+ All Categories
Home > Documents > The noun prefixes of New Benue-Congo

The noun prefixes of New Benue-Congo

Date post: 10-Oct-2016
Category:
Upload: kay
View: 214 times
Download: 2 times
Share this document with a friend
18
The noun prefixes of New Benue-Congo 1 KAYWILLIAMSON Abstract The pioneering work on noun class prefixes in Benue-Congo is de Wolfs The noun class System of Proto-Benue-Congo (1971). In the 1970s, Benue-Congo was defined äs the Benue-Congo ofGreenberg (1963), now referred to äs Old Benue- Congo. 1t consisted offour branches: Plateau, Jukunoid, Cross River, andBantoid. De Wolf discovered that a number oflanguages in two ofthese branches, Plateau andBantoid, had contrasts beween high (H) and low (L) tone prefixes;for example, the singular animal class (=Bantu class 9) showed low tone prefixes while theplural (-Bantu class 10) showed high ones. De Wolftherefore reconstructed this contrast back to Proto-Benue-Congo. Consequently, the Situation in Bantu, where all the nominal prefixes are low, was regarded äs an innovation; high tone nominal prefixes had merged with low tone ones, although the distinction was maintained in the pronominal prefixes. De Wolfs work has remained a reference pointfor scholars undertaking recon- struction ofvarious groups of Benue-Congo languages, but problems have been found in a number of South Bantoid (=Wide Bantu) languages (see,for example, Hombert 1976; Hyman 1980; and Watters 1990), for which varying explanations have been given while basically accepting de Wolfs reconstruction. In this paper l suggest that the reconstruction itselfneeds amendment. My starting-point will be a group oflanguages, Igboid, which was not considered by de Wolf (1971), because it was not then defined äs Benue-Congo. 1t is, however, included in the expanded New Benue-Congo (Williamson 1989) and therefore can provide evidencefor the reconstruction of Proto-Benue-Congo. 1. I am most grateful to Baudouin Janssens and Ciaire Gregoire for some crucial Stimulus and references in working out the role of the augment. An earlier version of the paper was presented at the 22nd Colloquium on African Languages and Linguistics at Leiden, 1992. Many thanks to the following for their comments, new Information, and corrections: Roger Blench, Bruce Connell, Thomas L. Cook, Jean-Marie Hombert, Gudrun Miehe, Pascale Piron, Thilo Schadeberg, and John Stewart. JALL 14 (1993), 29-45 0167-6164/93/0014005 © Walter de Gruyter Brought to you by | University of Wisconsin Madison Libraries 330 Memorial Librar Authenticated | 128.104.1.219 Download Date | 9/22/12 6:44 PM
Transcript

The noun prefixes of New Benue-Congo1

KAYWILLIAMSON

Abstract

The pioneering work on noun class prefixes in Benue-Congo is de Wolfs The nounclass System of Proto-Benue-Congo (1971). In the 1970s, Benue-Congo wasdefined äs the Benue-Congo ofGreenberg (1963), now referred to äs Old Benue-Congo. 1t consisted offour branches: Plateau, Jukunoid, Cross River, andBantoid.De Wolf discovered that a number oflanguages in two ofthese branches, PlateauandBantoid, had contrasts beween high (H) and low (L) tone prefixes;for example,the singular animal class (=Bantu class 9) showed low tone prefixes while theplural(-Bantu class 10) showed high ones. De Wolftherefore reconstructed this contrastback to Proto-Benue-Congo. Consequently, the Situation in Bantu, where all thenominal prefixes are low, was regarded äs an innovation; high tone nominalprefixes had merged with low tone ones, although the distinction was maintained inthe pronominal prefixes.

De Wolfs work has remained a reference pointfor scholars undertaking recon-struction ofvarious groups of Benue-Congo languages, but problems have beenfound in a number of South Bantoid (=Wide Bantu) languages (see,for example,Hombert 1976; Hyman 1980; and Watters 1990), for which varying explanationshave been given while basically accepting de Wolfs reconstruction. In this paperl suggest that the reconstruction itselfneeds amendment. My starting-point will bea group oflanguages, Igboid, which was not considered by de Wolf (1971), becauseit was not then defined äs Benue-Congo. 1t is, however, included in the expandedNew Benue-Congo (Williamson 1989) and therefore can provide evidencefor thereconstruction of Proto-Benue-Congo.

1. I am most grateful to Baudouin Janssens and Ciaire Gregoire for some crucial Stimulus andreferences in working out the role of the augment. An earlier version of the paper was presentedat the 22nd Colloquium on African Languages and Linguistics at Leiden, 1992. Many thanksto the following for their comments, new Information, and corrections: Roger Blench, BruceConnell, Thomas L. Cook, Jean-Marie Hombert, Gudrun Miehe, Pascale Piron, ThiloSchadeberg, and John Stewart.

