Abstract :
[en] It has been repeatedly observed, on the basis of typological ‘big data,’ that there is a worldwide preference for suffixes as opposed to prefixes. This can be explained in several ways. A possible explanation is that this feature is a world-wide retention from Proto-World, or is prone to diffusion through language contact. Another possible explanation is that suffixes are preferred for some reason in Universal Grammar or for hitherto unclear general cognitive reasons (Caballero et al. 2008). Yet another explanation is that suffixes are more prone to be created through regular processes of language change, e.g., grammaticalization (Bybee 1985), perhaps due to online usage factors (Himmelmann 2014). The explanation of this preference is directly relevant to a question highlighted in Good (2008), namely, the relationship between language universals and language change: do synchronic structural universals constrain change, or do diachronic universals, ultimately motivated by synchronic usage factors, give rise to synchronic universals? Kiparsky (2008) argues that the form of synchronic grammars constrains change, i.e., languages should not be able to change in such a way that they flout Universal Grammar. On the other hand, for Bybee (2008), the most robust universals are in fact universals of language change, and synchronic states are in a sense epiphenoma. For this question, apparently ‘counter-directional’ changes are crucial: why should language change lead to universally dispreferred distributions of linguistic structures? n this paper, we argue that universally dispreferred structures can and do arise as the result of regular language change, given the right background structures as the particular ‘ecology’ in which change takes place. Specifically, we show that Ancient Egyptian-Coptic (Afroasiatic), shows a long-term diachronic macro-change from mixed suffixing-prefixing to an overwhelming preference for prefixing. The empirical basis for this study is a comparison of ten typologically-significant parameters in which prefixing or affixing is at stake, based on Dryer’s (2013) 969-language sample. With its extremely high prefixing preference, Coptic belongs to the rare 6% or so of languages that are predominantly prefixing (Tables 1,2). Moreover, it has a higher prefixing index (11) than any other language in Dryer’s 969-language sample. The closest competitor is Hunde (Bantu; Democratic Republic of Congo), with a prefixing index of 9.5. In terms of areality, Coptic is an outlier: in mediterranean northern Africa, Coptic is the only language that is predominantly prefixing. We argue that each of the micro-changes implicated in this macro-change are better understood in terms of changes at the level of individual constructions, via grammaticalization, rather than in terms of a broad structural ‘drift.’ Crucially, there is nothing unusual about the actual processes of change themselves; what may be unusual, from a cross-linguistic point of view, is the length of uninterrupted documentation of a single language, which allows us to observe long-term changes with abundant evidence. In short, we argue that Ancient Egyptian-Coptic looks as though it is swimming against the typological tide, although it is constantly paddling along with the usual tides of language change.
