Startseite / Archive / 2013 / Valency patterns for bivalent verbs in two West African languages: Mandinka (Mande) and Jóola Banjal (Atlantic)
Artikelaktionen

 

1. Introduction

<1>

In this paper, we compare the valency properties of semantically bivalent verbs in Jóola Banjal, an Atlantic language spoken by approximately 7000 speakers in Casamance (Sénégal), and Mandinka, a Mande language spoken by approximately 1.5 million speakers in Casamance (Sénégal), The Gambia, and Guinea-Bissau. The main reference works on these two languages are Bassène (2007) for Jóola Banjal, and Creissels & Sambou (2013) for Mandinka. The genetic classification of Jóola Banjal and Mandinka can be summarized as follows:

Jóola Banjal

Jóola

Bak

Niger-Congo [1]

Mandinka

Manding

West Mande

Mande

⊂ (?)

Niger-Congo [2]

<2>

The contrastive analysis of Jóola Banjal and Mandinka bivalent verbs proposed in this paper is not based on a pre-established questionnaire. It relies on a list of 300 meanings referring to situations involving two essential participants for which we have found in both languages a bivalent verb encoding the situation in question in a construction whose only additional elements are two noun phrases representing the essential participants.  [3]

<3>

After providing the necessary information about the way verbal predication is organized in Jóola Banjal and Mandinka (Section 2), we discuss the following questions: the coding frames available for bivalent verbs and their relative frequency (Section 3), valency alternations (Section 4), the relationship between transitive verbs and monovalent verbs assigning a role similar to that assigned to the P argument of a transitive verb (Section 5), and mismatches between the morphological structure of verbs and their valency properties (Section 6). Section 7 puts forward some concluding remarks. [4]

2. Verbal predication

2.1. Transitive predication in Jóola Banjal

<4>

Transitive coding is defined with reference to core transitive verbs, i.e. bivalent verbs whose meaning is compatible with the maximum degree of semantic transitivity. A and P are abbreviations for terms of predicative constructions encoded like the agent and the patient of core transitive verbs.

<5>

In the transitive predication of Jóola-Banjal, there is no flagging of either A or P, but in addition to the rigid constituent order AVPX, there is a clear contrast between A and P in indexation. The A argument is obligatorily indexed by means of a verb prefix, and the presence of an NP co-referent with the A index prefixed to the verb is not obligatory – ex. (1), whereas the P argument can be indexed by means of a verb suffix, but P indexes are used only for topical P’s, and if a co-referent NP is present, it can only be in dislocated position – ex. (2).

(1)

Jóola Banjal

a.

Atɛjɔ

na-tɩñ-ɛ

sɩ-nnaŋ

sasʊ.

(cla)Atéjo

cla-eat-cmp

clsi-rice

clsi.def

‘Atéjo ate the rice.’

b.

Na-tɩñ-ɛ

sɩ-nnaŋ

sasʊ.

cla-eat-cmp

clsi-rice

clsi.def

‘(S)he ate the rice.’

c.

*Atɛjɔ tɩñ-ɛ sɩ-nnaŋ sasʊ.

(2)

Jóola Banjal

a.

Nɩ-tɩñ-ɛ

gʊ-mangɔ

gagʊ.

1sg-eat-cmp

clgu-mango

clgu.def

‘I have eaten the mangos.’

b.

Nɩ-tɩñ-ɛ-gɔ.

1sg-eat-cmp-clgu

‘I have eaten them (the mangos).’

c.

Nɩ-tɩñ-ɛ-gɔ*(,)

gʊ-mangɔ

gagʊ.

1sg-eat-cmp-clgu

clgu-mango

clgu.def

‘I have eaten them, the mangos.’

<6>

In addition to vowel harmony, the clearest evidence that P indexes are affixes (and not just pronouns occupying the same postverbal position as P NPs) comes from the fact that, with the inflected form of the verb expressing verb focalization, characterized by the reduplication of the stem, the P index is inserted between the stem and its copy – ex. (3).

(3)

Jóola Banjal

a.

Nɩ-tɩt-tɩñ

fʊ-mangɔ

fafʊ.

1sg-eat-eat

clfi-mango

clfi.def

‘I have EATEN the mango.’

b.

Nɩ-tɩf-fɔ-tɩñ.

1sg-eat-clfi-eat

‘I have EATEN it (the mango).’

<7>

Interestingly, in Jóola Banjal, a similar indexation mechanism is found with spatial phrases, even when they are clearly in adjunct role, as in ex. (4). There is however a clear distinction between the indexation of P arguments and that of spatial adjuncts: P indexes express the noun class of their antecedent, whereas the index for spatial adjuncts is invariably bo ~ bɔ. For example, town names belong to class E, which means that, in ex. (4), if Dakkar were in P role, it would not be represented by the locative index , but by the class E index . [5]

(4)

Jóola Banjal

a.

Nɩ-jʊg-ɩ

Dakkar.

1sg-see-2sg

Dakar

‘I saw you in Dakar.’

b.

Nɩ-jʊg-ɩ-bɔ.

1sg-see-2sg-there

‘I saw you there.’

2.2. Transitive predication in Mandinka

<8>

The transitive construction of Mandinka is characterized by a rigid APVX constituent order. Independent assertive and interrogative transitive clauses always include a predicative marker encoding TAM and polarity (yé ‘completive positive (transitive)’, má ŋ ‘completive negative (transitive)’, ka ́ ‘incompletive positive’, etc.), inserted between A and P. A and P are not flagged and cannot be indexed. Pronouns occupy the same positions as canonical NPs and have the same forms in all their possible functions – ex. (5).

