Home / Archive / 2019 / Conditional-Relatives in Senufo and Manding
Document Actions

 

1. Introduction

<1>

A number of West African languages have a type of relative clauses that show properties of both a relative and a conditional clause and that suggest a generalizing interpretation. The paper proposes a study of these so called conditional-relative clauses (CRCs) in the Gur languages Syer (officially called Western Karaboro)and Supyire and the Mande language Bambara. All of them have the obvious double marking in common, although the formal means used in the individual languages to mark relative and conditional clauses differ respectively. Simple relative clauses for instance differ with respect to the origin of the device marking a NP as relativized and with respect to the degree of subordination, some showing a hypotactic, others a more paratactic construction. What is common to the relative clauses in general in all discussed languages is the relation between the relative clause and the main clause in the resulting complex sentences: all the three languages are of the correlative type of relative clauses characterized by the presence of a NP that is anaphorically related to it in the main clause. The conditional clauses differ with respect to their characteristic marking. The Senufo languages use principally an auxiliary, whereas Bambara has the choice between a conjunction introducing the conditional clause and a conditional auxiliary.

<2>

The aim of this paper is to shed more light on this type of construction in the northeastern Senufo language Syer (Burkina Faso), for it has not been included in the recent grammar of this language (Dombrowsky-Hahn 2015). Further, showing that equivalent constructions exist in languages of the same region, even crosscutting language families, I provide evidence that the combination of features of relative and of conditional clauses to express a generalized clausal restriction of event participants is an areal phenomenon in West Africa. It seems to be a preferred strategy for generalizing relatives in languages which do not use representative-instance quantifiers equivalent to English any X, anything, anyone or whatever, whoever, etc. in this function.

I propose to study the structure of the conditional-relative as a special type of relative clauses by taking the form of each of the typical clauses it is based on as a starting point – the relative and the conditional. Subsequently the conditional-relative is presented. Assuming that similarities in form parallel those in function, this is followed by a discussion of the functions of the different constructions.

<3>

The paper is structured as follows. A section is consecrated to a language or dialect cluster respectively: Syer, Supyire, Bambara. Every section contains respectively a short typological introduction (resuming the larger group of Senufo); a presentation of the most typical relative clause, which is the RC modifying a definite NP, then a description of the hypothetical (and realis) conditional clause and finally, the conditional-relative clause. In a concluding section I try to explain how the linguistic means occurring in the CRC bring about the generalizing interpretation of this type of relative clause.

<4>

The analysis of the concerned sentences in the northeastern Senufo language Syer is based on data collected by myself during several research stays in the village of Ténguéréla between 2007 and 2012 [3]. For the other languages I cite published or submitted data and analyses. Thus, I refer to Carlson (1994, 1997) on Supyire (northwestern Senufo; Mali,), to Dumestre (2003) and Vydrin (2019) on Bambara (Manding; Mande; Mali). However, the construction is also attested in other, related languages, as found in Creissels & Sambou (2013) on Mandinka (Manding; Gambia, Senegal) and Creissels (2009a, 2009b) on Malinké of Kita (Manding; Mali).

Map 1: Situation of Bambara, Supyire and Syer in Mali and Burkina Faso

2. Senufo

2.1. Typological characterization of Senufo languages

<5>

In Senufo languages, syntactic relations are determined by word order, which is S (Aux) (O) V (X). X stands for obliques and peripheral participants, most frequently marked by means of adpositions. Most adpositions are postpositions as in (1), but there are also prepositions some of which are combined to a postposition respectively, resulting in a kind of extended exponent [4].

(1)

Syer. Word order determined by linear sequencing.

gbã̀n

cùru

la

kĩ̀

woman

rem

pot5

wash

1sgns

pp

S

Aux

O

V

X

‘The woman washed the pot for me’.

TAM values are encoded in three ways: first, one of two distinct verb forms - a perfective or an imperfective one - the latter of which is derived from the former; second, an auxiliary, and third, serial verb constructions (Carlson 1994, chap. 9). Several auxiliaries can be combined to a more complex tense-aspect value. In Syer, an auxiliary is not always present; instead, the tone on the subject pronoun and the tonal shape of the verb are decisive of some TAM values. For instance, the subject pronoun is L for the remote past, so that when the subject ‘woman’ in (1) is substituted with the corresponding pronoun, this pronoun gets a L tone: ù ní gbã̀n cùru la kĩ̀ ‘she washed the pot for me’. In spite of the importance of the subject pronoun as tone bearing element, only some TAM values require the resumption of the nominal subject with a pronoun (2).

(2)

Syer. Word order determined by linear sequencing.

nàà

cùru᷅-n

woman

pr1

prog

pr5

wash-ipfv

‘The woman used to wash / washes it regularly’.

<6>

Clauses can have a verbal or a nonverbal predicate. The latter is a locative phrase, a nominal (3) or an ideophone. In the abbreviations in the text excerpts the subject and the predicate of clauses with nonverbal predicates are subsumed under the cover term copula subject (cs) and copula complement (cc); the linking element is glossed cop.

(3)

Syer. Clause with a nonverbal predicate.

[la !

ninlɔ]cs

mɛ́

[wɔ̀dě-fwɔ̀]CC

1sgns

uncle

cop

money-owner

‘My uncle is a rich man.’

Senufo languages have an elaborated gender system. Carlson (2012) reconstructs eight noun classes for proto-Senufo; the actual languages show between six and nine noun classes forming three to six genders according to the particular language. For instance Supyire (North-west Senufo) has eight noun classes (Carlson 1994, chap. 3); the Nyɛnɛrɛ variety of Tyebari (Central Senufo) has seven (Rongier 2002), Kar (North-east Senufo) has six, while Syer (North-east Senufo) has nine, showing a supplementary class which is semantically associated with trees and bushes (cf. Dombrowsky-Hahn 2015, chap. 4).

(4)

Reconstructed proto-Senufo noun class system (using reconstructed pronouns to represent classes) (Carlson 2012)

<7>

Schema 4 shows the proto-Senufo system reconstructed by Carlson (2012). The numbers in this schema and in the glosses in this paper correspond to those used in Miehe et al. (2007) and Miehe et al. (2012) for the comparison of inflectional classes of nouns across Gur languages. The lines represent the linking of a singular with the corresponding plural class; classes *ti and *pi represent mass nouns, collectives, abstracts and liquids which do not usually partake in singular-plural pairings; the broken line stands for a supplementary association of some singular nouns of class *ki with the number value ‘collective’ as a special plural. Noun suffixes, pronouns and determiners which agree with the respective nouns bear a consonant characteristic for the respective noun class (in bold in tables 4 and 5) or a consonant closely related to it.

(5)

Syer. Illustration of the agreement targets (interrogative/relative pronouns and determiners)

The languages belonging to the Senufo branch of the Gur language family share a lexical basis and many grammatical features. Nevertheless, they differ with respect to other peculiarities of the grammar. For instance, with regard to the noun phrase, there is an important difference in the marking of definiteness. In Supyire it is encoded by means of definite suffixes that replace the basic nominal suffixes, however showing traces of the latter; in Syer definiteness is signalled (optionally) by articles preceding the nouns [5]. Similarly, the conditional-relative dealt with in this paper and the clause types it is based on show important differences in detail, justifying a separate account of the constructions in Supyire and in Syer.

2.2. Syer

<8>

Syer is a language spoken by about 25.000 speakers in southwestern Burkina Faso, in and around the town of Banfora. Known under the official name of Karaboro, it has several endonyms which reflect the name of the people who speak it and the place of settlement of these people. The data presented here have been recorded in tɛ̃́ɛ̃̀kùlu ‘Ténguéréla’, where the variety tɛ̃́ɛ̃̀ɲɛ̀r is spoken by the tɛ̃́ɛ̃̀ɲɛ̃᷅ɛ̃̀m  [6]. Other speakers prefer however for the language the superordinate term ʃyɛ̃̀r, of which tɛ̃́ɛ̃̀ɲɛ̀r is the most conservative variety. Syer is also known as Western Karaboro, which is part of the Karaboro subbranch of Senufo. A recent grammar of Syer (Dombrowsky-Hahn 2015) has not taken into account complex constructions, but Dombrowsky-Hahn (2018) describes the main types of relative clauses attested in the language. The description in the following section confines to the most typical relative clauses (RCs), the conditional clauses and the CRCs showing the cumulated properties of the two other constructions.

2.2.1. Typical relative clauses in Syer

<9>

A RC modifies or restricts the referent of a nominal constituent of another clause, the so called main clause (MC). Usually, only the MC is apt to figure as a simple independent clause, whereas the aptitude of a RC to occur as a simple clause depends on the degree of subordination of the RC, a point that will be addressed below. Only restrictive RCs are attested in Senufo languages.

The most frequently occurring RCs in Syer modify or restrict identifiable referents. The RC precedes the main clause (MC) and contains the head it modifies or a pronoun substituting a head. The RC does not directly assume a syntactic function in the MC which is rather indicated in it by means of a NP co-referentially related with the RC.

