Heine Bernd, Nurse Derek. A Linguistic Geography of Africa
Подождите немного. Документ загружается.
b. tse
´
di
nts
h
U
tse
´
di
`
q
h
Ul
U
tse
´
8/10:LINK 8/10:black 8/10:LINK 8/10:big 8/10:LINK
k
I-di
`
-re
´
ki
le
`
-
˛
S1SG-O3:8/10-buy-ANT-REL
‘the big black ones that I bought’
“Multifunction” linkers of the type illustrated by this Tswana example are
not common. By contrast, linkers used specifically to introduce postnominal
relatives are common throughout Africa; see section 4.14.3.
In the languages that have a construct form of nouns, the construct form may
appear in combination with a particular type of modifier only (typically, in
combination with genitival modifiers only), or with various types of modifiers.
Note that, in genitival constructions, the use of a construct form of the head
noun must be carefully distinguished from the obligatory use of possessive
affixes attached to the head noun, which is crosslinguistically a much com-
moner way of marking the genitival relation; see section 4.4.6.
The notion of construct form, traditional in Semitic linguistics, is rarely
explicitly mentioned in descriptive grammars of non-Semitic languages, but in
fact, morphological variations of nouns gover ned by the presence of (particular
types of) modifiers are found in many African languages that do not belong to
the Semitic family, for example Wolof, Tswana, Hausa, or the Cushitic lan-
guage Iraqw.
22
The following example is from Wolof.
(34) a. fas wu n
˜
uul
horse CL:LIN K be:black
‘black horse’
b. suma nijaay
1SG uncle
‘my uncle’
c. suma fas-u nijaay wu n
˜
uul
1SG horse-CSTR uncle CL:LINK be:black
‘my uncle’s black horse’
AccordingtoAndersen(2002 ), construct forms are common in Western
Nilotic languages (Dinka, Shilluk, DhoLuo, Pa
¨
ri,andAnywa).Inother
families, in particular in those incl uded in the Niger-Congo phylum, they
seem to be rather sporadic. Note however that in Tswana, the distinction
between the absolute form and the construct form of nouns is purely t onal,
and it may well be that a similar distinctionhasbeenmissedindescriptionsof
other languages.
Denis Creissels et al.122
4.4.6 The genitival modifier
It has been mentioned in section 4.4.5 that the genitival construction may
involve a special “construct form” of the head noun. The genitival construction
may also involve a genitive marker attached to the genitival modifier, or
so-called “possessive affixes.”
As regards the use of genitive markers, case-marked genitives are relatively
common in Africa, even in languages devoid of case contrast between subject
and object. By contrast, genitival constructions involving obligatory posses-
sive affixes even in the presence of a noun phrase in genitive function are not
very common, but they are found in different language families and in different
parts of the African continent, and the same can be said of genitival con-
structions involving a construct form of the head noun; see section 4.5.
Crosslinguistically, gender–number agreement of the genitival modifier is
not very common, but it occurs in the most typical Niger-Congo noun class
systems (particularly in Bantu languages).
A number of African languages have more than one possible way of combining
a noun with a genitival modifier, most commonly with a distinction in meaning so
that the variant with more morphological material (genitive markers or possessive
affixes) is used with “non-intimate” (or “alienable”) types of relations, and that
with less morphological material with “intimate” (or “inalienable”) types of
relations. As a rule, the inalienable form involves mere juxtaposition. This type of
distinction is particularly common in some language groups (e.g. Mande lan-
guages), but it is not really bound to particular families or areas.
For example, a distinction between alienable and inalienable possession
exists in many Chadic languages. In Lele, when the possessor is nominal, if the
possessum is alienable, the order is possessor possessum-pronoun, a pronoun
coding gender and number of the possessor being suffixed to the possessum
(Frajzyngier 2001):
(35) a. kı
´
we
´
dı
`
nga
`
w-ro
`
leopard ferocity-3FSG
‘the ferocity of the leopard’
b. gu
`
mno
´
dı
`
nga
`
wr-ı
`
y
buffalo ferocity-3MSG
‘the buffalo’s ferocity’
When the possessor is nominal, if the possessum is alienable, the order is
possessum–possessor ke
`
-pronoun:
(36) a. ku
´
lba
´
ca
`
nı
`
ge
´
ke
`
-y
cow Canige GEN-3MSG
‘cow of Canige’
Africa as a morphosyntactic area 123
b. gı
`
ra
`
ca
`
nige
´
ke
`
-y
dog Canige GEN-3MSG
‘dog of Canige’
c. gu
´
rba
´
lo
`
karma ke
`
-ge
`
cloth children GEN-3PL
‘children’s clothes’
d. kolo ye
´
-y ko
`
-ro
`
word mother-3MSG GEN-3FSG
‘news of his mother’
4.4.7 The adjectival modifier
We will return with this item in section 4.6.
