Heine Bernd, Nurse Derek. A Linguistic Geography of Africa
Подождите немного. Документ загружается.
languages), but we know of no Niger-Congo language whose case marking or
indexation system would justify the introduction of a grammatical relation
“indirect object” or “dative” similar to that traditionally recognized in
descriptions of European languages. In most language families included in the
Niger-Congo phylum, especially among Bantu and Atlantic languages, type
IIb (“double object constructions”) is predominant .
The situation is different in other phyla, particularly among Afroasiatic
languages. Double object constructions exist for example in Somali (Saeed
1987: 8), but among Afroasiatic languages, constructions of type I (including
the variant Ib, in which the recognition of a syntactic role “dative” is fully
justified) are also very common.
In Hausa (see (13) above), the dative is introduced by a special preposition
and precedes the direct object (if there is one), and a separate set of dative
pronouns can be recognized. More generally, Chadic languages have a
grammatical role “dative” both distinct from direct object and obliques, with
variations in the way individual Chadic languages distinguish it from other
grammatical roles (use of specialized adpositions, distinct sets of dative pro-
nouns, or other devices).
Berber languages have indexation systems including special dative markers
(see (15)).
Omotic languages tend to use a separate case distinction for the argument
expressing a recipient, benefactive or malefactive. A specialized dative case
exists also in some Cushitic languages (e.g. Oromo).
Several non-Niger-Congo languages have an intere sting type of mixed
construction belonging to type I as regards case marking, but to type II as
regards indexation: object markers identical to thos e representing the patient of
prototypical transitive verbs, when attached to verbs with an argument frame
similar to that of ‘give,’ exclusively refer to the recipient, but the noun phrase
representing the recipient receives a distinct dative (or dative- allative) case
marking. This situation is found in the Saharan language Kanuri, in the Ethio-
Semitic language Amharic, and in the Chadic language East Dangla (Shay
1999). In such systems, there is no problem in recognizing a syntactic role
“dative,” since case marking ensures the distinction between dative and direct
object, and indexation ensures the distinction between dative and obliques.
4.3 The verb
4.3.1 Presence vs. absence of a verbal inflection, and the use of
grammatical words expressing the same types of distinctions as
verbal inflection
In most languages, at world level as well as at the level of the African con-
tinent, the verb has an inflection considerably more complex than that of any
Denis Creissels et al.102
other category. The verb may have a rich inflection system even in languages in
which nominal inflection proper does not exist, in the sense that all of the
grammatical morphemes operating at noun-phrase level have the status of
phrasal affixes (or “clitics,” in more traditional terminology) rather than that of
affixes of the head noun. This situation is common for example among Kwa
languages.
Languages really devoid of verbal inflection are very rare in Africa, but the
available documentation on African languages may be misleading, since in
many descriptions of West African languages, as already mentioned in con-
nection with subject markers, verb prefixes are wrongly identified as free
morphemes, with the result that languages with an entirely prefixal verb
inflection (which is a fairly common situation among West African languages)
are wrongly presented as languages devoid of verbal inflection.
An interesting case is that of languages in which the word constituted by the
lexical verb and its inflectional affixes cannot be used alone (or only in very
restricted conditions) and must be accompanied by a grammatical word
expressing semantic distinctions that most languages tend to encode through
verb morphology (TAM and polarity distinctions, argument indexation). Such
languages are characterized by a complementarity (or sometimes redundancy)
between the distinctions expressed at the level of this grammatical word
(“predicative marker,” or “selector”) and those expressed at the level of the
lexical verb. In the borde rline case, illustrated by Zarma, all grammaticalized
TAM distinct ions appear at the level of the “predicative marker,” and the
lexical verb has no inflec tion proper.
In Africa, this type of organization of verbal predication is found mainly in
two groups of languages:
In West Africa, it is common among languages that have a constituent
order SOVX (such languages are found in a geographically compact area
including the languages of the Kru, Kwa, Gur , and Mande branches of
Niger-Congo, and Songhay);
In northeast Africa, it is common in some branches of Cushitic (see chapter 7).
