Heine Bernd, Nurse Derek. A Linguistic Geography of Africa

Подождите немного. Документ загружается.

this second slot Kimbu and Nilyamb a have Far Past #al

which might be a

reduced form of the Sukuma/Nyamwezi -al

‘past þ be.’ The Nyaturu evi-

dence for this shape is shaky; in one Nyaturu dialect it is said to carry the same

function as (low-toned) na

a Near Future in the main dialect. But it is high-

toned na

a in this second dialect which means Far Past: maybe this is a mistake

in the data? Furthermore this ali¼ari is not unique to our languages, being

widespread in the meaning Far Past in many eastern Bantu languages, as can be

seen in the examples in (7) and (8), taken from languages adjacent and to the

southwest of our target languages.

Similarly with other parts of the preverbal complex. Nyaturu (a

ja) and

Nilyamba (aza) have a cognate preverbal Near Past, which probably derives

from “pastþcome,” a common enough form of gramm aticalization, but not

attested in Kimbu or Sukuma/Nyamwezi. Finally, Nilyamba and Kimbu have a

Persistive morpheme -kyal i-, which can plausibly be analyzed as -k

þpastþbe.

While this -k

- might relate to the k



“persistive” of Nya turu, it should be noted

that Nyaturu -k

comes in the ﬁnal preverbal slot, not the initial one, and that

this k

(

) is not attested in Sukuma/Nyamwezi.

All this detail can be summarized thus. We would like to see a solid set of

categories and morphology in Nilyamba and Kimbu similar to those that occur

in the Nyaturu preverbal complex. While there are minor similarities, and

maybe some shared innovations, the evidence for such a set is simply not there,

and its absence is unlikely to be due to lacunae in the data.

On the other hand, in order to prove the preverbal clitic cluster to be an

internal development, it would also be necessary to demonstrate reas onably

that the Nyaturu preverbal complex is a reduc tion of structures inherited from

the earlier link with Sukuma/Nyamwezi. It should be possible to show that

Sukuma/Nyamwezi has the auxiliary structures that could have been fused to

produce something like what we ﬁnd in the preverbal complex in Nya turu.

Sukuma/Nyamwezi is very rich in auxiliary structures. We have counted

several dozen structures, most involving two verbs, a few involving three.

Further, as suggested in (5) and (10), Sukuma/Nyamwezi does fuse auxiliary

and main verbs. But with a few exceptions, all Sukuma/Nyamwezi auxiliaries

are forms of ‘be’ (either –l

or -



iiza). They do not resemble the morphemes

exempliﬁed in (4). And considering the reduction processes current in

Sukuma/Nyamwezi, indeed, considering any reduction processes, it is hard to

imagine how most of the components of the Nyaturu preverbal complex might

derive from anything in contemporary or older Sukuma/Nyamwezi.

The evidence does not support the hypothesis that the preverbal complex can

be directly derived from inherited structures. It suggests strongly the adoption

of the basi c preverbal structure, or at least large parts of it, from a neighboring

West Rift language. The structure itself can be interpreted as a transfer, the

constituent morphemes are a mixture of inherited and transferred items, the

Roland Kießling et al.202

grammatical categories are probably inherited. In other words, we have an

alien structure expressing inherited categories and using a mixture of trans-

ferred and inherited morphem es of sometimes irregular shape. Earlier Nyaturu

seems to have taken an alien structure to express familiar categories, using a

mixture of inherited and transferred morphemes. This suggests two possible

scenarios. Either Nyaturu, or one form of Nyaturu, was once used by a group of

bilingual West Rift speakers, or the Nyaturu, or a section of the Nyaturu, were

once bilingual in West Rift or one of its predecessors.

Gisamjanga Datooga (which might be representative of Datooga in general)

also has a preverbal clitic cluster, comprising four proclitic structural slots in

front of the verb, as shown in table 6.4 (columns 1–4). Again, we ﬁnd the

categories of sequential, future, persistive, and subordination among those

marked by proclitics.

The Gisamjanga Datooga examples in (12) and (13) illus trate the separ-

ability from the verb of the future and the sequential proclitic.