(5)

Mandinka

a.

Kambaan-óo

ye

saa

búsá

fál-óo

la.

boy-def

cmp.pos

snake.def

hit

stick-def

obl [6]

‘The boy hit the snake (with a stick).’

b.

Kew-ó

ka

a

téerímâa

máakóyí

kód-óo

to.

man-def

incmp.pos

3sg

friend

help

money-def

loc

‘The man helps his friend financially.’

c.

Kambaan-óo

maŋ

ber-ôo

fáyí

palantéer-óo

kaŋ.

boy-def

cmp.neg

stone-def

throw

window-def

on

‘The boy did not throw the stone into the window.’

d.

A

máŋ

a

fáyí

a

kaŋ.

3sg

cmp.neg

3sg

throw

3sg

on

‘He/she did not throw it on it.’

2.3. Jóola Banjal and Mandinka as obligatory A coding languages

<9>

In Jóola Banjal and Mandinka, every coding frame, without exception, must include a term with coding properties identical to those of the A term in transitive predication (obligatory indexation in Jóola Banjal, fixed position to the left of the predicative markers in Mandinka). Moreover, most syntactic operations treat the term showing these coding properties in a uniform way, which justifies the recognition of a grammatical relation subject contrasting with a grammatical relation object characterizing the P term of transitive predication.

2.4. The formal distinction between transitive and intransitive predication

<10>

In Jóola Banjal, there is no specific transitivity marking, and the presence vs. absence of a term in object role is the only difference between transitive and intransitive predication. By contrast, in Mandinka, three predicative markers have different forms in transitive and intransitive predication. For example, the completive positive, encoded by the predicative marker ye ́ in transitive predication, is encoded in intransitive predication by the verbal suffix ‑tá – ex. (6).

(6)

Mandinka

a.

Kambaan-óo

ye

ber-ôo

fáyi.

boy-def

cmp.pos

stone-def

throw

‘The boy threw the stone.’

b.

*Kambaan-óo

ber-ôo

fáyí-ta.

boy-def

stone-def

throw-cmp.pos

c.

Kambaan-óo

kumbóo-ta.

boy-def

cry-cmp.pos

‘The boy cried.’

d.

*Kambaan-óo

ye

kumboo. [7]

boy-def

cmp.pos

cry

3. Coding frames for bivalent verbs

3.1. Introductory remarks

<11>

Jóola Banjal and Mandinka have two possible types of coding frames for bivalent verbs: transitive coding, with the two participants encoded like the agent and the patient of core transitive verbs, and extended intransitivecoding, with one of the two participants encoded like the agent of core transitive verbs, the other being encoded as an oblique. As regards the coding frames for bivalent verbs, the main contrast is that the tendency to extend transitive coding to nearly all semantic types of bivalent verbs is much stronger in Jóola Banjal than in Mandinka.

3.2. The extension of transitive coding in Jóola Banjal

<12>

In Jóola Banjal, coding frames of the <A, X> type are regularly found with verbs belonging to one of the following two semantic classes, and only with such verbs:

– movement verbs that assign A coding to the term representing the figure and oblique-like coding to the term representing the source or destination of movement: -kkay ‘go to’, ‑alɔ ‘get down from’, ‑nɔgɛn ‘enter’, ‑jɩŋ ‘climb’, ‑pur ‘go out, leave’, ‑kak ‘step on’, ‑kɐgo ‘lean on’, ‑jɔw ‘go to’, ‑lɔ ‘fall / throw oneself into’.

– verbs encoding naturally reciprocal events; as a rule, such verbs have two synonymous constructions: a construction with two nominal terms, in which the second term is introduced by the preposition ni ‘with’, and a construction with just one nominal term representing a plural entity: ‑tɐj ‘fight against’, ‑yɐbo ‘marry (woman>man)’, [8] ‑ɛmɔr ‘meet’, ‑nogor ‘ressemble’, ‑rɛ ‘be equal’.

We have found just one bivalent verb with a coding frame of the extended intransitive type referring neither to movement nor to a naturally reciprocal event: ‑inen ‘believe in’.

3.3. The extension of transitive coding in Mandinka

<13>

In Mandinka, the same regularity is observed for bivalent verbs encoding movement and bivalent verbs encoding reciprocal events, but in contrast to Jóola Banjal, bivalent verbs with a coding frame of the <A, X> type are not exceptional among the verbs that do not belong to these two semantic classes. However, they are not very numerous, and no semantic generalization seems possible about them.

The list of the Mandinka verbs in our corpus that have a coding frame of the <A, X> type and correspond to a transitive verb in Jóola Banjal is given in (7).

(7)

Mandinka verbs with a coding frame of the <A, X> type
corresponding to transitive verbs in Jóola Banjal

sǒŋ

‘agree on sth.’

háwú

‘bark at’

síláŋ

‘be afraid of’

dalí

‘be familiar with’

járí

‘be worth, cost’

siibóo

‘dream about’

táa

‘fetch’

ñiná

‘forget’

yamfá

‘forgive’

bonó

‘lose’

suulá

‘need’

í jǐi

‘set down (burden), give birth to’

děŋ

‘share in common’

nóorá

‘smell’

suŋkáŋ

‘smell’

suŋkáŋ

‘smell of’

tambí

‘surpass, exceed, overtake’

jutú

‘underestimate’

<14>

Mandinka also has bivalent verbs with two possible constructions, a transitive one and an extended intransitive one, in which they assign the same semantic roles. Those included in our Jóola Banjal / Mandinka contrastive corpus are listed in (8). Their equivalent in Jóola Banjal is in most cases a transitive verb, with however two exceptions.

