Deriving classifier word order typology, or Greenberg’s Universal 20A and Universal 20

The word order typology of numerals (Num), classifier or measure word (C/M), and noun (N) put forth by Greenberg (1990 [1972], Numerical classifiers and substantival number: Problems in the genesis of a linguistic type. In Keith Denning & Suzanne Kemmer (eds.), , 166–193. Stanford, CA: Stanford University Press) can be reduced to a universal principle: N does not come between Num and C/M. Given the affinity between this universal and Greenberg’s Universal 20, which concerns the word order typology of D, Num, A, and N, the former is dubbed “Universal 20A” (Her et al. 2015). This paper first discusses, and ultimately rejects, the two alleged exceptions to Universal 20A, one in Ejagham, the other in some Tai-Kadai and Tibeto-Burman languages. Then, in light of Universal 20A, Cinque’s (2005, Deriving Greenberg’s Universal 20 and its exceptions. 36(3). 315–332) successful antisymmetric account of Universal 20 and all its exceptions is re-examined and shown to be inadequate for Universal 20A. The analysis I propose adopts Abels and Neeleman’s (2012, Linear asymmetries and the LCA. 15(1). 25–74) symmetric derivational account of Universal 20 and, crucially, takes complex numerals into consideration. The final account also integrates a multiplicative theory of C/M (Her 2012a, Distinguishing classifiers and measure words: A mathematical perspective and implications. 122(14). 1668–1691) and is able to explain the base-C/M harmonization, which was first discovered by Greenberg (1990 [1978]: 292, Generalizations about numeral systems. In Keith Denning & Suzanne Kemmer (eds.), On language: Selected writings of Joseph H. Greenberg, 271–309. Stanford, CA: Stanford University Press) but has since been overlooked in classifier research, and also offer a functional explanation for Universal 20A.