(12) Preverbal clitic cluster separated from verb by adverb (relative clause)

qwa



Ilo

oga

oha

at he

eda



I˛a

S3.erect cattle.enclosure S3Sg:increase place FUT.REL evening

gwa

lla

oga

S3.sleep.at goats

‘They built a cattle enclosure to increase the room for the goats and sheep to

sleep at night’ (Berger 1935/36)

(13) Preverbal clitic cluster separated from verb by adverb (relative clause)

k-a



Iis



Igu

urs-a

orje

eda

a-Ni

SEQ:AFF-FUT every person call:APPL-3 son-POSS.3SG

‘then everyone will cal l his son’ (Berger 1935/36)

Hadza shows a similar clustering of preverbal clitics for tense and subject

person in the sequential (Berger 1943; Wagner 1988; Bonny Sands, p.c.). Their

clitic status is manifest in (14), where the cluster is separated from the verb by

direct objects.

(14) Hadza: preverbal clitic complex separated from the verb by a direct object

yamo ts’okwana

ko el





ata ...

SEQ.3MSG giraffe create

‘he also created the giraffe ...’ (Berger 1943: 102f.)

yamo se

seme e

la¯ ...

SEQ.3MSG lion create

‘The lion was also created ...’ (Berger 1943: 103)

The Tanzanian Rift Valley area 203

Table 6.4 Preverbal clitic cluster and ﬁnite verb in Gisamjanga Datooga

Subordination Polarity Tense 3 Tense 2 Tense 1 Subject Verb stem Object

temporal relative G afﬁrmative- indicative

Ø unmarked Ø unmarked Ø unmarked aa-/da- 1SG -aan 1SG

g sequential m negative- indicative a

future gu

persistive nı

- perfect afﬁrmative ı

- 2SG -i

i  -ey 2SG

i conditional a

negative- subjunctive i- present negative (G)a- / Ø 3SG Ø 3SG

e-/si- 1PL -eesa 1PL

oo- 2PL -eegwa 2PL

e-/si- 3PL Ø 3PL

This morphophoneme indicates the alternation of the phonemes g and q.

In Sandawe, the movable person–gender–number marker (Elderkin 1989:

25ff.) bears some similarity to the preverbal clitic cluster in the other languages.

It can move from the end of the verb position to the end of the constituent

preceding the verb (object, adjunct) and as well to constituents preceding that

constituent, compare (15a and b).

(15) a. na

mso

mba

thı

-s



Nam ﬁsh cook-3FSG.IRR

‘She will cook Nam’s ﬁsh’ (Eaton 2002:92)

b. ijo:

jini˛-sa ka:

mother corn-3FSG plant

‘The mother planted corn’ (Kagaya 1990: 2, 4 quoted in

Eaton 2002: 92)

The function of the position of the person–gender–number marker is prag-

matic (focus). Thus the preverbal clitic cluster has the following characteristics

in common with West Rift and Nyatur u: (i) there is inﬂection including subject

marking separate from the verb, (ii) cliticization is to the left; (iii) the preverbal

clitic cluster has focus function (not in Nyaturu).

The differences with the preverbal clitic cluster in the other languages are

that the person–gender–number marker in Sandawe does not express tense–

aspect nor does it receive tense–aspect clitics, and that the Sandawe pers on–

gender–number marker does not have a ﬁxed position. In the Khoe languages

the person–gender–number markers (which seem cognate) are nominal suf-

ﬁxes which can be attac hed to verbs in order to nominalize them (Gu

ldemann

& Vossen 2000). Sandawe’s option of detaching and preposing the person–

gender–number marker is not too distinct from the behavior of person–gender–

number markers in other Khoisan languages where they are often repeated in

the same sentence. Although the structures of the preverbal clitic cluster in

West Rift and Sandawe are similar to some extent, the differences are

important enough not to assume contact inﬂuence. Both West Rift and San-

dawe inherited the feature, and the expansion of the preverbal clitic cluster in

West Rift can not be considered to be under inﬂuence of Sandawe becau se

Sandawe shows much more limitation in the variety of categories that is

expressed in this preverbal clitic cluster. At most, the structural similarity of

the neighboring languages may have reinforced this feature in both.

A historical interpretation can be given as follows. The preverbal innovation

of Bantu tense categories in Nyaturu with Bantu morphemes on the basis of a

Southern Cushitic syntactic pattern points to a period of massive language

shift, Southern Cushitic speakers shifting to Nyaturu and imposing their syn-

tactic patterns on the Bantu language. Similarly, speakers shifting from West

Rift to Pre-Datooga must have been responsible for the import of West Rift

The Tanzanian Rift Valley area 205

structures into Datooga. This scenario is suggested by the fact that in the

elaboration of preverba l clit ic clusters no markers seem to have been trans-

ferred directly; it is, rather, inherited material that has been reassembled

(sequential and persistive in Nyaturu) or internally motivated developments of

grammaticalization from inherited lexemes have been set into motion (future

and persistive in Datooga). The similarity with structures in Sandawe is

coincidence. The similarities with Hadza can only be judged properly once a

full description of Hadza is available.