(6)

Syer. Typical RC modifying an object; rel. determiner suffixed to nominal root

[[nã̀-tyìí]O

ni

nàà

Ø

m-pyi᷆-n]RC,

war-rel21

they

rem

prog

ip-make-ipfv

‘The war they were engaged in,

[[tigè

nda],

[ti]GEN

ju

ni

m-bugɔ̀]MC

d.dem21

deic.iden21

pr21

head

rem

ip-be.big

that’s it, its motives (its head) were important.’
Freely: ‘The war they were engaged in, had important motives.’ (Secoke 0465 18:23)

<10>

In the RC in (6), the relative morpheme -tyìí is suffixed to nã̂-, the stem of the head nã̂n ‘war’ with which it agrees in gender and number, exempt from the nominal suffix -n. The Syer relative morphemes are equivalent to the interrogative morphemes ‘which’ as illustrated in the question (7) [7]. The complete paradigm is given in (5) above.

(7)

Syer. Interrogative clause; interrogative morpheme suffixed to the nominal root

nã̀-tyìi

ntĩ̀

ni

nàà

m-pyi᷆-n?

war-inter21

iden21

they

rem

prog

ip-make-ipfv

‘Which war were they engaged in?’

Both the relative and the interrogative morphemes occur in free or bound form. As a free morpheme, they follow a full noun (8a) or replace it (8c, 15); as bound morphemes, they are suffixed to a stem. For instance, in (8b), the interrogative / relative nominal suffix is attached immediately to the stem, the basic nominal suffix -l of the full noun wugel does not occur. For other examples of nouns and noun stems, see Dombrowsky-Hahn (2015, chapters 2 and 4, and 2018).

(8)

Syer. Interrogative / relative morpheme as free and bound determiner, and as pronoun

a)

wugel dyii

‘which hole; the hole that’

b)

wuge-dyii

‘which hole; the hole that’

c)

dyii

‘which one; the one that’

<11>

The uses of the interrogative determiner in its free and bound form is illustrated in (9), of it functioning as relative morpheme in (10), the free relative determiner further in the attested example (11), the bound one in (6). The distribution of the free and the bound forms are not entirely clear, but see Dombrowsky-Hahn (2018) for a hypothesis.

(9)

Syer. Interrogative clause; bound and free forms of interrogative morpheme

ŋwùgɔ̀

Ø

wùgɛ̀l

wuge-dyii

~

wugel

dyii

nì?

snake15

prf

hide

hole-inter5

hole5

inter5

in

‘In which hole has the snake hidden?’

(10)

Syer. Relative clause; bound and free forms of relative determiner

[ŋwùgɔ̀

Ø

wùgɛ̀l

[wuge-dyii

~

wugel

dyii

nì]OBL]RC,

snake15

prf

hide

hole-inter5

hole5

inter5

in

‘The hole in which the snake has hidden,

[nɛ̃w

mɛ́

[li

nì]CC]MC

rat

cop

pr5

pp

there is a rat in it’ (Freely: ‘There is a rat in the hole in which the snake has hidden.’)

(11)

Syer. Relative determiner (free form) in the RC

[[yí

ŋwnɛ̃̀y

ʃĩ̀ɲ

yíí]o

na

Ø

ŋgbɔ̃]RC,

def4

stories4

two4

rel4

pr1

hest

ip?-tell

‘The two stories that she told yesterday,

[mɛ

laa

Ø

lɛ̀

[yi

nì]OBL]MC?

2sg

interior

prf

fall

q

pr4

pp

do you remember them? (Secoke 0963 35:39)

<12>

Due to the strong association of some genders with semantic categories, for instance humans with gender 1/2, artifacts with gender 15 / 4 / 21, activities, problems, affairs or topics with gender 5/6 [8], the interrogative / relative morpheme is used as a pronoun when the type of entity can be inferred from the linguistic or extra-linguistic context. (12) illustrates the interrogative pronoun of class 5 referring to a topic, (13) the relative pronoun of class 2 referring to human beings.

(12)

Syer. Interrogative pronoun (cl. 5) referring to a topic

wo

!

dyii

ǹ [9]?

1pl

fut

go

again

with

inter5

with

‘With what do we continue (lit.: we will go on with which [topic])?’ (Secoke 0611 24:05)

(13)

Syer. Relative pronoun (cl. 2) referring to human beings

[[plɛɛ̀]S

[pì]SRES

n-lì

ǹnɛ᷆

klu᷅

ǹnɛ᷆]RC

rel2

pr2

rem

ip-go.out

here

1pl

village

in

here

S

Aux

V

X

‘Those who originated from our village here,

[pigè]S

n-nìgɛ̀

...

d.dem2

rem

ip-be.many

they were many (Siwahii 0670 31:39,49)

<13>

There are two strategies available for the relativization of noun phrases in verbal clauses. In the first one, the in situ strategy, the constituents keep their unmarked order. It is the most frequent strategy attested for obliques (14), but is rather exceptional for objects.

(14)

Syer. In situ strategy, relativized oblique, suffixed relative determiner

dɔ̃᷇g

[ɲìgà

bi

kĩ̀

[nì

yò-tyii

nî,]OBL]RC

so

God

hod

them

present

with

language-rel21

with

S

Aux

O

V

Obl

‘So, the language God presented them with,

[[tigè

ǹ]O

bi

m-pàr-ni]MC.

d.dem21

iden21

they

ip-talk-ipfv

O

S

V

it is this one they speak > They speak the language that God gave them.’ (Madu 0643 22:44)

<14>

The second one is the fronting strategy; it comprises the displacement of an object to clause-initial position, leaving a gap in its usual slot, indicated by Ø in the RC of (11). The fronting is a focusing mechanism that can be formulated as follows: S (Aux) O V -> O S (Aux) V. The extraction of the relativized object entails marking of the out-of-focus part of the clause as intransitive. This happens by means of an intransitive prefix, which is always present in intransitive clauses marked for certain TAM values (i.e. remote, hodiernal, hesternal past, general imperfective). (15) illustrates the in situ strategy of an interrogative clause, (16) the fronting strategy.

Fronting is one of several possible mechanisms to focus a constituent. A fronted constituent can be additionally marked with a simple or deictic identifier (7, 16). Alternatively, when a relativized item is fronted, the accompanying relative morpheme is marked with a H tone (11). Focusing is possible in situ as well, for instance by means of the focus marker in (15).

(15)

Syer. Content question, interrogative pronoun in situ

ni

nàà

dyii

pyì-n?

2pl

rem

prog

inter5

dp

do-ipfv

S

Aux

O

V

‘What were you doing?’ (Siwahii 0028a 00:42,6)

(16)

Syer. Content question, interrogative pronoun fronted

dyii

ni

nàà

Ø

m-pyi᷆-n?

inter5

iden5

2pl

rem

prog

ip-do-ipfv

O

S

Aux

V

‘What were you doing?’ (elicited to Siwahii 0028a 00:42,6)

<15>

Fronting of obliques is possible, albeit rare. The fronting of an oblique does not involve preposition stranding nor pied piping in Syer. For subjects, only the fronting procedure is available. Since the subject is anyhow clause-initial in an unmarked clause it is not displaced. It is rather resumed by a simple pronoun which agrees with the head in gender and number. In (13), the simple pronoun pi of class 2 resumes the relative pronoun plɛɛ̀.

Constituents in clauses with nonverbal predicates cannot be fronted. For instance the locative predicate li nì in (10), designated as copula complement (CC) keeps its usual position following the copula [10].

<26>

Since the described RC in Syer corresponds to an independent simple interrogative clause, the entire complex sentence resembles a paratactic construction. Without a clear sign of subordination, the RC equivalent to a simple interrogative clause precedes the MC which equals an independent declarative clause [11]. In fact, the first clause can be qualified as RC only because of the relation that exists between it and a NP in the main clause that is coreferential with it, and because the whole functions as an intonational unit. Even if sometimes a very short suspensive pause can be perceived after the RC, there is neither an absolute pause, nor an utterance final lowering of pitch. These two features are present at the end of an independent clause.

<16>

In Comrie & Kuteva’s typology (2013b), a RC showing an element “which would not be present in the corresponding simple declarative sentence” equivalent to an interrogative pronoun is excluded from what they name the paratactic strategy. They reserve this category only to constructions in which RC and MC can function as independent declarative clauses. Within the authors’ typology based on relativizing strategies Syer RCs can rather be classified as a correlative RC. In this type, the RC modifies a NP that is a constituent in another clause (the MC), but it does not assume directly a syntactic role in it. Rather the MC contains an anaphoric NP that is coreferential with the RC and that indicates its syntactic role in the MC.

2.2.2. Conditional clauses in Syer

<17>

It is impossible to give an exhausting classification of conditional clauses for the languages discussed in the frame of the present contribution. Therefore, the description will be restricted to content conditionals [12], which are solely relevant for the conditional-relative constructions as the topic of this paper. The hypothetical conditionals (HC) and the course of events conditionals (CEC) are the most frequent content conditionals. They share the formal aspects in each language respectively. In this section the characteristics of Syer conditionals are presented.

(17)

Syer. Hypothetical conditional

[ǹ

jí]COND

[gi

!

la

gbu]MC

1sgs

cond

enter

pr15

fut

1sgns

kill

‘If I enter, it (ref.: the thing in the water) will kill me.’ (Madu 048 18:15,3)

<18>

In a hypothetical conditional construction, the apodosis follows the protasis, sometimes separated from it by a suspensive pause. There is no conditional conjunction; conditional marking is rather achieved through the combination of a L tone on the pronominal subject (17) (other NPs maintain their respective tune, as in (21)), the conditional auxiliary (glossed cond) and a HL tune [13]. The distribution of the HL tune is determined by two factors, first the presence or absence of an object following the auxiliary, and second, the tonal class of the verb. In an intransitive clause, when no object separates the auxiliary from the verb, the auxiliary is realized M and the verb shows HL, except when the verb belongs to the class of L verbs. The L tone is maintained on these verbs, and the H of the HL auxiliary moves to the auxiliary on its left. The realization of the conditional in intransitive clauses for all three tonal classes of verbs is illustrated in (18).