The following example is from the Songhay language Zarma (Oumarou
Yaro 1993):
(18) a. muusaa na feejoo wii
Moussa PERF:POS sheep:DEF kill
‘Moussa killed the sheep’
b. muusaa mana feejoo wii
Moussa PF:NEG sheep:DEF kill
‘Moussa did not kill the sheep’
Africa as a morphosyntactic area 103
c. muusaa
g
a feejoo wii
Moussa IPF:POS sheep:DEF kill
‘Moussa will kill the sheep’
d. muusaa si feejoo wii
Moussa IPF:NEG sheep:DEF kill
‘Moussa will not kill the sheep’
4.3.2 Types of distinctions expressed through verbal inflection
or variations of a grammatical word that obligatorily
accompanies the verb
The meanings most commonly encoded through verbal inflection are argument
indexation, TAM distinctions, and negation.
10
Argument indexation has already been dealt with in section 4.2. Verb
inflection systems in which argument indexation is expressed through varia-
tions of the verbal word, but TAM distinctions are expressed exclusively
through auxiliaries or adverbs, are very rar e; however, such systems are
attested in Sara lang uages (Central Sudanic).
The systems of TAM distinctions expressed through verbal inflection vary in
such a way that it does not seem possible to define an African type of TAM
system, or even types that would characterize languages spoken in particular
areas or belonging to particular families.
The TAM markers do not necessarily constitute one domain. In particular, in
Chadic languages, modality markers may be coded differently from tense and
aspectual markers. The deontic modality is often coded on the verb, while the
tense and aspectual markers may be coded on the verb or before and/or after the
verb. In some Chadic languages they are morphologically fused with subject
pronouns (e.g. Hausa). In other languages, they may occur in clause-initial
position (before subject pronouns), and in some languages, they may occur
before and after the verb (e.g. in the Central Chadic language Mina).
Systems of verbal inflection expressing negation are particularly common in
Africa. Very often, negation markers and TAM markers fuse together strongly,
and the TAM distinctions expressed by negative verb forms may be different
from those expressed by positive verb forms.
Among semantic distinctions less commonly expressed through verbal
inflection, a remarkable feature of African languages is the relatively high pro-
portion of systems of verbal inflection that directly express distinctions relating
to various types of focus phenomena, or interfere with other focus-marking
devices. Such systems seem to be very rare outside Africa. They are found in a
number of languages belonging to various branches of the Niger-Congo phylum
(Atlantic, Benue-Congo, etc.), and in Cushitic. For example, the Atlantic lan-
guage Wolof uses distinct verbal inflections to express focalization of the verb, of
the subject, or of a term of the construction of the verb other than the subject.
Denis Creissels et al.104
(19) a. gis na yaayam
see TAM:S3SG mother:3SG
‘He saw his mother’
b. moo gis yaayam
FOC:S3SG see mother:3SG
‘He saw his mother’
c. yaayam la gis
mother:3SG FOC:S3SG see
‘He saw his mother’
b. dafa gis yaayam
FOC:S3SG see mother:3SG
‘He saw his mother’
Somali als o has a prominent focus-marking system, as shown, for example,
by Tosco (2003). Main declarative (positive) sentences in Somali receive focus
marking by way of one of the two focus markers, baa and ayaa, which
immediately follow noun phrases. When a noun phrase is not focalized in such
a sentence, a “mood classifier” waa is used. Waa is basically a posi tive
declarative mood classifier, and its verbal-focus effect is ascribed to the
absence of nominal focus marking in sentences in which it occurs; see also
Creissels (1996) on Tswana.