6.2.4 Verbal plurality (G2)

The category of verbal plurality

in the sense of a “plural stem” is an areal

phenomenon of this contact zone, linking the West Rift languages of Southern

Cushitic (Alagwa, Burunge, Gorwaa, Iraqw) with Southe rn Nilotic (Datooga)

and Sandawe. All these languages display an alternation of singular vs. plural

verbal stem, triggered by the number feature in one of the core arguments of

the predication, and which operates independently of plural agreement or

plural concord as accomplished by inﬂectional morphemes which combine

person, gender, and number marking. Whereas the category of “pluractional”

is so common in Africa that it is not a good indicator for language contact, this

is decidedly different for the feature of “plural stem,” which is quite rare.

In Southern Cushitic, ﬁve verbal derivatio nal sufﬁxes together form an

inﬂectional system to mark the plurality of the subject or the object, depending

on the transitivity of the verb. This situation has been reconstructed for the

Proto-West Rift stage (Kießling 2002a: 315ff.), but within Cushitic as a whole

it seems to be unique. The sufﬁxes in question are the durative *-im, the

continuative *-it, the frequentative that operates an initial reduplication *CV-,

the progressive and the intensitive that operate sufﬁxal reduplications *-VVC

and *-aaC, respectively. They stand in quasi-complementary distribution

whose semantic ration ale has become blurred.

(16) Alagwa: plural stem formation by durative derivation in -im in saapis

‘move over’ vs. saapimis, triggered by plurality of the object

a. ?ana a-na saapis fa¿a

1SG S1/2-PERF move:SG:1SG porridge

‘I have moved over a portion of porridge’

b. handaa? a-na saapis-an fa¿a

1PL S1/2-PERF move:SG-1PL porridge

‘We have moved over a portion of porridge’

c. handaa? a-na saapimis-an fa¿oo

1PL S1/2-PERF buy:PL-1PL porridges

‘We have moved over portions of porridge’

Roland Kießling et al.206

(17) Burunge: plural stem formation by durative derivation in -im in hat

’is

‘ﬁll’ vs. hat

’imis, triggered by plurality of the object

a. ?ana ha hat

’is-a

yakwa

1SG S1/2 ﬁll:SG-1SG:IPERF calabash

‘I ﬁll a calabash’

b. ?ana ha hat

’imis-a

yakwaku

1SG S1/2 ﬁll:PL-1SG:IPF calabashes

‘I ﬁll the calabashes’

(18) Iraqw: plural stem formation by intensitive derivation in -aaC in hats’miis

‘ﬁll’ vs. hats’maamis, triggered by plurality of the object (Mous n.d.: 14)

a. ?anı

˛ at

upito?o

-r hats’miis

1SG S1/2 bottle-F ﬁll:SG:1SG:IPF

‘I am ﬁlling a bottle’

b. ?ate

upa a-na hats’maamı

1SG bottle O:3:FSG:-PAST ﬁll:PL:1SG:IPF

‘I ﬁlled the bottles’

Apart from these derivational strategies of forming plural stems, West Rift

marginally also uses suppletive verb stems to mark this category: gaas ‘kill,’

plural tsu

, gwaa

‘die,’ plural qat

’ kaka

(19) Burunge: alternation of suppletive stems

waa

‘die’ (SG) vs. kaka

(PL),

triggered by plurality of the subject

a. qwalara ya

a gwaa?-i

doctor S3:PAST die:SG-3MSG:PFV

‘The doctor died’

b. qwaleeri ya

a kaka¿ir-i

doctors S3:PAST die:PL-3MSG:PFV

‘The doctors died’

(20) Burunge: alternation of suppletive stems t

’aat

’af ‘cut down, fell’ (SG)

vs. tlaaq (PL), triggered by plurality of the object

a. dandiray ha-gutl’aatl’af-an-a

xa’imo

1PL S1/2-O3MSG cut.down-1PL-IPF tree

‘We cut down a tree’

b. dandiray ha-gitl’aaq-an-a

xa’i

1PL S1/2-O3PL cut.down:PL-1PL-IPF trees

‘We cut down trees’

The Tanzanian Rift Valley area 207

The Gisamjanga dialect of Southern Nilotic Datooga (Kießling 1998a) uses a

sufﬁx -ay(e en),-ac(een), probably cognate with the Proto-Nilotic frequen-

tative marker  -cVn (Reh 1991), for deriving a plural stem. In (21b) the plural

stem is ambigious, it may indicate that the action is repeated several times

while applied to the same object or that it is applied to a variety of differ ent

objects. In (22b), however, the plural stem is obligatory because of the

plurality of the direct nominal object.

