Current contributions and main research methods

Apart from obvious cases, language contact can work as a catalyst of convergence on levels that seem less spectacular and evident. This applies not only to convergence between varieties that have anyway been very close in areal and/or genealogical terms from the start, such as the levelling of dialect continue or the rise of urban koines. We should also be aware that macro-areal patterns (e.g., along the Pacific Rim, or in Eurasia) can only be explained if we account for them as results of long-term rapprochements between varieties that were dissimilar in areal and/or genealogical and, thus, also in structural terms; rapprochements occurred on many different local levels (regions) and summed up to a chain of family resemblances over a much larger area. In order to disclose such patterns, one must not only look at spectacular phenomena, and one has to account for distributional properties and their change (e.g., if minor patterns increase and gain in productivity).

Mechanisms of convergence: identification models

As concerns specific phenomena of convergence (and associated processes), several theoretical frameworks are on the market. All of them contrast model and replica languages (M and R, respectively),8 and all of them more or less concentrate on the moment of innovation, not on propagation (see Section 3):

Code Copying

This approach (Johanson, 2002) distinguishes the phonological (‘material’) shape and semantic aspects as well as combinatorial and frequential properties (the latter two probably tie up closely). Global copies are those which include all four aspects, copies are selective if only some of them apply.

PAT vs. MAT borrowing

PAT(tern) borrowing applies if R replicates a structure and/or function(s) from M without the phonological shape of items in M, while MAT(ter) borrowing includes the phonological shape (Matras and Sakel, 2007; Matras, 2009). This opposition is roughly identical to the more traditional difference between loan/borrowing and caique (cf. Matras, 2013, p. 68 for some terminological overview); it also considerably overlaps with the distinction between global and selective copies in Johanson's framework (see Section 4.1.1).


The mechanism of pivot-matching (Matras and Sakel, 2007; Matras, 2009, pp. 240-243, 2011, 2013. pp. 70-72) is subordinate to PAT-borrowing inasmuch as ‘pivotal features of the model construction are identified and replicated in the replica language’ (Matras, 2009, p. 241); ‘there is no one-to-one correspondence between the morphemes of the equivalent lexemes’; each pattern is created in its own self-contained system and selected according to communicative appropriateness, only the pivot is shared (Matras, 2009, p. 247, 2011, p. 151). Pivot-matching is itself insensitive to the substrate-adstrate difference (Matras, 2009, p. 258). This model, which is designed to comprise also contact-induced grammaticalization (see Section 4.1.4), allows for abrupt (not only gradual) changes, but simply because it focuses on the moment of creative invention and is agnostic as to whether the innovation further spreads, in particular grammaticalizes, or not. However, Matras (2011, p. 15 7f.) suggests some socio- linguistic conditions on which the spread of pivot-matched constructions depends, together with a chain-effect ‘across an entire network of contiguous or partly contiguous languages.’ Most crucial are Tax normative attitudes in a multilingual community with flexible identity boundaries’ (Matras, 2013, p. 72).

Contact-induced grammaticalization (CiGxn)

Essential for this line of research is the formulation of constraints on the directionality of changes from lexical to grammatical expression. Contact is recognized as a trigger of gram- maticalization (Matras, 2009, p. 238), but the processes are basically the same as in ‘internal grammaticalization.' Initially, Heine and Kuteva (2005) made a difference between Ordinary and Replica CiGxn: in the latter case, it was not only a pattern, or pivot, that was borrowed, but even the entir e gr ammaticalization process was assumed to be replicated by speakers of R. This idea was abandoned in Kuteva and Heine (2012), where the authors presented an integrative model of CiGxn by splitting the processes in two phases (propelling vs. accelerating forces).

Importantly, a more general model of pattern replication should account for cases which (at least superficially) violate unidirectionality, as assumed for grammaticalization, e.g., if categories are lost prior to a theoretical end of a grammaticalization process (Matras, 2013, p. 71).

Polysemy copying

A shift in meaning for some unit in R is inspired by M, but without a change in the gr ammatical status of the item in question. Usually, this leads to an extension of the lexical meaning (as in loan translations) or of the functional inventory of the respective unit. Note that the logical counterpart, namely monosernization in R on the basis of equivalents in M, occurs as well, e.g., in Molise Slavic (Breu, 2003, pp. 354-363).

Interlingual identification of linguistic subsystems

Interlingual identification of linguistic subsystems (both as process and result) includes signs and/or categories (Gast and van der Auwera, 2012). It is essentially a refined version of polysemy copying but was designed to cover different types of CiGxn as well (2012, pp. 389-390). See Breu (2018: §2) for a (somewhat implicit) application.

Comparison of approaches

The last three of the aforementioned mechanisms share (at least implicitly) assumptions known from semantic maps. Although the entities that are identified, copied, or assimilated differ in terms of complexity, these approaches face the same problem, namely: often it cannot be tested whether we are dealing with the replication of a grammaticalization process or with some kind of ‘wholesale copying' of functions which are anyway close in semantic space. Contiguity in cognitive space is what semantic maps are designed to test; implicational hierarchies, as assumed in grammaticalization dines, are based on the same contiguity assumptions. Now, since the functions of replica categories are very unlikely to leapfrog over contiguous domains of semantic space (or hierarchies), linguists would be unable to discern how a poly- semous replica came about (Wiemer and Walchli, 2012, pp. 27—44). Admittedly, examples in which such ‘jumps' obviously occurred do exist, for instance, the Estonian лглиш-future discussed by Metslang (2017). Less sure is the use of the HAVE+anteriority participle as a compound past in extinct Polabian Slavic varieties under the influence of German (Lotzsch, 1967). Yet another case is the MAT or PAT-replication of verb particles from German, e.g., in Sorb- ian varieties (Giger, 1998), e.g., Lower Sorbianpred-chytas (alongside wu-chytas) ‘reproach’ (Germ, vor-weifen), Upper Sorbian nutr widzec ‘understand, realize" (Germ, ein-sehen).

Furthermore, Ziegeler (2017) proposes to dissolve, or circumvent, naive implications carried by Replica CiGxn in Heine and Kuteva (2005), by assuming implicational hierarchies between more and less concrete meanings of a source expression:

The presence of an implicational hierarchy does not imply that the contact speaker has access to historical sources of a grammaticalizing item, only that the historical sources are often comparable with lexical roots of forms that may, at a particular tune, still be visibly co-existing with the more grammaticalized form. This produces a situation of polysemy enabling grammaticalization pathways to be reconstructed by the speaker in contact.

(Ziegeler, 2017, p. 343)

This land of polysemy is reminiscent of Hopper’s (1991) notion of layering: the lexical source is still ‘alive.’ It may, thus, supply a cue for contact speakers by which a function which is rather marginal in M can be retrieved and be grammaticalized, even ‘to a fr equency exceeding that of the source language’ (Ziegeler, 2017, p. 344).

Moreover, all processes mentioned previously often lead to isomorphism (Wiemer and Walchli, 2012, pp. 37—43), but they do not need to. The formation of isomorphism can be explained from the rise of diasystems (shared by a sufficient number of bilingual speakers for reasons of communicative economy), which then have to be accepted by monolingual speakers (or speakers with other ‘sets’ of multilingual competence); see Section 3.1.

< Prev   CONTENTS   Source   Next >