Navigation – Plan du site

AccueilNumérosMandenkan 71Causative in Gban

Causative in Gban

Le causatif en gban
Максим Леонидович Федотов. Каузатив в гбан
Maksim Fedotov
p. 3-38

Résumés

L’article traite des moyens d’expression du semantisme du causatif en gban : lexicaux (des paires supplétives et la labilité), périphrastiques et morphologique, plus précisément le suffixe dérivationnel productif ‑kà. Les causatifs suffixaux dérivés des verbes transitifs manifestent un modèle de codage atypique de « double objet direct ». Une hypothèse diachronique est proposée selon laquelle la construction causative suffixale aurait été grammaticalisée à partir d’une construction avec un verbe causatif et une nominalisation non marquée (en position de complément direct), qui pourrait elle-même avoir deux dépendants nominaux à gauche. Cette hypothèse est soutenue par plusieurs faits synchroniques indépendants, y compris les propriétés syntaxiques des nominalisations en gban.

Haut de page

Texte intégral

Introduction1

  • 1 I owe the original idea about the diachronic source of the “double object” construction to Denis Cr (...)
  • 2  I will understand causative in the broad sense (“causatives sensu latiore” in (Kulikov 2001: 888–8 (...)
  • 3  Gban (gbá̰; ISO 639-3 code: ggu, Glottocode: gagu1242) is a South Mande language spoken by ca. 60 (...)

1In this paper, I will present a description of the different ways in which causative2 can be expressed in Gban.3 Particular attention will be paid to the productive causative derivation with the suffix ‑kà and the associated “double object” construction. In considering these phenomena, I will adopt a primarily descriptive stance. Terminologically and theoretically, the description largely follows (Kulikov 2001) and (Dixon 2000).

2The paper is structured as follows. In Section 1, I give an overview of the ways of expressing causative in Gban: lexically, including suppletive pairs and lability (1.1), suffixally (1.2), and periphrastically (1.4). In 1.3, I provide a typological characteristic to the morphological subsystem of expressing (anti)causative in Gban based on (Haspelmath 1993). In Section 2, causative suffixation is discussed in more detail. In 2.1, I describe its morphological aspects: form, productivity, and combinability. In 2.2, I describe its semantics: different types of causation that can be expressed by suffixal causatives. And in 2.3, I describe the observed argument coding patterns, including the “double object” construction and its properties (2.3.2). In 2.4, I propose a diachronic hypothesis for the origins of the causative derivation that can explain the emergence of the “double object” construction.

1. Expressing causative in Gban: an overview

1.1. Lexical means: suppletion and lability

  • 4  As can be seen from the examples, gà ‘die’ cannot be used transitively (i.e. it is not labile), n (...)
  • 5  A peculiarity of the latter pair is that the anticausative member is already transitive: ‘learn so (...)

3Gban manifests at least two purely lexical (suppletive) causative-anticausative pairs: gà ‘die’ vs. zɛ̀ ‘kill’, illustrated below,4 and gɔ̰̋ ‘try; learn’ vs. gḭ̏ ‘show; teach’.5

  • 6  Gban has four level tones and two contour tones: ultra-high (), high (), low (), and ultra-l (...)

(1a)

Ba̋ba̋

ɛ́

gà.6

ram

3sg\pst

die[pfv.hod]

4‘The ram died.’

(1b)

Sɔ̀kȕ

ɛ́

ba̋ba̋

zɛ̀

/ *gà

/ *gà‑kȁ.

Soku

3sg\pst

ram

kill[pfv.hod]

die[pfv.hod]

die‑caus[pfv.hod]

5‘Soku killed the ram.’

  • 7  It is the number of confirmed anticausative-labile verbs in a studied sample of several dozen verb (...)

6At least for 19 verbs,7 causative-anticausative lability is attested:

(2a)

Wè

wṵ̋

ɛ́

ylé.

animal

oil

3sg\pst

melt[pfv.hod]

7‘The fat melted (in the sun).’

(2b)

Mɔ̰̈bì

ɛ́

wè

wṵ̋

ylé.

Monbi

3sg\pst

animal

oil

melt[pfv.hod]

8‘Monbi melted the fat.’

(3a)

Vle̋

ɛ́

gò.

door

3sg\pst

open[pfv.hod]

9‘The door opened.’

(3b)

Sɔ̀kȕ

ɛ́

vle̋

gò.

Soku

3sg\pst

door

open[pfv.hod]

10‘Soku opened the door.’

(4a)

Lȁdíó

ɛ́

wi̋.

radio

3sg\pst

break[pfv.hod]

11‘The radio broke.’

(4b)

Jȁlȅ

ɛ́

lȁdíó

wi̋.

Diale

3sg\pst

radio

break[pfv.hod]

12‘Diale broke the radio.’

  • 8  It should be noted that it is indeed not always easy to distinguish between a passive and an antic (...)

13It should also be mentioned — although this phenomenon no longer pertains to the domain of causativity — that active-passive lability (Letuchij 2009; Creissels 2014) has also been observed, for at least 17 verbs. In passive uses of such verbs, the agent participant is not (and cannot be) expressed in Gban.8 Some examples of passive-labile verbs are given in (5)–(7):

(5a)

Sɔ̀kȕ

ɛ́

kwa̰̋wȍɛ̏

kṵ̋.

Soku

3sg\pst

thief

catch[pfv.hod]

14‘Soku caught the thief.’

(5b)

Kwa̰̋wȍɛ̏

ɛ́

kṵ̋.

thief

3sg\pst

catch[pfv.hod]

15‘The thief got caught / was caught.’

(6a)

zǐ

bè

/ flà.

1pl\pst

sauce

eat[pfv.hod]

lick[pfv.hod]

16‘We ate (/ lapped up) the sauce.’

  • 9  Henceforth, a superscript “OK” before an example indicates that the entire example was constructed (...)

(6b)

OK Zǐ

ɛ́

bè

/ flà.9

sauce

3sg\pst

eat[pfv.hod]

lick[pfv.hod]

17‘The sauce was eaten (/ lapped up).’

(7a)

Ɔ́

s

gḭ̏

Zá̰

nɛ̰̏.

3pl\pst

house

show[pfv.hod]

Jean

at

18‘They (impers.) showed Jean the house.’

(7b)

Sa̋

ɛ́

gḭ̏

Zá̰

nɛ̰̏.

house

3sg\pst

show[pfv.hod]

Jean

at

19‘The house was shown to Jean.’

  • 10  Cf. Kakabe, where anticausative-labile verbs can also be passive-labile, too (Vydrina 2011: 205).

20In a couple of cases, both types of lability are attested for the same verb, cf. mḭ̋a̰̋ ‘split (tr.) / chop; split (intr.); be split / chopped’ (8a–c), tà ‘close (tr.); close (intr.); be closed’.10

(8a)

Yűkwȉ

ɛ́

mḭ̋a̰̋

(

blɛ̋

nɛ̰̏).

tree

3sg\pst

split[pfv.hod]

s/he/it

self

at

21‘The wood split (on its own) [e.g. the fruits on the tree had grown too heavy].’

(8b)

Yűkwȉ

ɛ́

mḭ̋a̰̋.

tree

3sg\pst

split[pfv.hod]

22‘The wood was split/chopped [by someone].’

(8c)

Zá̰

ɛ́

yűkwȉ

mḭ̋a̰̋.

Jean

3sg\pst

tree

split[pfv.hod]

23‘Jean chopped the wood.’

24Other types of lability (reciprocal, reflexive, or agent-preserving) have not been attested in Gban.

1.2. Suffixation

25As was already mentioned, Gban has a productive causative derivational suffix k/‑kȁ, which will be the main focus of this paper. Let us give here three preliminary illustrations of causative suffixation:

(9a)

Se̋

ɛ́

dő.

fire

3sg\pst

light_up[pfv.hod]

26‘The fire lit up.’

(9b)

Ḭ̋

se̋

dkȁ

1sg\pst

fire

light_up‑caus[pfv.hod]

27‘I lit the fire.’
(suffixal causative from an intransitive non-labile verb).

(10a)

Mṵ̋

dò

ɛ́

tȅnḭ̋

mɛ̰̏.

person

one

3sg\pst

child

beat[pfv.hod]

28‘Someone beat the child.’

(10b)

M

a̰̋

tȅnḭ̋

mɛ̰̏‑kà.

1sg\pst

s/he/it

child

beat‑caus[pfv.hod]

29‘I made him beat the child.’
(suffixal causative from a transitive non-labile verb).

(11a)

Gbő

ɛ́

ylɛ̰̀

yi̋

di̋.

canoe

3sg\pst

enter[pfv.hod]

water

under

30‘The canoe sank (lit. ‘entered the water’).’

(11b)

Mɔ̰̈bì

yë

ḭ̀

mɔ̰́

gbő

ylɛ̰̀

yi̋

di̋.

Monbi

3sg\pst

I

for

canoe

enter[pfv.hod]

water

under

31‘Monbi sank (lit. ‘[made] enter the water’) my canoe.’

(11c)

Mɔ̰̈bì

yë

ḭ̀

mɔ̰́

gbő

ylɛ̰̀-kȁ

yi̋

di̋.

Monbi

3sg\pst

I

for

canoe

enter‑caus[pfv.hod]

water

under

  • 11  Such cases of synonymy of labile simplex verbs and causatives will be discussed in more detail in (...)

32‘Monbi sank (lit. ‘made enter the water’) my canoe.’
(suffixal causative from an anticausative-labile verb).11

33All other aspects of causative suffixation, i.e. morphological, semantic, syntactic, and diachronic, will be discussed in detail in Section 2.

1.3. A typological characteristic of the system

  • 12  Only morphological means have been considered, leaving out the periphrastic constructions, which w (...)

34This subsection provides a typological characteristic of the morphological12 system of expressing causative/anticausative in Gban, using the methodology from (Haspelmath 1993).

35In Table 1, I give the Gban verbs corresponding to the 31 meaning pairs from (Haspelmath 1993: 97) and their characterization in terms of expression types (causative, labile, suppletive, etc.). The pairs are ordered by decreasing (presumed) likelihood of their spontaneous occurrence according to (Haspelmath 1993: 104) (more on that below).

Meaning pair

Gban verb(s)

Type

18. ‘boil (intr.)/(tr.)’

25. ‘freeze (intr.)/(tr.)’

wlɔ̰̋ / wlɔ̰̋-k

C

29. ‘dry (intr.)/(tr.)’

blɛ̏ / blɛ̏-k

C

1. ‘wake up (intr.)/(tr.)’

vòà / vòà-k

C

20. ‘go out/put out’

lȉbi̋

L

11. ‘sink (intr.)/(tr.)’

8. ‘learn/teach’

gɔ̰̋ / gḭ̏ ~ gɔ̰̋-kȁ

S/C

13. ‘melt (intr.)/(tr.)’

ylé (/ ylé-kȁ)

L/C

31. ‘stop (intr.)/(tr.)’

dɛ̰̀ / dɛ̰̀-k

C

23. ‘turn (intr.)/(tr.)’

lȉ / lȉ-k

C

26. ‘dissolve (intr.)/(tr.)’