(21) Gisamjanga Datooga: alternation of stems

ud ‘pierce’ (SG) vs.

uday

(PL), triggered by plurality of the object

a. na

a-˛u

S1SG:PF-pierce:1SG

‘I have pierced him (once)’

b. na

a-˛d-a

S1SG:PF-pierce-PL:1SG

‘I have pierced him (several times)’  ‘I have pierced them’

(22) Gisamjanga Datooga: alternation of stems lood ‘pull out’ (SG) v s. loot

(PL), triggered by plurality of the object

a. lo

oda

˛u

pull.out:IMP.SG spear

‘Pull out the spear!’

b. lo

ota

˛u

tka

aka

pull.out:PL:IMP.SG spears

‘Pull out the spears!’

In Gisamjanga Datooga the plural stem also has a holistic idea, as illustrated by

the contrast of rad ‘tie (one body part)’ and raday  rat ‘tie (the whole body)’:

(23) Gisamjanga Datooga: alternation of stems rad ‘tie’ (SG) vs. raday 

rat (PL)

a. na

a-ra

S1SG:PF-tie:1SG

‘I have tied him (partially, e.g. one of his arms)’

b. na

a-ra

d-a

S1SG:PF-tie-PL:1SG

‘I have tied him (entirely, i.e. all of his body)’

These semantic specializations of a Nilotic verbal derivational sufﬁx (Reh

1991) have not been reconstructed for the Proto-Nilotic period and must be

seen as a Southern Nilotic, probably Datoogan innovation.

Roland Kießling et al.208

Sandawe also ha s the c ategory of the plural verb stem (Kießling 2002b).

Actually this language lacks any me an s of nom inal number marking, plur-

ality of the core participants of a predicati on being exclusively marked on the

verb, either by deriving a plural stem by the plural stem sufﬁx -waa or by

replacing the singular stem by a suppletive plural stem. The morphosyntactic

conditioning of these plural stems reﬂects an ergative-like pattern in that

intransitive verbs are marked for the pl urality of the subject, whereas tran-

sitive verbs tend to be marked for the plural ity of the patient object, not that of

the subject. Besides this rather inﬂectional type of plural marking, there is

another more derivational type which coul d be labelled “plur ac tion al ,” fol-

lowing Newman (1980, 1990), i n that it indicates a plurality of action, mostly

repetition.

(24) Sandawe: plural stem formation by -waa in

o[we

] ‘milk’ (SG) vs.

o-wa

a (PL), triggered by the plurality of the direct object

a. ?u

mbu

-s ˛ko

o-we

cow-S1SG:PERF milk-SG

‘I have milke d a cow’

b. ?u

mbu

-s ˛ko

o-wa

cow-S1SG:PERF milk-PL

‘I have milke d cows’

(25) Sandawe: plural stem formation by -waa in t

’a

w[e

] ‘uproot’ (SG) vs.

’a

-wa

a (PL), triggered by the plurality of the direct object

a. ˛!

-s tl’a

w-e

root-S1SG:PERF uproot-SG

‘I have torn out a root’

b. ˛!

i-s tl’a

-wa

root-S1SG:PERF uproot-PL

‘I have torn out roots’

(26) Sandawe: plural stem formation by -waa in ma

‘eat’ (SG ) vs. ma

-wa

a (PL), triggered by the plural ity of the direct object

a. t

dı

-s ma

-a

I egg-S1SG:PERF eat-SG

‘I have eaten an egg’

b. t

dı

-s ma

-wa

I egg-S1SG:PERF eat-PL

‘I have eaten eggs’

The Tanzanian Rift Valley area 209

(27) Sandawe: alternation of suppletive verb stems t

as ‘die’ (SG) vs.

ate

(PL) ‘die’, triggered by the plurality of the subject

a. ˛je

etla

as-a

man die:SG-S3MSG:PERF

‘A man has died’

b. ˛jo

o la

ate

[ ...la

at-a

]

people die:PL [ ...die:PL-S3PL:PERF]

‘People have died’

It is not clear whether the plural stem category in Sandawe is to be regarded

as a Khoisan inheritance. Suppletive plural stems are found in Northern

Khoisan (Bernd Heine and Christa Ko

nig, p.c.) and in Southern Khoisan. The

situation in Hadza is not clear.

