Turkana (Nilotic, Nilo-Saharan) (for details see Dimmendaal 1983), Serer (Atlantic, Niger-Congo) (A. A. Kibrik, p. c.), Hindi (Indic, IE)(mentioned in Croft 1991: 127) have genders in nouns but none in pronouns .Teop (Bougainville, Austronesian) (Mosel & Spriggs 2000) has a gender contrast in nominals but not in pronouns. Obviously it is the locus of gender marking, i.e., the article (used with nouns, adjectives, and numerals), which prevents pronouns from having gender. Gender itself is an innovation in Teop and its closest relatives (Nehan-North Bougainville network). The proposed grammaticalization path leads from demonstratives differing in their deictic value to gendered articles (e.g., proximal deixis developed into the “human” gender, the connection being “proximity” either of the speaker or of the speaker’s most prominent property). Note also that Teop gender contradicts the claim of #1676 by utilizing almost complete polarity.
1. Cf. #256. 2. Greenberg’s assertion is less restrictive insofar as (read literally) it permits DIFFERENT genders in nouns and pronouns, while Uspensky requires the SAME genders.
1. Cf. #256. 2. Greenberg’s assertion is less restrictive insofar as (read literally) it permits DIFFERENT genders in nouns and pronouns, while Uspensky requires the SAME genders.