3. ‘burn (intr.)/(tr.)’

kȅ / kȅ-k

C

14. ‘be destroyed/destroy’

kpě

L

27. ‘fill (intr.)/(tr.)’

fà (/ fà-k)

L/C

22. ‘finish (intr.)/(tr.)’

vò / vò-k

C

7. ‘begin (intr.)/(tr.)’

10. ‘spread (intr.)/(tr.)’

kpɔ̋ (/ kpɔ̋-k)

L/C

24. ‘roll (intr.)/(tr.)’

blǒ tɔ́ (/ blǒ-k tɔ́)

L/C

16. ‘develop (intr.)/(tr.)’

15. ‘get lost/lose’

gṵ̀à̰ / gṵ̀à̰-k

C

21. ‘rise/raise’

dȉȁ / dȉȁ-k

C

28. ‘improve (intr.)/(tr.)’

lè ɔ̰́ / lè-k (w)ɔ̰́

C

19. ‘rock (intr.)/(tr.)’

fȉfi̋ / fȉfi̋-k

C

17. ‘connect (intr.)/(tr.)’

dɔ̀ kèɛ̀

L

12. ‘change (intr.)/(tr.)’

lȉ zȁȍ / lȉ-k zȁȍ

C

9. ‘gather (intr.)/(tr.)’

bɔ̀ kèɛ̀ / bɔ̀-kȁ kèɛ̀

C

5. ‘open (intr.)/(tr.)’

gò

L

2. ‘break (intr./(tr.)’

wi̋

L

6. ‘close (intr.)/(tr.)

tà

L

30. ‘split (intr.)/(tr.)’

mḭ̋a̰̋

L

4. ‘die/kill’

gà / zɛ̀

S

36Table 1. A sample of Gban verbs and their expression types, based on (Haspelmath 1993)

37Comments on Table 1.

381) “C” = causative pair; “L” = [anticausative-]labile verb; “L/C” = both a labile verb and a derived causative verb available; “S” = suppletive pair; “S/C” = both a suppletive verb and a derived causative verb available.)

392) For the meaning pairs marked with “—”, no suitable verbal equivalents have been attested. For example, ‘begin’ is expressed in Gban by a fixed “noun + verb” combination bò bɛ̋ <head take>, in which the patient is expressed within the NP headed by bò (cf. Bȅa̋ bò ɛ́ bɛ̋ ‘The work began’, lit. ‘The head of the work was taken’ and Kȁȍkḭ̏lè ɛ́ bȅa̋ bò bɛ̋ ‘The farmer started the work’, lit. ‘The farmer took the head of the work’). ‘Boil’ and ‘dissolve’ are translated with similar expressions. ‘Sink’ is translated as lit. ‘(make) enter water’, cf. (11). For ‘develop’, the closest attested equivalent is ‘change [not necessarily for the better]’.

  • 13  Non-directed pairs are all pairs except causative and anticausative (directed) ones: labile, equip (...)

40The counts for different expression types and their ratios in Gban are as follows: 26 total, 0 A[nticausative], 15.5 C[ausative], 0 E[quipollent], 9 L[abile], 1.5 S[uppletive]; A/C = 0; % non-dir[ected]13 = 40.

  • 14  It seems that M. Haspelmath used 0.5 instead of zeros in all A/C calculations to avoid dividing by (...)

total

A

C

E

L

S

A/C14

%non-dir.

Russian

31

23

0

5

0

3

46.00

26

German

31

14.5

0

4

11.5

1

29.00

53

Greek

31

13.5

0

0

16.5

1

27.00

56

Rumanian

30

24

1

0

3

2

24.00

17

French

31

20.50

2

0

7.5

1

10.25

27

Lithuanian

31

17.5

6

6

0.5

1

2.92

24

Hebrew

31

20.5

7.5

2

1

0

2.73

10

Arabic

31

17

8.5

3

1

1.5

2.00

18

Georgian

31

9

4.5

15.5

0

2

2.00

56

Armenian

31

16

8.5

5.5

0

1

1.88

21

Swahili

31

11

11

8

0

1

1.00

29

Finnish

28

12

13.5

0.5

0.5

1.5

0.88

9

Udmurt

31

10.5

12.5

4.5

2.5

1

0.84

26

Hungarian

31

7

9

12

0

3

0.78

48

Lezgian

31

8

12

6

5

0

0.66

35

Hindi-Urdu

31

7.5

14

7.5

2

0

0.54

31

Turkish

30

9

17.5

2.5

0

1

0.51

12

Mongolian

31

6

22

2

0

1

0.27

10

Indonesian

31

0

14

17

0

0

0.04

55

Gban

26

0

15.5

0

9

1.5

0.00

40

English

31

2

0

1

25

3

94

Japanese

31

3.5

5.5

20.5

0.5

1

71

total

636

243

164.5

128.5

69

310

41Table 2. A table of expression types by language from (Haspelmath 1993: 101), with the addition of Gban

42As can be seen, the anticausative/causative ratio (“A/C”) is zero, which puts Gban into the very bottom of the table of languages that was given in [Haspelmath 1993: 101] and is reproduced in Table 2, with the addition of Gban. It finds itself together with Indonesian, which also has only the causative directed expression type (and on the opposite end from Russian, which has (almost) exclusively the anticausative directed expression type).

43The percentage of non-directed pairs (“% non-dir.”) for Gban is relatively high at 40%, which puts it in the upper half of the languages from M. Haspelmath’s sample: between Lezgian with 35% and Hungarian with 48% (the lowest percentage being 9% for Finnish and the highest 94% for English).

44Among non-directed pairs, Gban prefers labile verbs (to equipollent and suppletive pairs), similarly to English, Greek, and German.

45Finally, in terms of the preferences of different verbs for different expression types, the Gban data seem to conform to or at least not to contradict the general tendency described in (Haspelmath 1993: 103–106) that causative marking is more associated with situations that usually occur spontaneously in the nature (and anticausative marking with those that are usually brought about by an external force). In Table 1, the pairs are ordered by decreasing presumed likelihood of their spontaneous occurrence, based on the table in (Haspelmath 1993: 104). We can see that the upper half of the table (not counting the blanks) contains more causatives than the lower half: 11 vs. 7. In particular, the top three pairs are all causative, while the bottom five are all non-causative.

1.4. Periphrastic constructions

46In addition to the morphological means described above, Gban has several periphrastic constructions, also expressing causative semantics.

471. There are two constructions with seemingly broad causative semantics: “X yɛ̀kɛ̋, Y P” (lit. ‘X does it, Y P’) (12) and “X nɔ̰̀, Y P” (lit. ‘X gives it, Y P’) (13)–(14). These two constructions are frequent and seem to be the main competitors to causative suffixation in the grammatical system of Gban.

(12)

[Zá̰á̰

y

yɛ̀kɛ̋]

[mṵ̋

dò

yë

ḭ̀

Jean

3sg\pst

s/he/it

do[pfv.hod]

person

one

3sg\pst

I

yènḭ̋

mɛ̰̏].

brother

beat[pfv.hod]

48‘Jean made a man beat my brother [provoked him].’ (lit. ‘Jean did it, a man beat my brother.’)

(13)

[Y

nɔ̰̀]

[ḭ̋

Zá̰

zɛ̀].

3sg\pst

s/he/it

give[pfv.hod]

1sg\pst

Jean

kill[pfv.hod]

49‘He made me kill Jean.’ (lit. ‘They gave it, I killed Jean.’)

(14)

[W

nɔ̰̀]

[ḭ̋

wè

bȅ].

3pl\pst

s/he/it

give[pfv.hod]

1sg\pst

animal

eat[pfv.hod]

50‘They made/helped me eat some meat [e.g. bought it or gave me money].’

512. There is a construction based on the verb kpɔ̋ ‘spread (out), lay (down)’ with the semantics of coercion, forced causation. It has two variants: with a subordinate clause marked with nḭ́: “X kpɔ̋=le̋ Y nɛ̰̀ nḭ́ Y P[sbjv]” (lit. ‘X lays it on Y, so that Y P’) (15) and with a nominalization in =le: “X [V]=le kpɔ̋=le̋ Y nɛ̰̀” (lit. ‘X lays P=ing on Y’) (16):

(15)

Bààkḭ̏lè

y

kpɔ̏=lȅ

mṵ̋

fǒfǒ

nɛ̰̏

chief

3sg

s/he/it

ipfv\spread=loc2

person

all

at

[nḭ́

ɔ̏

klě

wò].

that

3pl

hat

put[sbjv]

52‘The chief compels everybody to wear hats.’ (lit. ‘The chief lays it on everybody, so that they wear hats.’)

(16)

Sɔ̀kȕ

ɛ́

[sà

sḭ̋=le̋]

kpɔ̋=le̋

ḭ̀

nɛ̰̀.

Soku

3sg\pst

song

sing=nmlz

spread[pfv.hod]=loc2

I

at

53‘Soku forced me to sing [e.g. by threatening to kill me].’ (lit. ‘Soku laid the song-singing on me.’)

  • 15  It should be noted that permissive semantics is closely related to causative proper, but is not st (...)

543. There are several constructions with permissive semantics ‘let Y P / allow Y to P’:15

  • 16  And its anticausative version: “di̋ yèè Y kɔ̰́, nḭ́ Y P[sbjv]” <road be Y on that Y P[sbjv] (li (...)

55a) “X () di̋ kpɔ̋ Y kɔ̰́, nḭ́ Y P[sbjv]” <X (s/he/it) road spread Y on that Y P[sbjv]> (lit. ‘X puts (its) way on Y that Y P’) ‘X allows Y to P’16 (17);

56b) “X di̋ nɔ̰̀ Y nɛ̰̏, nḭ́ Y P[sbjv]” / “X di̋ nɔ̰̀ Y nɛ̰̏, Y P[ind]” <X road give Y at that Y P[sbjv]> / <X road give Y at Y P[ind]> (lit. ‘X gives way to Y that Y P’ / ‘X gives way to Y, Y P’) ‘X allows Y to P’ (18);

57c) “X ȁ gɔ̰̋, (nḭ́) Y P[sbjv]” <X s/he/it accept, (that) Y P[sbjv]> (lit. ‘X accepts it, (that) Y P’) ‘X allows Y to P’ (19);

58d) “X Y tő yɔ̰̏nḭ́, (nḭ́) X P[sbjv]” <X Y leave so that Y P[sbjv]> (lit. ‘X leaves Y this way, (that) Y P’) ‘X lets Y P’ (20).

(17)

Sɔ̀kȕ

dɛ̌=nṵ̋

kè

di̋

kpɔ̋

kɔ̰́

Soku

father=pl

3pl‑ind.neg

s/he/it

road

ipfv\spread

s/he/it

on

nḭ́

ɛ̏

Mɔ̰̈bì

bɛ̋.

that

3sg

Monbi

take[sbjv]

59‘Soku’s parents don’t allow him to marry Monbi.’

(18)

Ɛ́

di̋

nɔ̰̈

nḭ̋-lè

nɛ̰̏

3sg\pst

road

give\pfv.preh

s/he/it

child-woman

at

nḭ́

ɛ̏

bò

sɔ̋=ő, <…>

that

3sg

leave[sbjv]

yard=in

60‘He allowed his daughter to leave the house {because she asked him.}’

(19)

M

a̰̋

gɔ̰̋

nḭ́

ɛ̏

yà.

