The Classification of Languages 211
collectivity, of a big number, a liquid, and also of things which occur
in pairs, e g ma-naka (horns of an animal) The prefix ka- (Ganda)
corresponding to a plural prefix tu-3 denotes small size, e g ka-ntu
(small man)., tu-ntu (small men) With the prefix bo- (Duaia), abstract
nouns are formed, derived from adjectives, verbs and names for things,
eg bo-nyaki (growth, from nyaka} grow) The prefix ku- (Ganda)
serves for the formation of verb-nouns or infinitives, e g ku-lagira (to
command, or commanding)
Since there is no precise parallel to this type of concord in our own
language,, we must fall back on an artificial model to illustrate what it
involves Let us first suppose that every English noun had one of
twenty prefixes analogous to the suffix ~er common to the occupational
fisher-wnter-builder class We may also suppose that the words dog and
sheep respectively earned the prefixes be- and rrf- If English also had
the same concord system as a Bantu dialect, the sentence hungry dogs
sometimes attack young sheep would then be be-hungry be-dogs sometimes
be~they-attack nf-young iri-sheep.
The origin of the Bantu classifiers is not above dispute It is
possible, though not conclusively proved, that they weie once inde-
pendent words with a concrete meaning, standing for groups of allied
objects, such as human beings, trees, liquids, things long or short, big
or small, weak or strong When associated with other words they
originally marked them as members of one class According to this view,
be-dog and nf-sheep of the parable used above would be what remains
of beast-dog and meat-sheep Subsequently the outlines of once-distinct
classes became blurred through contamination and fusion, and the
classifier sank to the level of a purely grammatical device. If so, the
original plan has survived only in the first two classes. With few excep-
tions these signify human beings
Only in a relatively static society at a primitive level of culture with
little division of labour could classificatory particles retain a clear-cut
function Migration and civilization biing human beings into new
situations which call for new vocables. These do not necessarily fall
into any pre-existing niche of a classificatory system. In fact, languages
of the classificatory type are confined to communities which used
neither script nor the plough before contact with white men. The
surmise that Bantu classifiers were once concrete words suggests
analogy with the numeratwes which the Chinese and Japanese almost
invariably insert between figures and things counted, as when we speak
of three head of cattle. Thus the Chinese say two piece man (=== two men),
three tail fish (= three fish), four handle knife (=fow kmves^five orna-