(8)

Mandinka verbs whose second argument can be encoded either as the object in a
transitive construction or as an oblique in an extended intransitive construction

a.

The equivalent in Jóola Banjal is a transitive verb

wóŋwóŋ

‘bark at’

takí

‘bump’

juŋkú

‘bump’

teyí

‘cross’

kacáa

‘discuss’

wúlúu

‘give birth to’

jélé

‘laugh at’

kumbóo

‘mourn for s.o.’

muñá

‘put up with’

b.

The equivalent in Jóola Banjal has an extended intransitive construction

daŋkeneyáa

‘believe in’

selé

‘climb’

4. Valency alternations

4.1. Valency operations involving changes in the verb stem

<15>

Mandinka has two morphological derivations encoding valency changes: causativization and antipassivization. Ex. (9) illustrates the use of the causative suffix ‑ndí.

Mandinka

(9)

a.

Díndíŋ-o

lá

dendik-ôo

nóo-ta.

child-def

gen

shirt-def

get_dirty-cmp.pos

‘The child’s shirt got dirty.’

b.

Díndíŋ-o

yé

a

lá

dendik-ôo

nó-ndi.

child-def

cmp.pos

3sg

gen

shirt-def

get_dirty-caus

‘The child soiled his shirt.’

<16>

Mandinka has a suffix ‑ri (‑diri in combination with stems ending with a nasal) which can be analyzed as an antipassive marker, since it is found exclusively with transitive verbs in constructions in which the P argument is left unexpressed, cannot be identified to the referent of a noun phrase included in the same construction, and is interpreted as non-specific. However, in other respects, ‑ri has properties quite unusual for an antipassive marker, since with just one exception (dómó ‘eat’), ri-forms cannot be used as the verbal predicate of finite clauses, and are mainly found in an antipassive periphrasis in which the antipassive form of a transitive verb used nominally is the object of ké ‘do’ – ex. (10).

(10)

Mandinka

a.

Mus-óo

ye

maan-óo

tuu.

woman-def

cmp.pos

rice-def

pound

‘The woman pounded the rice.’

c.

Mus-óo

ye

tuu-r-ôo

ké.

woman-def

cmp.pos

pound-antip-def

do

‘The woman pounded.’
lit. ‘The woman did the pounding.ANTIP.’

<17>

Jóola Banjal has a causative suffix (‑ɛn) – ex. (11) – and four valency-decreasing suffixes: passive ‑ɩ – ex. (12), reflexive ‑ɔrɔ – ex. (13), middle ‑ɔ – ex. (14), and middle ‑ɔr – ex. (15). [9]

(11)

Jóola Banjal

a.

A-ññɩl

akʊ

na-rɛm-ɛ

bʊ-bbʊn

babʊ.

cla-child

cla.def

cla-drink-cmp

clbi-medecine

clbi.def

‘The child drank the medecine.’

b.

Nɩ-rɛm-ɛn-ɛ

bʊ-bbʊn

babʊ

a-ññɩl

akʊ.

1sg-drink-caus-cmp

clbi-medecine

clbi.def

cla-child

cla.def

‘I made the child drink the medecine.’

(12)

Jóola Banjal

a.

Atɛjɔ

na-jʊg-ɛ

fɩgɛn

si-be

sasʊ.

Atéjo

cla-see-cmp

yesterday

clsi-cow

clsi.def

‘Atéjo saw the cows yesterday.’

b.

Si-be

sasʊ

sʊ-jʊg-ɩ

fɩgɛn.

clsi-cow

clsi.def

clsi-see-pass

yesterday

‘The cows were seen yesterday.’

(13)

Jóola Banjal

a.

Atɛjɔ

na-ttɛp-ɛ

pay-ɔl

y-aŋ.

Atéjo

cla-build-cmp

father-cla

cle-house

‘Atéjo built a house for his father.’

b.

Atɛjɔ

na-ttɛp-ɔrɔ-ɛ

y-aŋ.

Atéjo

cla-build-refl-cmp

cle-house

‘Atéjo built a house for himself.’

(14)

Jóola Banjal

a.

Atɛjɔ

na-fʊm-ɛ

ga-rafa

gagʊ.

Atéjo

cla-break-cmp

clgu-bottle

clgu.def

‘Atéjo broke the bottle.’

b.

Ga-rafa

gagʊ

gʊ-fʊm-ɔ-ɛ.

clgu-bottle

clgu.def

clgu-break-mid-cmp

‘The bottle broke.’

(15)

Jóola Banjal

a.

Atɛjɔ

na-ssaf-ɛ

Gɐleto.

Atéjo

cla-greet-cmp

Galéto

‘Atéjo greeted Galéto.’

b.

Atɛjɔ

Gɐleto

gʊ-ssaf-ɔr-ɛ.

Atéjo

and

Galéto

clgu-greet-mid -cmp

‘Atéjo and Galeto exchanged greetings.’