1sg\pst

s/he/it

accept[pfv.hod]

that

3sg

go[sbjv]

61‘I allowed him to go.’

(20)

Lɛ̌

ɔ̏

tő

yɔ̰̏nḭ́,

therefore

2pl

they

leave[sbjv]

so

ɔ̏

blɔ̋

vi̋

wȍ

kpȍ

ve̋i̋i̋, <…>

they

self

whole

3pl

we

call[sbjv]

before

62‘So let them themselves call us first, {then we’ll go.}’

  • 17  Note, however, that the true copula yè() ‘be’ cannot attach the causative suffix (23e).

634. Finally, denominal causatives (e.g. ‘make clear’, ‘make king’) are also expressed in Gban by means of analytical constructions. More precisely, they are formed on the basis of the corresponding quasi-copular constructions (21)–(22), (24)–(25): as a transitive use of a labile quasi-copula (23a), (26a) or as a causative derived from the quasi-copula (23b–d), (26b–c).17 Denominal causative constructions are illustrated below first for the adjective zlȅi̋i̋ ‘clear’ and then for the noun blȅgbḭ̏ ‘king’.

(21)

Yi̋

yȅȅ

zlȅi̋i̋.

water

3sg

ipfv\be

clear

64‘The water is clear.’

(22a)

Yi̋

ɛ́

yɛ̀kɛ̋

zlȅi̋i̋.

water

3sg\pst

do[pfv.hod]

clear

(22b)

Yi̋

ɛ́

gȉȅ

zlȅi̋i̋.

water

3sg\pst

become[pfv.hod]

clear

65‘The water became clear.’

(23a)

Zá̰

ɛ́

yi̋

yɛ̀kɛ̋

zlȅi̋i̋.

Jean

3sg\pst

water

do[pfv.hod]

clear

(23b)

Zá̰

ɛ́

yi̋

yɛ̀kɛ̋-k

zlȅi̋i̋.

Jean

3sg\pst

water

do-caus[pfv.hod]

clear

(23c)

OK Zá̰

ɛ́

yi̋

gȉȅ-kà

zlȅi̋i̋.

Jean

3sg\pst

water

become-caus[pfv.hod]

clear

(23d)

[Zá̰

y

nɔ̰̀]

[yi̋

yȅ

zlȅi̋i̋].

Jean

3sg\pst

s/he/it

give[pfv.hod]

water

3sg

ipfv\be

clear

(23e)

* Zá̰

ɛ́

yi̋

yè-k

zlȅi̋i̋.

Jean

3sg\pst

water

be-caus[pfv.hod]

clear

66‘Jean made the water clear.’

(24)

Ɛ̏

yȅȅ

blȅgbḭ̏.

3sg

ipfv\be

king

67‘He’s king.’

(25a)

Ɛ́

yɛ̀kɛ̋

blȅgbḭ̏.

3sg\pst

do[pfv.hod]

king

(25b)

Ɛ́

gȉȅ

blȅgbḭ̏.

3sg\pst

become[pfv.hod]

king

68‘He became king.’

(26a)

W

yɛ̀kɛ̋

blȅgbḭ̏.

3pl\pst

s/he/it

do[pfv.hod]

king

(26b)

W

yɛ̀kɛ̋-k

blȅgbḭ̏.

3pl\pst

s/he/it

do-caus[pfv.hod]

king

(26c)

W

gȉȅ-kà

blȅgbḭ̏.

3pl\pst

s/he/it

become-caus[pfv.hod]

king

69‘They made him king.’

2. Causative suffixation

2.1. Morphology: form, productivity, and combinability

  • 18  In rapid speech it is reduced to ‑à/-ȁ.
  • 19  A foot, understood as a prosodic unit larger than syllable and lesser than a phonological word, pr (...)
  • 20  Cf. the same behaviour for a causative from a |CV1.V2| verb: sȉȅkà ‘spoil (tr.)’ (from sȉȅ(...)

70The causative suffix k/‑kȁ18 is an inseparable morpheme; it is the only true verbal suffix attested in Gban. The lexical tone of the suffix is low after an ultra-low tone of the stem and ultra-low after other tones. Cf. lȉ ‘turn (intr.)’ lȉ-kà ‘turn (tr.)’; wlɔ̰̋ ‘cool down (intr.), freeze (intr.)’ → wlɔ̰̋-kȁ ‘cool (tr.), freeze (tr.)’; yà ‘go’ → yà-kȁ ‘make go / let go’. When attached to a verb, the suffix forms an additional prosodic foot19 and participates in the tonal grammatical modifications as part of the verb. That is, a suffixal causative derived from a one-foot verb behaves tonally in the same way as a simplex two-foot verb. Cf. vò‑kȁ ‘finish (tr.)’ (from vò ‘finish (intr.), come to an end’): vȍ‑kȁ ‘finish-caus\pfv.hest’, vȍkä ‘finish‑caus\pfv.preh’, bȅa̋ vőkȁ ‘work ipfv\finish‑caus’.20 Compare a simplex two-foot verb yɛ̀kɛ̋ ‘do’: yɛ̏kɛ̏ ‘do\pfv.hest’, yɛ̏kɛ̈ ‘do\pfv.preh’, nɔ̰̌ yɛ̋kɛ̏ ‘thing ipfv\do’.

  • 21  They are very similar to the so-called phrasal verbs, or particle verbs, in English, such as give (...)
  • 22  In these alternative forms, inflectional tonal changes continue to occur on the verbal part (verb (...)

71With verbs that contain a detachable adverbial-like particle, such as yɛ̏ tɔ̰́ ‘sit down’ or lȉ kɔ̰́ ‘return (intr.)’,21 the causative suffix normally attaches to the verbal (first) element: yɛ̏kà tɔ̰́ ‘seat, make sit down’, lȉ-kà kɔ̰́ ‘return (tr.), make return’. However, alternative forms (said to be more characteristic of younger speakers) are also possible, with the suffix attached only to the particle: yɛ̏ tɔ̰́-kȁ, lȉ kɔ̰́-kȁ, or with the repetition of the suffix: yɛ̏kà tɔ̰́-kȁ, lȉ-kà kɔ̰́-kȁ.22

72In general, causative suffixation can be considered productive. It is possible with borrowed verb roots, cf. vòtè ‘vote’ (< Fr. voter) vòtè‑kȁ ‘make vote’. Overall, more than 90 causative lexical items have been attested in my data.

73At the same time, some suffixal causatives or combinations with them are idiomatic, cf. lá-kȁ ‘release, let escape; miss’, which seems to be derived from lá ‘be soiled, not clean’. Cf. also ȁ yȉ‑kà X nɛ̏ ‘notify X (in advance, the day before)’ < yȉ ‘pass a night’ — lit. ‘make it pass a night on X’. Also, for several attested suffixal causatives, no corresponding simplex anticausatives can be established, cf. kpȁ(-)kà ‘grill, roast’ in the absence of *kpȁ.

  • 23  With the exception of zè say’, which seems to allow a synonymous causative zè-kȁ ‘say’ (i.e. a (...)

74Causative suffixation is generally possible with both intransitive verbs, like dő ‘light up (intr.)’ dkȁ ‘light (tr.)’ (9a–b), and (mono)transitive verbs, like mɛ̰̏ ‘beat’ mɛ̰̏-kà ‘make beat’ (10a–b) (their argument coding patterns will be discussed in 2.3). This also includes labile verbs, like lȉbi̋ ‘go out; put out’ lȉbi̋-kȁ ‘make put out’ (more details below). At the same time, causative suffixation seems to be impossible for ditransitive verbs: nɔ̰̀give’, bȁ ‘give back, hand over’, gḭ̏ ‘show’.23 This can be compared to what is reported for, e.g., Abkhaz, Basque, Songhay, and Rawang, where causatives too can be derived from intransitive and transitive but not ditransitive verbs (Kulikov 2001: 888; Dixon 2000: 57, 65).

  • 24  However, gà-kȁ <die-caus> is nevertheless possible in the idiomatic expression gbí…gà <breast…die> ‘forget (...)

75Additionally, there are some cases of seemingly arbitrary (lexical) restrictions. Causative suffixation is found to be impossible with some non-labile transitive (cf. zɛ̀ ‘kill’), some non-labile intransitive (cf. gà ‘die’),24 and some labile verbs (cf. passive-labile bɛ̋ ‘take’).

76With regard to causatives from transitive verbs, it should be noted that my second consultant, for some reason, tended to avoid them, often judging such causatives to be less than fully acceptable and, in particular, not allowing causatives from transitives with animate patients: mɛ̰̏ ‘beat’, wlé ‘pierce’, ki̋ő ű ‘bite’. However, my first consultant freely produced and allowed causatives from transitives, including those with animate patients, cf. mɛ̰̏ ‘beat’ (10b).

77Labile verbs, as has already been said, in general allow causative suffixation. Interestingly, it seems that for (anticausative-)labile verbs usually only one of the two logically possible interpretations of the suffixal causative is available: either a causative-of-intransitive interpretation or a causative-of-transitive interpretation, but not both.

78For instance, on the one hand, we attest only the causative-of-transitive interpretation in case of the suffixal causative from kpě ‘collapse, come to ruin; destroy’ kpě-kȁ ‘make destroy’, not *‘destroy’ (27a–d), and in case of the suffixal causative from tà ‘close (intr.); be closed; close (tr.)’ tà‑kȁ ‘make close’ and not *‘close (tr.)’ (28a–d).

(27a)

Kǒ

[ɛ́]

kpě.

fence

3sg\pst

collapse[pfv.hod]

79‘The fence collapsed.’

(27b)

Ɔ́

kǒ

kpě.

3pl\pst

fence

collapse[pfv.hod]

80‘They destroyed the fence.’

(27c)

* Ɔ́

kǒ

kpě-kȁ.

3pl\pst

fence

collapse-caus[pfv.hod]

81Int.: ‘They destroyed the fence.’

(27d)

OK Yë

ḭ̀

kǒ

kpě-kȁ.

3sg\pst

I

fence

collapse-caus[pfv.hod]

82‘He made me destroy the fence.’

(28a)

Vle̋

ɛ́

tà.

door

3sg\pst

close[pfv.hod]

831. ‘The door closed [because of the wind].’

842. ‘The door was closed [by someone].’

(28b)

Zá̰

ɛ́

vle̋

tà.

Jean

3sg\pst

door

close[pfv.hod]

85‘Jean closed the door.’

(28c)

* Zá̰

ɛ́

vle̋

tà-kȁ.

Jean

3sg\pst

door

close-caus[pfv.hod]

86Int.: ‘Jean closed the door.’

(28d)

OK Yë

ḭ̀

vle̋

tà-kȁ.

3sg\pst

I

door

close-caus[pfv.hod]

87‘He made me close the door.’

88On the other hand, we attest only the causative-of-intransitive interpretation in case of the suffixal causative from ylɛ̰̀ ‘enter; put (inside), insert’ ylɛ̰̀-kȁ ‘put (inside), insert’, not *‘make put (inside)’ (11a–c), (29), and in case of the suffixal causative from ná̰(=é) ‘stick (to) (intr.); stick (tr.), glue’ ná̰-kȁ(=è) ‘stick (tr.), glue’, not *‘make glue’ (30a–d).