(18)

Syer, TAM and conditional marking for three tonal classes of verbs

tonal class of verb

M

HL

L

ji

'enter'

yírì

'get up'

‘fall’

conditional

ù mɛ jî

‘if s/he enters’

ù mɛ yírì

‘if s/he gets up’

ù mɛ́ tò

‘if s/he falls’

The distinction between M and HL verb classes is neutralized in the conditional, but in the L class of verbs, which is the most stable one, the verb maintains its L tone. When an oblique follows a verb of the M or the HL class, the verb is realized H instead of HL. For instance the verb klô, which is realized HL when it stands in utterance final position is realized H in nonfinal position in (19). Sometimes, the H realization is found also when the suspensive pause between the protasis and the apodosis is almost absent as in (17), stressing so the intonational unity of the complex sentence.

(19)

Syer. Conditional: H realization when the verb (HL class) is not clause-final

mɛ̀

kló

la

ni

...

2sg

cond

stay

with

1sgns

with

‘If you stay with me ...’

<19>

In a transitive clause, when an object precedes the verb, the distribution is more complex. It depends on the tone of the object. For instance, with a nominal object bearing an initial L tone (sã̀sugu in (15)) the characteristic HL is reduced to H, associated to the auxiliary. The tonal realization of the verb then depends on the tone of the object NP (cp. Dombrowsky-Hahn 2015, chap. 3, section 3.5.3).

(20)

Syer. Ambiguous interpretation: realis CEC and hypothetical HC

mɛ̀

mɛ́

sã̀sugu

wɛ̂r,

!

wɔ̂de᷅

ɲã̀ã̀.

2sg

cond

corn

cultivate

you

fut

money

see

‘If you cultivate corn you (will) get money.’ ‘If one cultivates corn one gets money.’ (Misc 0270 2011.02.25)

<20>

Content conditionals can be distinguished according to the speaker’s attitude towards the likelihood of the realization of the event, ranging from factual at the one end to counterfactual conditionals (in the past) at the other, passing through less likely or unlikely conditionals. HCs are potentially real or unreal, whereas CECs are considered to be real. (17) is an example of a hypothetical conditional, in which “the consequent is somehow predicted to follow from satisfying the ‘condition’ expressed in the antecedent” (Athanasiadou & Dirven 1997:62). In such cases a non-factual hypothetical situation is expressed in the antecedent and therefore also in the consequent.

In a course of events conditional two events co-occur, and the one is dependent on the other. They express a factual ‘whenever’ relationship. According to Athanasiadou & Dirven (1997:71 ff), the CEC and the HC differ with respect to the speaker’s commitment to the realization of the situation: in a CEC the speaker makes a commitment as to the actual, frequent or general realization of the two situations, whereas in the HC the speaker distances her-/himself from any commitment to real occurrence but considers potential realities. In both cases the conditional marking creates a hypothetical world, but CECs differ from HCs among others by the use of ‘higher-order entities’ (Langacker 1997), such as for instance ndye ‘person’ in (21). Tense-aspect categories in the main clause contribute to a factual interpretation, too. Thus, the imperfective in (21) is interpreted as a habitual and consequently the entire sentence describes a generally observed situation [14].

(21)

Syer. Course of events (‘whenever’) conditional

ndye

nã̀ã̀ni,

person

cond

stop

you

next.to

‘If someone comes to stand next to you,

ú

si

ŋ-klɛ̀-n

u

s/he

neg

ip-look.for-ipfv

s/he.subj

go.away

again

s/he doesn’t want to leave again’ (Misc 0093)

2.2.3. Conditional-relative clauses in Syer

<21>

CRCs are constructions that cumulate relative and conditional marking. An example is the elicited sentence (22). The relative pronoun in the RC and the simple pronoun coreferential with it in the MC are the elements characterizing a relative, and the conditional auxiliary and the HL tune on the verb the conditional clause.

(22)

Syer. CRC, relativized subject

[[yii]S

wɛ̂r]CRC,

[[ú]S

kĩ̀]MC

rel1

cond

cultivate

pr1.hab

you

present

‘‘Whoever / anyone who cultivates, offers you gifts’.

Natural data from texts are usually more complex. Thus, in (23), the relative is separated from the main clause by the additional clause ‘the young is with her’. It expresses another condition, however without showing conditional marking.

(23)

Syer. CRC, relativized subject, no following resumptive pronoun

If you see a frightened fisherman fleeing in his boat, it is the following:

[[yii]S

mɛ́

ʃyè]CRC

plòge

mɛ́

wu

ye,

[[wu]S

rel1

cond

give.birth

young

cop

pr1

with

pr1

‘Anyone (ref. hippo-mare) that has given birth (to a young), (if) the foal is with her,

yɛ́

wu

nàà

nã̂n

pyì-n

n᷅.]MC

defeat

she.subj

prog

war

make-ipfv

with

2pl

with

(she) can start fighting with you.’ Freely: ‘A(ny) hippo-mare that has given birth can start fighting with you,
if her young is with her.’ (Madu 0207 07:47,85)

<22>

This type of clause shares two features with referential definite relatives: first, the marking with a relative morpheme (yii in (22) and (23)), which determines or replaces a head, signaling that the so modified or restricted NP is a constituent in a following main clause. Second, the presence of a NP in the MC that is coreferential with the RC and which indicates its syntactic function in the MC. Insofar, the CRC is a correlative construction like the RC. The CRC shares with the conditional clause its marking by means of the conditional auxiliary and the tune characteristic of the conditional [15] (cf. 2.2.2). This conditional marking makes the CRC an unequivocally subordinated clause, whereas the referential RC resembles an independent interrogative clause. While two strategies are possible in the referential definite RCs (at least for objects and obliques), only the in situ strategy is available in the CRCs (and in the conditional). No displacement of constituents is possible; accordingly the fronting strategy does not apply. That is, the subject in (22) and (23) cannot be resumed by a simple pronoun respectively, nor can the object in (24) be fronted.

(24)

Syer. CRC, relativized object, canonical position before the verb [16]

[mɛ̀

mɛ́

[fàwɔ̀-yii]O

ɲã̀ã̀

lùgɔ̀

la᷆ŋ]CRC

[pi

ŋwròm

you

cond

dirt-rel1

see

water

in

def22,23

waste22,23

S

Aux

O

V

S

‘Any dirt you may see in the water, it is the waste

pìgè

mpĩ̀

káá

pyì

[ki

ʃi᷆]CC]MC.

d.dem22,23

iden22,23

pr22,23

it(go)

be

pr15

sort

that became like this.’ (Madu 0619 21:43,9)

<23>

The examples in (25) through (27) illustrate a) a referential definite RC in the perfect aspect (Ø auxiliary), making use of the in situ strategy; b) same as a) but using the fronting strategy; c) a CRC and d) the unavailable fronting strategy within this type of relatives.

(25)

Relativization of S; in situ and fronting strategies in RC + CRC; M tone verb

a)

RC, S relativized, in situ impossible

*yii wɛr

‘the one who cultivated’

b)

RC, S relativized, fronting

yíí ù wɛr

‘the one who cultivated’

c)

CRC, S relativized

yii mɛ wɛ̂r

‘whoever cultivates’

d)

CRC, S relativized, fronting impossible

*yii ù mɛ wɛ̂r ....

‘whoever cultivates’

(26)

Relativization of S; in situ and fronting strategies in RC + CRC; L tone verb

a)

RC, S relativized, in situ impossible

*yii ʃyè

‘the one who has given birth’

b)

RC, S relativized, fronting

yíí ù ʃyè

‘the one who has given birth’

c)

CRC, S relativized

yii mɛ́ ʃyè

‘anyone who has given birth’

d)

CRC, S relativized, fronting impossible

*yii ù mɛ́ ʃyè

‘anyone who has given birth’

(27)

Relativization of O; in situ and fronting strategies in RC + CRC

a)

RC, O relativized, in situ

mɛ̀ fàwɔ̀-yii ɲã̀ã̀ lùgɔ̀ la᷆ŋ

‘the dirt you saw in the water’

b)

RC, O relativized, fronting

fàwɔ̀-yíí mɛ̀ ɲã̀ã̀ lùgɔ̀ la᷆ŋ

‘the dirt you saw in the water’

c)

CRC, O relativized

mɛ̀ mɛ́ fàwɔ̀-yii ɲã̀ã̀ lùgɔ̀ la᷆ŋ

‘any dirt you see in the water’

d)

CRC, O relativized, fronting impossible

*fàwɔ̀-yíí mɛ̀ mɛ́ ɲã̀ã̀ lùgɔ̀ la᷆ŋ

‘any dirt you see in the water’

The CRC shares with both the relative and the conditional its intonational unity with the main clause, especially the absence of a final lowering and a suspensive pause at the frontier between the CRC and the MC.

2.2.4. Summary: the constructions in Syer

<24>

In this section I have discussed three constructions: the RC, the conditional and the CRC, the last of which brings together the characteristic features of the two previous constructions.