The exact extent of this phenomenon is howe ver difficult to evaluate. This is
not only a question of documentation. It seems common that the same mor-
phological devices are used to foca lize the choice of a particular verbal lexeme
and to put emphasis on the assertive value of declarative sentences. Moreover,
focalization may interfere with operations on valency or with tense/aspect in a
way that sometimes makes it difficul t to identify the precise nature of mor-
phemes involved in the expression of such distinctions. In particular, Gu
¨
lde-
mann (2003b) presents evidence for a relation between verb focus and
progressive aspect in Bantu, and the possibility of a relation between verb
focus and antipassive should perhaps be considered.
11
4.3.3 Auxiliary verbs and compound verb forms
Complex verb forms of the type commonly encountered in European languages
(i.e. those consisting of an auxiliary verb and a dependent or nominalized form of
another verb) are quite common in African languages too, and comparative data
very often suggest that TAM markers that synchronically have the status of
inflectional morphemes originate in ancient auxiliary verbs (future markers
originating in an auxiliarized form of a verb ‘come’ or ‘go’ are particularly
Africa as a morphosyntactic area 105
common in African languages). Auxiliary verbs are also a plausible source of
grammatical words obligatorily accompanying the verb and expressing mean-
ings commonly encoded through verb morphology.
A very general characteristic of Niger-Congo and Nilo-Saharan languages is
that they tend to have auxiliary verbs expressing meanings commonly taken up
by adverbial expressions in European languages, i.e. auxiliary verbs with
meanings such as ‘to do first,’ ‘to do again,’ ‘to do often,’ ‘to have previously
done,’ ‘to have done the day before,’ ‘not to have done yet,’ etc., as illustrated
by the following examples from Tswana.
(20) a. k
Un
Up
O e
´
I-ts
h
Ila
`
U-q
h
a
´
Uv-i
le
`
9:button 9:DEM S3:9-live S3:9-tear-ANT
‘This button is always torn’
b. r
I-a
`
ti
sa
`
v
U-ja
`
ke
´
re
`
ke
´
-˛
S1P-increase INF-go 9:church-LOC
‘We often go to church’
c.
U-ra
´
t-i
le
´
v
U-t
h
u
`
ba
`
pi
tsa
´
ja
`
m
I
S2SG-like-ANT INF-break 9:pot 9:my
‘You nearly broke my pot’
d. pu
´
la
´
I-le
´
ts-
I
I-na
`
9:rain S3:9-spend the night-ANT S3:9-fall (rain)
‘It rained last night ’
e. ts
h
i
m
U e
´
I-(tla
´
a
`
-)ts
Uv-a
´
I-l
I˛wa
´
9:field 9:DEM S3:9-(FUT-)rise S3:9-plough:PSV
‘This field will be ploughed tomorrow’
f. ts
h
i
m
U e
´
I-di
k-i
le
`
I-l
I˛wa
´
9:field 9:DEM S3:9-surround-ANT S3:9-plough:PSV
‘This field was cultivated last year’
Similarly, there are a number of verbs in various Chadic languages whose
function is to code the manner and the temporal characteristics of an event,
such as ‘spend the day doing,’ ‘keep on doing,’ ‘start doing,’ ‘cease, stop
doing,’ ‘finish doing,’ ‘do a lot.’
A typical feature of languages in northeastern and north-central Africa is the
use of compound verb formations that consist of an invariable non-verbal
component determining the lexical meaning of the compound, and of an
inflected form of a verb with the inde pendent lexical meaning ‘say’ or ‘do’
functioning in such compounds as a mere bearer of inflectional marking
(cf. Dimmendaal, this volume chapter 9, for further details). A similar use of a
verb with the meaning ‘say,’ restricted however to combinations with ideo-
phones, is sporadically found in Niger-Congo languages, as in the following
example from Tswana; see Creissels (2003a) for more details.
Denis Creissels et al.106
(21) a.
