorical scenarios may differ, languages like Anglo-Romani and Ma'a appear to have arisen through the same linguistic processes that led to the genesis of Media Lengua and Michif.