JALL 14 (1993), 29-45 0167-6164/93/0014005© Walter de Gruyter

Brought to you by | University of Wisconsin Madison Libraries 330 Memorial LibraryAuthenticated | 128.104.1.219

Download Date | 9/22/12 6:44 PM

30 K.Williamson

1. Noun prefixes in Igboid

It might seem str nge to seek for evidence of noun class prefixes in a languagegroup which has no functioning noun class System left. There are, however,abundant relics of such a System having been formerly in Operation (seeWilliamson and Ohiri Aniche, forthcoming for details). Here I draw attention tojust two: the presence of an otherwise unexplained downstep in one tone classof nouns, and the occurrence of a few CV prefixes in a small pari of the Igboidarea.

The downstep is found in the class of Igbo nouns which in Isolation bear Htones:

(1) isi 'head' έwύ 'goat'The downstep occurs when two apparently HH nouns are linked by anAssociative Particle which is generally agreed to consist of a floating H tone:(2) N! AP N2

isi ' ewu=> isi έΝνύ 'head of goat'

The problem is to explain why a construction in which, apparently, all theconstituents consist of H tones should result in a downstep, which is usuallybelieved to reflect an underlying L tone. I have argued previously (Williamson1986) that this downstep must be the remains of an old low tone nominal prefixin nouns. In most environments it is now preceded by a Η-tone prefix, the sourceof the first H in such nouns. There is another construction which differs from theassociative and which Green and Igwe (1963: 20) call the Genitive of Personifi-cation, since it is found most frequently with proper names; in some dialects itoccurs in other contexts and has been called the Specific Construction. In thisconstruction the new H-prefix is absent, and all nouns normally beginning withH tones therefore fall together with those with initial L. In (3), the tones of N2are shown for a noun which is HH in Isolation (£wu 'goat') and for one whichis LH in Isolation ( ke 'rat')· The resulting tone pattern differs in the associativeand the specific for the HH noun but not for a LH one.

(3) a. i -si + + E -wuH L - H + H + H L - Hi si e1 wu'head of goat'i -si + + O -keH L - H + H + L L - Hi si { ke'head of rat'

Brought to you by | University of Wisconsin Madison Libraries 330 Memorial LibraryAuthenticated | 128.104.1.219

Download Date | 9/22/12 6:44 PM

The noun preßxes ofNew Benue-Congo 31

b. i -si + + E -wuH L -H + H + L -Hi si 1E -wu'Mr Goat's head'i -si + + O -keH L -H + H + L -Hi si <0 ke'Mr Rät's head'

The fact that the new H-prefix is absent in the specific construction implies thateven at the Proto-Igboid level the two prefixes must have been separable. Theevidence of CV-prefixes comes from the southwest corner of the Igboid area;Epeye shows 6V-prefixes, corresponding in part to wV-/vV- in Ndele andwV-/bV- in Ogbakjri. Lects such äs Ndele and Akpo in the same area showrV-prefixes, and also a downstep, even in Isolation, in words such äs 'head'.Words such äs 'goat', on the other band, appear with no CV-prefix but still showthe downstep in Isolation.

(4) Ndele ri^si 'head' o*wü 'goat'Akpo rfa 'head' 6lwu 'goat'

I am thus led to reconstruct the words in Proto-Igboid äs in (S), where I omit thearguments justifying the segmental reconstructions.

(5) Proto-Igboid *<fi-N-tui * —

A natural question to ask is for the source of these H-prefixes, and for a possibleanswer I turn to Bantu.

2. Prefixes in Bantu

In Bantu, Meeussen (1967) reconstructs pronominal prefixes äs H, except inclasses l and 9, and all nominal prefixes äs L. Omitting the locative classes andadding the tones to bis reconstructions, the possible sequences of a pronominalprefix followed by a nominal prefix are äs in Table 1. I have added forcomparison those pronominal prefixes and nominal prefixes reconstructed by deWolf (1971) for Proto-Benue-Congo and those reported for Amo (an EastemKainji language described by Anderson 1980) for which there are Bantucognates.

Brought to you by | University of Wisconsin Madison Libraries 330 Memorial LibraryAuthenticated | 128.104.1.219

Download Date | 9/22/12 6:44 PM

32 K.Williamson

Table l. Proto-Bantu, Proto-Benue-Congo, andAmo prefixes

Class Proto-Bantu Proto-Benue-Congo Amo

1234566a78910111213141516

PP

jü-bä-gu-gi-di-gä-gä-ki-bi-ji-Ji-du-kä-tu-bu-ku-

-

NPmü-bä-mü-mi-!-mä-mä-kl-

bi-n-n-dü-kä-tü-bü-kü-Pi-

PPgwu-ba-u-, gu-i-,(zi-?>zi- (?)ga-, a-ma-, nä-ki-bi-zi-i-,(zi-?)lu-ka-ti-bu-ku-[pi-

NPii-.ö-bä-u-i-(=10)li-ä-mä-, nä-ki, ke-bi, be-e-, i-i-(=4)lu-kä-ti-bü-ku-bi-]

PPu-a-u-?le-a-mi-ki-??i-?ka-te-??fe-

NPü- (=3)ä- (=6)ü- (=1)?le-ä- (=2)Ski-?9

i-?kä-te-??fe-

PP = pronominal prefixNP = nominal prefix

Meeussen (1967: 99) noted the presence in Bantu of a pre-prefix to the nounknown äs the augment:

The augment, appearing äs a partial pre-doubling of nominal and prenominalprefixes in many languages, would seem to have been a separate word, identical inform with the pronominal prefix, and used äs a weak demonstrative, or ratheranaphoric, in affirmative non-predicative constructions, with definite meaning.