The RC in Syer is a correlative construction; that is, the RC does not constitute directly a nominal constituent in the MC; it is rather represented in it by means of a coreferential NP. The relativized noun is marked or replaced by a relative morpheme which originates in the interrogative morpheme ‘which’ agreeing in gender and number with the antecedent. On a purely structural level, the complex sentence is a paratactic construction, for both the RC and the MC (exempt the characteristic intonation making them a coherent unit) can occur as independent interrogative and declarative clauses respectively. Nevertheless, in the current typology, the Syer RC is not considered to belong to the paratactic type which requires both the RC and the MC clauses to be apt to figure as independent declarative clauses. Two strategies are possible in typical RCs: the in situ and the fronting strategy. The latter serves to particularly focus on the relativized item

<25>

Syer conditionals are not introduced by a conjunction; they are marked by means of an auxiliary, which is a subordinator at the same time. Unable to figure as an independent clause, the conditional clause is unequivocally subordinated. Due to the presence of the conditional auxiliary, the CRC is subordinated, too. The CRC is, like the RC, a correlative construction. However, unlike RCs, CRCs only admit the in situ strategy.

2.3. Supyire

<26>

Supyire is a language of the northwestern subgroup of Senufo, spoken in southern Mali. Minute information on Supyire relatives is found in Carlson’s (1994) grammar, and differences between relatives in the northeastern Senufo (to which Syer belongs) and northwestern Senufo are put together in Carlson (1997). The following discussion is based on these documents.

2.3.1. Typical relative clauses in Supyire

<27>

Closely related, the northwestern Senufo language Supyire and its northeastern relative Syer share many features, but they also show dissimilarities. Those concerning relative clauses will be discussed in the following. The commonalities are the order RC – MC and the choice between the in situ and the fronting strategies. The fronting strategy, which is more common, includes the resumption of a relativized subject by a simple pronoun (28) and the dislocation of a relativized object (29) to clause-initial position. Like in Syer, a gap is left in the object’s usual position (signaled as Ø in 29).

(28)

Supyire. RC with relativized subject, fronting strategy, optional marking with dem

[yaagé [17]

(ŋké)]S

[k’]SRES

a

ù

ké]RC

thing.def15

rel15 [18]

pr15

prf

pr1

kill

rm

‘(Lit.: The thing that killed him,

[mu

a

[kùrù]

cé]MC

you

prf

d.dem15

know

‘you know it.) You know the thing that killed him.’ (Carlson 1997:32, ex. 25a.)

(29)

Supyire. RC with relativized object, fronting strategy

[[myàhíí]o

u

a

Ø

cèè

gé]RC,

song.def6

pr1

prf

sing

rm

O

S

Aux

V

‘The songs which she sang,

[[ci]CS

náhá

mìì

fúnŋí

í]MC

pr6

be.here

my

inside

in

they are here inside me.’ Freely: ‘I remember the songs which she sang.’
(Carlson 1994:492)

<28>

Like in Syer, the MC in Supyire includes a NP coreferential with the RC (the discourse demonstrative kuru in (28) and in (30), the simple pronoun ci in (29)). Although both Syer and Supyire relatives belong to the correlative type in Kuteva & Comrie’s (2005) and Comrie & Kuteva’s (2013a, 2013b) typology, they differ with respect to the degree of subordination. As shown above, the typical Syer RC is part of a paratactic construction [19], whereas the Supyire RC bears traces of nominalization, and hence is in a more hypotactic relation to the MC (Carlson 1997:32). The sign of nominalization in Supyire RCs is the invariant clause-final particle . The form of this morpheme resembles the nominal suffix of nouns belonging to the singular class 15 (Carlson’s gender 2 singular), -gV and its definite counterpart -ge or -ke. This parallel between the clause-final marker and a nominal suffix is interpreted by Carlson as a sign of former nominalization and hence subordination of the RC. The nominalization hypothesis is corroborated by data from the central Senufo language Senari (Cebaara) where the corresponding clause-final morpheme is , showing the consonant l (cf. section 2.1) characteristic of the noun class 5 (Carlson’s gender 3 singular). The author therefore assumes that originally the final marker might have been variable, showing agreement with the head.

<29>

The nominalization of a clause is considered as the process by which a finite verbal clause is converted into a noun phrase. Usually, in the course of this process the verb becomes a head noun, the tense-aspect-modal morphology is lost, and subject or object acquires genitive case-marking, among others (Givón 2001:24). None of these adjustments is presently observable in Supyire (or Cebaara), considered by Givón (2001:29) to be an extreme finite (non-embedding) language. Nevertheless, the clause-final particle in Supyire relatives is never present in an independent clause and can be interpreted as a signal of subordination. Thus, the classification as correlative construction says nothing about the degree of subordination in the arrangement of the relative and main clauses, and both the paratactic construction in Syer and the more hypotactic construction in Supyire can be qualified as correlatives. That correlative relative clauses can be paratactic or hypotactic constructions has been shown also for different branches of Mande languages (Creissels 2009a, Nikitina 2012).

<30>

A major difference between Syer and Supyire is the marking of the head of a RC. While in Supyire the clause-final particle is obligatory in all RCs, the marking of the head NP as relativized is optional. This marking consists of a morpheme agreeing in gender and number with the head. In (28) its optional presence is indicated in brackets; in (29) the option is not indicated. The relative determiner corresponds to the deictic demonstrative; however, unlike the relative determiner, the demonstrative precedes the noun (i.e. ŋ̀ké yaagé ‘this thing’, but yaagé ŋké ‘the thing that’). In clauses with an unmentioned head, the same morphemes function as relative pronouns.

In the in situ version of the RC, the relative determiner is obligatory, and it is often extended by the suffix -mu. Like in the neighbor language Minyanka spoken in Mali (Dombrowsky-Hahn 1998), the form of this suffix concurrent with the maintenance of the canonical word order (or application of the in situ strategy) result from contact with Bambara, the lingua franca used in the region (Carlson 1997:33, and below). In Bambara, the word order is extremely strict, and in relative clauses the relative morpheme mîn [mĩ]~ mûn [mũ] (which is the source of the Supyire suffix -mu) follows the head or replaces it.

(30)

Supyire. RC modifying an object, in situ strategy,
obligatory marking with demonstrative-relative morpheme and the borrowed suffix -mu

[ali

níɲjáà

jínàŋa

à

[yaagé

ŋ̀ké-mù]O

kàlìfǎ

even

today

jinn.def1

prf

thing.def15

rel15-suf

entrust

‘Even today the jinn which thing entrusted to him, it is there.’

ú

ge]RC,

[[kuru]CS

na

aní]MC

pr1

pp

rm

d.dem15

prog

be.there

there

Freely: ‘Even today the thing which the jinn entrusted to him is there’. (Carlson 1994:498-499)

2.3.2. Conditional clauses in Supyire

<31>

Content conditionals show the conditional auxiliary (or its allomorph ahá) in the protasis, and future tense in the apodosis. This is illustrated in (23).

(31)

Supyire. Hypothetical conditional

[mu

ahá

mìì

bó]COND,

[mu

tacwóŋi

ŋ̀kwû]MC

you

cond

me

kill

your

fiancée.def1

fut

fp-die

‘If you kill me, your fiancée will die.’ (Carlson 1994:571, extract of ex. 55)

2.3.3. Conditional-relative clauses in Supyire

<32>

Carlson (1994:507-509) deals with the conditional-relative clause in chapter 13 on relative clauses in a section entitled “conditional-relative clauses”, and, in chapter 15 on conditionals, he mentions them under the heading “other uses of the conditional” (1994:580). He notes that the “‘whoever’ meaning” is obtained by using the conditional auxiliary in a relative clause with a nonreferential relativized NP. (32) shows such a sentence with a relativized subject, (33) with a relativized object.

(32)

Supyire. CRC, relativized subject

[[nàŋi

ŋ̀gé-mù]S

ḿ-pá

ge]CRC,

man.def1

rel1-suf

cond

ip-come

rm

‘Whichever man comes,

[wyɛ́rɛ́ŋi

kan

[ura

à]OBL] MC

money.def1

give

d.dem1

to

give him the money.’ (Carlson 1994:580)

<33>

The account of the relative clauses on the one hand and of conditional clauses on the other hand shows that the conditional-relative clauses ((32) and (33)) combine features of the two previous constructions. They share the auxiliary with the conditional and the marking of the head NP by means of the demonstrative-relative determiner and the clause final relative marker ge with the relative. Furthermore, the main clause comprises a NP co-referentially related with the CRC (ura in (32) and (33)).

However, the CRC differs from the RC by the impossibility to dislocate the relativized item to the left; i.e. the fronting strategy does not apply with the CRC. Correspondingly, the relativized NP is always marked by means of the extended relative determiner rel+suf (Carlson 1994:507). The obligatory use of this determiner and the resemblance of this construction to similar sentences in Bambara (Manding, see below) suggest here again a certain influence of the lingua franca on the Senufo language. However, as shown in section 2.2, a very similar construction exists in Syer without any borrowed form. Syer seems generally to have undergone much less influence by the Dyula variety of the Manding cluster used as lingua franca in southwestern Burkina Faso, so that it is difficult to judge whether we are dealing with an instance of metatypy here as well, or whether it is a phenomenon of larger areal distribution.

