The Innovative Character of Object-Verb Word Order in Nen (Bantu A44, Cameroon) Maarten Mous 1. Introduction Nen (A44) has some features that are unusual for a Bantu language: Object-Verb word order and full ATR vowel harmony. These features are to some extent shared by the neighbouring and related Mbam Bantu languages (A40/60). The Mbam Bantu languages of Nen (A44), Nyokon (A45), Mande (A46), Gunu (A62a), Yambasa (A62), and Baca (A62e) have been established as a subgroup in the Atlas linguistique du Cameroun (Dieu & Renaud 1985) and Mous and Breedveld (1986). Lexicostatistic classifications, such as Bastin and Piron (1999), invariably treat this group of languages as one of the primary branches of Bantu. In view of their geographical location on the borders of the Bantu/Bantoid area and their classification as one of the primary branches of Bantu (or just outside of Bantu), and taking into consideration that the features mentioned above are found in non-Bantu Niger Congo, it had been proposed that these features are retentions and thus evidence for the conservative nature of these languages. The purpose of this article is to argue against this claim. In particular, I argue that the Object-Verb word order is an innovation. With regard to vowel harmony, Stewart (2000), using de Blois (1981), has already given evidence for the innovative character of the full vowel harmony of Nen (A44) 1 . No historical analysis of the aberrant OV word order in Nen (A44) has been proposed so far. Claudi (1993) deals with Nen (A44) in a footnote and considers Nen (A44) to have undergone a change from VO to OV within her general framework of possible word order shifts from VO to OV in Niger Congo; I discuss this in section 3. In section 2, I address the historical origin of word order by comparing Nen (A44) with Nyokon (A45), which also shows OV word order, and with Mande (A46) and Gunu (A62a), where there is no OV word order. In section 1. Taking a Proto-Bantu vowel system i, I, E, a, O, U, u as his point of departure, Stewart proposes a first innovation of +ATR spread from the high + ATR vowels i and u, which introduced full harmony, and a subsequent merger of I > e and U > o and, in a push-chain, e > 2, and subsequently e > E in stems. Other Mbam Bantu languages such as Gunu (A62a), Kalong (A62d), Mande (A46), and Nyokon (A45) show the same full vowel harmony and the same characteristic alternations of i M E, u M o , and o M O, which suggests that the same two basic innovations apply to them also. See Hyman (2003) for an analysis of Kalong (A62d) vowel harmony. A third conservative feature has been claimed: Van Leynseele and Stewart (1980) proposed a retention of a fortis/lenis distinction in the double reflexes of proto Bantu sounds in Nen (A44). Janssens (1993) shows that the double reflexes cannot be fully explained by the fortis/lenis distinction. He proposes conditioning by vowel quality and lexical diffusion. Botne (1992) accepts the validity of a retention of a fortis/lenis distinction in some of the double reflexes, which he reanalyses as a glottal/pulmonic distinction, but with subsequent innovations in the glottal (lenis) consonants conditioned by high tone.