(29)

* M

a̰̋

gbő

ylɛ̰̀-kȁ

yi̋

di̋.

1sg\pst

s/he/it

canoe

enter‑caus[pfv.hod]

water

under

89Int.: ‘I made him sink my canoe (“made him put my canoe under the water”).’

(30a)

Vɛ́lȅ

ɛ́

ná̰=

ta̋blȍ

là.

glass

3sg\pst

stick[pfv.hod]=loc1

table

on

90‘The glass stuck to the table [on its own].’

(30b)

Zá̰

ɛ́

vɛ́lȅ

ná̰=

ta̋blȍ

là.

Jean

3sg\pst

glass

stick[pfv.hod]=loc1

table

on

91‘Jean glued the glass to the table.’

(30c)

Zá̰

ɛ́

vɛ́lȅ

ná̰-kȁ=

ta̋blȍ

là.

Jean

3sg\pst

glass

stick-caus[pfv.hod]=loc1

table

on

92‘Jean glued the glass to the table.’

(30d)

* Ḭ̋

Zá̰

vɛ́lȅ

ná̰-kȁ=

ta̋blȍ

là.

1sg\pst

Jean

glass

stick-caus[pfv.hod]=loc1

table

on

93Int.: ‘I made Jean the glue glass to the table.’

94However, there also seem to exist rarer cases where both causative-of-intransitive and causative-of-transitive interpretations are available, cf. fà ‘be/become many/much; fill’ fà-kȁ ‘fill; make fill’:

(31a)

Yi̋

ɛ́

fà

vȕkɔ̏

nɛ̰̏.

water

3sg\pst

be_many[pfv.hod]

pot

at

95‘Water filled the pot (lit. ‘Water became much/full in the pot’).’

(31b)

Mɔ̰̈bì

ɛ́

yi̋

fà

vȕkɔ̏

nɛ̰̏.

Monbi

3sg\pst

water

be_many[pfv.hod]

pot

at

96‘Monbi filled the pot with water.’

(31c)

OK Mɔ̰̈bì

ɛ́

yi̋

fà-kȁ

vȕkɔ̏

nɛ̰̏.

Monbi

3sg\pst

water

be_many-caus[pfv.hod]

pot

at

97‘Monbi filled the pot with water.’

(31d)

OK Ḭ̋

Mɔ̰̈bì

yi̋

fà-kȁ

vȕkɔ̏

nɛ̰̏.

1sg\pst

Monbi

water

be_many-caus[pfv.hod]

pot

at

98‘I made Monbi fill the pot with water.’

99In cases such as (30b–c) and (31a–d) (also (11b–c) above), we observe synonymy of a simplex labile verb in its transitive use (i.e. fà ‘fill’, ná̰=é ‘glue’) and a cognate suffixal causative in a causative-of-intransitive interpretation (i.e. fà-kȁ ‘fill’, ná̰-kȁ=è ‘glue’). Sometimes they appear to be fully synonymous (as in the examples above), but at least sometimes there seem to be subtle semantic differences between the two verbs; this will be discussed in 2.2.

  • 25  Unlike Kakabe (Vydrina 2011: 191) or Mandinka (Creissels & Sambou 2013: 387–388), in which derived (...)

100Suffixal causatives themselves apparently do not allow intransitive uses in Gban:25 intransitive clauses with suffixal causatives are absent from texts and are judged infelicitous by the language consultants (cf. * Ḭ̋ yà-kȁ <1sg\pst go-caus[pfv.hod]>, * Ḭ̋ ylɛ̰̀-kȁ <1sg\pst enter-caus[pfv.hod]>).

101No double suffixal causatives (*Verb-ka-ka) have been attested.

2.2. Semantics

102Suffixal causatives can express a broad spectrum of semantic types of causation: direct (immediate physical, contact causation) (32)–(34); indirect verbal (35)–(37); indirect mediated: by a third person (35) or by supernatural forces (38); assistive: help (39)–(40) or teaching (41); permissive (35). Apart from animate volitional causers, suffixal causatives can also be used for situations with animate non-volitional (42)–(43) as well as inanimate causers (44)–(46):

(32)

Ɛ̀kȅ

yȍkȁ

ɛ̀

kwí=

nḭ̋nɛ̰́, …

2sg‑cond.aff

s/he/it

ipfv\pass‑caus

thou

back=on

now

103‘If you (sg.) pass it (the calabash of wisdom) to your back, {won’t it be easier for you to climb?}’ (lit. ‘If you make it pass to your back…’)

(33)

Sɔ̀kȕ

ɛ́

yűkwȉ

lɛ́kȁ.

Soku

3sg\pst

tree

bend‑caus[pfv.hod]

104‘Soku bent a twig.’

(34)

Mɔ̰̈bì

yi̋

főa̋kȁ.

Monbi

3sg

water

ipfv\heat‑caus

105‘Monbi heated the water.’

(35)

Ḭ̋

Zá̰

yàkȁ.

1sg\pst

Jean

go‑caus[pfv.hod]

106‘I made Jean go.’ The following readings are available:

1071) ‘I sent Jean, I told Jean to go [e.g. instead of me to some event].’

1082) OK ‘I had Jean go [by asking my friend to ask him].’

1093) OK ‘I chased/turned Jean out [against his will].’

1104) OK ‘I allowed Jean to go.’

(36)

Sɔ̀kȕ

yë

ḭ̀

sà

sḭ̋kȁ.

Soku

3sg\pst

I

song

sing‑caus[pfv.hod]

111‘Soku made me sing’ (OK asked me’; * ‘forced me by menacing me’).

(37)

Sɔ̀kȕ

ɛ́

mṵ̋=nṵ̋

mɛ̰̌mɛ̰̌

bɔ̀kȁ.

Soku

3sg\pst

person=pl

many

arrive‑caus[pfv.hod]

kè=ɛ̀

sɔ̋=.

each_other=at

s/he/it

gen

yard=in

112‘Soku invited (lit. ‘made come together’) many people to his house.’

(38)

Yë

ḭ̀

gwà̰kȁ

gbɔ̀=.

3sg\pst

I

get_lost‑caus[pfv.hod]

bush=in

113‘He made me [by sorcery] get lost in the bush.’

(39)

M

zȉkà

lɔ̌

nɛ̰̏.

1sg\pst

s/he/it

descend‑caus[pfv.hod]

palm_tree

at

114‘I helped him/her get down from the palm tree.’

(40)

Ḭ̋

kǎkwȉnḭ̋

ta̋kȁ.

1sg\pst

old‑dim

walk‑caus[pfv.hod]

115‘I helped the old man walk [I was supporting him as he walked].’

(41)

Ɔ̏

Zá̰

vútúkȁ

(yi̋=ő).

3pl

Jean

ipfv\swim‑caus

water=in

116‘They teach Jean to swim.’ / ‘Jean is taught to swim.’

(42)

Mḭ̋

ḭ̀

dɛ̌

dȅ

bòkȁ.

1sg\pst

I

father

passage

remove‑caus[pfv.hod]

117~‘I made my father cut a path (through the woods).’ / ~‘Because of me, my father cut a path (through the woods).’ (He had to do so because of the very fact that I was with him.)

(43)

Sɔ̀kȕ

‑lì

gwà̰kȁ.

Soku

3sg-foc[pst]

we

get_lost‑caus[pfv.hod]

118We got lost because of Soku [he didn’t succeed as our guide].

(44)

Yèvě

yi̋

főa̋kȁ.

sun

3sg

water

ipfv\heat‑caus

119‘The sun is heating the water.’

(45)

Dȉ

bȅbèɛ̀

dò

yë

ḭ̀

vòà-kȁ

wà

dḭ́á̰.

noise

bad

one

3sg

I

get_up‑caus[pfv.hod]

hour

middle

120‘A strange noise woke me up in the middle of the night.’

(46)

Gbíàlȉȅlè

tȅnḭ̀

nɛ̰́

yě

ka̋-kȁ.

joy

3sg

child

this

laugh

ipfv\put‑caus

121Joy is making the child laugh.’

122At the same time, coercive interpretation (in which causation clearly contradicts the will of the causee) appears to be uncharacteristic of suffixal causatives. It is sometimes accepted, usually not as a first reading (35), but more often it is rejected by the speakers (36). However, this interpretation can be lexically coerced (pun not intended) by the addition of the adverbial …glȉgbó yɛ̋ ‘by force / under compulsion’ (36′). In general, it is more appropriate in these cases to use a periphrastic construction specialized for expressing coercion: the construction with the verb kpɔ̋ ‘spread’, see ex. (15)–(16) in 1.4.

(36′)

Sɔ̀kȕ

yë

ḭ̀

sà

sḭ̋kȁ

#(glȉgbó

yɛ̋).

Soku

3sg\pst

I

song

sing‑caus[pfv.hod]

force

with

123‘Soku made me sing [by force].’

  • 26  This discussion and the suggested semantic components/restrictions listed below are preliminary; t (...)

124As was mentioned in 2.1, in the case of causatives from labile verbs we sometimes observe synonymy between a simplex labile verb and a cognate suffixal causative (as in e.g. fàfà-kȁ ‘fill’, ná̰=éná̰-kȁ=è ‘glue’). In some of these cases, the two verbs seemed to be completely synonymous in my data, as in the examples (11b–c), (26a–b), (30b–c) and (31a–d) above. In some other cases, however, there were some semantic contrasts between the two: while the simplex verb behaved neutrally, the causative verb seemed to express additional semantic components or impose additional semantic restrictions.26

125One of the possible components/restrictions introduced by the causative marker concerns the definiteness/deicticity of the patient (note the use of the postpositive demonstrative ~ definite quasi-article bɛ̏):

(47a)

Zá̰

ɛ́

wè

wṵ̋

(bɛ̏)

ylé.

Jean

3sg\pst

animal

oil

there

melt[pfv.hod]

126‘Jean melted the fat.’

(47b)

Zá̰

ɛ́

wè

wṵ̋

?(bɛ̏)

ylé‑kȁ.

Jean

3sg\pst

animal

oil

there

melt‑caus[pfv.hod]

127‘Jean melted the~that fat.’ (Omission of bɛ̏ was judged less felicitous; a postpositive demonstrative was suggested in the French translation to highlight the difference in meaning: …la graisse-là.)

(48a)

Zá̰

ɛ́

lȁdíó

wi̋.

Jean

3sg\pst

radio

break[pfv.hod]

128‘Jean broke the radio.’

(48b)

Zá̰

ɛ́

lȁdíó

*(bɛ̏)

wi̋‑kȁ.

Jean

3sg\pst

radio

there

break‑caus[pfv.hod]

129‘Jean broke the~that radio.’ (Omission of bɛ̏ was judged infelicitous; a demonstrative was used in the French translation: …la radio qui est là là ‘that radio that’s there’.)

130Another attested effect was related to quantification. In (49), the translations suggest that the causative version conveys an additional idea of ‘much’. In (50), the causative (with a causative-of-intransitive interpretation) was allowed only together with the semi-auxiliary verb yɛ̰̀ ‘do all, do completely’ (note that in this case the addition of bɛ̏ was not enough to make it felicitous).

(49a)

OK Zá̰

ɛ́

tɔ̏‑nḭ̋

kpɔ̋=è

ta̋blȍ

lȁ.