(33)

Supyire. CRC, relativized object

[u

ahá

[pyàŋi

ŋ̀gé-mù]O

ké]CRC

pr1

cond

child.def1

rel1-suf

get

rm

‘She would get whatever child,

[[ura]S

asì

ǹ-tòrò]MC

d.dem1

hab.seq

ip-pass

it would die.’ Freely: ‘Whatever child she got would die.’ (Carlson 1994:508)

2.3.4. Summary: the constructions in Supyire

<34>

RCs in Supyire are correlative constructions. Both strategies, the in situ and the fronting strategy are allowed in typical RCs. All RCs are obligatorily marked with the immutable clause-final marker ke or ge, which is interpreted as a trace of nominalization of the clause and which functions as subordinator. There is also a relative determiner marking a head as relativized and showing agreement with it. It follows the definite form of the head when used as determiner; the same form replaces the head when used as pronoun. The forms of the relative determiner correspond to deictic demonstratives, but their position with respect to the head differs, for the deictic demonstratives usually precede the head. Relative determiners are optional when the relativized item is fronted to the initial position of a RC. But when the in situ strategy is used, they are obligatory and, additionally, they show a suffix -mu added to the demonstrative element.

Conditional clauses are encoded by means of a conditional auxiliary. The same auxiliary is found in CRCs, which show at the same time features that characterize RCs. Allowing only the in situ strategy, relativized items in CRCs obligatorily bear the extended rel+suf marker.

3. Manding

<35>

Similar conditional-relative clauses have been described for Malinké of Kita (màningakan, southeastern Mali) by Creissels (2009a, 2009b), for Mandinka (mandinkakáŋo (Gambia, Senegal)) by Creissels & Sambou (2013), and recently for Bambara (bámanankan, Mali) by Vydrin (2019). All these are varieties of the Manding dialect cluster belonging to the family of Mande languages. Although a discussion of the constructions would be worthwhile for all three languages, the limited space allows me considering only one of them. Therefore, I restrict the following presentation to Bambara [20], focusing, after a short typological introduction, on typical relative clauses, typical conditionals and conditional-relative clauses.

3.1. Bambara illustrating Manding

3.1.1. Typological characterization

<36>

Senufo and Mande languages share the word order S Aux (O) V (X). In Bambara, the auxiliary is a constitutive and obligatory part of an independent clause, which fuses a TAM value with either affirmative or negative polarity [21]. An exception is the perfective value in intransitive affirmative clauses for which the auxiliary is replaced by the inflectional morpheme -ra (or its allomorphs -la, -na) suffixed to the verb. This results in the schematic representation S V- ra for intransitive affirmative clauses in the perfective aspect. The difference can be seen comparing an intransitive affirmative clause in the imperfective (34) and in the perfective aspect (35).

In Bambara (and Manding in general), the verb does not show agreement with the subject or the object. Syntactic relations are determined by word order, which is much more strict than in Syer and Supyire (or Senufo in general). So, no dislocation of a constituent is possible in Bambara, for instance as a focusing strategy.

<37>

Next to clauses with verbal predicates, Manding varieties dispose of several types of clauses with nonverbal predicates.

(34)

Bambara. Intransitive affirmative clause, imperfective aspect

dén

bɛ́

kási

child

ipfv.aff

cry

S

Aux

V

‘The child is crying ~ cries.’

(35)

Bambara. Intransitive affirmative clause, perfective aspect

dén

kási-ra

child

cry-pfv.aff

S

V-ra

‘The child cried.’

<38>

A main difference vis-à-vis Senufo or Gur in general is the absence of a gender system in Mande. There is a unique nominal marker in Bambara, called the definite marker or definite modality by Dumestre (2003:137) and ‘definite’ article by Vydrin (2019:33), but ‘specific article’ in the Corpus bambara de référence  [22]. It has the form of a floating Low tone (noted `) causing a downstep of a following H, except for a few words in which it is rather perceived as part of the tonal contour of the word. Among the few words there is the relative morpheme mîn (Vydrin 2019). No thorough study of the role of the article for the marking of referentiality and definiteness in Bambara being available up to now, the author only adds that it occurrs more frequently than the definite article in French, and refers to Creissels’ (2009b) description of Kita Malinke where its use is very similar to that in Bambara. In both varieties it is the absence of the article that is marked, not its presence. Resuming and interpreting the most important of Creissels’ findings, it is possible to say that the article is present in the citation form of nouns and other generic contexts (for instance objects in affirmative clauses), and with nouns that have a referential-definite and a referential indefinite [23] interpretation. It is absent with non-referential NPs among others in such negative clauses as mùso tɛ́ ‘This is not a woman’.

<39>

In Bambara, there is a unique plural morpheme -u (written in the orthography as -w, i.e. mùso ` ‘woman’; mùsoẁ ‘women’) suffixed to the lexeme. In a noun phrase, the plural morpheme is added to the last element, for instance in a NP comprising a noun + an attribute, such as for instance an adjective, the plural is suffixed to the adjective which follows the noun (mùso ɲùman` ‘a nice woman’; mùso ɲùmanẁ ‘nice women’).

The tonal article is associated to the plural suffix -u, resulting in [ù] ( in the orthography) (as in mùso ɲùmanẁ ‘nice women’). There is also the more recently developed ‘definite’ article ìn  [24] (40), originating in the deictic demonstrative nìn but whose distribution/function and delimitation vis-à-vis the tonal definite/specific article is not yet understood.

3.1.2. Typical relative clauses in Bambara

<40>

The typical relative clause in Bambara  [25] precedes the MC. The relativized item is followed by the tonal article in form of a floating low and marked by the relative morpheme mîn  [26]. As shown by Creissels (2009a:46), mîn originates from a demonstrative morpheme, whose demonstrative function got lost in Bambara but that has been maintained in Koyaga of Mankono (Ivory Coast), another Manding variety. In this variety it fulfills both the demonstrative and the relativizing functions. The same form is used as determiner and, when the head remains unmentioned, as pronoun. It varies only with respect to number, the plural being mínnù or múnnù. Like in Senufo, the RC and the MC constitute a correlative construction. This means that the MC shows a NP coreferential with the RC, frequently in form of the (discourse) demonstrative (d.dem) pronoun ò (36), or another pronoun, or the repetition of the head determined by the discourse demonstrative morpheme (37).

(36)

Bambara. RC modifying an object; coreferential NP in MC = ò

táa-ra

[dén

fàsa

mîn]O

fìli

kúngɔ

3sg

go-pfv.aff

child

emanciated

rel

discard

wilderness

‘The emaciated child that she threw away in the wilderness,

rɔ́]RC

[jínɛ-w

nà-na

[ò]O

tà]MC

ò

lámara

pp

djinn-pl

come-pfv.aff

d.dem

take

inf

d.dem

raise

djinns came and took it and raised it.’ (Freely: Djinns came and took the emaciated child that she had thrown
away in the wilderness, and cared for her.’ (simplified extract of text dén fàsafili, in Dumestre (1989:49))

(37)

Bambara. RC modifying an oblique; coreferential NP in MC = ò + repeated head

[npògotigi-w

táa-ra

[síra

mîn

fɛ̀]OBL]RC,

girl-pl

go-pfv.aff

way

rel

pp

‘The way on which the girls went, they cut leaves and left them on this way.’

fúra

kári

k’

à

bìla

síra

kàn]OBL]MC

3pl

pfv.aff

leaf

cut

inf

3sg

leave

d.dem

way

on

Freely: ‘The girls cut leaves and left them on the road that they took’.
(Bailleul, Vydrin & Dumestre 1992:1, npògotigiw ní bílisiw)

3.1.3. Conditional clauses in Bambara

<41>

There are two different conditional constructions in Bambara distinguished formally: first, a construction displaying the conjunction ‘if’, and second, a clause showing the auxiliary mána  [27]. Clauses introduced by the conjunction occur most often initially; in rare cases the order is inverted, so that they follow the MC (Dumestre 2003:380). The conjunction can be considered as a subordinator; a clause it introduces cannot function as independent clause. On the contrary, the MC corresponds without any change to an independent clause. If the subordinate part is a verbal clause, the choice of the TAM category depends on the aspectual relation between the conditioning event and the predicted event. If both the event encoded in the conditional and the one in the MC take place at the same time, the conditional contains an imperfective predicate marker (38). If the prediction (expressed in the MC) is such that the conditioning event is expected to be completed before the MC event takes place, the conditional clause bears perfective and the MC imperfective aspect (39), (40), or future tense. As the comparison of (38) with (39) illustrates, the difference of perfective vs imperfective marking in the conditional clause does not correspond to the distinction between course of events conditionals (or the realis type) and hypothetical conditionals (or the irrealis type).