De Blois (1970) surveyed the distribution of the augment in Bantu languages,working on the basis of Meeussen's hypothesis. Thilo Schadeberg (personalcommunication) points out that although Meeussen and de Blois propose that theaugment originated äs "identical in form with the pronominal prefix", theynowhere explicitly claim that Proto-Bantu had a low tone in the augments ofclasses l and 9, and that it is likely that the augment was high in all classes.

Thus, for example, in Class 5 the augment plus the nominal prefix would givea sequence *di-i. Applied to the Common Bantu root for 'head', this yields aremarkable convergence with the Igboid form.

Brought to you by | University of Wisconsin Madison Libraries 330 Memorial LibraryAuthenticated | 128.104.1.219

Download Date | 9/22/12 6:44 PM

The noun prefixes ofNew Benue-Congo 33

(6) Common Bantu *di-j-tue 'head' (Meeussen 1980)Proto-Igboid *rfi-Mui 'head'

This root for 'head' is usually found in 3/4 in Bantu, though Guthrie recordsoccasional 5/6 cases. De Wolf (1971: 54) reconstructs it in li/a, correspondingto 5/6, which is in line with the Proto-Igboid form.

While this word gives a very striking correspondence to Proto-Bantu,particularly in the matching of both prefixes, there is an indication that theremay even be a third prefix. When forms with CV-prefixes occur in secondPosition in the associative construction, they are preceded by a downstep,suggesting that a low-toned vowel originally preceded the CV-. In two otherIgboid words in rfV- which are cognate with reconstructions of de Wolf in li/a,an initial vowel or syllabic nasal actually survives in an occasional lect, therebyenabling us to make the revised reconstructions in (7).(7) 'head' 'tongue' 'breast'Common Bantu *di-|-tufc *di-j-diini *di-i-befcdfeProto-Benue-Congo *li- -to *li- -lemi *li- -baniProto-Igboid *x-rfi-Mui *i-dV-dö ^-(fe-v-mäThilo Schadeberg (personal communication) has drawn my attention to theargumentation developed by A.E. Meeussen, and Kadima (1969). They note thatthe class 5 prefix, äs reconstructed by Meeussen, is the only one consisting ofa vowel alone. This has been responsible for the class 5 augment being re-interpreted in many languages äs the nominal prefix, leading to CV-reconstruc-tions for the nominal prefix such äs Guthrie's *di- or *di-, after which it isnatural for Speakers, by analogy with the other classes, to add a "false augment"i- or |- at the beginning. Such a false augment will explain the extra initial prefixof Igboid. It will also explain why other New Benue-Congo languages, such äsEbira, classified in the Nupoid group, show a number of nouns which begin witha VrV- structure, representing a frozen false augment (V-) plus an old trueaugment (-CV-): - -sü 'head' (Adive 1989: 110).

3. Augment plus nominal prefix coalescing into a new nominal prefix

The augment is usually thought of äs a Bantu phenomenon. Doneux (1975),however, reconstructs it in Proto-North-Atlantic, though without discussing itstone. It is, therefore, not surprising to find it appearing in any language grouplower than Proto-Atlantic-Congo and higher than Bantu in the Niger-Congofamily tree, such äs Igboid. De Wolf (1971) does not report the augment in anyof the languages he surveys. Its apparent absence can be explained by thehypothesis that, äs proposed above for Igboid and Ebira, in many languages ithas been prefixed to the nominal prefix and has coalesced with it.

Brought to you by | University of Wisconsin Madison Libraries 330 Memorial LibraryAuthenticated | 128.104.1.219

Download Date | 9/22/12 6:44 PM

34 K.Williamson

Hombert (1976: 15) has already proposed such an explanation for the originof non-low nominal prefixes in Ngie, a Momo language within the Grassfieldsbranch of South Bantoid. He shows that Ngie mid (M) tone noun stemscorrespond to Proto-Bantu and Proto-Grassfields HL stems äs the result of aprocess of coalescence of the two tones into an intermediate one. He furtherClaims, contrary to Meeussen (1967) and de Blois (1970), that the augment wasH in all classes. Normally, the H of the augment plus the L of the nominalprefix coalesce into a M just äs in the noun stems. But in both class 9 and class10 the new prefix is, in all cases, L rather than the expected M. He attributesthis to the depressing effect of a nasal (i.e. the original nominal prefix). Thesequence N (from the original prefix) and C (the initial consonant of the stem)"would have had a depressing effect on the fundamental frequency of neigh-boring vowels which resulted in keeping a L tone on the vowel" (Hombert1976: 15). The process could, however, be obscured by the subsequent loss ofthe N which had caused the lowering. Hombert's scenario suggests interestingpossibilities, which are discussed below, using 'goat' äs an example (Table 2).