U-ne
`
a
`
-tsa
´
ja
´
m
´
-mi
di
a
`
-
U-r
I v
Ur
U fa
´
fa
`
ts
h
I
S3:1-AUX S3:1:
SEQ-take
3:maize S3:1:SEQ-
O3:3-say
IDEO PRE on the
ground
‘He then took the maize and poured it out on the ground’
b. no
´
ts
h
I ja
´
-m
U-r
I po
´
mo
´
tse
`
be
´
-˛.
9:bee S3:9:SEQ-
O3:1-say
IDEO PRE ear-LOC
‘The bee stung him on the ear”
4.3.4 Dependent verb forms
African languages commonly have dependent verb forms morphologically
distinct from those that fulfill the predicate function in independent clauses.
Many of these forms have uses broadly similar to those of the non-finite verb
forms encountered in European languages (infinitives, participles, gerundives).
In addition to that, Niger-Congo languages often have special “consecutive”
or “sequential” verb forms that characterize non-initial clauses in sequences of
clauses reflecting a chronological presentation of events, and such dependent
verb forms are also attested in Nilo-Saharan languages. The following example
is from Tswana:
(22) l
Ipo
`
di
si
l
I-ts
h
we
´
r
I ma
´
vo
`
du
`
la
´
-a
`
-tse
´
fia
´
5:policeman S3:5-catch:ANT 6:thief S3:5:SEQ-O3:6- put
dı
`
ha
´
ka
`
bo
´
i
10:handcuff
la
´
-a
`
-i
sa
´
S3:5:SEQ-O3:6-go:CAUS
t
Uro
´
˛ko
´
˛
9:jail:LOC
‘The policeman caught the thieves, put them in handcuffs, and took them
to the jail’
A common clause-chaining strategy of Afroasiatic languages in Ethiopia as
well as of Nilo-Saharan languages to the west of this region (with an extension
into Chad) is the use of so-called converbs in non-final clauses. These verb
forms usually have a reduced system of inflectional morphology. Thus, in
Maale, converbs lack the inflectional features that characterize independent
verb forms (Amha 2001):
(23) ?i
Zi
mi
s’-o
´
tik’-a
´
??o makiin-aa c’aan-e
´
-ne
3MSG:SUBJ tree-ABS cut-CNV car-LOC load-PERF-POS:DECL
‘Having cut the wood he loaded it on a car’
Dependent verb forms such as the converb in this Maale example are “non-
finite” in the sense that they lack some inflectional features characterizing the
Africa as a morphosyntactic area 107
independent verb forms of the same language, but dependent verb forms with
an inflection showing the same distinctions as that of independent verb forms
are attested. This is the case with the sequential form in the Tswana example
above, and Tswana has another set of stric tly dependent but morphologically
finite verb forms, used in temporal subordinate clauses.
12
Many Chadic languages have two tense/aspect systems, where there may
exist two perfective, two imperfective, and two progressive/habitual aspects,
as well as two future tenses. One system codes pragmatically independent
clauses, i.e. clauses that can be interpreted on their own, and the other codes
pragmatically dependent clauses, i.e. cla uses that must be interpreted in con-
nection with another clause. Pragmatically independent clauses are morpho-
logicaly marked because they occur less frequently than pragmatically
dependent clauses. Pragmatically independent clauses typically involve
affirmative clauses, yes/no questions, matri x clauses in complex sentences, and
comments on topicalized constituents. Pragmatically dependent clauses may
involve specific questions (wh-questions), comments on focused elements,
temporal sequential clauses , and relative clauses, i.e. comments on the head of
the relative clause. Some negative clauses may also be coded as pragmatically
dependent.
The use of bare verbs in contexts in which most languages tend to use
morphologically marked dependent verb forms is typically found in languages
commonly identified as “serializing” languages (see section 4.3.6), but also in
Wolof and in many Bantu languages.