(38)

Bambara. Imperfective aspect in both, the conditional (CEC) and the MC

[ní

í

bɛ́

fíyen

wéle]COND,

if

2sg

ipfv.aff

blind.person

call

‘If you call a blind person,

bɛ́

mɔ̀gɔ

fìla

wéle]MC

2sg

ipfv.aff

person

two

call

you call two people.’ (Bailleul 2005:196, proverb n°1944)

(39)

Bambara. Perfective in conditional (CEC), imperfective in MC

sábu

[ní

cɛ̀

mùso

jɛ̀]COND,

because

if

man

and

woman

pfv.neg

have.sexual.relations

‘Because, if a man and a woman don’t have sexual relations

[dén

tɛ́

sɔ̀rɔ]MC

child

ipfv.neg

get

a child cannot be born.’ [entretien sida 1994_04_10.dis.html, #369611]

(40)

Bambara. conditional-perfective in conditional (HC), imperfective in MC

...[ní

ìn

kɛ̀lɛ]COND,

if

drought

def

pfv.neg

combat

‘...if we do not combat the drought (lit.: if the drought is not combatted),

[jàmana

tɛ́

yíriwa]MC.

country

ipfv.neg

defeat

inf

progress

the country will not be able to progress.’ (Dumestre & Maïga 1993:7)

<42>

The alternative conditional clause displays the conditional auxiliary mána. Its use is restricted to affirmative clauses with verbal predicates, whereas the conjunction introduces both affirmative (38) and negative ((39), (40)) clauses, and both clauses with a verbal and a nonverbal predicate (42d) [28]. In the conditionals with mána, no conjunction or other temporal or aspectual marker or copula can be added, and the negative counterpart of the auxiliary has almost fallen into oblivion [29]. To express negative parallels speakers of standard Bambara rather use conditionals with the conjunction described above.

(41)

Bambara. Conditional clause with auxiliary mána

màrifa

mána

tíɲɛ]COND,

2sg

con

rifle

cond

break

‘If your rifle breaks,

bɛ́

k’

ò

dìla]MC

3sg

ipfv.aff

defeat

inf

d.dem

repair

he can repair it.’ (Bird, Hutchison & Kante 1977)

<43>

The copula of a clause with a nonverbal predicate (42a) cannot simply be combined with or substituted by the auxiliary mána (42b). mána can be used, when the copula is replaced by the verb kɛ́ ‘happen, become, do’ (42c); otherwise the alternative construction bearing the conjunction is used (42d).

(42)

a)

Bambara. independent clause with nonverbal predicate and copula bɛ́ [b’]

kìbaru

b’

í

fɛ̀

news

cop.loc.aff

2sg

pp

‘you have got news’

b)

impossibility to use mána with a nonverbal predicate

*kìbaru

mána

(b’)

í

fɛ̀

news

cond

cop.loc.aff

2sg

pp

c)

mána-conditional clause with nonverbal predicate; copula replaced by verb kɛ́

kìbaru

mána

kɛ́

í

fɛ̀

news

cond

do

2sg

pp

‘if you have got news’

d)

conditional with the conjunction , clause with nonverbal predicate

kìbaru

b’

í

fɛ̀

if

news

cop.loc.aff

2sg

pp

‘if you have got news’

3.1.4. Conditional-relative clauses in Bambara

<44>

Although generalizing relative clauses are frequent in Bambara, the construction was not expressly discussed in Dumestre’s (1987, 2003) grammar. The first examination of the topic is found in Vydrin’s (2019) “Cours de grammaire bambara”. Vydrin follows Creissels (2009) and Creissels & Sambou (2013) who studied the corresponding construction in Malinké of Kita and Mandinka calling it “construction relative généralisante” (generalizing relative construction). The generalizing relative is described as a construction combining features of a relative and a conditional and meaning ‘the one who’, ‘any X who’ or ‘if there is a X that …’ (Vydrin 2019). The feature characteristic of conditionals found in this construction is the marking by means of the conjunction (43-46) or the auxiliary mána (47); the feature characteristic of relative clauses, a relativized NP marked by the relative morpheme mîn, and a NP coreferential with the RC in the MC.

<45>

Showing the deictic demonstrative ò, the coreferential NPs in the MCs are similar to those in sentences showing typical relatives. Furthermore, some show an element stressing the general meaning, expressing a higher-order category, i.e. tìgi ‘person, agent’ in (43), or mùso ‘woman’ in (45), always accompanied by the deictic demonstrative ò. Alternatively the second person singular pronoun í stands for an impersonal pronoun (44) ‘the one, the person concerned’. A plural head is sometimes referred to by the pronoun òlu ‘they’ (46).

(43)

Bambara. (-) CRC modifying the subject, coreferential in MC: ò tìgi

[ní

[mîn]S

dɔ̀nkili

dɔ́rɔn]CRC,

if

rel

pfv.aff

song

sing

only

‘Whoever only sings the song,

[[ò

tìgi]S

bɛ́

fàga]MC

d.dem

agent

impf.aff

kill

must die (lit.: the one will be killed).’ [npogotiginin kokɔrɔbɔla]

(44)

Bambara. (-) CRC modifying the subject mɔ̀gɔ ‘person’, coreferential in MC: í

[ní

[mɔ̀gɔ

mîn]S

kɛ́ra

dònso

yé]CRC

if

person

rel

become-pfv.aff

hunter

pp

say

‘Whoever ~ anyone who becomes a hunter,

[ɲáma-gɛn-fura

fɛ̀OBL]MC

negative.force-chase-medicine

subj

see

2sg

pp

should be in possession of a medicine to chase away negative forces.
(lit.: a remedy chasing negative forces should be seen with you’) (Dumestre 1979:386)

(45)

Bambara. (-) CRC modifying the subject, coreferential in MC: mùso (superordinate category [30])

[ní

[mîn]S

à

ù

wólo

kúngo

kɔ́nɔ]CRC

if

rel

say

3sg

pfv.aff

3pl

give.birth

forest

in

‘Whoever says she has born them in the wilderness,

...

[[ò

mùso]CS

ù

yé]MC

d.dem

woman

cop

2sg

mother

pp

this woman is their mother.’ Freely: ‘The woman who pretends to have born them
in the wilderness is their mother.’ (Görög-Karady, 1979:52 ff.)

(46)

Bambara. (-) CRC modifying the oblique; coreferential in MC: pronoun òlu

[ní

wùlu

dòn-na

mín

kɔ́nɔ],

b’

if

dog

enter-pfv.aff

compound

rel

in

3sg

ipfv.aff

‘Whatever compound the dog enters, he directly goes toward

í

ɲɛ́sin

[òlu]GEN

súrɔfana

sìgi-len

mà]

refl

orient

3pl.emph

con

dinner

put-part

pp

their (the compound’s members’) served dinner’ (Dunbiya & Sangare 1996:53)

<46>

The coreference in the last example builds on the metonymic relation between ‘compound’ as a place where the dog enters and ‘family’, represented by òlu ‘they’ as the people living in one compound and who the dinner has been prepared for.

(47)

Bambara. (mána-) CRC modifying one conjoint of the subject; coreferential in MC: pronoun ò

[kúngo-taa-la

mîn]CONJ

mána

bɛ̀n

síra

fɛ̀,

1sg

and

wilderness-go-ag

rel

cond

meet

road

pp

‘I explain it to every person walking in the wilderness I meet

ń

b’

à

ɲɛ́fɔ

yé]OBL

...

1sg

ipfv.aff

3sg

explain

d.dem

pp

(lit. me and whoever walks around in the wilderness meet on the road,
I explain it to him/her’ [dumestre-manigances_2003_09-dis.html, #320942]

3.1.5. Summary: the constructions in Bambara

<47>

Like in the Senufo languages discussed above, Bambara relatives are correlative constructions. The typical RC precedes the MC. The head is marked by means of the relative morpheme mîn / mínnù ~ múnnù originating in a deictic demonstrative, which is still in use in this function in another Manding variety. In a Bambara RC, mîn has subordinating function, and its position signals the syntactic role relativized. The main difference vis-à-vis the Senufo languages is the strict word order in Manding. A coreferential NP in the MC usually consists of or is determined by the discourse demonstrative ò.

<48>

There are two types of conditionals, one introduced by the conjunction ‘if’, possible with all clause types and both polarity values, and another one that shows the (affirmative) conditional auxiliary mána, which is restricted to verbal clauses.

CRCs are possible with either conditional marking. The other features of the clause correspond to those of RCs.

4. Conclusion: the constructions and their meaning

<49>

In the previous sections of this paper I have shown that several Gur and Mande languages spoken in large areas in West Africa have a type of relative clause that is additionally marked as conditional. These languages are Syer and Supyire, two varieties belonging to the Senufo group (Gur), and Bambara as a representative of Manding (Mande). According to formal reasons the relevant construction was called the conditional-relative clause. From a functional point of view these constructions are said to be nonreferential (Carlson 1994) or generalizing relative clauses (Creissels 2009, Creissels & Sambou 2013, Vydrin 2019). Such ‘generalizing’ relative clauses translate into English with a pronoun that comprises the element -ever as in whoever, whatever, related to the representative-instance quantifier every, or any X that, comprising another quantifier of the same type, any.

The question to be answered in this section is how the generalizing function of CRCs is brought about in the three languages presented above. Therefore, it is worthwhile resuming the functions of each of the formal components making up the conditional-relative clauses.

<50>

On the one hand, the conditional expresses a hypothetical event, i.e. an event abstracted from a concrete, real situation. It creates a hypothetical mental space, which is distinct from current reality and establishes a particular viewpoint by removing the situation from the current reality of the actual speech event (Langacker 1997:220-221). The two main subtypes of conditionals discussed above are the hypothetical conditionals (HCs) and the course of events conditionals (CECs). The HCs represent hypothetical non-factual events. The CECs, bearing the same hypothetical encoding, reflect habitual or generic situations resulting from former real instances of an event type that have occurred at specific points in time. The similarity of encoding of these two subtypes provides evidence of the possibility for hypothetical events to be interpreted as general, habitual or universally valid events, provided they are combined with some higher-order terms and/or the appropriate tense-aspect marking.