First, the Common Bantu reconstruction shows a HL tone pattern on the stem,which I assume to be original, and a non-syllabic nasal nominal prefix, forminga single, prenasalized NC with the following consonant and associated with a"semi-floating" low tone (Thilo Schadeberg, personal communication) for boththe Singular (class 9) and the plural (class 10). This reconstruction is supportedby that for Proto-Mbam-Nkam (Elias et al. 1984) and by the synchronic form forEjagham (Watters 1980), both showing loss of the original C2 but retaining thetone pattern. Ejagham has an altemation in the stem between -b- after the nasalin the Singular and -ß- after the vowel in the plural; Watters (1980: 117) showsthat the class 14 plural is a recent innovation because with numerals the nounretains its Singular form, interpreted äs a relic of the old class 10, which wasphonologically only minimally different from class 9. In Proto-Manenguba thestem has been simplified from HL to H. Hedinger (1987: 101 ff.) observes thatdiachronically, there was certainly a L nasal prefix for classes 9/10; syn-chronically the L tone remains in all cases, but before voiceless consonants thenasal has been lost, while before voiced stops the nasal can be analyzed äs thefirst part of a prenasalized stop.

Following Stewart (1983) in claiming that the N-prefix originales from anasalized vowel, but assuming a process of nasal epenthesis rather than hispreferred stop epenthesis, these processes are summarized in Table 3.

Brought to you by | University of Wisconsin Madison Libraries 330 Memorial LibraryAuthenticated | 128.104.1.219

Download Date | 9/22/12 6:44 PM

O O Tj-

ΟΝ ΟΝ ΟΝΟ Ο Ο ΟΟΝ ΟΝ ΟΝ ΟΝ ΟΝ

a

Ί11

8,

tαll

υ

l

,3 ^ ε ;

11!' • f^

.» l S00 S O

l

Brought to you by | University of Wisconsin Madison Libraries 330 Memorial LibraryAuthenticated | 128.104.1.219

Download Date | 9/22/12 6:44 PM

X

l

cn m cn"

ώ Ιω Ιω W ύ ύ α) Iw Iw H > Ι

1.~ Ό

J?IJ.Js l -

Brought to you by | University of Wisconsin Madison Libraries 330 Memorial LibraryAuthenticated | 128.104.1.219

Download Date | 9/22/12 6:44 PM

The noun pre xes ofNew Benue-Congo 37

Table 3. Derivationofdass9/10prefixesin CommonBantu,Proto-Mbam-Nkam,and Proto-Manenguba

Singularplural

*r-*r- > IN-> tN-Nasalepenthesis

> ft-> s.Voweldeletion

> N-> N-L tonedeletion

Voorhoeve (1980) describes dere, a northern Mbam-Nkam language, in whichclasses 6, 9, 10 and 19 have a homorganic, non-syllabic nasal following theprefix and preceding the stem. The nasal does not function synchronically s aprefix. I Interpret this, in agreement with Miehe (1991: 60, 63), s an inter-mediate stage before the loss of the nasal postulated by Hombert in languageslike Ngie. Classes 9/10 have a zero prefix preceding this nasal, and the pluralis reduplicated. The HL stem pattern coalesces into M.

Table 4. Derivation ofdass 9110 prefixes in dere

Singularplural

*ί-*ί-

> ίΝ-> ΙΝ-Nasalepenthesis

> Ν-> Ν-Voweldeletion

> Ν-> Ν-L tonedeletion

> 0> 0-mbu-Prefixation

In Ngie, in a modification of Hombert's analysis, I assume that a H augment hasbeen added to the noun classes which had the shape V, and that the H of theaugment coalesced, s suggested by Hombert, with the L of the nominal prefixto produce a M prefix. The augment was not, however, added to 9/10, which hadthe shape IN- and were interpreted s a combination i-ft- parallel to the V-V- ofthe other classes; hence the absence of a H in these classes. The apparentcombination i-ft-, however, did not arise from a combination of two prefixes,but, s proposed above, from an earlier nasalized vowel prefix *ί-, which gaverise to an epenthetic nasal between itself and the stem. This originally epentheticnasal, the source of the nasal for classes 9/10 in Bantu, was later deleted inNgie, leaving the present oral prefix. As proposed by Hombert, the stemcoalesces its HL pattem into mid.

