^{page 111 note 1} Published by the Crown Agents on behalf of the Government of Nigeria, London, 1940. This supersedes his earlier The grammar of Tiv. The Tiv people, now numbering three-quarters of a million, occupy a large area of Benue Province, Northern Nigeria, most of them being south of the Benue river.

^{page 111 note 2} A dictionary of the Tiv language, and A Tiv reader for European students, both published as above. Other works on Tiv by the same author, including Tiv phonetic and tonal principles, 1940, are frequently referred to in the above books, but if published do not appear to be generally available.

^{page 111 note 3} Especially Mr. J. Akperan Orshi and Mr. J. I. Abuul. I am also indebted to my colleague, the Rev. A. M. Jones, for assistance with tonal analysis.

^{page 111 note 4} Principles, chap. v. The tones given here are those indicated by Abraham, though he uses a different notation. In this article tones are indicated as follows: High /; Mid ; Low /; Falling tones: High-Mid ; Mid-Low ; High-Low ; Rising tone, Low-High ; //, II, and // indicate a continuing High, Mid, or Low tone on the whole of a long vowel, which is treated in this analysis as a succession of two vowels (cf. section 8 (c) below). See also from section 26 onwards.

^{page 111 note 5} The digraph gh is the conventional representation of the voiced fricative .

^{page 112 note 1} There are a very few apparent exceptions in everyday use, such as bee ‘it is finished’, kuma ‘that's enough!’ (cf. kuma ‘be sufficient’). But these should perhaps be treated as interjections rather than as true verbal forms.

^{page 112 note 2} A verbal complex of this kind occurs as a self-standing piece chiefly, but not exclusively, in response.

^{page 112 note 3} Stable groups such as tòr rùmùn ‘the chief agreed’, ìWá yèm ‘the dog went away’, etc., in which the first item is an Independent Nominal, are less suitable for the present purpose, since the verbal form in such cases is restricted to certain tenses. The corresponding forms in other tenses (e.g. tòr rùmùn ‘the chief will agree’, ìwá ìá yèm ‘the dog will go away’) always have a Subject Pronoun between nominal and verbal forms, i.e. the sentence consists of Nominal plus Verbal Complex. Sentences such as tor rnmun, however, may also be analysed as Nominal plus Verbal Complex, the Subject Pronoun being zero; in that case they are covered by the present study of Verbal Complexes, and do not need special treatment. The same is true of narrative sequences such as inyon va saŋge, ya ‘the birds came and picked (them) up and ate (them)’, where va, saŋge, and ya can each be analysed as a Verbal Complex with zero Subject Pronoun. The question whether the Subject Pronoun would be better treated as a Verbal Prefix is outside the scope of this article, and the conventional practice of writing it separately is followed here.

^{page 113 note 1} Abraham treats ki, mba, etc. in the Continuous (‘ Present’) tense as a copula, and the verbal form as a verbal noun; but for a description of the verbal system it seems more convenient to treat ki, mba, etc. as subject pronouns of a distinct series and the verbal form as a base including a suffix -n.

^{page 116 note 1} Principles, pp. 2, 3.

^{page 116 note 2} It should be noted that in the case of type (v) verbs having bases with back vowels, alternative forms with a voiced velar fricative are recorded by Abraham and by Malherbe (Tiv-English dictionary, with grammar notes and index. Lagos, Govt. Printer, [1931 ?]), e.g. ndzughur as well as ndzuur, SUghur as well as suur. But it seems that the two forms are used by different speakers, so that the existence of the alternative forms need not be taken into account here.

^{page 116 note 3} I and r are often interchangeable; cf. Abraham, Principles, p. 5, and Malherbe. Further investigation may reveal some regularity in the distribution of these two variants.

^{page 116 note 4} The voiced velar fricative is conventionally represented by the digraph gh, which is retained in the examples in this article.

^{page 116 note 5} Sometimes, but by no means always associated with the proximity of an i or u vowel. This y- and w-coloration occurs even more frequently, and with more consonants, in nominals.

^{page 117 note 1} And also, incidentally, to the final consonant of nominals.

^{page 118 note 1} Malherbe, op. cit., mentions two different pronunciations of E-vowels, but uses a single symbol for both, as do all publications of the Dutch Reformed Church Mission, and the Tiv broadsheet Mwanger u Tiv published by the Gaskiya Corporation. Both distinguish the two O-vowels by the symbols o and ô.

^{page 118 note 2} Not in CVV structures.

^{page 118 note 3} In CVCVC and CVVC structures only, where there is often an alternative form having o instead of u.

^{page 118 note 4} Not in CVVC or CVCVC structures.

^{page 118 note 5} -OUgh being the realization of o + gh, see section 11, second paragraph.