<18>

Ex. (14) illustrates the anticausative use of the middle suffix ‑ɔ, and ex. (15) illustrates the reciprocal use of the middle suffix ‑ɔr. These are their most productive uses, but each of these two suffixes is found in other middle functions, and their use must be viewed as lexicalized to a considerable extent. Our contrastive corpus includes 8 cases in which the addition of ‑ɔ has an anticausative function, but also 4 cases in which the addition of ‑ɔ triggers an autocausative (rather than anticausative) reading, and 8 cases in which the addition of ‑ɔr triggers an anticausative or autocausative (rather than reciprocal) reading:

(16)

Anticausative and autocausative uses of the suffix ‑ɔ in Jóola Banjal

-fʊm

‘break (tr.)’

>

-fʊmɔ

‘break (intr.)

-ppɛk

‘close (tr.)’

>

-ppɛgɔ

‘close (intr.)

-kkɐsul

‘reduce’

>

-kkɐsulo

‘decrease’

-fɔk

‘extinguish’

>

-fɔgɔ

‘go out (fire)’

-ppegul

‘open (tr.)’

>

-ppegulo

‘open (intr.)

-gɩs

‘split (tr.)’

>

-gɩsɔ

‘split (intr.)

-jal

‘untie’

>

-jalɔ

‘come undone, break loose’

-liw

‘wake’

>

-lio

‘wake up’

-ssɩm

‘dress (so.)’

>

-ssɩmɔ

‘dress (o.s.)’

-kkɔp

‘hide (sth.)’

>

-kkɔpɔ

‘hide (o.s.)’

-wwʊ

‘wash (so.)’

>

-wwʊɔ

‘wash (o.s.)’

-ccik

‘shave (so.)’

>

-ccigo

‘shave (o.s.)’

(17)

Anticausative and autocausative uses of the suffix ‑ɔr in Jóola Banjal

-lutten

‘make dirty’

>

-luttenor

‘become dirty’

‑ɔmɛn

‘gather (tr.)’

>

-ɔmɛnɔr

‘gather (intr.)’

-vvɩs

‘scatter (tr.)’

>

-vvɩsɔr

‘scatter (intr.)’

-fful

‘drag’

>

-ffulor

‘crawl’

-fʊt

‘cover, wrap’

>

-fʊlɔr

‘cover o.s., wrap o.s.’

-bbɔtɛn

‘coil’

>

-bbɔtɛnɔr

‘coil up’

-jjeben

‘wet’

>

-jjebenor .

‘get wet’

-bɛt

‘spread (tr.)’

>

-bɛlɔr

‘spread (intr.)’

<19>

Note also that, in addition to its use as a detransitivizing operator, ‑ɔr has a sociative use (‘do something together’) in which it combines with transitive as well as intransitive verbs without changing their valency, as in -tɛy ‘run’ > -tɛyɔr ‘run together’, or ‑tɛbɛn ‘raise, lift up’ > ‑tɛbɛnɔr ‘raise together, lift up together’.

4.2. Valency operations involving no change in the verb stem

4.2.1. Intransitive use of transitive verbs without change in the semantic role assigned to the subject

<20>

In Jóola Banjal, transitive verbs are very commonly used intransitively with a subject representing the same participant as in the transitive construction, the absence of an object phrase being interpreted as indetermination about the participant that could be encoded as the object. We have observed no obvious restriction on this intransitive use of transitive verbs.

<21>

By contrast, in Mandinka, the ability for a transitive verb to be used intransitively with a subject representing the same participant as in the transitive construction is limited to a subset of about thirty verbs. As illustrated in ex. (18) by teyí ‘cross’, with most verbs having this ability, the participant encoded as the object of the transitive construction can be left unexpressed, but it can also be encoded as an oblique. A list of Mandinka verbs lending themselves to this alternation has already been given in (8).

(18)

Mandinka

a.

Mǒo-lu

ye

báa

teyi.

person.def-pl

cmp.pos

river.def

cross

‘The people crossed the river.’

b.

Mǒo-lu

teyi-ta

(báa la).

person.def-pl

cross-cmp.pos

river.def

obl

‘The people crossed (the river).’

<22>

Mandinka also has a few transitive verbs with an intransitive construction in which the argument expressed as the object of the transitive construction is normally left unexpressed.

(19)

Mandinka transitive verbs with an intransitive construction in which the
second argument is normally not expressed and is interpreted as non-specific

lóŋ

‘know’

karáŋ

‘learn’

<23>

Mandinka also has a limited subset of transitive verbs whose use with a reflexive pronoun in object position expresses a valency operation of the antipassive type, as mǐŋ ‘drink’ in ex. (20).

(20)

a.

Kew-ó

ye

jíy-o

miŋ.

man-def

cmp.pos

water-def

drink

‘The man drank water.’

b.

Kew-ó

ye

miŋ

(jíy-o

la).

man-def

cmp.pos

refl

drink

water-def

obl

‘The man drank (water).’

<24>

In our corpus, the same alternation is possible with dáŋkú ‘answer (a call)’. A similar alternation is also found with jé ‘see’, with the difference that, with jé ‘see’, the second argument is obligatorily left unexpressed and interpreted as non-specific.

Our data also includes a Mandinka verb (í jǐi ‘set down (burden), give birth to’) whose construction includes a reflexive pronoun but cannot be explained synchronically as resulting from the reflexivization or antipassivization of a transitive construction, since jǐi is an intransitive verb meaning ‘go down’.