Table 5. Derivation ofdass 9/10 prefixes in Ngie

Singularplural

*r-*ί- > IN-> IN-Nasalepenthesis

> IN- > i-N-> IN- > 1-N-Vowel Reanalysisdenasalization

> 1-> 1-Nasaldeletion

Brought to you by | University of Wisconsin Madison Libraries 330 Memorial LibraryAuthenticated | 128.104.1.219

Download Date | 9/22/12 6:44 PM

38 K.Williamson

Obviously, the process suggested for Ngie results in ambiguity between classes9 and 10. If other Southern Bantoid languages underwent a similar process, thenthe ambiguity was likely to be resolved in different ways in different languages,äs indeed it is; the plural (class 10) is distinguished from the Singular (class 9)by reduplication (äs already seen in Ädere), by suffixation, äs in Babungo bi'goat', b^-sÄ 'goats' (Schaub 1985), or by prefixation of the augment, äs seenin the following languages. Thilo Schadeberg (personal communication) pointsout that there are clear parallels within (Narrow) Bantu for the use of theaugment for distinguishing the plural class 10 from the Singular class 9,especially in Zones R and S, e.g. UMbundu (R. 11):

(8) Singular Pluralo- N- o- I- o- N-H L H L L Laugment nominal prefix 1 2 3 4o^mbämbi olo^mbämbi 'gazelle (small species)'

1 = the synchronic augment, o- for all classes2 = the old augment for class 103 = the frozen augment of the class 9 nominal prefix4 = the nominal prefix class 9/10In Esimbi (Stallcup 1980) I assume that an augment has been added to theoriginal nominal prefix in the plural only. The Singular prefix iN- caused alowering of the following stem to L in the Singular, whereas the H in the pluralsequence *i-IN- did not. The HL stem coalesced to M, äs in Ngie. Finally, thenasal of the nominal prefix is lost, just äs in Ngie, while the vowel of theoriginal nominal prefix is retained in the Singular but lost after the identicalvowel of the augment in the plural.

Table 6. Derivation of class 9/10 prefixes in Esimbi

Singularplural

*i- > IN-* - > IN-

Nasalepenthesis

> IN-> IN-Vowel

>> i-i-N-Augment

> i-> i-iN-deletion

denasalization prefixation

In the Beboid language Noni, Hyman (1981) reconstructs the root with a Lprefix in the Singular and a H one in the plural, äs shown in Table 2; the surfaceforms are bie and ble respectively. Both the stem and the augment with thenominal prefix coalesce into M, äs in Ngie.

Brought to you by | University of Wisconsin Madison Libraries 330 Memorial LibraryAuthenticated | 128.104.1.219

Download Date | 9/22/12 6:44 PM

The noun prefixes ofNew Benue-Congo 39

Table 7. Derivation ofclass 9/10 prefixes in Noni

Singular *ί- > iN- > IN- > > l- > x-plural *ί- > iN- > IN- > WN- > T- > ~-

Nasal Vowel de- Augment -N-deletion Vowelepenthesis nasalization prefixation coalescence deletion

In the Platoid branch, the Plateau language Eggon, described by Sibomana(1985), shows contrasts between L and M in Singular prefixes but none in pluralprefixes, which are all M. This can be interpreted s in Esimbi (augment addedin plural but not in Singular), if the Singular vowel prefix becomes e-, either bysound change or by analogy with the plural.

Table 8. Derivation ofclass 9/10 prefixes in Eggon

Singular *ί- > l- > fe-plural *ί- > i- > έ-l- > e- > e-

Vowel Augment Tone coalescence, Vowel heightdenasalization prefixation i-deletion lowering

Alternatively, a L augment was added in the Singular and a H one in the plural(Table 9).

Table 9. Alternative derivation ofclass 9/10 prefixes in Eggon

Singularplural

*ί-*r- > i-> 1-Voweldenasalization

> έ-ϊ-> έ-ι-Augmentprefixation

> έ-> e-Tone coalescence,i-deletion

The second alternative is simpler and, s will be seen, is in line with otherlanguages. The stem has simplified to H.

In Proto-Upper Cross, Sterk (no date, a) reconstructs the stem of 'goat' witha HL tone and with a preceding floating L tone. I widerstand the floating L torepresent the tone of the prefix, which will be the same for Singular or plural,whereas the vowel may vary. Unfortunately I do not have his reconstructedprefixes, and therefore write only V. In many stems referring to animals,including 'goat', the first consonant is represented s doubled, that is "fortis"or long. I propose that the length of the consonant is the result of the originalnominal prefix, perhaps a nasal, being assimilated to the following stem-initialconsonant. (Janssens 1991 has proposed that "strong" [fortis] reflexes in Bantuare often the result of consonants following a nasal.) If this is correct, then the

Brought to you by | University of Wisconsin Madison Libraries 330 Memorial LibraryAuthenticated | 128.104.1.219

Download Date | 9/22/12 6:44 PM

40 K.Williamson

original nominal prefix is represented by two survivals: the low tone of thecurrent prefix and the lengthening of the initial consonant. The vowel of thecurrent prefix represents the augment.

Table 10. Derivation ofclass 9/10 prefixes in Proto-Upper Cross

singularplural

*t-*ΐ-

> IN-> IN-Nasalepenthesis

> iC-> iC-Denasal-ization

> V-iC-> V-iC-Augmentprefixation

> Y-C-> V-C-i-deletiontone shift

In Proto-Lower Cross, Connell (1991) reconstructs a H tone stem, again asimplification of the original HL, and H tone prefixes έ-/ί-. These prefixes arehere inteφreted s secondary prefixes originating from the augment; this timethe H of the augment is retained rather than the L of the augment.