^{page 119 note 1} cf. Abraham, Principles, sections 190, 192, 157C.

^{page 120 note 1} This CVCV variant of the type (ii) structure also occurs, in the Past tense and other tenses where CVC is usual, when the base is followed by the first person object pronoun of class 1 (-m ‘me’) and occasionally the corresponding second person (-U ‘you’); cf. Principles, section 40. In this case, where the first vowel is a, e, i, or U, the vowel sequence is the same as above, but some speakers use 0-U, ɔ-U in place of 0-0, ɔ0, and ɔ-U in place of ɔ-ɔ; e.g. a kerem ‘he aimed at me’, a karem ‘he missed me’, ivim ‘fill me’ (seen on a bag for soiled linen!), a SUghum ‘he greeted me’, Una kɔrom or Una kɔrum ‘he will catch hold of me’ (past tense bases ker, kar, iv, sugh, kɔr).

^{page 120 note 2} Abraham, presumably for similar reasons, calls it the ‘Special Tense’—see chapters xxii, xxiii, etc.

^{page 121 note 1} -OUgh. being the realization of o + gh; for the vowel alternation of u-a with o, see section 18 (b).

^{page 121 note 2} tsugh, dugh, dzough, sough could be regular RP tense basest corresponding to type (iv) Past tense bases with a velar fricative between the two vowels—٭tsughe, ٭dughe, ٭dzugha, ٭SUgha; but in present-day speech at any rate there is no such consonant. I do not even hear a. semi-vowel sound after the u vowel of tsue, dzua, etc., any more than after the i vowel of hia, cie, etc.

^{page 123 note 1} Only two instances of type (iv) verbs with i-a sequence have been observed, yira (quoted above) and nyima (an alternative of the more usual nyuma ‘bite’) which has nyem in the RP tense. There seem to be no verbs with this sequence and a ‘non-final’ consonant in position C². If there were, one would expect them to have CeCe pattern in the RP tense on the analogy of venda and gegh, though type (iii) radicals with i-a sequence in the Past tense have i-e sequence in the RP tense—see section 18 (b) below.

^{page 123 note 2} Usage appears to differ here. Both Abraham's dictionary and my informants agree on himbi, jiŋgi, ciŋgi, hidi, idi as RP bases corresponding to himbe, jiŋge, ciŋge, hide, ide in the Past. On the other hand Abraham gives rihi and fise, while my informants give rihe and fishi. But i-i sequence certainly predominates in the RP tense.

^{page 123 note 3} Abraham gives kuse, puse, Use, bunde as RP bases used in RP as well as Past tense, but my informants give kuSU, pUSU, USU, bundu for the RP tense.

^{page 124 note 1} -OUgh being the realization of o + gh.

^{page 126 note 1} And a common pattern CeCe-m with the object pronoun -m, e.g. a kerem ‘he aimed at me’ (ker), and a neŋgem ‘he saw me’ (neŋge); cf. n. 1, p. 120.

^{page 126 note 2} And also in Past and RP tenses with the object pronoun -m; cf. n. 1, p. 120. The absence of any CVC_nfV counterpart to yira and nyima with sequence i-a is probably fortuitous.

^{page 127 note 1} I hear kema, kpema, mena, nyoho in place of Abraham's kama, kpama, mana, nyɔho.

^{page 127 note 2} The facts described are compatible with a hypothesis that historically all bases of type (ii) as well as type (iv) verbs at one time had CVCV structure in all tenses (plus -n suffix in the Continuous), but that in course of time the second vowel disappeared in all circumstances where a consonant in the ‘ final’ category occurred in the middle of the sequence a-e or any sequence of identical vowels, except where it was ‘protected’ by a suffixed -n or -m. A similar hypothesis for types (v) and (vii) and types (vi) and (viii) is also consistent with the known facts. But of course these can be no more than hypotheses in the absence of historical evidence to support or disprove them.

^{page 128 note 1} Further investigation might indicate that the Mid tone is not in fact a distinct tone, but could best be treated as a High tone realized at an intermediate level in certain cases.

^{page 128 note 2} The description of the RP tense tone pattern of type (iv) verbs is also simplified if CVC structure is regarded as equivalent to two tone-bearing units. For instance, the bases in á hídi ‘he returned’ and á ‘he asked’—RP tense complexes of the type (iv) verbs hìdè and pìnè —have different structures, but on the principle suggested would each have two tone-bearing units; and, as will be seen later, both complexes can then be described in terms of a common Tone Sequence High High Mid.

^{page 129 note 1} cf. Abraham's distinction between (a) Low Tones and (V) Mid followed by Low Tone, quoted in section 2 above.

^{page 132 note 1} There will, of course, also be similar structures without an initial consonant; cf. section 8 (b) above.