¹ Guid' A. O. F. (Dakar, 1948). The material for this study was gathered in August and September 1949 at Nikki, Dahomey, French West Africa. The investigation was made possible by a grant from the American Council of Learned Societies, and by the cooperation of the Sudan Interior Mission and its staffs at Nikki, Parakou, and Kandi.

² There is apparently no reference to Bariba in Trager's Bibliographical classification system for linguistics and languages, SIL 3.54–108 (1945) and 4.1–50 (1946).

³ Cf. Welmers, Notes on two languages in the Senufo group: I. Senadi, Lg. 26.126–46; II. Sup'ide, Lg. 25.494–531 (1950).

⁴ These names are chosen (1) for quick reference, (2) because the top level is absent in the Kandi dialect, being replaced there by high, and (3) because the top level is rather uncommon even in the Nikki dialect. High is left unmarked because it seems to be the most frequently occurring level.

⁵ At least in rather slow speech. In rapid speech it sometimes seems to begin and end somewhat lower.

⁶ Yoruba, an adjacent but not closely related language which is the source of several loanwords, has initial /a/ but no /p/.

⁷ In the slow speech of one informant, this is a glottal stop followed by /a/, which would be interpreted as external open juncture. In this study, word division is written arbitrarily. There is no immediately apparent evidence of other phonemic junctures.

⁸ In the Kandi dialect, which has no phonemic top tone, high tone is raised to a non-phonemically higher level under this condition.

⁹ There is undoubtedly a difference in meaning (i.e. usage) associated with the addition of , but it has not been isolated.

¹⁰ */kpeē/ is reconstructed as the stem of /kperū/ ‘stone’ by analogy with the first two examples and with the form /kpèè té/ ‘that stone’.

¹¹ For the details of stem formation, apart from vowel length, see Section 4 below.

^11a Since this article was written, I have been informed by the Rev. J. B. Williams of Parakou that he has found an eighth noun class, with the identifying demonstrative /pí/.

¹² The /w/ in this form is suspect. The stem is /goō/.

¹³ Cf. §3.1 above. Some informants say /wíya/ and /wii/.

¹⁴ The fact that some of these nouns are compounds and a few perhaps phrases does not affect the analysis.

¹⁵ For many words like this, the plural is far more common, and indicates an indefinite amount (e.g. ‘millet‘) rather than a number of discrete items.

¹⁶ A few recorded exceptions to this statement are suspect, usually for other reasons as well.

¹⁷ On the basis of my experience with a number of different types of West Sudanic languages, and of conversations with Professor J. Berry of the University of London, I believe it is safe to say that West Sudanic languages tend to have aspect rather than tense systems. Most of these languages, however, have one form which is specifically past.

¹⁸ This name is chosen because the form is often used as the second verb in a sentence: /ko n tasu dūūrē/ ‘I will plant yams’, /na kíⁿ n tasu dūūrē/ ‘I want to plant yams’, /ko n kpiⁿ n tasu dūūrē/ ‘I can't plant yams’.

¹⁹ A negative command is expressed by the negative Consecutive: /a kú tasu dūūrē/ ‘Don't plant yams’.

²⁰ See footnote 9.

²¹ Or perhaps ‘I haven't planted yams yet’.

²² The precise distinction between these forms and the corresponding forms without /rǎ/ is not clear.

²³ The verb /kpê/ ‘finish’ is tentatively put into this class although a Past */kpē/ is not recorded.

²⁴ For the vowel length, see §2.2.

²⁵ A stem such as /soku/ ‘call’ may be put into either group, since the final vowel of the Consecutive is /u/ to begin with.

²⁶ No further statements are required for high-top-mid or high-low-mid. The second tone remains unchanged in all forms.

²⁷ Top-mid in la does not seem to correspond to mid-mid in 1. A form /wúràmā/ ‘come back’ is included in this group.

²⁸ 2-u, all mid, and 4, all mid, are the only possible combinations in which all forms must be different. Only one verb 2-u, all mid, and two verbs 4, all mid, are recorded.

²⁹ The conditions under which some adjectives take /ɔ/ and some /o/ are not known. But all of these adjectives seem to have identical forms for the g-class and the w-class.