Table 11. Derivation ofclass 9/10 prefixes in Proto-Lower Cross

singularplural

*f-*ί-

> i-> 1-Voweldenasalization

> έ-1-> ί-1-Augmentprefixation

> έ-> ί-1-deletion

In Nupoid, the only functioning noun class System is that of Gade: "Nounprefixes generally have a low tone. A few noun prefixes have a high tone andsome have a mid tone. These have to be marked in the lexicon" (Sterk, no date,b: 9). This Statement suggests that in Gade, s in Amo, the original nounprefixes of Proto-Benue-Congo have largely survived without combination withthe augment. The singular prefix f - of Sterk's class 9 is an innovation, perhapscognate with Bantu class 19; the plural i- could be interpreted s either adenasalized reflex of the old nominal prefix of class 10 or s an augment whichhas lost its high tone and taken on the L of the old nominal prefix. Given theapparent absence of the augment in Gade, the first Option seems more likely.

Table 12. Derivation ofclass 9/10 prefixes in Gade

singular *i- > i- > f -i- > f -plural *i- > i- > i-

Vowel Re-prefixation i-deletiondenasalization

Brought to you by | University of Wisconsin Madison Libraries 330 Memorial LibraryAuthenticated | 128.104.1.219

Download Date | 9/22/12 6:44 PM

The noun prefixes ofNew Benue-Congo 4 1

(In vowel harmony languages, such äs Gade, the prefixes are given in their non-expanded form harmonizing with the stem.) The stem in Gade shows a coalescenceof HL to M äs in Ngie. Asu also shows a L prefix, but this time with a L tonestem äs well, perhaps to be explained äs in Esimbi (Table 6). Finally, Ebirashows an apparent coalescence of H augment and L nominal prefix in the prefix,and a simplification of the HL stem to H.

Table 1 3. Derivation of class 9 prefix in Ebira

singular *i- > i- > - -Vowel Augmentdenasalization prefixation

> 8-Tone coalescence,i-deletion

The lack of unity in the Nupoid reflexes is noticeable.In Idomoid, the Yala singular prefix clearly shows the old augment with its

initial consonant (cf. Table 1), either with L tone or with its H tone replaced bythe old nominal prefix L. In the other Idomoid languages, the initial y- has beenlost. The tones on the noun stem coalesce to M.

Table 14. Derivation of class 9/10 prefixes in Yala (Ikom)

singular * -plural * -

> i- > ye-1-> i- > - -Vowel Augmentdenasalization prefixation

> yfc-> fe-1-deletiontone shift

Idoma is similarly derived. Igede is interesting; the stem (unmarked in Table 2)bears a lowered mid tone, whereas both the singular and plural prefixes bear araised mid tone (Igede has a four-tone System). This suggests that the stem Hplus L coalesced into M, which was then lowered by the original L nominalprefix. The augment in both singular and plural was H; the H augment plus theL nominal prefix coalesced to give M (which had no preceding L to lower it) onthe new prefix. Omitting the prefix-stem interactions, both singular and pluralprefixes developed äs in Ebira (Table 13).

The M tone of the singular prefix in Eloyi Mbeji might be taken äs evidencefor a similar coalescence of a H augment and a L nominal prefix; but there isthen a difficulty in explaining why the same coalescence does not take place inthe plural prefix, where H contrasts with the M of the singular. It is simpler toassume that it was originally L, äs in Etoyi Mbekyi, and later underwent apartial assimilation to the following H (Table IS).

Brought to you by | University of Wisconsin Madison Libraries 330 Memorial LibraryAuthenticated | 128.104.1.219

Download Date | 9/22/12 6:44 PM

42 K.Williamson

Table 15. Derivation ofclass 9/10 prefixes in Eloyi Mbeji

Singularplural

*r-*r- > \-> i-Vowel

> έ-ι-> έ-ί-Augment

> έ-> έi-deletion

denasalization prefixation

> ε-

L-toneassimilation

The initial nasal consonant of the stem in both Eloyi and Igede reflects theoriginal nasality of the nominal prefix.

In Igboid the tone of the stem is HL in some lects, simplified to H in themajority, just s in *head'. A downstep when the noun is in second position inthe associative construction, s in 'head', indicates the low tone of the originalnominal prefix. The H prefix έ-, s in Proto-Lower Cross, is derived from theaugment (Table 16).

Table 1 6. Derivation of class 9 prefix in Proto-Igboid

Singular *ί- > l- > έ-1-Vowel Augmentdenasalization prefixation

> έ-Ν-i-deletion

The tones of Proto-Edoid have not been reconstructed, and looking at the varietyof tone pattems cited here from Elugbe (1989) it is clear that it is a formidabletask if undertaken solely from inside the group. Tentatively, I will try to Interpretthese forms in line with the other languages so far examined. I note first thatthere is no overt trace of the original nominal prefix, nor has it left anysegmental trace by making the initial consonant of the root fortis; Elugbereconstructs it s lenis. I assume, however, that it was present because of itsapparent effect on the tone of the prefixes, which are the same for both Singularand plural. There are three sets of tonal reflexes which, incidentally, do notcoincide with the internal subgrouping of Edoid. In Degema, Urhobo and Edothe prefixes are low, which is a reflex of the L of the original nominal prefix;the stem simplifies its original HL to H.