Thus, the Bambara proverb (38) ‘If / when(ever) you call a blind person, you call (in fact) two people’ results from the observation of several instances of a scene where a blind person comes in the company of a child guiding him or her to the one who requests him/her to come. Nevertheless, it is also valid when the speaker him/herself never observed a concrete event of this type. Building up a hypothetical mental space (Fauconnier 1995, Langacker 1997) triggered by the conditional auxiliary in Syer, Supyire and Bambara or by means of the conjunction ‘if’ in Bambara, the speaker detaches the situation from all concrete occurrences and thereby generalizes it.

<51>

On the other hand, a RC modifies clausally a participant of an event that has a syntactic role in the MC. But this syntactic role is only indirectly represented in it by means of a coreferential NP. The relative morpheme corresponding to the interrogative in Syer, to the (extended) demonstrative in Supyire and the unique relative morpheme, originating in a presently obsolete demonstrative in Bambara signals which argument of the clause will be co-referred to in the MC, i.e. about which one the speaker wants to tell something in the MC. The RC modifies and thereby restricts the referent among other entities belonging to the same type. The type of entity is either specified lexically, when the head is overtly mentioned, or only schematically, when the head is replaced by the relative pronoun. The event expressed in the MC holds for the entity restricted by the event expressed in the RC.

<52>

The CRC makes use of both the function of the conditional as a means to abstract from concrete instances of an event type and of the relative as a means to restrict the referent of a type of entity. In the CRC, the conditional encoding conveys the modifying event a hypothetical property; a particular type is restricted by the hypothetical property. By a clause meaning ‘hypothesize an entity X restricted by the hypothetical event expressed in the RC’, abstraction is made from a concrete individual, because an entity that is modified by a non-actual event is itself non-actual. This means it can be any entity of the given type, under condition that the hypothetical restriction applies. The coreferential NP in the MC refers to such a hypothetical or non-actual entity and the MC predicts something upon such a non-actual and consequently generalized entity.

In English generalizing relatives every X that, any X that the generalization is achieved through the representative-instance quantifiers [31] every, any, a, which stand for any member of a particular type restricted or modified by the event expressed in the RC. In the three languages discussed above generalization of a property attributed to a type is achieved by presenting this property as hypothetical.

Abbreviations

Numbers before an abbreviation indicate 1st or 2nd (or, in Bambara 3rd) person pronouns, numbers following an abbreviation stand for the noun class of the referent.

1sgs

first person singular subject

inter

interrogative morpheme

1sgns

first person sing. non-subject

ip

intransitive prefix

aff

affirmative

ipfv

imperfective

ag

agent

it

itive

Aux

auxiliary

L

low tone

con

connective (genitive)

M

mid tone

cond

conditional auxiliary

mc

main clause

cond

conditional clause

neg

negative

conj

conjoint

O

object

cop

copula

obl

oblique

cop.loc

copula in clause with predicate locative

part

participle

CEC

course of events conditional

pfv

perfective

CRC, crc

conditional-relative clause

pl

plural

def

definite

pp

postposition

d.dem

discourse demonstrative

pr

simple pronoun

deic.iden

deictic identifier

proh

prohibitive

dem

demonstrative

prf

perfect

dp

discourse particle

prog

progressive

emph

emphatic

q

question marker

fp

future prefix

RC, rc

relative clause

fut

future

refl

reflexive

hab

habitual

rel

relative morpheme (determiner or pronoun)

hab.seq

habitual sequential

suf

relativizing suffix

HC

hypothetical conditional

rem

remote past

hest

hesternal past

res

resumptive subject pronoun

HL

high-low tone

rm

clause-final relative marker

hod

hodiernal past

S

subject

gen

genitive

sg

singular

iden

identifier

subj

subjunctive

ind

indefinite morpheme

V

verb

inf

infinitive

X

oblique

References

Athanasiadou, Angeliki and René Dirven 1997

'Conditionality, hypotheticality, counterfactuality'. In: Athanasiadou, Angeliki and René Dirven (eds.) On conditionals again, pp.61-96. Amsterdam: Benjamins

Bailleul, Charles 2005

Sagesse bambara. Proverbes et sentences. Bamako: Donniya

Bailleul, Charles, Gérard Dumestre, and Valentin Vydrin 1992

Npogotigiw ni bilisiw. Sankt-Petersburg: Gafedilanso Eropukènè

Bird, Charles, John Hutchison, and Mamadou Kanté 1977

An ka bamanankan kalan. Introductory Bambara. Bloomington: Indiana University Linguistics Club

Carlson, Robert J. 1994

A grammar of Supyire. Berlin: Mouton de Gruyter

Carlson, Robert J. 1997

The Senufo languages. Gur Papers / Cahiers voltaïques 2:23-42

Carlson, Robert J. 2012

'Proto-Senufo noun phrase syntax and morphology'. Proto-Niger-Congo. Comparison and reconstruction. International congress, Paris, 18-21 September 2012 (abstract and handout)

Comrie, Bernard and Tania Kuteva 2013a

'Relativization strategies'. In: Dryer, Matthew S. and Martin Haspelmath (eds.) The World Atlas of Language Structures Online. Leipzig: Max Planck Institute for Evolutionary Anthropology. http://wals.info/chapter/s8, (accessed 2015-10-02)

Comrie, Bernard & Tania Kuteva 2013b

'Relativization on subjects'. In: Dryer, Matthew S. and Martin Haspelmath (eds.) The World Atlas of Language Structures Online. Leipzig: Max Planck Institute for Evolutionary Anthropology.  http://wals.info/chapter/122, (accessed 2017-08-11)

Creissels, Denis 2009a

'Les relatives corrélatives: le cas du malinké de Kita'. Langages 174:39-52.

Creissels, Denis 2009b

Le malinké de Kita: un parler mandingue de l'ouest du Mali. Cologne: Köppe

Creissels, Denis & Pierre Sambou. 2013

Le mandinka: phonologie, grammaire, textes. Paris: Karthala

Dombrowsky-Hahn, Klaudia 1998

'Contact minyanka-bambara: l'exemple de la relativisation'. Gur Papers / Cahiers voltaïques 3:27-34

Dombrowsky-Hahn, Klaudia 2015

A grammar of Syer: phonology, morphology and argument realization (Western Karaboro, Senufo). Cologne: Köppe

Dombrowsky-Hahn, Klaudia 2018

'Relative clauses in Syer (Western Karaboro, Senufo, Gur) '. In: Kramer, Raija and Roland Kießling (eds.) Current approaches to Adamawa and Gur languages. To the Honour of Raimund Kastenholz. Afrika und Übersee Beiheft 34, pp.127-170. Cologne: Köppe

Dumestre, Gérard 1979

La geste de Ségou racontée par des griots bambara. Paris: Armand Colin

Dumestre, Gérard 1989

La pierre barbue et autres contes du Mali. Edition bilingue bambara - français. Angers: Ville d'Angers, Bibliothèque Municipale

Dumestre, Gérard 2003

Grammaire fondamentale du bambara. Paris: Karthala

Dumestre, Gérard 2011

Dictionnaire bambara - français suivi d'un index abrégé français - bambara. Paris: Karthala

Dumestre, Gérard and Ismael Maïga 1993

Baabu ni baabu. Paris: Éditions du MRAP/Différences

Dunbiya, Siyaka and Bakari Sangare 1996

An ka yɛlɛ dɔɔnin. Bamako: Jamana

Fauconnier, Gilles 1995

Mental spaces: aspects of meaning construction in natural language. Cambridge: Cambridge University Press

Givón, Talmy 2001

Syntax: an introduction. Volume II. Amsterdam: John Benjamins

Görög-Karady, Veronika 1979

Contes Bambara du Mali recueillis et présentés par Veronika Görög. Transcription et traduction, en collaboration avec Abdoulaye Diarra. Dessins de Abdoulaye Diarra. Mali bamannsiirinw (vol. 1). Paris: Publications Orientalistes de France

Kuteva, Tania and Bernard Comrie 2005

'The typology of relative clause formation in African languages'. In: Voeltz, Erhard F. K. (ed.) Studies in African Linguistic Typology, pp. 209-228. Amsterdam: Benjamins

Langacker, Ronald W. 1997

'Generics and habituals'. In: Athanasiadou Angeliki and René Dirven (eds.) On conditionals again, pp. 191-222. Amsterdam: Benjamins

Miehe, Gudrun, Brigitte Reineke, and Kerstin Winkelmann 2012

Noun class systems in Gur languages. Vol. II. North Central Gur languages. Cologne: Köppe

Miehe, Gudrun and Kerstin Winkelmann (eds.) 2007

Noun class systems in Gur languages, Vol. I. Southwestern Gur languages (without Gurunsi). Cologne: Köppe

Nikitina, Tatiana 2012

'Clause-internal correlatives in Southeastern Mande: a case for the propagation of typological rara'. Lingua 122:319-334

Sweetser, Eve 1990

From etymology to pragmatics. Metaphorical and cultural aspects of semantic structure. Cambridge: Cambridge University Press

Traugott, Elizabeth C. 1985

'Conditional markers'. In: Haiman, John (ed.) Iconicity in syntax: proceedings of a Symposium on Iconicity in Syntax, Stanford, June 24-26, 1983. Amsterdam: Benjamins

Vydrin, Valentin (2019)

Cours de grammaire bambara. Ouvrage en réalité augmentée. Paris: Presses de l'INALCO

Syer data are taken from recorded texts indicated in parentheses at the end of an example. Examples not showing a source are elicited. The sources of Bambara data indicated in [ ] stem from the Corpus bambara de référence, an online ressource under: http://cormand.huma-num.fr/noms_fichiers.html or other internet sources. Among them are the following:

[diarra-ntalen.dis.html] Jara, Umaru Ɲanankɔrɔ. 2012

Ntalen. Manuscrit; http://cormand.huma-num.fr/noms_fichiers.html (accessed 2017-09-08)

[dumestre-manigances_2003_09-dis.html]

http://cormand.huma-num.fr/noms_fichiers.html (accessed 2017-09-08)

[entretien sida 1994_04_10.dis.html] – Entretiens sur le sida: Zacharia avec A., 10.04.1994

http://cormand.huma-num.fr/noms_fichiers.html (accessed 2017-09-08)

[npogotiginin kokɔrɔbɔla]

https://www.indiana.edu/~afrifolk/bam_folk/ftale/ftale_bam1.pdf (accessed 2017-09-08)

Examples without references were elicited.