Jean

3sg\pst

cloth‑dim

spread[pfv.hod]=loc1

table

on

131‘Jean spread a/the [piece of] cloth on the table.’

(49b)

OK Zá̰

ɛ́

tɔ̏‑nḭ̋

kpɔ̋‑kȁ=è

ta̋blȍ

lȁ.

Jean

3sg\pst

cloth‑dim

spread‑caus[pfv.hod]=loc1

table

on

132‘Jean spread the rags / [much] cloth on the table.’

(50a)

Zá̰

ɛ́

yűkwȉ

mḭ̋a̰̋.

Jean

3sg\pst

tree

split[pfv.hod]

133‘Jean chopped the wood.’ (= (8c))

(50b)

* Zá̰

ɛ́

yűkwȉ

(bɛ̏)

mḭ̋a̰̋‑kȁ.

Jean

3sg\pst

tree

there

split-caus[pfv.hod]

134Int.: ‘Jean chopped the wood.’

(50c)

Zá̰

ɛ́

yűkwȉ

mḭ̋a̰̋‑kȁ

yɛ̰̀.

Jean

3sg\pst

tree

split‑caus[inf]

do_all[pfv.hod]

135‘Jean finished chopping the wood.’ ~ ‘Jean chopped all the wood.’

136Less reliably, several other factors were observed that facilitated the use of the causative version (while being irrelevant for the simplex version). These include the presence of an overtly expressed instrument (‘Jean pierced the snake (with a stick)’), a benefactive participant (‘I bought some food (for myself)’), or an external cause/causer (‘(Because of me,) Jean broke the radio’).

2.3. Argument coding. The “double object” construction

137Let us now turn to the ways in which arguments are coded in the clauses with suffixal causatives. It is important to mention in advance that Gban has a very rigid S‑PM‑O‑V word order: the subject is the NP that immediately precedes the predicative markers (PMs) and the direct object is the NP that immediately precedes the predicate.

1382.3.1. Clauses with suffixal causatives derived from intransitive verbs (51b) are identical to plain transitive clauses (52). The causer NP occupies the subject position and the causee NP the direct object position.

(51a)

Sa̋ka̋

ɛ́

fóá.

rice

3sg\pst

heat[pfv.hod]

139‘The rice is hot [= has heated].’

(51b)

Sɔ̀kȕ

ɛ́

sa̋ka̋

fóá‑kȁ.

Soku

3sg\pst

rice

heat‑caus[pfv.hod]

140‘Soku heated the rice.’

(52)

Sɔ̀kȕ

ɛ́

ba̋ba̋

zɛ̀.

Soku

3sg\pst

sheep

kill[pfv.hod]

141‘Soku killed the sheep.’

  • 27  No parallel for such “split” of the preverbal position is found in other Mande languages (Vydrin 2 (...)

1422.3.2. Clauses with suffixal causatives derived from transitive verbs exhibit, however, a peculiar pattern. In such clauses, instead of a single preverbal position for a direct object NP, two preverbal positions are observed. The first one is occupied by the causee NP (“former subject”) and the second by the NP expressing the original O-participant (“former direct object”).27

  • 28  As noted in 2.1, such causatives from transitive verbs tended not to be fully acceptable for my se (...)

143Two illustrations are given below: a sentence with a suffixal causative mɛ̰̏kà ‘make beat’ derived from transitive mɛ̰̏ ‘beat’ (53b) and a sentence with a suffixal causative bȅkà ‘make eat, feed’ derived from transitive bȅ ‘eat’ (54b).28

(53a)

Mṵ̋

dò

ɛ́

tȅnḭ̋

mɛ̰̏.

person

one

3sg\pst

child

beat[pfv.hod]

144‘Someone beat the child.’

  • 29  In this section, I will mark the boundaries of the two NPs with square brackets for clarity.

(53b)

Zá̰á̰

ɛ́

[mṵ̋

dò]29

[tȅnḭ̋]

mɛ̰̏‑kà.

Jean

3sg\pst

person

one

child

beat‑caus[pfv.hod]

145‘Jean made somebody beat the child.’

(54a)

Ḭ̋

wè

bȅ.

1sg\pst

meat

eat[pfv.hod]

146‘I ate meat.’

(54b)

Wö

[̀]

[wè]

bȅ‑kà.

3pl\pst

I

meat

eat‑caus[pfv.hod]

  • 30  For this single verb (therefore this is perhaps a case of lexicalization; this coding was rejected (...)

147‘They let me eat meat ~ They fed me meat.’30

  • 31  Instead of one of them occupying a more peripheral position, as, for example, in French: Je ferai (...)

148In other words, in such clauses both the former direct object (original O-participant) and the former subject (causee) occupy direct-object-like positions.31 This corresponds to what have been observed in some other languages and what has been called “double object” strategy in (Polinsky 1995), who discusses Tamil and Kinyarwanda, or syntactic doubling in (Kulikov 2001: 890–891). See also the “type (iii)” of argument marking in morphological causatives from transitives in (Dixon 2000: 50–52), who discusses Hebrew, Oromo, Yagua, and some other languages.

149Cf. an example from (Polinsky 1995: 142, 164):

150Kinyarwanda:

(55a)

Umukoobwa

y--uhagi-ye

umwaana.

girl

3sg-pst-wash-pfv

child

151The girl washed the child.’

(55b)

Umugore

y--uhag-iish-ije

umukoobwa

umwaana.

woman

3sg-pst-wash-caus-pfv

girl

child

152The lady made the girl wash the child.

153Polinsky (1995) shows that even in such cases syntactic tests may reveal certain differences in behaviour between the two NPs, showing that only one of them has a true direct object status. The same is discussed in (Dixon 2000: 50–52), who claims that this is the case for most languages with this pattern. He also notes that in these cases it is generally the causee (and not the original O-participant) which behaves as a true direct object.

154Unfortunately, most of the proposed tests — those involving passivization and reflexives — are inapplicable to Gban, due to its grammatical structure.

155Note: Polinsky (1995) also employed tests involving control of coreference. In Gban, such tests proved mostly inconclusive, except perhaps for control of the depictive: the latter seems to be possible for the causee (57) but not for the original O-participant (56b).

(56a)

Ḭ̋

vȕȅ

bȅ

kpṵ̏.

1sg\pst

banana

eat[pfv.hod]

unripe

156‘I ate the banana green.’

(56b)

??? M

[a̰̋]

[vȕȅ]

bȅ‑kà

kpṵ̏.

1sg\pst

s/he/it

banana

eat‑caus[pfv.hod]

unripe

157Int.: ‘I made him eat the banana green.’

(57)

OK Ḭ̋

[Zá̰]

[mà̰]

bȅ‑kà

gbɛ̏ɛ̏.

1sg\pst

Jean

food

eat‑caus[pfv.hod]

dirty

158‘I made Jean eat the food dirty [It was Jean who was dirty: I didn’t let him wash himself before eating].’

  • 32  It is important to rule out this possibility because the nouns/NPs inside the genitive constructio (...)

159There is strong evidence that these two preverbal nominals indeed occupy two separate clause-level positions and not a single one, i.e. they are not a head and its (preposed) dependent in a genitive construction,32 neither is the second nominal incorporated into the verb.

  • 33  This postposition is polysemous and also has benefactive and recipient-benefactive uses (‘for’).

1601) It is impossible to insert the genitive postposition mɔ̰́33 between the two nominals (58), which is optional but possible in the true genitive construction (cf. ḭ̀ mɔ̰́ gbő ‘my canoe’ in (11b)).

(58)

Ɔ́

[Sɔ̀kȕ]

(*mɔ̰́)

[wè]

bȅ‑kà.

3pl\pst

Soku

for

meat

eat‑caus[pfv.hod]

161‘They fed Soku meat.’

1622) Not only the first (which is insignificant) but also the second position can host focus-marked NPs (59), NPs with modifiers (60), specific definite/indefinite NPs (61), including proper names (60).

(59)

Mḭ̋

[ḭ̀

dɛ̌]

[dȅ

lì]

bòkȁ.

1sg\pst

I

father

passage

foc_obj

remove‑caus[pfv.hod]

163It was a path (through the woods) that I made my father cut.’

(60)

Zá̰á̰

ɛ́

[mṵ̋

d]

Jean

3sg\pst

person

one

[Mɔ̰̈bì

tȅnḭ̋

kléènḭ̋

*

*ḭ̀]

mɛ̰̏kà.

Monbi

child

small

s/he/it

I

beat‑caus[pfv.hod]

164‘Jean made someone beat Monbia small child*him*me.’

(61)

Zá̰á̰

yë

[ḭ̀]

[mṵ̋

dò]

mɛ̰̏kà.

Jean

3sg\pst

I

person

one

beat‑caus[pfv.hod]

165‘Jean made me beat someone.’

1663) Most importantly, a clause-level particle vȍ ‘already, quickly, early’ can be placed between, before and after both positions (62a–с).

(62a)

Ɔ́

[Sɔ̀kȕ]

[blú

bɛ̏]

vȍ

bȅkà.

3pl\pst

Soku

bread

there

quickly

eat‑caus[pfv.hod]

(62b)

Ɔ́

[Sɔ̀kȕ]

vȍ

[blú

bɛ̏]

bȅkà.

3pl\pst

Soku

quickly

bread

there

eat‑caus[pfv.hod]

(62c)

Ɔ́

vȍ

[Sɔ̀kȕ]

[blú

bɛ̏]

bȅkà.

3pl\pst

quickly

Soku

bread

there

eat‑caus[pfv.hod]

167‘They quickly made Soku eat bread.’

168Compare the behaviour of the particle with a true genitival NP in the direct object position (63a–c).

(63a)

Ḭ̋

[Zá̰

sa̋]

vȍ

yò wa̰̋.

1sg\pst

Jean

house

quickly

see[pfv.hod]

(63b)

* Ḭ̋

[Zá̰

vȍ

sa̋]

yò wa̰̋.

1sg\pst

Jean

quickly

house

see[pfv.hod]

(63c)

Ḭ̋

vȍ

[Zá̰

sa̋]

yò wa̰̋.

1sg\pst

quickly

Jean

house

see[pfv.hod]

169‘I already saw Jean’s house.’

  • 34  Unfortunately, I was able to test the infelicitousness of genitival NPs with nouns as possessors i (...)

170At the same time, the second preverbal position turns out to have some syntactic limitations that the first one does not. As demonstrated in (60) above, the second position cannot be occupied by a personal pronoun. It also cannot host a genitival NP (neither with a personal pronoun nor with a noun as possessor),34 cf. (64)–(65).

(64)

* Zá̰á̰

ɛ́

[mṵ̋

dò]

[ḭ̀

yènḭ̋]

mɛ̰̏‑kà.

Jean

3sg\pst

person

one

I

sibling

beat‑caus[pfv.hod]

171Int.: ‘Jean made somebody beat my brother.’

(65a)

Ḭ̋

[Sɔ́kȕ]

[mlɛ̰̋]

kɛ̀-kȁ.

1sg\pst

Soku

hair

make‑caus[pfv.hod]

172‘I made Soku do [my~his] hair.’