Table 1 7. Derivation of class 9/10 prefixes in Degema

singular *i-plural *i-

> i- > c-i-> i- > i-i-Vowel Augmentdenasalization prefixation

> έ-> i-i-deletiontone drift

Brought to you by | University of Wisconsin Madison Libraries 330 Memorial LibraryAuthenticated | 128.104.1.219

Download Date | 9/22/12 6:44 PM

The noun prefixes ofNew Benue-Congo 43

In Aoma, Uhami and Ukue the prefixes are high, reflecting the H of the augment(cf. Proto-Lower Cross, Table 11), while the L of the nominal prefix has beenrealized on the stem. Finally in Ghotuo and Ehueun a coalescence of the new Hand the old L on the two prefixes has yielded a M which cannot be interpretedäs an assimilation to a following H, because the stem is L äs the result of aperseverative assimilation, äs in Esimbi and Asu. Thus these apparently chaotictonal correspondences can be understood if set in a larger Benue-Congo context.

After this, the Yoruboid correspondences are child's play. The stem hassimplified HL to H, and in Ife (Togo) and Yoruba the H and L of the twoprefixes have coalesced into M, äs in Ebira (Table 13). In Igala the H prefix hasdisplaced the L; it is possible the L remains underlyingly, äs in Igboid, becausethere are tonal alternations in the Igala noun in various dialects. It should benoted that throughout Yoruboid there is a three-way contrast (H,M,L) in nounstems but only a two-way contrast in noun prefixes: M or L, äs in Yoruba, orH or L in Igala.

4. Conclusion

In the search for a historical origin for the complexity of tones in Igboid nouns,at least a partial answer has been found by suggesting that in many New Benue-Congo groups the present noun prefixes are to be regarded äs a complex of two(or possibly on occasion even three) original prefixes. By comparison withBantu, I have identified the first of these with the augment and the second withthe nominal prefix. Similar conclusions have already been drawn within SouthBantoid by Hombert (1976). The nominal prefixes were all originally L, äs theystill are in Amo (Table 1), the augments were probably all H in Proto-Bantu, butelsewhere at least the class 9 augment was apparently L in some groups.

The combination of a H and a L prefix into a single one has led to a varietyof reflexes. First, the new prefix occurs äs a H, sometimes followed by a floatingL which surfaces either äs a downstep in particular grammatical circumstances,or äs a total lowering of the noun stem. Secondly, äs a L, explained äs the L ofthe original nominal prefix replacing whatever tone the augment bore. Thirdly,äs a M, explained äs a coalescence of the H and L into an intermediate tone.

The tones of originally HL noun stems (exemplified by 'head' and 'goat' inthese examples) are remarkably stable once the effects of the prefix in particularlanguages are recognized. Bantu appears particularly archaic, both in the tonesof its noun stems and in attaching only the nominal prefix to them in manylanguages. The addition of the augment takes place independently in differentBantu groups, and thus provides a model for what happened in other branchesof Benue-Congo.

Miehe (1991) has already shown that the nasal prefixes of Bantu, proposed byGreenberg (1963) äs an Innovation, in fact surface in many Benue-Congo and

Brought to you by | University of Wisconsin Madison Libraries 330 Memorial LibraryAuthenticated | 128.104.1.219

Download Date | 9/22/12 6:44 PM

44 K.Williamson

even Kwa languages; she suggests that they go right back to Proto-Niger-Congoand explains them, äs here, äs having been disguised by re-prefixing andtherefore surfacing only in an apparently irregulär fashion, whereas they remainopenly accessible in archaic Bantu. Accepting Stewart's hypothesis that theprefixes of classes 9 and 10 were originally close nasalized vowels rather thanhomorganic nasals, it is somewhat easier to explain why these old prefixessurface sometimes äs close vowels, sometimes äs homorganic nasals, andsometimes äs both.

There are still places where alternative processes can be postulated to explainthe modern forms. The reconstruction of the prefixes of Benue-Congo by deWolf (1971) should now be re-examined in the light of this hypothesis to obtainthe simplest overall picture and, if possible, evidence of distinctive innovationswhich will improve the iriternal classification of New Benue-Congo.

University ofPort Harcourt andWolfson College, Oxford

References

Anderson, Stephen C. (1980). The noun class System of Amo. In Noun Classes in the GrassfieldsBantu Borderland, Larry M. Hyman (ed.), 155-178. Los Angeles, CA: Department ofLinguistics, University of Southern California.

Armstrong, R. G. (1965). Comparative wordlists of two dialects of Yoruba with Igala. Journal ofWest African Languages 2: 51-78.(1983). The Idomoid languages of the Benue and Cross River valleys. Journal ofWest AfricanLanguages 13: 91-149.

