Counterexample to Unidirectionality
Elizabeth Closs Traugott
In recent years a frequently-asked question has been whether unidirectionality is of theoretical significance in grammaticalization (e.g. Newmeyer 1998, Campbell 2001 argue it is not, but others argue it is, e.g. van Kemenade 1999, Haspelmath 1999). Many alleged counterexamples to GR are not legitimate, but some are.
One interesting set of possible counterexamples REGRAMMATICALIZATION/EXAPTATION: A LEGITIMATE involves a type of change that has been called "regrammaticalization" (Greenberg 1991), "exaptation" (e.g. Lass (1990, 1997), Greenberg (1991), Vincent (1995), Giacalone Ramat (1998), Norde (2001), or "hypoanalysis" (Croft 2000) (henceforward EXAPT). The purpose of the present paper is to assess the extent the extent to which EXAPT challenges the role of unidirectionality in grammaticalization (GR).
I take GR to be the subset of cross-linguistically recurring changes that involve correlations across time between semantic, morphosyntactic, and sometimes also phonological changes. And I take unidirectionality in GR to be a robust hypothesis about a strong tendency whereby lexical items and constructions come in certain linguistic contexts to serve grammatical functions or grammatical items develop new grammatical functions.
Although there are several differing views of EXAPT, what all
have in common is that it involves "the opportunistic co-optation of a
feature whose origin is unrelated or only marginally related to its later use"
(Lass 1990: 80) and "disjunctive semantic change" (Greenberg 1991:
301). EXAPT is best conceptualized as exhibiting:
o cooption of grammaticalized material that has become marginal to more grammatical function (contrast GR in which constructions involving major categories (N, V, A) acquire more grammatical/"minor category" function)
o semantic/functional discontinuity (contrast GR, in which earlier and later meanings are different, but clearly related, at least pragmatically)
o non-decategorialization (contrast GR, in which decategorialization is prototypical)
o irregularity (cross-linguistic replication such as is evidenced for GR has not been observed to date).
While they are legitimate counterexamples to GR because they concern the emergence of grammatical function, cases of EXAPT do not undermine the robustness of the hypothesis of unidirectionality because of their infrequency and irregularity. This conclusion must of course be empirically tested cross-linguistically.
Sources Cited:Andersen, Henning, ed. 1995. Historical Linguistics 1993: Selected Papers from the 11th International Conference on Historical Linguistics, Los Angeles, 16-20 August 1993. Amsterdam: Benjamins.