(21)

A

ye

jii

díŋkée

la.

3sg

cmp.pos

refl

set_down

son.def

obl

‘She gave birth to a son.’

<25>

A historical explanation can however be proposed, according to which the synchronically anomalous construction of í jǐi ‘set down (a burden), give birth to’ results historically from the antipassivization of a construction in which jǐi was used transitively with the meaning ‘make / let go down’.

4.2.1. Intransitive use of transitive verbs with a change in the semantic role assigned to the subject

<26>

Mandinka has a relatively productive causative / anticausative alternation. In this alternation, a verb that can be used transitively also has an intransitive construction which does not imply the involvement of a participant with the semantic role assigned to the subject of the transitive construction, the referent of the subject of the intransitive construction being presented as undergoing the same process as the object of the transitive construction, but without any hint at a possible external cause – ex. (22).

(22)

a.

Máŋk-óo

jolón-tá

baŋk-óo

to.

mango-def

fall-cmp-pos

ground-def

loc

‘The mango fell on the ground.’

Kew-ó

ye

mur-óo

jolóŋ

baŋk-óo

to.

man-def

cmp.pos

knife-def

drop

ground-def

loc

‘The man dropped the knife on the ground.’

<27>

The existence of an unrestricted active / passive alternation giving rise to morphologically unmarked passive constructions, as in ex. (23) and (24), constitutes the most original aspect of the argument structure in Mandinka and other Manding varieties. In spite of the absence of anything that could be analyzed as passive morphology, the construction illustrated by sentences (23b) & (24b) is passive in the sense that the patient is the subject of an intransitive construction in which the agent is syntactically demoted without however being deleted from argument structure. A decisive proof of the passive nature of the intransitive constructions involved in this alternation is their ability to include an agent-oriented adverb, such as feereetoo‑ ‘cleverly’ in (24b).

(23)

a.

Kew-ó

ye

wot-ôo

dádaa.

man-def

cmp.pos

car-def

repair

‘The man has repaired the car.’

b.

Wot-ôo

dádáa-ta.

car-def

repair-cmp.pos

‘The car has been repaired.’

(24)

a.

Kambaan-óo

ye

nás-óo

feereetoo-bóŋ

kolóŋ-o

kóno.

boy-def

cmp.pos

magic_water-def

cleverly-pour

well-def

inside

‘The boy cleverly poured the magic water into the well.’

b.

Nás-óo

feereetoo-bón-tá

kolóŋ-o

kóno.

magic_water-def

cleverly-pour-cmp.pos

well-def

inside

‘The magic water was cleverly poured into the well.’

<28>

There is no lexical restriction on this active / passive alternation: all Mandinka verbs found in the coding frame <A, P> can also be used in an intransitive construction with a passive reading. The causative / anticausative alternation is quite obviously much less productive. However, in a language in which all transitive verbs can be used intransitively with a passive reading, it is very difficult to propose a precise delimitation of the set of the transitive verbs that can be used intransitively with an anticausative reading. The relatively uncontroversial cases found in our contrastive corpus are listed in (25):

(25)

Mandinka verbs lending themselves to a morphologically
unmarked causative / anticausative alternation

jolóŋ

‘drop’ (intr. ‘fall’)

jáakálí

‘bother’

dáamutá

‘begin’

kafú

‘add’

těe

‘break (reduce to fragments)’

kátí

‘break sth. off, fracture’

fǎa

‘kill’ (intr. ‘die’)

kontáaní

‘please’

táwúŋ

‘close’

kuntú

‘cut’

tíñáa

‘damage, ruin, spoil’

ké

‘do’ (intr. ‘happen’)

tará

‘find’ (intr. ‘be somewhere’)

lábáŋ

‘finish’

sotó

‘have, ‘receive’ (intr. ‘be available, exist’)

nukúŋ

‘hide’

tú

‘leave’ (intr. ‘remain’)

málá

‘light’

yelé

‘open’

sǒo

‘pierce’

ñori

‘push’ (intr. ‘move’)

fátá

‘separate’

fárá

‘split, tear’

deté

‘tighten’

kúníŋ

‘wake’

<29>

In contrast to Mandinka, Jóola Banjal does not have morphologically unmarked passive constructions. As regards the existence of a morphologically unmarked causative / anticausative alternation, the verbs in (26) are the only ones in our corpus that have this ability.

(26)

Jóola Banjal verbs lending themselves to a morphologically
unmarked causative / anticausative alternation

-jjʊ

‘begin’

-jakalɩ

‘bother

-kajɛn

‘damage, ruin, spoil’

-nɔw

‘file’ (intr. ‘be blunt’)

-lu

‘pierce’

tʊj

‘break’

5. The relationship between transitive verbs and monovalent verbs assigning a role similar to that assigned to the P argument of a transitive verb

5.1. Introductory remarks

<30>

An important aspect of the transitivity system of languages is the manifestation of the semantic relationship between transitive verbs and monovalent verbs assigning a role similar to that assigned to the P argument of a transitive verb – in particular, between monovalent verbs encoding processes that can be conceptualized as occurring more or less spontaneously, or at least without a clearly identified instigator, and bivalent verbs encoding the same processes triggered by the action of an agent. As discussed in detail by Nichols & al. (2004), there may be no formal relationship between such pairs of verbs, but they may also be related in various ways, among which the following ones are attested in Mandinka and/or Jóola Banjal:

  • the transitive use of an ambitransitive verb may imply the involvement of an active participant that is not included in the argument structure of the same verb used intransitively;

  • the transitive verb may be morphologically derived from its intransitive counterpart (augmentation);

  • the intransitive verb may be morphologically derived from its transitive counterpart (reduction);

  • the transitive verb and its intransitive counterpart may be both derived from an abstract root that has no independent existence as a verb stem (double derivation).