4.4.8 Noun phrase coordination
In most African languages, the same morpheme serves as a comitative adpo-
sition (with) and as a noun phrase coordinator with an additive meaning (and).
The sample of languages used in Stassen’s typological study of noun phrase
coordination (Stassen 2000) includes a number of African languages, and
provides interesting insights on their status in a typology of coordination.
4.5 Adpositions
Most African languages have relatively few monomorphemic words acting
exclusively as adpositions, and make a wide use of adposition-like words or
expressions that clearly have a nominal or verbal origin. This pattern is
widespread, for example, in the Nilo-Saharan phylum, where adpositions
derived from nouns frequently serve to specify the search domain for an object
(‘on top of,’ ‘underneath’), whereas non-derived adpositions tend to introduce
other types of adjuncts, for example phrases introducing semantic roles such as
instrument, beneficiary, or recipient.
A feature common to languages that have obligatory applicatives and to
languages that have the type of complex predicates presented in section 4.3.6 is
that, in comparison with other languages, they make only a very limited use of
adpositions, since adpositions typically encode the semantic role of obliques, and
both mechanisms result in giving the status of direct objects to various semantic
types of complements that in other languages tend to be treated as obliques.
23
Adpositions derived from nouns (usually referring to body parts) may take
pronominal markers identical to the possessive affixes attached to nouns in order
to express ‘next to / on top of / behind / ... him/her.’ Similarly, adpositions
derived from verbs may take pronominal affixes reflecting a former mechanism
of object indexation on the verb. For example, in Gidar (Central Chadic), the
Denis Creissels et al.124
dedicated dative and associative prepositions code the gender and number of the
complement noun, which may indicate the verbal origin of both prepositions. An
adposition inflected for gender has been signaled in the South Khoisan language
!Xo
˜
o
˜
, and the Kuliak language Ik has case-inflected adpositions (Ko
¨
nig 2002).
However, on the whole, inflected adpositions seem less widely attested in Africa
than in some Eurasian or American language families.
Some languages, e.g. in the Kwa branch of Niger-Congo (see Ameka 2003
on Ewe), have both prepositions and postpositions, each originating from
different sources.
4.6 The adjective
The languages of the world greatly differ in the way they treat adjectival con-
cepts (i.e. concepts that tend to be expressed through adjectival lexemes in
languages in which the existence and the delimitation of a category “adjective”
are not controversial). In principle, one can imagine languages having a class of
adjectives whose behavior would be equally distinct from that of verbs and from
that of nouns, and some languages (e.g. Japanese, Bambara) stand fairly close to
this ideal type, but in most languages, the grammatical behavior of lexemes
expressing typically adjectival concepts is more or less similar, and sometimes
identical, either to that of nouns, or to that of verbs. See Creissels (2003b)fora
general introduction to the question of adjectives in African languages.
In some languages (e.g. Mupun, West Chadic), even if adject ives behave
like verbs in predicative constructions, they behave differently from verbs in
attributive constructions, and therefore do constitute a separate lexical cat-
egory (Frajzyngier 19 93). A similar situation is found, for example, in Songhay
(Creissels 2003b).
Languages do not necessarily treat typically adjectival concepts in a uniform
way: some languages (e.g. Baule) use both noun-like adjectives and verb-like
adjectives to encode even the most typical adjectival concepts.
Contrary to what is observed in other areas, both languages with a predom-
inance of noun-like adjectives and languages with a predominance of verb-like
adjectives are common all over Africa. A number of African languages have a
category that consists of a very small number of non-derived adjectives
(sometimes less than ten), and that cannot be expanded by derivation from other
categories. For example, Igbo is famous for having a class of adjectives con-
sisting of eight members, semantically four pairs of antonyms. In Africa, this
situation does not seem to be restricted to particular language families or areas.
In many African languages, true adjectival modifiers (i.e. adjectival modi-
fiers distinct from relativized verbs) have very limited possibilities of expan-
sion: they can combine with single words (adverbs or ideophones) expressing
intensity, but cannot be used as the head of adjective phrases similar to those
Africa as a morphosyntactic area 125
encountered in European languages, in which adjectives take full phrases
(“maximal projections”) as their complement (as in a man [proud [of his son]],
or a man [ready [to fight]]). Turkana is however an exception: in this language,
adjectives can be expanded with a prepositional phrase (Dimmendaal 1983b:
334–5); this phenomenon may be more widespread, but the literature on this
topic is rather scanty.
