1 What diachronic typology can tell us about language universals and variation Andrea Sansò UŶiǀeƌsità dellIŶsuďƌia, Coŵo 1. Stop worrying and love diversity! Universals, variation, and explanation in a diachronic perspective Functional approaches account for language universals and patterns of cross-linguistic variation by deriving them from more general aspects of language use. The most important factors that are responsible for uni- versals and variation include: (i) frequency (ofteŶ deĐliŶed as ǁoƌld frequency, ďut possiďlLJ to be conceived of in terms of speeĐh fƌeƋueŶĐLJ, Haspelmath 2008: 45); (ii) economy ;ǁhat is pƌediĐtaďle ƌeĐeiǀes less ĐodiŶg thaŶ ǁhat is Ŷot, HaiŵaŶ ϭϵϴϯ: ϴϬϳͿ; (iii) iconicity (linguistic distance reflects conceptual distance; Haiman 1983: 782); (iv) computational efficiency (preferences in performance data are mirrored in constraints in gram- mar and variation patterns, Hawkins 2004); (v) system pressure (defined as the teŶdeŶĐLJ of gƌaŵŵatiĐal ĐodiŶg to target entire classes of iteŵs, Haspelmath 2014: 197); (vi) harmony (Croft 2003: 62ff.; also prominent in Optimality Theory; see foƌ iŶstaŶĐe AisseŶs ϮϬϬϯ OT treatment of Differential Object Marking). Glossing over the different weights that scholars assign to (i)-(vi), these factors may either conspire towards the same results or be in competition with one another (Du Bois 1985; Mac Whinney et al. 2014). Competition is held to be responsible for both principled cross-linguistic variation and universal tendencies (i.e., frequency of language types; Croft 2003: 64). The competing motivations model has been variously criticized. On the one hand, as argued for instance by Maslova (2000: ϯϬϴͿ aŵoŶg ŵaŶLJ otheƌs, there seem to be no criteria that would allow for an empirical distinction between genuine distributional universals and accidental statistical tendencies. This ŵeaŶs that the skewing of linguistic traits may be historically determined rather than produced by functional pressures on speakers (see also Daniel 2010: 61): rare features, for instance, are more likely to be found in languages spoken in mountainous areas than in languages spoken in the plains (Nichols 1992). This suggests that lin- guistic (and, more generally, cultural) features spread more easily in the latter kind of environment, for ob- vious reasons. On the other hand, a more far-ranging criticism has been levelled against the nature itself of such expla- nations. More or less implicitly, the factors in (i)-(vi) are generally operationalized at the synchronic level: whenever a given grammatical pattern or construction conforms to what is predicted by a given universal motivated in terms of one of the factors in (i) to (vi), it ĐaŶ ďe ĐoŶsideƌed as edžplaiŶed, iƌƌespeĐtiǀe of its individual history (Haspelmath 2008: 43; Moravcsik 2010: 68; Cristofaro 2014). Thus, for instance, stating that English has the order Preposition + Noun because this order is harmonic with the head-dependent order of constituents in other phrase types amounts to explaining why English prepositional phrases are the way they are. This is a weak kind of explanation, however, and this weakness is not without consequences. Lin- guistic typology is often blamed for the coarse-grained nature of its explanations, especially when contrasted with the fine-grained explanations of language-particular facts carried out by exponents of the opposite camp of linguistic thinking. The criticism is not entirely without merit: explanations based on (i)-(vi) may fail to target the full complexity of language-specific and construction-specific facts, operating only at a rather low degree of granularity.