Gur-Oti-Volta-Western Nominal Classes

Nominal Declensions in Gur > Oti/Volta > Western > Southeast > Mampruli (/Dagbani/Kusaal…)

Tony Naden  :: G.I.L.L.B.T.

July 28, 2008

[ This is not the assured results of scholarship; the reason I am trying to get on with the Comparative Dictionary is because we need a better quantity and quality of data before we can go further than we have reached so far. This is an ad hoc summary thrown together in response to a request from Hannes Hirzel. It is what I know about the subject without doing further specific research, but I suppose I know as much as most people. I am presenting the system as primarily represented by Mampruli which I know best and which is also a fairly conservative language – it is more convincing (to me!) to derive the data for the other languages from something like the Mampruli system, and it also ties the immediate cluster into the less-closely-related languages better. Anyone working solely within the data of a single language, whether by limitation or conviction, would not see all of this, and would probably come up with a more-complex, less-tidy analysis which would, however, not postulate so many abstract underlying forms and processes (Geoffrey Hunt on Hanga, Paul Schaefer on Safaliba). Read John Callow, 1965, Kasem nominals - a study in analyses : JWAL II(1) : 29-36 several times, slowly, concentrating on the principle expounded, not so much the data. This presentation fits Mampruli, and KusaalAgole (allowing for the peculiarities of its (morpho-)phonology, and apart from a few aberrations Dagbani and Hanga (/KaMara/Kantoonsi). It should also fit Talni and Nabit (data insufficient, but Talni is like Mampruli and Nabit like KLToende). Farefare, Moore and Wali have essentially-similar systems, as does Safaliba. In the Dagaari-Birifor cluster the sound-changes have muddied the system to the point where it is possible but of largely theoretical interest to link it to this system: Manessy includes them in the ‘langues voltaïques sans classes’.

Declensions and Classes

We are primarily concerned with suffixes to the nominals which indicate the categories Singular, Plural and Non-Count. It is largely arbitrary which singular suffix a given stem takes, but, given the singular suffix, the one used for the plural is largely predictable. This is usually referred to as ‘Noun Class System’, and the term ‘class’ had been applied either to each separate suffix, or to the pairing of a singular with a plural; in the former case the sg./pl. pairings are sometimes called ‘gender’. Given that in most of these languages there is no concord with other elements in the phrase or clause, and the pairings do not systematically correlate with any semantic categories, let alone ‘masculine’ and ‘feminine’, I prefer to avoid these usages and talk of ‘declensions’ as is used for similar phenomena in Latin or Greek. If you want to call them ‘classes’ or ‘genders’ that’s fine with me, just make it clear how you are using your terms.

Semantic criteria: The idea of semantic correlation is unhelpful. Apart from a declension which contains only (but by no means all) human referents, and those referring to liquids/masses/ abstracts, which are as much syntactic as semantic categories, attempts to define other declensions by semantic criteria are either very full of exceptions, or have to associate disparate definitions (‘domestic animals and long-and-thin objects’), or are so abstract as to be vacuous. What can be said is that some groups referents, for no perceivable reason, stay together in the same declension as each other as one passes from language to language, even when the form, and function in the system, of the declension changes. Also that sub-regularities develop in different languages, such as having a group of roots which in one declension name a species of tree and in another the fruit, nut or seed of the said tree.

Presentation: There are seven nominal declensions to be considered. I will discuss each one under the formula for its pair of suffixes, or suffix. In my convention *NN refers, not to a formally-established Ur-form, but to a summary formula which will be recognised as, or as sufficiently like, the actual form in any language one is looking at. Within languages *nn is a comparable abstract or summary form representing a morpheme which may appear in different allomorphs. Thus “the suffix *-DI is represented in Dagbani by *-li which may appear as –ni”. I number the declensions with Roman numerals; this is an arbitrary device, but one needs some way of referring to a particular declension. The order is quasi-arbitrary: everyone starts with the persons declension and ends the core group with the liquid/mass one – the two with some semantic basis. The three which contain most count nouns follow, people differ in their order. I then put the smaller and rather irregular ‘cow, horse, snake and money’ group before the liquids (in earlier presentations the order of thse two was reversed). It is also usual to list the odd or rare groups at the end: the only one of these in our languages I list as seventh.

It is a matter of taste where one draws the line between very-small classes and irregularities. Some items are regularly irregular, in many cases in the same way from language to language (but not the same as each other) – ‘guineafowl’, ‘hand’, and ‘sheep’, for instance have normal suffixation but abnormal sg./pl. pairing, ‘man’, ‘small’ and ‘money’ have a different form of the root for singular and plural (dropping or adding a consonant), ‘house’ may collapse its plural to a single syllable (*yi- + *-a > yiya > ya) and sometimes retains its suffix in compounded forms.