4.7 Adverbs and ideophones
Among the various types of words traditionally called adverbs, languages often
have a productive class of “manner adverbs” mainly derived from adjectives,
or adjectives that can be used as verb modifiers to express the same kind of
meaning. But in many African languages, the possibility of deriving manner
adverbs from other categories, or to use adjectives as verb modifiers, is very
limited. There are however exceptions to this generalizatio n. In particular,
some Chadic languages have a large class of modifiers that can serve as both
adjectives and adverbs, and a common feature of many Chadic languages is the
derivation of adverbs through the reduplication of verbs, nouns, and even
numerals (e.g. in Mina, Gidar, Central Chadic).
In languages that have only limited possibilities of deriving manner adverbs
from adjectives or to use adjectives as manner adverbs, the meaning commonly
conveyed by manner adverbs in other languages may be expressed either by
nouns in oblique function, or by cognate object constructions (e.g. the
equivalent of He walked slowly may be something like He walked with
slowness or He walked a slow walk). But there is also considerable overlap
between the role fulfilled by manner adverbs in European languages and that
fulfilled in African languages by ideophones.
In descriptions of African languages, ideophones are often recognized as a
category of words that does not correspond to any of the categories trad-
itionally recognized in descriptions of European languages. But it may well be
that this is a matter of descriptive traditi on, rather than a characteristic feature
of African languages. Ideophones are generally described as having not only
phonological properties that set them apart from other categories (which is
commonly viewed as the manifestation of a semantic feature of expressivity),
but also distributional characteristics that justify analyzing them as constitut-
ing a distinct syntactic category.
There are however important crosslinguistic variations in the distribution of
ideophones:
(a) It may happen that the occurrence of ideophones is conditioned by indi-
vidual verbs or adjectives, each ideophone combining only with a par-
ticular verb or adjective (or with a very limited set of verbs/adjectives).
Denis Creissels et al.126
(b) It may happen that all the ideophones combine with a verb meaning ‘be,’
‘do,’ or ‘say.’ In such combinations, the verb in question loses the meaning
and the subcategorization properties it has when it functions by itself as the
predicative cent er of a clause, and the meaning as well as the subcat-
egorization properties of the combination be/do/say þ ideophone are
entirely determined by the ideophone.
(c) It may happen that ideophones occur in verbless clauses in which they
determine the semantic roles assigned to the noun phrases with which they
combine.
The use of ‘say’ as a light verb introducing ideophones is particularly
widespread in Nilo-Saharan and Afroasiatic languages ranging from Ethiopia
to Lake Chad, but the same phenomenon is sporadically attested in other areas,
e.g. in Wolof, or in Tswana (Creissels 2001; 2003a; 2003b ).
For a recent survey, see the studies in Voeltz and Kilian-Hatz (2001).
4.8 Word-order typology
Among the logically possibl e clause-constituent orders, the most commonly
used in the languages of the world with the status of basic constituent order are
those in which the subject precedes both the verb and the object, i.e. SOV and
SVO. Constituent orders with the verb in first position (VSO and VOS) are less
commonly used as basic constituent order, and languages having a basic
constituent order with the object in first position are very rare.
The family in Africa with the most important variations in constituent order
seems to be Kordofanian. Some Kordofanian languages are recorded with
SVO, some with SOV, at least one with VSO, and at least one (Lafofa) with
OSV in a series of tense–aspect forms (Thilo C. Schadeberg, p.c.). This is an
area that deserves more investigation.
African languages confirm the strong predo minance of subject-initial orders
(SOV and SVO), and the proportion of verb-initial Afri can languages is
roughly comparable to that observed at world level. But in some respects,
clause-constituent order is a domain in which the diversity observed at the
level of the African continent differs from that observed at world level.
The proportion of African languages with a particularly rigid clause-con-
stituent order is relatively high. Sandawe, a poorly documented language spoken
in Tanzania, has been reported to have a relatively flexible clause-constituent
order (see Dalgish 1979), but such cases are exceptional in Africa, and none of
the well-documented African languages exhibits a “free” clause-constituent
order (i.e. a constituent order pragmatically rather than syntactically determined)
of the type encountered in Russian, Hungarian, or in some Australian languages.