Connell, Bruce (1991). Phonetic aspects of the Lower Cross languages and their implications forsound change. Unpublished Ph.D. dissertation, Edinburgh.

de Blois, K. F. (1970). The augment in the Bantu languages. Africana Linguistica 4: 87-165.de Wolf, Paul (1971). The Noun Class System ofProto-Benue-Congo. The Hague: Mouton.Doneux, J.L. (1975). Hypotheses pour la comparative des langues atlantiques. Africana Linguistica

6: 41-129.Elias, Philip, Jacqueline Leroy and Jan Voorhoeve (1984). Mbam-Nkam or Eastern Grassfields.

Afrika und Übersee 67: 31-107.Elugbe, Ben Ohiomamhe (1989). Comparative Edoid: Phonology and lexicon. (Delta Series, 6.) Port

Harcourt: University of Port Harcourt Press.Greenberg, Joseph H. (1963). The languages of Africa. Internationaljournal of American Linguistics

29.2.1.Guthrie, Malcolm (1967-1971). Comparative Bantu. Farnborough: Gregg.Hedinger, Robert (1987). The Manenguba Languages (Bantu A.15, Mbo Cluster) of Cameroon.

London: School of Oriental and African Studies.Hombert, Jean-Marie (1976). Noun classes and tone in Ngie. In Studies in Bantu Tonology, Larry M.

Hyman (ed.), 1-21. Los Angeles, CA: Department of Linguistics, University of SouthernCalifornia.

Hyman, Larry M. (1980). Babanki and the Ring Group. In Noun Classes in Grassfields Bantu, LarryM. Hyman and Jan Voorhoeve (eds.), 193-226. Paris: CNRS.

Brought to you by | University of Wisconsin Madison Libraries 330 Memorial LibraryAuthenticated | 128.104.1.219

Download Date | 9/22/12 6:44 PM

The noun preßxes ofNew Benue-Congo 45

Hyman, Larry M. (1981). Noni Grammatical Structure. (Southern California Occasional Papers inLinguistics, 9.) Los Angeles, CA: Department of Linguistics, University of Southern California.

Janssens, Baudouin (1991). Doubles räflexes apparents en ewondo ou les chass6s-crois€s de laderivation. Pholia 6: 155-190.

Meeussen, A. E. (1967). Bantu grammatical reconstructions. Africana Linguistica 3: 79-128.Miene, Gudrun (1991). Die Präfixnasale im Benue-Congo und im Kwa. (Sprache und Oralität in

Afrika.) Berlin: Dietrich Reimer.Schaub, Willi (1985). Babungo. London: Croom Helm.Scholz, Hans-Jörgen and Chris Scholz (1972). Let'sGoForward—Let'sReadandWriteEbira. Zaria:

Institute of Linguistics, Ahmadu Bello University.Sibomana, Leo (1985). A phonological and grammatical outline of Eggon. Afrika und Obersee 68:

43-68.Stallcup, Kenneth L. (1980). Noun classes in Esimbi. In Noun Classes in the Grassfields Bantu

Borderland, Larry M. Hyman (ed.), 139-153. Los Angeles, CA: Department of Linguistics,University of Southern California.

Sterk, Jan. (No date, a). Reconstruction of Proto Upper Cross. Unpublished manuscript.(No date, b). Gade-English dictionary, including English-Gade reference dictionary andsummary of Gade grammar. Unpublished manuscript.

Voorhoeve, Jan. (1980). Noun classes in Ädere. In Noun Classes in the Grassfields Bantu Borderland,Larry M. Hyman (ed.), 57-70. Los Angeles, CA: Department of Linguistics, University ofSouthern California.

Watters, John R. (1980). The Ejagam noun class System: Ekoid Bantu revisited. In Noun Classes inthe Grassfields Bantu Borderland, Larry M. Hyman (ed.), 99-137. Los Angeles, CA: Departmentof Linguistics, University of Southern California.(1990). Reduplication and the origin of high tone in noun prefixes in Ejagham. Journal of WestAfrican Languages 20: 105-117.

Williamson, Kay (1986). The Igbo associative and specific constructions. In The PhonologicalRepresentation of Suprasegmental: Studies in African Languages Offered to John M. Stewart onhis 60th Birthday, Koen Bogers, Harry van der Hülst and Maarten Mous (eds.), 195-208.Dordrecht: Foris.(1989). Benue-Congo overview. In The Niger-Congo Languages, John Bendor-Samuel (ed.),246-274. Lanham: University Press of America.

Williamson, Kay and C. Ohiri Aniche. (Forthcoming). Comparative Igboid.

Brought to you by | University of Wisconsin Madison Libraries 330 Memorial LibraryAuthenticated | 128.104.1.219

Download Date | 9/22/12 6:44 PM

Brought to you by | University of Wisconsin Madison Libraries 330 Memorial LibraryAuthenticated | 128.104.1.219

Download Date | 9/22/12 6:44 PM


Recommended