Stem/Suffix Boundary: The stems onto which these forms are suffixed may end with a vowel, a nasal, or the consonants –B-, –D-, -G-, -L-, -S- (voicing non-contrastive). It is the interaction between stem-ending and suffix-beginning that makes the actual data so much more complex than the basic system I postulate here. The overall system for Mampruli can be seen in an item I omitted to add to the references and hyperlinks of ‘Gur Morphology’ – the grammar sketch prefaced to Unit III of the language-learning course pp. xxviii ff. < TN-MP-GrammarU.DOC#Nouns > . Various expositions and examples of parts of the system or relevant data will be found in the other items in this folder and mentioned and hyperlinked in Gur Morphology < TNgurMorphology.doc >.

General Observations: A number of the singular suffixes are most likely to be obscured by morphophonemics, so it is often more useful to look at the plurals first. Therefore, when collecting survey or sample data, you must get plurals where applicable. You can usually get the stem form by asking for the noun followed by an adjective. There is a widely-found rule that lengthens a stem-final vowel before a consonant-initial suffix. There is a tendency to drop final vowels, very common in Agole Kusaal, Talni, and the Dagaari cluster (also Buli). Sometimes the lost suffix-final vowel is echoed or metathesised to before the consonant (this is where SVMR is trying to get to < TN-SVMRu.DOC >).

The Declensions

I – *-A~O~Ø / -BA  :

The *BA element for persons plural goes righhttp://www.typecraft.org/w/skins/common/images/button_bold.pngt down to Bantu. It may assimilate a preceding nasal to make -m+b-, and may be in turn cross-assimilated to the nasality to give –m+m- The resulting geminate may be degeminated -m- .

“strangers” *SAAN- + -BA  : DB saamba MP saamma KLA saam(a)

Following stem-final b the ‘devoicing’ rule may apply

“men” *DAB- + *-BA : MP dɔppa KLA dap(pa) cf./ct. DB daba (~dabba~dɔbba)

The singular is usually *-A in this group, but *O and zero are found.

Founder-members of the declension are ‘man’, ‘woman’, ‘person’ and ‘stranger’: other basic nouns are few, but forms in this declension are common as it is a productive or widely-lexicalised way of forming agent nouns from the imperfective forms of verbs <Bayreuth paper TN-VNderivn06u.doc#Agent >.

One special item which is found throughout is ta(a)ba “together, each-other” : the singular, presumably *ta.a, is rarely found as such, unless in compounds like (MP/DB) nyintaa (KLA nintaa) “co-wife”, dataa “wife’s lover”, but these are now in Decl.3 (pl. –taas(i) ) :

“peer, equal, ‘fellow-friend’, neighbour”

MP taraana KLA tiraan HG to ~ todaana pl. (all) taaba : DB taba

II – *-DI / *-A :

The plural suffix *-A derives from something like *ŊA seen in the archaic survival in the Dagbani indefinite shɛŋa “some, certain ones” (and, e.g. in Buli). Interestingly parallel to the historic development of that *Ŋ in initial position, the plural suffix has an allomorph *YA when following a stem-final vowel.

“roads” *SO- + *-A  : DB/MP sɔya HG soya KLA sueya

After all the consonant-final stems it just forms with the consonant a final CV syllable. One can therefore fairly confidently isolate the stem by removing the final -a.

It is in the singular that the fun starts. The most obvious (post-vocalic) form of II.sg. and IV.pl./collective in Mampruli is identical, -ri (In all the languages [r] is an allophone of /d/ but speakers familiar with English have insisted in representing it orthographically.)

The corresponding pronoun, historically and in systems which still (again?) have class-concord , is *DI (See table < TNnClas1.doc >)

In several languages the pronoun has an allomorph li (DB di subject/possessive, li object : KLA di and li in free, or (idio-?)dialectal variation). The suffix is basically *-li in DB and some other languages:

“road” *SO- + *-DI  : DB soli MP/HG soori KLA suer(e)

Following a nasal the *D becomes n and/or preceding the *D a stem-final nasal may become n : the cross-assimilation obscures the contrast of stem-final nasals which is visible before the vocalic plural suffix:

“lion” *GBIGM- *-DI / -*A  :

MP gbigimni or gbiginni / gbigima

DB gbuɣinli / gbuɣima

Following an l a double ll results in Mampruli as much as in Dagbani:

“egg” *GEL- + *-DI  : DB galli/gala MP gyɛlli/gyɛla HG jilli/jila KLA  gel(le)/gela

When the resulting geminates are degeminated, the singular suffix suffix looks like just *-I  :

“tooth” – *ÑIN- + *-DI/*-A:

MP/HG nyinni / nyina KLA nyin(e) / nyina DB nyini / nyina

Following b, g, and s the normal morphophonology applies (may be epenthetic vocoid, s may be h in some contexts). Following another d~r, instead of the plain devoicing various things can happen (note that three results are possible with the same word in MP):

“baboon” *KPAD- + *-RI  :

1) No change : DB kparili

2) Cross-assimilation : MP kpalli

3) Dissimilation (metathesis?) : MP kparili1

4) Devoicing and retention : HG/MP kpatiri KLA kparit

Membership has all sorts of animals (wild and domestic), body parts (single and paired) and inanimate objects, in many cases with semantically-comparable roots being found in II or IV. In DB and MP (? partially in the others) ‘tree’ stems in III refer to the tree and in II to the fruit.