5.2. Transitive verbs and monovalent verbs assigning a role similar to that assigned to the P argument of a transitive verb in Mandinka

<31>

Mandinka has two main ways of making apparent this kind of relationship between transitive and intransitive verbs: the transitive verb and its intransitive counterpart may be identical, or the transitive verb may derive from its intransitive counterpart. Both strategies are relatively productive. A list of 25 Mandinka verbs lending themselves to the causative / anticausative alternation, among those included in our contrastive data, has been given in (25) above. Our contrastive corpus also includes 32 transitive verbs regularly derived from intransitive verbs via causative derivation, plus two verbs whose transitive use is optionally marked by the causative suffix.

(27)

Mandinka transitive verbs derived from intransitive verbs via causative derivation

jimindí

‘bend (tr.)’

<

jimí

‘bend (intr.)’

fájíndí

‘boil (tr.)’

<

fájí

‘boil (intr.)’

kúpíndí

‘capsize (tr.)’

<

kúpí

‘capsize (intr.)’

nóndí

‘make dirty’

<

nóo

‘become dirty’

tájíndí

‘discourage’

<

tájí

‘become discouraged’

jindí

‘dismount’

<

jǐi

‘go down’

borindí

‘drive’

<

borí

‘run’

jandí

‘make dry’

<

jǎa

‘become dry’

fándí

‘fill’

<

fáa

‘be full’

bendí

‘gather (tr.)’

<

běŋ

‘gather (intr.)’

fúunúndí

‘inflate’

<

fúunú

‘swell’

kanandí

‘let go’

<

kaná

‘escape’

ñorindí

‘move (tr.)’

<

ñorí

‘move (intr.)’

maamandí

‘move (tr.)’

<

maamáŋ

‘move (intr.)’

kéñéndí

‘prepare’

<

kéñé

‘be ready’

kúyándí

‘put on bad terms’

<

kúyáa

‘be on bad terms’

fítínándí

‘put on bad terms’

<

fítína

‘quarrel’

malundí

‘put to shame’

<

malú

‘be ashamed’

wúlíndí

‘erect, raise, lift up’

<

wúlí

‘get up, stand up, rise’

báláfándí

‘sadden’

<

báláfáa

‘be sad’

kíisándí

‘save’

<

kíisá

‘escape’

jáñjándí

‘scatter, spread (tr.)’

<

jáñjáŋ

‘scatter, spread (intr.)’

boyindí

‘set down’

<

boyí

‘fall’

jindí

‘set down’

<

jǐi

‘go down’

dáa díyándí

‘sharpen’

<

dáa díyáa

‘be sharp’

fátíyándí

‘stir up’

<

fátíyáa

‘be ardent’

londí

‘stop (tr.)’

<

lǒo

‘stop (intr.)’

suusundí

‘suckle’

<

suusúu

‘suck (breast)’

batandí

‘tire’

<

batáa

‘get tired’

yelemandí

‘turn sth. over’

<

yelemá

‘turn over (intr.)’

kandindí

‘warm sth. up’

<

kandí

‘be warm’

sínándí

‘wet’

<

sínáŋ

‘get wet’

tóorándí ~ tóorá

‘torment’

<

tóorá

‘suffer’

ñíkíndí ~ ñíkíŋ

‘soak (tr.)’

<

ñíkíŋ

‘soak (intr.)’

<32>

A third possible strategy is the use of the reflexive pronoun in autocausative function, which can be viewed as a particular type of reduction strategy. It is found with three of the Mandinka verbs included in our contrastive corpus.

(28)

The autocausative use of the reflexive pronoun in Mandinka

míníŋ

‘coil’

>

í míníŋ

‘coil up’

kǔu

‘wash (so.)’

>

í kǔu

‘wash (o.s.)’

líi

‘shave (so.)’

>

í líi

‘shave (o.s.)’

5.3. Transitive verbs and monovalent verbs assigning a role similar to that assigned to the P argument of a transitive verb in Jóola Banjal

<33>

Jóola Banjal shows more diversity in the possible relationships between transitive verbs and monovalent verbs assigning a role similar to that assigned to the P argument of a transitive verb.

Our contrastive data includes 15 such pairs in which the transitive verb derives from the intransitive verb via causative derivation, but 20 in which the intransitive verb derives from the transitive verb. The pairs illustrating the reduction strategy have been listed in (16) and (17) above. Those illustrating the augmentation strategy are listed in (29).

(29)

Jóola Banjal transitive verbs derived from intransitive verbs via causative derivation

-ʊnɛn

‘announce’

<

-ʊn

‘hear’

-labɛn

‘boil (tr.)’

<

-lab

‘boil (intr.)’

-tɛyɛn

‘drive’

<

-tɛy

‘run’

-kayɛn

‘dry’

<

-kay

‘become dry’

-mmɛŋɛn

‘fill’

<

-mmɛŋ

‘be full’

-saɛn

‘light’

<

-sa

‘burn’

-llɩmɛn

‘lose’

<

-llɩm

‘get lost’

-tɔsɛn

‘move (tr.)’