Syntactically conditioned variations in constituent order are not rare among
African languages. Most of the time, they are trigered by the TAM value of the
Africa as a morphosyntactic area 127
verb or by negation. For example, a few Central Chadic languages (e.g. Hona,
Ga’anda) have a different word order in the perfective (verb-initial, VSO) and
in the imperfective (SVO), and several Kwa languages have a variation
between SVO and SOV with a similar conditioning. For variations in the
const ituent order trigge red by negation, see section 4.12 .
The proportion of subject-initial languages (SOV or SVO) taken as a whole is
roughly comparable to that observed at world level, but the proportion of lan-
guages with a basic SVO order is considerably higher (and the proportion of
those with a basic SOV order considerably lower) among the languages of Africa
than at world level, at least from a strictly numerical point of view (genetically
balanced samples show a higher proportion of verb-final languages).
Heine (1976) argues that a strict dichotomy between an SVO and an SOV
type cannot be held in African linguistics, and that a satisfying account of the
patterns of word-order variation observed in the African languages requires
recognizing four main types, in the definition of which the position occupied
by the genitive modifier and by noun phrases in oblique functi on is more
important than that of the object.
Heine’s type A corr esponds to what is often considered as the “consistent”
SVO type. Languages of this type have a basic SVOX clause-constituent order,
prepositions and, within the noun phrase, all of the modifiers (including the
genitival modifier) follow the head noun. In Africa, this type is found in all
phyla, but it is particularly predominant in Niger-Congo. In particular, virtu-
ally all Atlantic and Benue-Congo languages belong to it.
A minor type , viewed as a subtype of type A, differs from type A proper in
the position of the adjectival modifier: in this subtype, the adjectival modifier
precedes the head noun (but all other kinds of modifiers follow it, as in type A
proper). This subtype is found in a geographically defineable area including
mostly Adamawa-Ubangian languages, but also som e Benue-Congo and
Chadic languages (in particular Hausa).
Type B is defined by the following characteristics: languages of this type have
postpositions; within the noun phrase, the genitival modifier precedes the head
noun, but all other kinds of modifiers follow it; at clause level, type B languages
may have an SVOX order, or an SOVX order, or both. In other words, this type
groups together SVO languages that differ from the “consistent” SVO type in
putting the genitival modifier before the noun it modifies, and SOV languages
that differ from the “consistent” SOV type by putting the obliques after the verb.
The reason for grouping together these two sets of languages and for considering
as irrelevant the position of the object is that
(a) in all languages that have both constituent orders SVOX and SOVX in
complementary distribution, the genitival modifier precedes the noun it
modifies, which makes it possible to explain the emergence of an
Denis Creissels et al.128
alternative SOVX order in languages that originally have a rigid SVOX
order;
24
(b) languages with SOVX as the only possible clause constituent order share
many more properties with SVO languages than with “true” SOV lan-
guages (i.e. consistently verb-final languages): in consistent SOV lan-
guages, heads systematically follow their modifiers, whereas in type B
languages, most constructions conform to the head þ modifier order:
attributive adjectives follow the noun, auxiliaries precede the main verb,
complementizers precede the complement clause, etc.
Type B is widespread in Africa, but relatively uncommon outside the con-
tinent. It is mainly located in West Africa, in a geographically compact area
including the languages of the Kru, Kwa, Gur, and Mande branches of Niger-
Congo and Songhay, but it occurs also sporadically in others parts of the
African continent. Mande languages, with their rigid SOVX clause-constituent
order, can be viewed as its most typical representatives.
Heine’s Type C corresponds to the type known in the literature as VSO. It
differs from type A in the position of the subject only. It shares with it the use of
prepositions and the position of the noun modifiers (including the genitival
modifier) after the noun they modify. Apart from Ancient Egyptian and a few
Chadic languages, uncontroversial examples of this type are largely confined
to the Eastern Sudanic group of Nilo-Saharan. It has been reported to exist also
in the Berber and Semitic branches of Afroasiatic, but this question is some-
what controversial. The point is that there may often be hesitations in identi-
fying the basic constituent order of individual languages as SVO or VSO, due
to the fact that languages that have a basic VSO constituent order always have
at least SVO as a possible alternative.
25
Reh (1983) describes an exceptional
case of a verb-initial language with postpositions.
Type D is mainly found in East Africa, and is particularly uncommon among
Niger-Congo languages (Ijo is the only Niger-Congo language belonging to
this type). This language type is further discussed elsewhere in this volume
(chapter 9).
Note that this typology was elaborated prior to the reanalysis of some
Western Nilotic languages as having a coding of syntactic roles of the ergative
type (see section 4.2.2 above), and therefore does not take into account recent
analyses according to which these languages have a basic OVA order in
transitive clauses, and SV in intransitive clauses.