(65b)

* Ḭ̋

[Sɔ́kȕ]

[Zá̰

/ ḭ̀

/ ȁ

/ ȁ

mlɛ̰̋]

kɛ̀-kȁ.

1sg\pst

Soku

Jean

I

s/he/it

s/he/it

hair

make‑caus[pfv.hod]

173Int.: ‘I made Soku do Jean’s/my/his hair.’

1742.3.3. Additionally, clauses with suffixal causatives from transitives sometimes exhibit a different pattern, which may be described as causee omission or demotion. In such (rarer) cases, there is only one preverbal position, and it is occupied by the NP expressing the original O-participant (“former direct object”), while the causee NP (“former subject”) is either completely missing or is demoted to an oblique position.

175(An inverse pattern for causatives from transitives, i.e. a single preverbal position occupied by the causee NP, has not been attested and has been judged infelicitous, cf. the last interpretation in (66).)

  • 35 I should mention that this coding pattern/construction is rare in my data, it was not always judged (...)

176In the first group of cases,35 the causee NP is missing and the causee participant is construed as impersonal (or known from the context); this pattern is similar to French J’ai fait réparer ma voitureI had my car repaired, I made [someone] repair my car’.

(66)

Mɛ̰̋

ɛ̰̀

mɛ̰̏-kà.

1sg\pst

thou

beat‑caus[pfv.hod]

1771. ‘Because of me, you were beaten [by somebody].’ ~ ‘I had you beaten.’

1782. ‘I made him beat you.’

1793. * ‘I made you beat somebody.’

(67)

Zá̰á̰

y

nḭ̋

bàtízé-kȁ.

Jean

3sg\pst

s/he/it

child

baptize‑caus[pfv.hod]

180‘Jean had his son baptized.

(68)

M

ḭ̋

sa̋

kɛ̀‑kȁ.

1sg\pst

I

house

make‑caus[pfv.hod]

181‘I had my house built / a house built for me [It was built by workers].

182In the second group of cases, the causee NP is not omitted, but demoted to an oblique position: a postverbal position marked by the comitative~instrumental postposition yɛ̋ ‘with’. In all of such examples in my data, the demoted causee is inanimate. Expression of an animate agentive causee in that position was judged infelicitous, cf. (70c).

(69a)

Yi̋

ɛ́

tȅnḭ̋

dò

zɛ̀.

water

3sg\pst

child

one

kill[pfv.hod]

183‘A child drowned.’ (lit. ‘Water killed a child.’)

(69b)

Y

zɛ̀-kȁ

yi̋

yɛ̋.

3sg\pst

s/he/it

kill‑caus[pfv.hod]

water

with

184‘S/he drowned him/her.

(69c)

# Y

zɛ̀

yi̋

yɛ̋.

3sg\pst

s/he/it

kill‑caus[pfv.hod]

water

with

185Int.: ‘S/he drowned him/her.’ (But possible with the literal reading: ‘S/he killed him with water [by giving too much water to drink].’)

(70a)

Gwlɛ̰́

ɛ́

Zá̰á̰

mɛ̰̏.

rain

3sg\pst

Jean

beat[pfv.hod]

186‘Jean got caught / soaked in the rain.’ (lit. ‘Rain hit Jean.’)

(70b)

Zá̰á̰

ɛ́

tȅnḭ̋

mɛ̰̏‑kà

gwlɛ̰́

/ nḭ̋nḭ̋

/ yèvě

yɛ̋.

Jean

3sg\pst

child

beat‑caus[pfv.hod]

rain

cold

sun

with

187‘Jean made it so that the child got hit by the rain [i.e. was soaked] / the cold [i.e. froze] / the sun [i.e. got hot] [It may have been unintentional].

(70c)

* Zá̰á̰

ɛ́

tȅnḭ̋

mɛ̰̏‑kà

mṵ̋

dò

yɛ̋.

Jean

3sg\pst

child

beat‑caus[pfv.hod]

person

one

with

188Int.: ‘Jean made somebody beat the child.’ (cf. (53b))

(70d)

Y

blɛ̋

mɛ̰̏

gwlɛ̰́

yɛ̋.

3sg\pst

s/he/it

self

beat[pfv.hod]

rain

with

189‘He got himself soaked in the rain [e.g. by forgetting his umbrella].

190(There is also a very similar series of examples in my data with the expression “wḭ́ X zɛ̀” (lit. ‘wine kills X’) ‘X gets drunk’.)

  • 36  Cf., however, (69c).

191As can be seen from (70d), an identical postpositional coding is in fact available in some of these cases even with the corresponding simplex verb without the causative suffix.36 The inanimate participant (‘rain’) can no longer be analyzed here as a causee; it probably should be analyzed as an instrument/means.

2.4. Possible diachrony

  • 37  Comparative data have been of little help, unfortunately. Although some other Mande languages also (...)
  • 38 I am grateful to Denis Creissels [p. c.] for the original idea.

192A diachronic hypothesis based primarily on internal reconstruction37 can be proposed to explain the observed phenomena, in particular the “double object” construction.38

193The hypothesis can be outlined as follows:

1941) The causative suffix ‑kà was once a causative verb (similar to English make or French faire).

1952) This causative verb took as its object an (unmarked) nominalization of the content verb and it was possible for this nominalization (in the case of transitive content verbs) to have two left-side nominal dependents of its own.

196So the whole construction would look as in (80) (the superscript asterisk denoting reconstruction).

197(80) Zá̰á̰ ɛ́ [[mṵ̋ d] [Mɔ̰̈b] mɛ̰̏nmlz] ka

198lit. ‘Jean [somebody’s Monbi’s beating] caused.’

1993) Later, this construction was reanalyzed as a single finite causative-suffixed verb (into which the nominalized content verb and the causative verb merged) with two NP positions, now clause-level.

200That would give us the modern construction in (81).

201(81) > Zá̰á̰ ɛ́ [mṵ̋ d] [Mɔ̰̈b] mɛ̰̏‑kà

202lit. ‘Jean [somebody] [Monbi] beat-caused.’

203(i.e. ‘Jean made somebody beat Monbi.’) (= (60))

204Let us now try to substantiate the proposed scenario. Interestingly, almost all the necessary ingredients can be found independently even synchronically, in present-day Gban:

2051. Gban, like some other Mande languages (cf. Mano below), has periphrastic causative constructions, including one based on nominalization (albeit a marked one): the coercive construction with kpɔ̋ ‘spread’ (see 1.4). Let us reproduce here example (16) in (82) as an illustration:

(82)

Sɔ̀kȕ

ɛ́

[sà

sḭ̋=le̋]

kpɔ̋=le̋

ḭ̀

nɛ̰̀.

Soku

3sg\pst

song

sing=nmlz

spread[pfv.hod]=loc2

I

at

206‘Soku forced me to sing.’ (lit. ‘Soku laid the song-singing on me.’)

  • 39  So one of its varieties is similar to the variety of the kpɔ̋-construction with a subjunctive subo (...)

207Mano, another South Mande language, has as its default means of expressing causative a periphrastic construction based on a transitive causative verb kɛ̄ (< ‘do’) and a subordinate clause (with subjunctive, conjoint, or agreeing marking);39 the direct object position of the causative verb is filled with a dummy pronoun or with the original S- or O-participant. Cf. (83)–(84) (Khachaturyan 2015: 140–141).

(83a)

Māà

kɛ̄

āà

wɛ́ì

dɔ̀kɛ̄

kɛ̀lɛ̀.

1sg.perf>3sg

do

3sg.perf

money

give

thou

to

208I made him give you money. (lit. ‘I did it, he gave you money.’)

(83b)

Māà

wɛ́ì

kɛ̄

dɔ̀kɛ̄

kɛ̀lɛ̀.

1sg.perf

money

do

3sg.sbjv>3sg

give

thou

to

209I made him give you money. (lit. ‘I did the money so that he gave it to you.’)

(84)

Lɛ́ɛ̀

yílí

kɛ̀

ɓā.

3sg.ipfv

tree

do:ipfv

3sg.sbjv

bear_fruit

210He makes the tree bear fruit. (lit. ‘He does the tree so that it bears fruit.’)

211A grammaticalization path from a causative verb to a causative suffix has been proposed by Denis Creissels for three other Mande languages. He argues that the causative suffixes ‑ndí (Mandinka, Soninke) and ‑ní (Bozo) grammaticalized from periphrastic causative constructions based on the verb *tin ‘do’ (Creissels 2021).

  • 40  The vowel change *ɛ > a seems unmotivated. However, a similar vowel alternation is found in Gban i (...)
  • 41  However, a path from ‘put’ to causative is not implausible. Consider the close meaning ‘spread (on (...)

212As for the exact lexical origin of a causative verb *ka that could serve as a source for the suffix ‑kà in Gban, there are currently two hypotheses. First, modern Gban has a verb kɛ̀ ‘make, construct; hunt, court; say, want’ (cognate to the Mano verb kɛ̄ discussed above), which is a perfect match semantically, but problematic phonologically.40 Second, there is a verb ka̋ ‘put (several objects), pour (tr.); fall (several objects), pour (intr.), enter (several objects), gather (intr.); feed, breed’, which is a perfect match phonologically, although is less straightforward semantically.41

2132. Gban has unmarked nominalizations (although they are syntactically more restricted than marked nominalizations with =le) and constructions with semi-auxiliary verbs (other than causative) that take such unmarked nominalizations as their direct objects.

  • 42  The unmarked nominalization is glossed here as ‘inf’, but it remains an open question whether it s (...)

214Cf. an example with the semi-auxiliary verb dɔ̀ ‘do well’:42

(85)

Ɛ̏

[fḭ̏ȁ̰=nṵ̏

dȁ]

dɔ̀.

3sg

tale=pl

compose[inf]

ipfv\do_well

215‘He tells tales well.’

  • 43  It is known to be possible in some other South Mande languages [Valentin Vydrin, p. c.], including (...)
  • 44  As can be seen from (86), the omissibility of the postpositional marking in the first dependent va (...)

2163. Nominalizations in Gban, as well as in some other Mande languages,43 can have two left-side nominal dependents, both of which may be unmarked. In Gban, it is attested for the marked nominalization in =le:44

(86)

[[Fíɛ́lȅ

OK/?? (mɔ̰́)]

[mṵ̋]

zɛ̏=l]

Pierre

for

person

kill=nmlz

217‘murder of a personO by PierreA

218In fact, the more precise syntactic analysis of this structure should be that the second nominal is the inner argument (direct object) of the nominalized clause, while the first nominal is an external genitival dependent in the NP headed by the nominalization:

(87)

[[Fíɛ́lȅ

(mɔ̰́)]

[[mṵ̋]

zɛ̏=l]]

Pierre

for

person

kill=nmlz

219‘murder of a personO by PierreA

220This analysis of nominalizations in Gban is corroborated by the behaviour of the clause-level particle vȍ ‘already, quickly, early’. This particle can be inserted before the second nominal (88) or before a single nominal corresponding to an O-participant (89). But it cannot be inserted (while still modifying the situation expressed in the nominalized clause) before the first nominal (88) or before a single nominal corresponding to an S-participant (90):

(88)

(* vȍ)

Fíɛ́lȅ

mɔ̰́

(OK vȍ)

mṵ̋

(OK vȍ)

zɛ̏=l

quickly

Pierre

for

quickly

person

quickly

kill=nmlz

221‘(quick) murder of a personO by PierreA

(89)

(OK vȍ)

mṵ̋

(OK vȍ)

zɛ̏=l

quickly

person

quickly

kill=nmlz

222‘(quick) murder of a personO

(90)

(* vȍ)

mṵ̋

(mɔ̰́)

(OK vȍ)

zɛ̏=l

quickly

person

for

quickly

kill=nmlz

223‘(quick) death of a personS

  • 45  They can, however, take genitival NPs beginning with nouns:

2244. Nominalizations in Gban (again, marked nominalizations in =le) for some reason cannot take a personal pronoun or an NP beginning with a personal pronoun45 as their second dependent (93)–(94) (only as their first (91) or their only (96) dependent), similarly to causatives from transitives, cf. (64)–(65) above.