Regarding the historical interpretat ion, it seems that pre-Sandawe is the

source of the spread of the feature of verbal plurality to the West Rift Southern

Cushitic languages and to Southern Nilotic Datooga. In both language groups,

West Rift Southern Cushitic and Southern Nilotic, there is no speciﬁc need for

it, since a differentiation of number in the nominal arguments of the verb is

already taken care of by the complex morphological apparatus of nominal

plural and singulative deri vation. Since this is different in Sandawe, which

lacks a number distinction in nouns, it must be suspected that the strategy of

head marking the number of verbal arguments on the verb itself was spread

together with its principal conditioning factors originating from a Sandawe

predecessor to the neighboring languages of the area. In this context it should

also be noted that the Sandawe iterative marker -im is very close – formally as

well as semantically – to the common West Rift Southern Cushitic verbal

extension for the durative *-im (Kießling 2002a: 296), as it is in its modern

reﬂexes in Alagwa (16) and Burunge (17).

The descriptions of the Bantu languages of the area occasionally mention the

common Bantu habitual or pluractional marker -aga,-anga, though it is not

clear if it takes over functions comparable to the plural stem. This needs further

investigation.

6.2.5 The category of ventive marking (G3)

The presence of the category of ventive marking entails that there is a marker

(verbal afﬁx or clitic) for indicating that the event or action is directed towards

a pre-established deictic center.

Datooga has an opposition of a centrifugal (andative, itive) vs. centripetal

(ventive) markers in verbal derivation (Rottland 1982: 184) as part of its

Nilotic heritage (Rottland 1983; Dimmendaal 1981; Reh 1996).

Roland Kießling et al.210

Although paralleled in Somali (Tosco 1997: 96; Saeed 1999: 126; Bourdin

2006), this kind of directional marking on the verb does not seem to be typical

of Cushitic languages. All modern West Rift languages, though, come up with

a verbal proclitic ventive marker which is reconstructed as *ni for PWR

(Kießling 2002a: 368). It is suspected that the development of this verbal

category is a structural transfer from a Southern Nilotic source. The long form

of the ventive sufﬁx in Southern Nilotic contains an n (-aan or -n). But this

sufﬁx was well integrated in lexical verbal morphology. We do not propose a

direct borrowing of form and function. At most the similarity in form (the

presence of n) may have helped the emergence of a independent ventive

marker in West Rift. A possible source of the proclitic ventive marker is the

ﬁrst-person singular pronoun ani. Sandawe has a ventive morpheme that is

almost identical in form, n

; see (28). There is no comparative evidence to show

that this sufﬁx is inherited in Sandawe. Khoe languages have derivational

sufﬁxes that express direction (Vossen 1997a: 354) but no obvious cognates

with ventive n

. One option is transfer from Sandawe into West Rift, but if the

concept was indeed a result of Datooga inﬂuence as we suggested above, the

transfer of the form from Sandawe to West Rift at the times when Datooga was

inﬂuential is unlikely. Transfer from West Rift to Sandawe is also unlikely in

view of the difference in position: verbal sufﬁx in Sandawe, preverbal clitic in

West Rift. The similarity in form and function is best regarded as chance.

(28) tha

-n

-ko

‘run to this place’ (Steeman 2003)

tha

-ko ‘run away’

The marker ni in the preverbal clitic complex in Nyaturu, see table 6.3 above,

is similar in form but different in function. At most, one could suggest that

the frequent use of proclitics in n in Cushitic facilitated the use of a Bantu

copula ni as a proclitic in the preverbal clitic complex in the Nyaturu speech

of former Cushitic speakers.

There is no information on Hadza for this feature. The concept is an early

Southern Nilotic inﬂuence on West Rift. In West Rift the ﬁrst-person singular

pronoun has been remolded into this function; while the function itself was taken

from Southern Nilotic. This suggests Southern Nilotic speakers shifting to West

Rift. Sandawe inherited or acquired the ventive sufﬁx n

independently.

6.2.6 Two or more past tenses (G5); one or more future tenses (G6)

Tense is typically a highly differentiated category in Bantu languages in

general. Meeussen’s (1967: 109) reconstruction of Proto-Bantu includes two

morphemes with past reference, a- (recent or hodiernal) vs. a

- (hesternal),

but at the same time he observes that both markers are usually embedded in a

The Tanzanian Rift Valley area 211