4.3.5 Transitivity and morphological coding of operations
on verb valency
Verbs used in transitive constructions may be strictly transitive, if their use in
intransitive constructions is restricted to elliptical constructions, but they may
also have non-elliptical intransitive uses, with two possibilities: the subject of
an ambitransitive verb used in an intransitive construction may receive the
same semantic role as the subject of the same verb in a transitive construction
(as in English eat: We ate fish vs. We ate slowly), or it may receive a role
identical or similar to that of the object of the transitive construction (as in
English break: He broke the plate vs. The plate broke). Both types of ambi-
transitivity are widely attested in African languages, and some language
families may have marked preferences for one type or the other, but none can
be said to be predominant for Africa as a whole. For example, the Mande
language Bambara represents the extreme case of a language in which virtually
every transitive verb has the behavior illustrated above by break, whereas
the type of ambitransitivity illustrated by eat is almost non-existent in this
language; by contrast, in Tswana, the type of ambitransitivity illustrated by
Denis Creissels et al.108
break is almost non-existent, but transitive verbs generally accept intransitive
uses of the type illustrated by eat.
Many Niger-Congo languages have rich systems of verb affixes encoding
operations on verb valency, i.e. operations that modify the correspondence
between semantic roles and grammatical relations. A particularly rich system
is found in Wolof: Nouguier (2002) identifies ten verbal suffixes productively
used in Wolof to code operations on valency, some of them being used to code
two or more different types of operations. A typical Niger-Congo system of
morphologically marked valency changes includes passive, reciprocal, dec-
ausative,
13
causative, and applicative derivations. In addition to that, reflexive
object markers commonly tend to acquire a variety of uses that cannot be
reduced to the notion of coreference, which may justify reanalyzing them as
middle-voice markers.
The typologically salient feature of Niger-Congo systems are:
the existence of distinct reciprocal and decausative forms (cross-
linguistically, reciprocal and decausative meanings are commonly
expressed by middle-voice forms that by themselves have a more general
and abstract meaning);
a particularly wide use of applicatives and, in particular, of obli gatory
applicatives:
14
in typical Niger-Congo systems, several semantic types of
complements that crosslinguistically tend to be treated as obliques can
occur only as (direct) objects of applicative verb forms, and the semantic
interpretation of an applicative construction entirely depends on the lexical
meaning of the verb and of the noun involved in the construction, since by
itself, the applicative marker gives no clue to the semantic role of the
object it licenses, as in the following example from Tswana:
(24) a. k
I-bi
d-i
tse
´
ba
`
na
´
/k
I-bi
d-i
tse
´
˛a
´
ka
`
S1SG-call-ANT 2:child S1SG-call-ANT 9:doctor
‘I h ave called the children’ / ‘I have called the doctor’
b. k
I-bi
l-e
´
d-i
tse
´
ba
`
na
´
di
d
Z
O
S1SG-call-APPL-ANT 2:child 8/10:meal
‘I h ave called the children to eat’
c. k
I-bi
l-e
´
d-i
tse
´
ba
`
na
´
˛a
´
ka
`
S1SG-call-APPL-ANT 2:child 9:doctor
‘I h ave called the doctor for the children’
In the other phyla, morphological coding of changes in the valency of the
verb is also quite common. In particular, applicatives are widespread all over
Africa (applicatives are found for example in Ik, Amharic, etc.). Their iden-
tification is most of the time straightforward, but in some cases their status
Africa as a morphosyntactic area 109
from the point of view of a general typology of valency-changing devices is
problematic, because they depart more or less from the prototype. For
example, in Ik, the term licensed by the applicative derivation is treated as an
oblique rather than a direct object; similarly, in North Khoisan languages, the
presence of valency-external noun phrases may trigger a morphologic al
marking on the verb, as in typical Niger-Congo obligatory appl icatives, but this
morphological marking may be triggered by the presence of terms that do not
show the characteristics of objects (Gu
¨
ldemann & Vossen 2000). In Tswana,
applicative derivation may change the semantic role of locative complements
of motion verbs without changing the construction from a formal point of view
(see section 4.15), or even simply put into focus a locative complement
(Creissels 2002), etc. In many languages, the distinction between applicative
and causative derivation is not clearcut: some languages, e.g. Wolof (Nouguier
2002), are described as using the same derivatives to code valency operations
of the causative and of the applicative type, and in some others, derived
forms of the verb currently identified as applicatives may have causative-like
uses, e.g. in Kanuri (Hutchinson 1981), and vice versa, e.g. in Tswana
(Creissels 2002).