(91)

[[

mɔ́]

[la̋lȁ]

kṵ̋a̰̋=le̋]

s/he/it

for

money

steal=nmlz

225‘theft of moneyO by him/herA

(92)

[[Fíɛ́lȅ

mɔ́]

[Zá̰]

zɛ̏=lè]

Pierre

for

Jean

kill=nmlz

226‘murder of JeanO by PierreA

(93)

* [[Fíɛ́lȅ

mɔ́]

[]

zɛ̏=lè]

Pierre

for

s/he/it

kill=nmlz

227Int.: ‘? hisO murder by PierreA

(94)

[[Fíɛ́lȅ

mɔ́]

[(*)

yènḭ̋]

zɛ̏=lè]

Pierre

for

s/he/it

sibling

kill=nmlz

228‘murder of his brotherO by PierreA

(95)

OK [[Zá̰

/ mṵ̋

(mɔ̰́)]

zɛ̏=lè]

Jean

for

kill=nmlz

229‘murder of JeanO/*A

(96)

OK [[

(mɔ̰́)]

zɛ̏=lè]

s/he/it

for

kill=nmlz

230‘hisO/*A murder’

  • 46  It should be noted that (marked) nominalizations of transitive verbs can also, at least sometimes, (...)

2315. Cases of causee omission are, too, compatible with the hypothesis. As was said in 2.3.3, with suffixal causatives from transitives we sometimes observe a single preverbal position, and if it is the case, it corresponds to the original O-participant (‘I made [somebody] beat you’) and not to the causee (* ‘I made you beat [somebody].’). The same is true for nominalizations: if there is a single nominal dependent, it must correspond to the O-participant, cf. (95)–(96) above.46

232So it is possible to directly derive the variety of the modern construction with causee omission (97b) from a variety of the reconstructed construction containing a nominalization with a single dependent (97a).

233(97a) Mɛ̰̋ [ɛ̰̀ mɛ̰̏nmlz] ka

234lit. ‘I [yourO beating] caused.’

235(97b) > Mɛ̰̋ ɛ̰̀ mɛ̰̏‑kà

236lit. ‘I [youO] beat-caused

237(i.e. ‘I had you beaten.’) (= (66))

2386. Finally, the proposed hypothesis fits well with the view that the Mande languages have historically undergone a mass reanalysis of nominalization-based constructions into finite ones, where clause-level NPs in a similar way originated in nominal dependents of nominalizations (Nikitina 2011; Khachaturyan & Nikitina 2016).

Conclusions

239The paper described how causative semantics is expressed in Gban: lexically (suppletive pairs and causative-anticausative lability), periphrastically, and, most importantly, morphologically: by the productive derivational suffix kà, with broad causative semantics.

240In the case of suffixal causatives from transitives, a peculiar “double object” coding pattern is observed, with two preverbal clause-level NP positions (the second of which cannot host a personal pronoun or a genitival NP, but can be separated by a clause-level particle).

241The suffixal causative construction is hypothesized to have grammaticalized from a construction based on a causative verb and an unmarked nominalization (as its object), which could itself have two left-side nominal dependents.

242The hypothesis is supported by a number of independent synchronic facts, including the syntactic properties of present-day (marked) nominalizations and the similar way in which personal pronouns behave in the suffixal causative construction and with nominalizations. The hypothesis also fits well with the general view that Mande languages have undergone a mass reanalysis of nominalization-based constructions into finite ones.

Abbreviations

2431/2/3 — 1st/2nd/3rd person

244aff — affirmative

245ag — agent noun

246caus — causative

247cond — conditional

248ctnv — continuative

249d — default determiner (Mandinka)

250dim — diminutive

251foc — subject focus

252foc_obj — direct object and nominal dependent focus

253gen — genitive

254hest — hesternal

255hod — hodiernal

256ind — indicative

257inf — infinitive

258ipfv — imperfective

259loc1/loc2 — locative applicative

260neg — negation

261nmlz — nominalization

262(n)pst — (non-)past

263perf — perfect

264pfv — perfective

265pl — plural

266poss — possessive

267preh — pre-hesternal

268purp — purpose

269refl — reflexive

270sbjv — subjunctive

271sg — singular

Haut de page

Bibliographie

References

Creissels, Denis. 2014. P-lability and radical P-alignment. Linguistics 52(4). 911–944.

Creissels, Denis. 2021. Antipassive derivation in Soninke (West Mande). In Katarzyna Janic & Alena Witzlack-Makarevich (eds.), Antipassive: Typology, diachrony, and related constructions, 293–314. John Benjamins. https://doi.org/10.1075/tsl.130.09cre.

Creissels, Denis & Pierre Sambou. 2013. Le mandinka. Phonologie, grammaire, textes. Paris: Karthala.

Dixon, Robert Malcolm Ward. 2000. A typology of causatives: form, syntax and meaning. In Robert Malcolm Ward Dixon & Alexandra Aikhenvald (eds.), A typology of causatives: form, syntax and meaning (Changing Valency: Case Studies in Transitivity), 30–83. Cambridge: Cambridge University Press.

Fedotov, Maksim. 2016. Adnominal predicative possessive construction and pragmatically “flexible” noun phrases in Gban. In Issledovanija po jazykam Afriki 6 (Исследования по языкам Африки. Выпуск 6) [Studies in African languages 6], 320–345. Moscow: Kliutch-S.

Fedotov, Maksim. 2017. Gban jazyk (Гбан язык) [Gban]. In Valentin Vydrin, Yulia Mazurova, Andrej Kibrik & Elena Markus (eds.), Jazyki mira: Jazyki mande (Языки мира: Языки манде) [Languages of the world: Mande languages], 902–999. St. Petersburg: Nestor-Historia.

Haspelmath, Martin. 1993. More on the typology of inchoative/causative verb alternations. In Bernard Comrie & Maria Polinsky (eds.), Causatives and transitivity (Studies in Language Companion 23), 87–120. Amsterdam: John Benjamins. https://doi.org/10.1075/slcs.23.05has.

Khachaturyan, Maria. 2015. Grammaire du mano. Mandenkan 54. 1–252.

Khachaturyan, Maria & Tatiana Nikitina. 2016. Syntactic reanalysis of nominalizations in Mande and its historical consequences. Presented at the Syntax of the World Languages VII, Mexico.

Kulikov, Leonid. 2001. Causatives. In Martin Haspelmath, Ekkehard König, Wulf Oesterreicher & Wolfgang Raible (eds.), Language typology and language universals: An international handbook, vol. 2, 886–898. Berlin: Walter de Gruyter.

Kuznetsova, Natalia. 2007. Le statut fonctionnel du pied phonologique en gouro. Mandenkan (43). 13–45.

Kuznetsova, Natalia & Olga Kuznetsova. 2021. Dictionnaire gouro-français avec un index français-gouro. Mandenkan 66. 3–186. https://doi.org/10.4000/mandenkan.2599.

Letuchij, Alexander. 2009. Towards a typology of labile verbs: Lability vs. derivation. In Alexandre Arkhipov & Patience Epps (eds.), New challenges in typology. Transcending the borders and refining the distinctions, 223–244. Berlin: Mouton de Gruyter.

Nikitina, Tatiana. 2011. Categorial reanalysis and the origin of the S-O-V-X word order in Mande. Journal of African Languages and Linguistics 32(2). 251–273.

Polinsky, Maria. 1995. Double objects in causatives: Towards a study of coding conflict. Studies in Language 19(1). 129–221.

Vydrin, Valentin. 2008. Metricheskaja stopa v jazykakh mande (Метрическая стопа в языках манде) [Metric foot in Mande languages]. In Alexander Arkhipov, V. Zakharov, Andrej Kibrik, Alexander Kibrik, Irina Kobozeva, Olga Krivnova, Ekaterina Ljutikova & Olga Fedorova (eds.), Fonetika i nefonetika. K 70-letiju Sandro Vasiljevicha Kodzasova (Фонетика и нефонетика. К 70-летию Сандро Васильевича Кодзасова) [Phonetics and non-phonetics. To the 70th anniversary of Sandro V. Kodzasov], 308–317. Moscow: Jazyki slav’anskikh kul’tur.

Vydrin, Valentin. 2017. Mande jazyki (Манде языки) [The Mande languages]. In Valentin Vydrin, Yulia Mazurova, Andrej Kibrik & Elena Markus (eds.), Jazyki mira: Jazyki mande (Языки мира: Языки манде) [Languages of the world: Mande languages], 16–45. St. Petersburg: Nestor-Historia.

Vydrin, Valentin. 2018. Mande languages. Oxford Research Encylopedias (Linguistics). https://oxfordre.com/linguistics/abstract/10.1093/acrefore/9780199384655.001.0001/acrefore-9780199384655-e-397?rskey=QOE1wp&result=43.

Vydrina, Alexandra. 2011. Лабильность в языке какабе [The distribution of lability in the verbal lexicon of Kakabe]. In Nikolaj N. Kazanskiy & Valentin Vydrin (eds.), Mandeica Petropolitana II (Acta Linguistica Petropolitana. Труды Института лингвистических исследований), vol. VII(2), 174–217. Санкт-Петербург: Наука.

Haut de page

Notes

1 I owe the original idea about the diachronic source of the “double object” construction to Denis Creissels. I would like to express my gratitude to my language consultants, Taki Oya Robert and Kouadio Magni Julienne. I am grateful to Valentin Vydrin and two anonymous reviewers for their constructive comments and criticism on the initial version of the paper. Finally, I would like to thank all my fellow Mande field team members.

2  I will understand causative in the broad sense (“causatives sensu latiore” in (Kulikov 2001: 888–889), i.e. including lexical causatives and biclausal causative constructions.

3  Gban (gbá̰; ISO 639-3 code: ggu, Glottocode: gagu1242) is a South Mande language spoken by ca. 60 000 people in the central part of Côte d’Ivoire. This study is based on the author’s field data from 2011–2019. A preliminary version of this study was published as part of the grammar sketch (Fedotov 2017).

4  As can be seen from the examples, gà ‘die’ cannot be used transitively (i.e. it is not labile), nor can it attach the causative suffix.

5  A peculiarity of the latter pair is that the anticausative member is already transitive: ‘learn something’. See a brief discussion in (Haspelmath 1993: 110).

6  Gban has four level tones and two contour tones: ultra-high (), high (), low (), and ultra-low (); rising () and mid-rising ().