<

-tɔs

‘move (intr.)’

-ssʊɛn

‘put to shame’

<

-ssʊ

‘be ashamed’

-pagɛn

‘save’

<

-pak

‘escape’

-ñɛgɛn

‘sharpen’

<

-ñɛk

‘be sharp’

-bugen

‘soak (tr.)’

<

-buk 

‘soak (intr.)’

-rafɛn

‘suckle’

<

-raf

‘suck (breast)’

-yɔgɛn

‘tire’

<

-yɔk

‘get tired’

-supen

‘warm sth. up’

<

-sup

‘be warm’

<34>

Our corpus also includes 5 such verb pairs in which both members of the pair are morphologically marked: the transitive verb results from the addition of the causative suffix to an abstract root, whereas the corresponding intransitive verb results form the addition of a detransitivizing suffix (‑ɔ, ‑ɔr, or ‑i) to the same abstract root.

(30)

Transitive/intransitive verb pairs with both members equally marked in Jóola Banjal

-lɐgen

‘hang (tr.)’

-lɐgo

‘hang (intr.)’

-ɩlɛn

‘put up, erect, stop’

-ɩlɔ

‘stand up’

-alɛn

‘set down’

-alɔ

‘go down’

-bɐken

‘turn sth. over’

-bɐko

‘turn over (intr.)’

-jjeben

‘wet’

-jjebi

‘get wet’

Finally, we have found 6 verbs lending themselves to a morphologically unmarked causative / anticausative alternation – see (26) above.

5.4. Summary

<35>

The following chart summarizes the number of verbs in our contrastive corpus involved in one of the possible strategies making explicit the relationship between transitive verbs and monovalent verbs assigning a role similar to that assigned to the P argument of a transitive verb:

augmentation

reduction

double

derivation

ambitransitivity

ambitransitivity

~ augmentation

Jóola Banjal

15

20

5

6

Mandinka

32

3

25

2

6. Mismatches between the morphological structure of verbs and their valency properties

6.1. Non-canonical uses of the causative suffix in Mandinka

<36>

Our contrastive corpus includes 34 Mandinka transitive verbs regularly derived from intransitive verbs via causative derivation, but also 4 transitive verbs with the ending ‑ndí that do not show the expected relationship with the verb from which they seem to derive morphologically.

<37>

In the case of síláŋ ‘be afraid of’ / sílándí ‘frighten’, the causative suffix does not have its usual valency-increasing effect, since both verbs are semantically bivalent. However, this use of the causative suffix can be considered regular in the sense that it encodes a change in the semantic role of the subject consistent with its use in canonical causative derivation. This analysis is confirmed by the fact that, cross-linguistically, the same relationship between ‘be afraid of’ and ‘frighten’ can be found in a number of languages having a causative derivation. In particular, the same explanation applies to the double derivation characterizing the corresponding verb pair in Jóola Banjal: -kolen ‘frighten’ / -koli ‘be afraid of’.

<38>

In the case of báyí ‘chase away’ / báyíndí ‘follow’, both verbs are transitive, and nothing in the meanings they express seems to justify deriving ‘follow’ from ‘chase away’ via the addition of a causative suffix.

In the case of kíitíi ‘go to court’ / kíitíndí ‘judge’, the meaning of the derived verbs is not exactly the meaning expected from the general meaning of causative derivation.

Finally, the transitive verb terendí ‘come across’ is synonymous with teréŋ, which expresses the same meaning in an extended intransitive construction in which it assigns the same role to its subject.

6.2. Non-canonical uses of valency-changing suffixes in Jóola Banjal

<39>

A striking contrast between Mandinka and Jóola Banjal is that the non-canonical uses of the only valency-changing derivation of Mandinka are statistically marginal, whereas the verbal lexicon of Jóola Banjal includes a relatively high proportion of verbs with endings that could represent valency-changing operators, but for which, either no possible source of the derivation can be identified, or a possible source can be identified, but the semantic relationship between the two verbs is not in accordance with the general meaning of the suffix in question.

<40>

We have in our contrastive corpus 15 Jóola Banjal transitive verbs analyzable as regularly derived from an intransitive verb via causative derivation, as already mentioned above, but we also have 53 transitive verbs ending with ‑ɛn that cannot be analyzed as regularly derived from an intransitive verb. The existence of such a high number of verbs with the same ending ‑ɛn is probably not due to mere chance. However, for 40 of them, we have not been able to identify a possible source. For 13 of them, a possible source is attested, but with a meaning that excludes a straightforward explanation in terms of causative derivation. For example, ‑tɛbɛn ‘raise’ is probably cognate with -tɛb ‘carry’, but these two verbs are equally transitive, and the semantic roles they assign do not differ in a way that could justify the use of a causative suffix for ‘raise’ but not for ‘carry’.

<41>

A plausible explanation is that the causative derivation of Jóola Banjal is an ‘old’ derivation that was more active at a former stage in the evolution of the language. The situation we observe in the present state of the language suggests that, for most of the verbs that have been formed according to this derivation, either the source has been lost, or the relationship with the source has been blurred by changes in the lexical meaning of either the derived verb or its source.