In some Chadic languages, verbal morphology codes distinctions of “point
of view” and/or “goal orientation.” The representation of an event from the
point of view of the subject invites the listener to consider what effects the
event has on the subject, without indicating any specific effects. The subject in
such a representation is in the scope of the event, which implies that no other
argument (including the object) is there. The representation of an event from
the point of view of the object means that it is the object that is in the scope of
the result of the event. In some Chadic languages (e.g. Hausa), the point of
view of the subject is the default point of view for the majority of transitive
verbs, and putting the object in the scope of the result of an event involves the
use of special morphological markers. Goal orientation in Hausa can be illu-
strated by the distinction between the verb ‘to fall’ (unmarked form) and ‘to
fall onto something’ marked by the goal-oriented marker -a added to the verb.
It appears that in Gidar, the category of goal orientation is in contrast with the
point of view of the subject: the distinction between ‘eat’ and ‘eat something’
is marked by the suffix -a. In the following example from Gidar, the omission
of the marker -a in sentence (b) codes the event from the point of view of the
subject, and the addition of -a in sentence (d) codes object orientation; specific
object coding on the verb, as in sentence (e), implies that the object is known,
determined by deictics, or present in the speech environment:
(25) a. k@
-vr-a
´
-n-k @
fa
´
S2PL-hit-OBJOR-PL-PERF man
‘You hit somebody’
Denis Creissels et al.110
b. k@
-vr@
-n-ka
`
S2PL-hit-PL-PERF
‘You hit’
c. a
`
-kk@
-ka
`
S3M-polish-PERF
‘He polished’
d. a
`
-kk@
-a
´
-n-ka
`
S3M-polish-OBJOR-PL-PERF
‘They polished something’
e. a
`
-kk@
-@
-ka
`
S3M-polish-O3M-PERF
‘He polished it (M)’
It is likely that in some languages morphemes that are not currently ana-
lyzed as valency operators attached to the verb should perhaps be reanalyzed.
For example, Somali grammars mention the existence of four “prepositions”
that “differ from E nglish prepositions in not being next to t he noun they
govern but placed before the verb, regardless of the position of their noun”
(Saeed 1987), which suggests the possibility of reanalyzing these
“prepositions” as obligatory applicative markers. In Amharic, prepositions
have been incorporated into the verb, and thus can be ana lyzed as having
given rise to applicative suffixes (Am berber 2000). However, the analysis of
incorpora ted prepositi ons as app lica tives may not be justified, at least in
some languages; moreover, this possibility of analysis largely depends on the
precise way the definition of “applicative” is formulated, so that we prefer to
leave this question open.
In this connection, antipassive is rarely mentioned in description of African
languages, but this question should perhaps be reconsidered too. Not surpris-
ingly, reanalysis of some Western Nilotic languages as having an ergative
coding of core syntactic roles implies the recognition of an antipassive der-
ivation that was not identified as such in previ ous analyses. Derived forms of
transitive verbs used to avoid the mention of the object of transitive verbs (or of
one of the two objects of ditransitive verbs) are attested even in languages that
show no trace of ergativity in the way they code core syntactic roles, as in the
following example from Wolof (Nouguier 2002).
(26) a. Xaj bii du ma
`
tt-e
dog DEM NEG.S3SG bite-AP
‘This dog does not bite’
b. Alal du jox-e ma
`
qaama
wealth NEG.S3SG give-AP pres tige
‘Wealth does not give prestige’
Africa as a morphosyntactic area 111