7  It is the number of confirmed anticausative-labile verbs in a studied sample of several dozen verbs. The actual number of anticausative-labile verbs is presumed to be higher. The total number of verbs in my lexical database of Gban is 277.

8  It should be noted that it is indeed not always easy to distinguish between a passive and an anticausative interpretation, especially given the inability to express the agent. In most cases, however, the context can be used to determine whether an agent is still present in the situation (passive) or whether it is agentless (anticausative).

9  Henceforth, a superscript “OK” before an example indicates that the entire example was constructed by me in Gban and then approved by a speaker as felicitous (while an asterisk “*” or from one to three “?” indicate a greater or lesser degree of infelicitousness). Examples without “OK” were translated from French into Gban, suggested by a speaker in the course of the discussion, or taken from texts. The symbol “#” before an example indicates that the sentence is either a) grammatically correct, but was judged to be nonsensical, or b) cannot be used with the given interpretation (but may be felicitous with another interpretation, irrelevant to the discussion).

10  Cf. Kakabe, where anticausative-labile verbs can also be passive-labile, too (Vydrina 2011: 205).

11  Such cases of synonymy of labile simplex verbs and causatives will be discussed in more detail in 2.1 and 2.2.

12  Only morphological means have been considered, leaving out the periphrastic constructions, which will be discussed in 1.4.

13  Non-directed pairs are all pairs except causative and anticausative (directed) ones: labile, equipollent, and suppletive pairs (Haspelmath 1993: 91).

14  It seems that M. Haspelmath used 0.5 instead of zeros in all A/C calculations to avoid dividing by zero and zero ratios, hence the unexpected A/C values for Russian, German, Greek, and Indonesian. If we followed the same practice, the Gban ratio would be 0.03 instead of 0.00 (the ordering remains the same either way).

15  It should be noted that permissive semantics is closely related to causative proper, but is not strictly speaking one of its subtypes, since it involves causation only in an indirect way: ‘allow to P’ ≈ ‘not(cause(not(P))’ (Kulikov 2001: 892). It also remains to be determined whether all (or any) of the constructions listed in this paragraph necessarily convey a factual component ‘P took place’.

16  And its anticausative version: “di̋ yèè Y kɔ̰́, nḭ́ Y P[sbjv]” <road be Y on that Y P[sbjv]> (lit. ‘The way is on Y that Y P’) ‘Y is allowed to P’.

17  Note, however, that the true copula yè() ‘be’ cannot attach the causative suffix (23e).

18  In rapid speech it is reduced to ‑à/-ȁ.

19  A foot, understood as a prosodic unit larger than syllable and lesser than a phonological word, proved to be relevant for the phonological systems of some tonal languages, notably Mande languages (Kuznetsova 2007; Vydrin 2008). In Gban, the foot is the domain of tone and nasalization. It can be monosyllabic of the types |V|, |CV|, |ClV| or disyllabic of the type |CV1.V2| (with two different vowels; all tested words with two identical vowels turned out to be composed of two feet: |CV1|V1|).

20  Cf. the same behaviour for a causative from a |CV1.V2| verb: sȉȅkà ‘spoil (tr.)’ (from sȉȅ ‘spoil (intr.), deteriorate’): sȉȅkȁ ‘spoil-caus\pfv.hest’, sȉȅkä ‘spoil‑caus\pfv.preh’, wȍtló síékȁ ‘car ipfv\spoil‑caus’.

21  They are very similar to the so-called phrasal verbs, or particle verbs, in English, such as give up and find out.

22  In these alternative forms, inflectional tonal changes continue to occur on the verbal part (verb ± causative suffix), while the adverbial-like part (particle + causative suffix) remains invariable.

23  With the exception of zè say’, which seems to allow a synonymous causative zè-kȁ ‘say’ (i.e. a causative-of-intransitive, but not a causative-of-transitive).

24  However, gà-kȁ <die-caus> is nevertheless possible in the idiomatic expression gbí…gà <breast…die> ‘forget’. Similarly, zɛ̀‑kȁ <kill-caus> is possible in the idiomatic light-verb expression yì zɛ̀ <sleep kill> ‘sleep’.

25  Unlike Kakabe (Vydrina 2011: 191) or Mandinka (Creissels & Sambou 2013: 387–388), in which derived causatives can be passive-labile.

26  This discussion and the suggested semantic components/restrictions listed below are preliminary; the data available at the moment are rather scarce.

27  No parallel for such “split” of the preverbal position is found in other Mande languages (Vydrin 2017; Vydrin 2018). In Kakabe, for example, with prefixal causatives from transitives only the former subject occupies a preverbal (i.e. direct object) position, while the former direct object is in the postverbal position, marked with the locative postposition là (Vydrina 2011: 177).

28  As noted in 2.1, such causatives from transitive verbs tended not to be fully acceptable for my second consultant, who (while still producing or allowing them in a sufficient number of cases) often rejected them and provided periphrastic causative constructions with nɔ̰̀ or yɛ̀kɛ̋ (see 1.4) as more appropriate or even the only possible alternatives. For my first consultant, however, suffixal causatives from transitives were generally perfectly acceptable.

29  In this section, I will mark the boundaries of the two NPs with square brackets for clarity.

30  For this single verb (therefore this is perhaps a case of lexicalization; this coding was rejected for dà̰‑kȁ ‘drink-caus’), an alternative coding strategy was attested: with the causee NP as the only preverbal NP (i.e. the direct object) and the NP expressing the original O-participant in a peripheral position with the instrumental postposition yɛ̋ ‘with’:

(i)

Ḭ̋

[Zá̰]

bȅ-kà

[blú]

yɛ̋.

1sg\pst

Jean

eat‑caus[pfv.hod]

bread

with

I fed Jean (lit. ‘with’) bread.’

31  Instead of one of them occupying a more peripheral position, as, for example, in French: Je ferai manger les gâteaux à Jean (Dixon 2000: 35). In Gban, such a peripheral position would be a postverbal position, cf. (i) in footnote 30.

32  It is important to rule out this possibility because the nouns/NPs inside the genitive construction have been shown in Gban to be unusually “independent” in terms of semantics and information structure (Fedotov 2016). For example, recipient and recipient-benefactive participants can be coded adnominally inside the direct object NP (lit. ‘He gives my money’ or ‘He gives me money’ or lit. ‘I bought his clothes’ for ‘I bought clothes for him’) (Fedotov 2016: 342–343). And the possessor in one of the predicative possession constructions can be coded adnominally within the subject NP, together with the possessee (lit. ‘John’s house exists’ for ‘John has a house’) (Fedotov 2016: 336–341).

33  This postposition is polysemous and also has benefactive and recipient-benefactive uses (‘for’).

34  Unfortunately, I was able to test the infelicitousness of genitival NPs with nouns as possessors in this position only with my second consultant (while the infelicitousness of those with personal pronouns was confirmed by both consultants).

35 I should mention that this coding pattern/construction is rare in my data, it was not always judged felicitous, and its usage remains to be studied in more detail.

36  Cf., however, (69c).

37  Comparative data have been of little help, unfortunately. Although some other Mande languages also have causative affixes (some have prefixes/preverbs: most Manden languages lá‑, Mokole languages la‑, Susu ra‑, Kla-Dan le̋-; some other suffixes: Mandinka ‑ndí, Soninke ‑ndí, Bozo ‑ni, Bobo ‑bɛ̄) (Vydrin 2017: 37), none of those are cognate with ‑kà. As for verbs that might have been the source of the suffix, I could not find any other plausible candidates in the dictionaries of closely related languages apart from cognates of the same two Gban verbs kɛ̀ ‘make’ and ka̋ ‘put’ (discussed below).

38 I am grateful to Denis Creissels [p. c.] for the original idea.

39  So one of its varieties is similar to the variety of the kpɔ̋-construction with a subjunctive subordinate clause, cf. (15), and another to the yɛ̀kɛ̋- and nɔ̰̀-constructions with agreeing clauses, cf. (12)–(13) in 1.4.

40  The vowel change *ɛ > a seems unmotivated. However, a similar vowel alternation is found in Gban in the agent noun clitic =ɛ/=ɛ̰/=a/=a̰ag’. In the closely related and neighbouring language Guro, the cognate verb k(ɛ̄)lɛ̄ ‘do, make’ has an imperfective form k()lā, produced by ablaut ɛ > a (Kuznetsova & Kuznetsova 2021: 11, 94–95).

41  However, a path from ‘put’ to causative is not implausible. Consider the close meaning ‘spread (onto)’ which did grammaticalize to causative in the kpɔ̋-construction (see 1.4). It may also be of relevance that there are a number of lexicalized “noun + verb” combinations with ka̋ as a transitive light verb: yě ka̋ ‘laugh’ (yě ‘laughter’), yḭ̏ ka̋ ‘cry’ (yḭ̏ ‘tear’), fȉȅ ka̋ ‘tell lies’ (fȉȅ ‘lie’).

42  The unmarked nominalization is glossed here as ‘inf’, but it remains an open question whether it should be lumped together with the infinitive proper (used in postverbal position) or treated as a separate construction.

43  It is known to be possible in some other South Mande languages [Valentin Vydrin, p. c.], including Mano (Khachaturyan 2015: 149–151), and in Mandinka (Creissels & Sambou 2013: 89–91). Cf. an unmarked nominalization in Mandinka: alikáal-óo la i dóoyaa <chief-d gen they humiliate.d> ‘their humiliation by the chief’ (Creissels & Sambou 2013: 91) and an unmarked nominalization in Mano: ē dàā là ká lɔ́ <s/he/it.refl father s/he/it.poss house buy> ‘the purchase of the house by his father’ (Khachaturyan 2015: 150–151).

44  As can be seen from (86), the omissibility of the postpositional marking in the first dependent varied between my two consultants.

45  They can, however, take genitival NPs beginning with nouns:

(ii)

OK [[Fíɛ́lȅ

mɔ́]

[Zá̰

yènḭ̋]

zɛ̏=lè]

Pierre

for

Jean

sibling

kill=nmlz

‘murder of Jean’s brotherO by PierreA

46  It should be noted that (marked) nominalizations of transitive verbs can also, at least sometimes, have no nominal dependents, cf. zɛ̏=lè ‘murder’, mɛ̰̏=lè ‘beating’. In this respect, there is no parallelism with suffixal causatives from transitives: suffixal causatives do not allow intransitive uses.

Haut de page

Pour citer cet article

Référence papier

Maksim Fedotov, « Causative in Gban »Mandenkan,  | -1, 3-38.

Référence électronique

Maksim Fedotov, « Causative in Gban »Mandenkan [En ligne],  | 2024, mis en ligne le 26 décembre 2024, consulté le 08 décembre 2025. URL : http://journals.openedition.org/mandenkan/3324 ; DOI : https://doi.org/10.4000/130bd

Haut de page

Auteur

Maksim Fedotov

CNRS-LLACAN / INALCO, Paris

Haut de page

Droits d’auteur

CC-BY-NC-SA-4.0

Le texte seul est utilisable sous licence CC BY-NC-SA 4.0. Les autres éléments (illustrations, fichiers annexes importés) sont susceptibles d’être soumis à des autorisations d’usage spécifiques.

Haut de page
Rechercher dans OpenEdition Search

Vous allez être redirigé vers OpenEdition Search