The patterns of word-order variation presented above call also for the
following remarks:
(a) In a number of African languages with an SVOX or SOVX clause-
constituent order, the genitival modifier is the only type of modifier that
Africa as a morphosyntactic area 129
precedes the noun, whereas at world level, the tendency of the genitival
modifier towards preceding the noun is not particularly strong in com-
parison with other types of modifiers such as numerals or demonstratives.
(b) In some other languages with an SVOX or SOVX clause-constituent
order, the adjectival modifier is the only type of modifier that precedes
the noun, which contradicts the fact that in the languages of the world,
the adjectival modifier is among the types of modifiers that exhibit a
marked tendency towards following the noun they modify.
As regards adpositions, by and large, African data confirm that languages in
which the genitival modifier precedes the noun generally tend to have post-
positions, whereas those in which the genitival modifier follows the noun tend
to have prepositions. The obvious explanation for this correlation is that,
historically, the reanalysis of genitive þ noun constructions is universally an
important source of noun phrase þ adposition constructions. However, the
reanalysis of object þ verb constructions is another widely attested source, and
this may explain the development of both postpositions and prepositions, for
example, in Kwa languages in which the genitive precedes the noun it modi-
fies, but the object follows the verb; see in particular Heine, Claudi, and
Hu
¨
nnemeyer (1991) and Ameka (2003) for Ewe.
4.9 Non-verbal predications and copulas
Non-verbal predications are commonly encountered in clauses expressing
identification, existence, location, or attribution of qualities, but are often
restricted to clauses expressing a TAM value identical to that expressed in
verbal predication by the verb tenses commonly labeled indicative present.
Uncontroversial examples of non-verbal predications are those involving mere
juxtaposition of non-verbal words or constituents devoid of any predicative
marking (noun phrases, adposition phrases, adverbs), as in the following
example from Kanuri:
(37) a. bi
ntu fe
´
ro
Bintu girl
‘Bintu is a girl’
b. mu
´
sa ka
´
no-lan
Musa Kano-LOC
‘Musa is in Kano’
c. nyi
ku
´
ra
2SG big
‘You are big’
Denis Creissels et al.130
This type exists also, for example, in the equational clauses of some Chadic
languages, but on the whole, it is not particularly frequent in Africa.
Copulas with clearly verbal properties are also attested – for example, in
Baule (Kwa), in the Khoisan language !Xun, and in the Kuliak language Ik –
but this is not a very common situation either.
By contrast, in clauses expressing identificat ion, existence, location, or
attribution of qualities, constructions involving predicative words with very
few verbal prope rties, or even entirely devoid of morphological prope rties that
would identify them as verb forms, are particularly common all over Africa.
Certain descriptive traditions tend to analyze such predicative words as
irregular verbs, but labels such as “non-verbal predicative” or “non-verbal
copula” are also current in descriptions of African languages. For example, b
‘be located at’ is analyzed as a non-verbal predicative in most recent
descriptions of Bambara; this analysis accou nts for its limited combinability
with TAM morphemes , but its synt actic distribution is clearly that of a verb,
and it combines with a past marker; consequently, one may prefer to view it as
an irregular verb. Historically, there is in many cases comparative evidence
that such predicatives origina te in decategorialized verbs, but this is not their
only possible source (in particular, they may originate also in demonstratives).
Frajzyngier, Krech, and Mirzayan (2002) propose a correlation between the
occurrence of the copula in equational clauses and the form of the modifying
construction involving two nouns.
Some African languages also have what may be analyzed as a predicative
inflection of nouns. For example, in many Bantu languages, a prefix identical
to the verbal prefix expressing agre ement with the subject in positive tenses of
the indicative is attached to nouns in predicate function. However, in Bantu
languages, the predicative inflection of nouns is always limited to person–
number(–gender), and possibly negation. A similar phenomenon occurs in the
Nilotic language Luo, where it may have originated as part of a more general
typological shift towards a Bantu-type of language with noun classes and tense
marking o n verbs (Dimmendaal 2000). Compa re the parallel inflection of verbs
and predicative nouns in Luo – data from Tucker (1994):
(38) a. a
´
nra
´
wra
1SG boy
‘I am the boy (e.g. they were looking for)’
b. a-ra
´
wra
S1SG-boy
‘I am a boy’
c. a-˛e
´
yo dho
´
-lu
´
o
ˆ
S1SG-speak mouth-Luo
‘I speak/know Luo’
Africa as a morphosyntactic area 131