[1] This paper was presented at the 2nd Symposium on West African Languages in Vienna. I am very thankful to the organizers of SyWAL 2016 for the opportunity to participate in the event and the participants for their questions and suggestions. I would also like to express my gratitude to two anonymous referees whose comments helped me to improve this contribution. All shortcomings are of course my responsibility.

[2] I am much indebted to Salabary Doumbia and Valentin Vydrin for their help in writing this Bambara summary. While basic vocabulary for grammar lessons in primary schools is available thanks to the cooperative work by Diallo et al. (2006), many terms (e.g. relative clause, conditional clause) have not been established yet in Bambara. Therefore, I take up Valentin Vydrin’s suggestions for these terms hoping to provoke the necessary discussion.

[3] The research was made possible thanks to a project grant by the German Research Foundation DFG, project numbers Mi 289/15-1 and Mi 289/15-2. I am very grateful to the foundation, and also to my language assistants and the speakers in the village of Ténguéréla who participated in the research.

[4] In Syer, I know of the existence of only one extended exponent, while there are several in Supyire.

[5] Instances of the opposition of a neutral form of the noun and its definite counterpart in Supyire: yaaga ‘thing’, yaagé ‘the thing’; in Syer kla ‘thing’, ki kla ‘the thing’. However, in Syer the interpretation of the simple noun is ambiguous between indefinite and definite, and explicit marking of definiteness by means of the article is not obligatory.

[6] Tone in Syer and Supyire is marked on every vowel as follows: the acute accent is used to mark High tone, grave accent to mark Low tone. Absence of a tone on a vowel corresponds to Mid tone. Contour tones on short vowels are marked as a combination of register tones: a haček stands for a LH (), a circumflex for a HL (â), the grave accent + hyphen for Low-Mid (a ), and on long vowels by L on the first vowel and Ø on the second (àa).

[7] This question shows a kind of fronting of the (asked for) focused item + an identification morpheme. It is a sort of cleft-like construction. In the RCs the final high tone on the relative morpheme has a similar function as the identification morpheme and is occasionally also found in questions.

[8] For more information on the form and semantics of the noun classes and genders in Syer see Dombrowsky-Hahn (2015, chap. 4).

[9] nì ... ni, or its abbreviated form nì ... ǹ is a discontinuous adposition with comitative/instrument meaning.

[10] Fronting of constituents is common in main clauses of complex sentences as well, provided they have a verbal predicate.

[11] In Dombrowsky-Hahn (2018) I discuss the tendency toward a more grammaticalized construction in those cases where a relativized item is marked with the definite article, for the interrogative morpheme usually does not cooccur with the article in an interrogative clause. Nevertheless, only a small portion of the attested constructions shows this marking.

[12] In Sweetser’s (1990) semantically based typology, content conditionals are opposed to the epistemic and speech-act conditionals.

[13] I assume that the conditional auxiliary is related to the existential copula mɛ̂, used to connect diverse nonverbal predicates to the subject. The HL is charasteristic of both the copula (cf. Dombrowsky-Hahn 2015, chap. 9) and the conditional. However, in the conditional the tone is dislocated to the right, a phenomenon that I am not yet able to explain.

According to Traugott (1985), existential copulas are a possible source of conditional markers attested in different (African) languages, among others in Swahili ikiwa ‘it being so > if’. However, for Syer it is difficult to imagine a concrete scenario of a development from a copula normally occurring only in clauses with nonverbal predicates to a conditional auxiliary admitting exclusively verbal predicates.

[14] The ambiguous interpretation of (20) as HC or CEC is pragmatically conditioned: it derives from the use of 2SG pronoun as a well defined conversation partner and as an equivalent of the impersonal ‘one’.

[15] In the intransitive clause in (22) the M verb wɛr is realized HL and the auxiliary is M, in (23) the verb belongs to the class of L verbs, which are realized L, whereas the H passes on to the auxiliary.

[16] In this sentence, there is no true coreferential NP in the MC. ki ʃi᷆ ‘like this’ is rather an adverbial phrase exemplifying a less explicit representation of the RC in the MC. For different degrees of explicitness of the coreferential NP see Dombrowsky-Hahn (2018).

[17] The gender marking is adapted to the one used for Syer in the above sections. It differs from Carlson‘s as follows: his G1S and G1P are classes 1/2, G2S and G2P classes 15/4, G3S and G3P classes 5/6, G4 class 21 and G5 class 22,23.

[18] At this point I change the glossing to be consistent with the one used for the other languages: Carlson uses dem for the relative morpheme (determiner or pronoun) reflecting its origin and rel for the suffix probably borrowed from Bambara. I gloss the former rel and the latter suf. The clause final relative marker is glossed rm.

[19] As discussed above, the terminology may be confusing. Most Syer relatives are a sequence of an independent interrogative and an independent declarative clause (hence a paratactic construction), unified by the characteristic intonation and the coreferential NP in the MC. Although the typology by Kuteva & Comrie (2005) and Comrie & Kuteva (2013a, 2013b) includes a paratactic type of relative, the Syer constructions which use an interrogative morpheme as an element marking the relativized constituent cannot be so classified, because the precondition that both clauses be declarative clauses is not met.

[20] There is no standardized notation system for tone in Bambara. In the present contribution, I use the conventions established by Dumestre in most of his publications (cf. 1979, 2003, which however differs from the notation in 2011), noting only the first tone of a word. Unfortunately, the L article is not or not consistently noted in most texts (for instance in the Bambara Reference Corpus, the article is noted when it is manifest in the contour tone of a word such as mîn, but it is not noted when it is floating).

In this section, I mark the article only to illustrate its description; I do not add it where it is not marked in the source. The lack of notation of the floating L article is regrettable for many reasons, among others because it inhibits a fine-grained study of the nominal properties of referentiality and definiteness in Bambara.

[21] This statement is simplified and overgeneralized, for the Manding varieties have also other means, for instance an extended exponent of a hortative modality used in blessings. In Bambara it comprises the auxiliary and the verbal suffix -ra. as in ála má síra nɔ̀gɔya-ra, lit.: ‘May God facilitate the way’ used to wish somebody a good trip.

[23] Thus, it occurs in a noun phrase with the indefinite determiner dɔ́ as in wáati ` dɔ́ ‘a certain time, some time, another time’

[24] In the Corpus bambara de référence, it is called “le nouvel article défini” http://cormand.huma-num.fr/gloses.html

[25] In Bambara (and in the other Manding varieties) there are also other types of RCs; among others the RC modifying an indefinite NP. In this case, the RC follows the MC, which contains the head NP. This is one of the rare contexts where a NP is not accompanied by the tonal article, except if it contains the indefinite modifier dɔ́, for in this case the preceding noun shows it. Another type is a RC that has the function of a complement clause of a verb of cognition etc. occurring in the initial MC. However, the RC modifying definite referents is not only the most frequently occurring one. It is also the one that, from a structural and functional point of view, resembles most the construction that is the main topic of this paper.

[26] I adopt here Vydrin’s writing of mîn, pl. mínnù. The circonflex representing a HL contour tone signals that that the relative morpheme (which is H) is like the head itself followed by the tonal article, a floating L.

[27] In Mandinka, no mention is made of a conditional auxiliary, but Malinké of Kita also has two alternative constructions parallel to those found in Bambara (Creissels 2009b:203-205).

[28] Followed by a word with an initial vowel, the vowel of the auxiliary bɛ́ elides and is realized b’ in (35).

[29] Thanks to Valentin Vydrin for directing my attention to the existence of bìlén ~ bìlé ~ bèlé, the negative counterpart of mána. This is an archaique morpheme represented by not more than one example in the Corpus bambara de référence.

[30] The CRC includes a clause that is the complement of the verb kó.

[31] The representative-instance quantifiers every, any, a quantify a single, arbitrary instance of the nominal category, which is, however, a representantive of it. By using them, generalization is achieved, because, “although just one member of the class is actually mentioned and ascribed a property, what holds for a representative member must hold for the others as well” (Langacker 1997:217).

License

Fulltext

Comments

There are no comments yet.

Do you have any additions or comments?

Submit comment.