<42>

As regards detransitivizing derivation, a somewhat similar situation is found with the suffix ‑ɔr. As already mentioned, this suffix has productive reciprocal and sociative uses, and is also used to form the decausative or autocausative counterpart of a small number or transitive verbs (see (17) above). But our data also includes 12 transitive verbs with the ending ‑ɔr for which, either no possible source of the derivation can be identified, or a possible source can be identified, but the semantic relationship between the two verbs is not in accordance with the general meaning of the suffix ‑ɔr. In some cases (‑ɛmɔr ‘meet’, ‑nogor ‘ressemble’ ‑nappɔr ‘share in common’), the meaning suggests a sociative / reciprocal derivation whose source has been lost, but this explanation can hardly apply to cases such as ‑jimor ‘forget’ or ‑ppɛgɔr ‘give birth to’. In such cases, an investigation of possible cognates in other Jóola varieties and in other Bak languages would be necessary before deciding whether the ending ‑ɔr can be viewed as a frozen suffix or not.

7. Conclusion

<43>

The main contrasts we have observed in the organization of the valency properties of bivalent verbs in Jóola Banjal and Mandinka can be summarized as follows:

  • Both languages are obligatory A coding languages with a very strong tendency to generalize transitive coding to nearly all semantic types of bivalent verbs, but this tendency is particularly strong in Jóola Banjal; in both languages, movement verbs and verbs encoding naturally reciprocal events constitute the only systematic exceptions to this tendency.

  • In Mandinka, the ability to be used intransitively with a non-specific reading of the P argument is restricted to a limited number of transitive verbs, and this restriction is compensated by a productive antipassive derivation, whereas in Jóola Banjal, there seems to be no particular restriction on the intransitive use of transitive verbs with a non-specific reading of the P argument.

  • Mandinka has a cross-linguistically rare type of passive construction with no specific marking, whereas in Jóola Banjal, transitive verbs in passive constructions are marked by a passive suffix.

  • As regards the relationship between transitive verbs and monovalent verbs assigning a role similar to that assigned to the P argument of a transitive verb, when this relationship is formally marked, the preferred strategies are (1) augmentation and (2) ambitransitivity in Mandinka, (1) reduction and (2) augmentation in Jóola Banjal.

  • The verbal lexicon of Jóola Banjal includes a particularly high proportion of verbs with endings that could represent valency-changing operators, but for which, either no possible source of the derivation can be identified, or a possible source can be identified, but the semantic relationship between the two verbs is not in accordance with the general meaning of the suffix in question.

Abbreviations

ANTIP

antipassive

CAUS

causative

CL

noun class

CMP

completive aspect

DEF

definite

GEN

genitive

INCMP

incompletive aspect

LOC

locative

MID

middle

NEG

negative

OBL

oblique

PASS

passive

PL

plural

POS

positive

REFL

reflexive

SG

singular

References

Bassène, Alain-Christian 2007

Morphosyntaxe du jóola banjal, langue atlantique du Sénégal. Köln: Rüdiger Köppe

Creissels, Denis and Pierre Sambou 2013

Le mandinka (mandiŋkakáŋo): phonologie, grammaire, textes. Paris: Karthala

Nichols Johanna, David Peterson, and Jonathan Barnes 2004

‘Transitivizing and detransitivizing languages.’ Linguistic Typology 8:149–211



[1] Traditionally, Bak languages are considered a subgroup of an Atlantic family included in the Niger-Congo phylum, but the opinion that prevails now is that Atlantic is not a genetically valid grouping, and must be viewed as a purely areal grouping of several independent branches of Niger-Congo.

[2] The Mande language family was included by Greenberg in the Niger-Congo phylum, but the evidence supporting this decision is rather slim, and the Niger-Congo affiliation of Mande is considered questionable by many specialists.

[3] We would like to express our thanks to William Diatta for his help with Jóola Banjal data.

[4] This paper has benefited from the support of the French National Research Agency (ANR) within the frame of the ‘Sénélangues’ project (ANR-09-BLAN-0326).

[5] The locative index coincides with the P index of class BI, and this coincidence is probably not fortuitous, but synchronically, the indexation of P arguments and the indexation of spatial adjuncts are two distinct mechanisms.

[6] In the Mandinka examples below, postpositions marking oblique arguments are glossed according to the meaning they typically express as heads of postposition phrases in adjunct function, with three exceptions: lá, má, and tí, for which the generic gloss OBL is used. The reason is that the analysis of the uses of these three postpositions as extensions of some ‘central’ or ‘prototypical’ meaning is particularly problematic.

[7] This sentence would be acceptable with the meaning ‘The boy must cry’, since there is a homonymous predicative marker yé ‘subjunctive positive’ compatible with intransitive predication, but it is impossible with yé ‘completive positive’.

[8] Morphologically, ‑yɐbo ‘marry (woman>man)’ is cognate with ‑yab ‘receive’, also used with the meaning ‘marry (man>woman)’. ‑Yab ‘receive’ is a transitive verb, and ‑o is a detransitivizing suffix. The change in the quality of the vowel is however irregular.

[9] Each of these suffixes has an allomorph with a +ATR vowel, and another with a –ATR vowel. The allomorph with a –ATR vowel is used for quotation, since in the vowel harmony system of Jóola Banjal, –ATR is the unmarked value of the ±ATR feature.

Lizenz

Kommentare

Es liegen noch keine Kommentare vor.

Möchten Sie Stellung zu diesem Artikel nehmen oder haben Sie Ergänzungen?

Kommentar einreichen.