purpose oblique case marker' or valence-external marker (Dickens n.d.), unique ..... Western Nilotic Päri has somewhat unordinary OVS basic word order in ...
On typological peculiarities of grammatical structure in African linguistic contact areas1
Kwon, Myong-Shik2
Contents 1.
Introduction
2.
Amharic inconsistent PP word order in Ethiopian linguistic area
3.
Hausa predicative marker of Western Sudan belt area
4.
Sandawe verbal categories at NP and verbal plurarity in East African Rift Valley Area
5.
Päri ergative and marked nominative alignment pattern in Western Nilotic area
6.
Kx’a/Tuu features of VE/MPO with Khoe-Kwadi ACC-marker in Kalahari basin
7.
Bantu conjoint/disjoint distinction for clause structure
8.
Concluding remarks
1. Introduction Concerning the typological peculiarities of African languages, there have been introduced a lot of descriptions and discussion in the studies of African languages. First of all it concerns with the phonologically unusual sounds like clicks, labial-velar, implosive, nasal vowel, lateral fricative, tone, ejectives, advanced tongue root (ATR) feature and so on (Maddieson 1984, Clements & Rialland 2008). Of these phonological features we could conclude that some of them are the result of areal diffusion between neighboring linguistic groups, for example clicks in Nguni Bantus (Gunnink et al. 2015) or ATR feature along the West-East Sudan belt area across, which occur widely across the 1
This paper will be read first on the occasion of the 8th Institute of African Studies HK International Conference, at Global campus of Hankuk University of Foreign Studies, Korea on 25th October, 2018. 2
Professor of the division of African languages at the Hankuk University of Foreign Studies, Korea. 1
genealogical boundaries (Güldemann 2008). At the morpho-syntactic level of word or clause structure, all possible types of complexity occur throughout the continent from simple Yoruba type of so called ‘isolating’, over agglutinating Bantu and to a Berber type of inflectional one. Even though the northern Afro-asiatic phylum displays mostly inflectional fused word/clause structure (Hayward 2000), in other sub-Saharan Niger-Congo shows different varieties of word/clause structure from Mande, Kwa or Benue-Congo to Bantu to the South (Bendor-Samuel 1989). In this paper I would like to deal with “strange” types of expressions for the grammatical relations, which are reflected in a given construction with morpho-syntactic devices. In particular, these are typologically inconsistent, or of peculiar patterns from a general point of view about the language structure, for example on constituents order or grammatical marking patterns (Greenberg 1966, Nichols 1986). Moreover those structural and typological peculiarities which will be discussed in this paper seem to be more of the result of linguistic contact and diffusion, rather than from genetic inheritance, or coincidence. Differentiated from the genealogical study of language divergence, we are dealing more with structural convergence3 induced by the language contact. In this areal study of African languages, there have been known to us several linguistic areas in which common structural features are assembled. The mostly known African linguistic contact areas are Chad-Ethiopia4, Sahara spread zone5, Tanzanian Rift Valley, ‘Macro Sudan Belt’ region and Kalahari Basin (cf. Greenberg 1959, 1983, Heine 1976, Güldemann 2007, 2010, Clements & Rialland 2008). Through the contact of more than two typologically different languages, it seems to arise an unusual pattern of structure in a given language areas like the following contacts of languages; Amharic/Cushitic, Songhai/Hausa/Mande, Sandawe/Bantu or Kx’a/Khoe-Kwadi/Tuu, and Nilotic Päri/Cush-Omotic or substrates. Compared to genetically same languages in other areas these contact languages seem to show quite different pattern of construction. For example two different linear3
“Convergence refers to a coming together in geographical space of communities speaking two or more distinct dialects or languages which results, in the course of history, in a greater degree of linguistic uniformity” Dimmendaal 2011: 204 4
“One of the first linguistic areas or convergence areas of this type identified for the African continent was the so-called “Ethiopian language area”. Leslau (1945, 1952) argues that Ethiopian Semitic languages underwent strong influence from Cushitic languages in Ethiopia.” Dimmendaal 2011: 204. 5
According to the distinction between spread vs. residual/accretion zones by Nichols (1992), Dimmendaal (2011: 329) introduced some areas where are still much more diversified genetically than others; namely the Nigerian river belt, the Nuba Mountains (Sudan), the Ethiopian Highlands and Southern Africa. It is not clear, however, how far the grammatical convergence or change for typological peculiarities in these residual zones also might have happened. 2
ordering patterns at different level of constituents are found in Ethiopian Semitic Amharic (Ahland 2009, Leyew 2003). In our three other example languages we find the so called ‘predicative marker’ (Abdoulaye 2011, Hayward 1984), verb-plural marker (Eaton 2002, Kießling 2000) and ‘multipurpose oblique case marker’ or valence-external marker (Dickens n.d.), unique accusative case suffix {-a} of Khoe-Kwadi (Kilian-Hatz 2013) and Nguni conjoint/disjoint alternation (Van der Spuy 1993, Buell 2005, Abloiu & Avery 2009). The last feature seems to be the areal innovation among Bantu including Nguni with or without contact-resulting consequences with neighboring Khoisan or with disappeared substrates. Based on the Western Nilotic Päri (Andersen 1988), I would like to introduce another peculiar alignment pattern of case marking for expressing grammatical relations. As treated more in detail in section 5, two odd patterns of ergative alignment and marked nominative system occur in main and in subordinate clause each other separately.
2. Inconsistent word order in Amharic adpositional phrase The Semitic languages are known as a head initial type.6 So the verb comes first before subject and object in intransitive and also in transitive sentences as the following examples of standard Arabic. (1) Standard Arabic a.
daxala
l-walad-u
entered.3M.SG7 the-child-NOM 'The child came.'
b.
katab-uu
l-kitaab-a
Aoun et al. 2010:18
ʔams
wrote-2M.PL the-book-ACC yesterday 'They wrote the book yesterday.'
Aoun et al. 2010:22
Verb-initial ordering is widely spread across the languages of Afro-asiatic family. Kabyle, one of Berber dialect and Old Egyptian show also VSOx8 constituents order as the following examples show.
6
“The reconstructed default word order for Proto-Semitic is Verb-Subject-Object (VSO), possessed-possessor, and nounadjective order (Hetzron 1997). In Classical and Modern Standard Arabic, this is still the dominant order, but VSO has given way in most modern Semitic languages to SVO.” Dimmendaal 2011: 204. 7
Used abbreviations will be given at the end of the paper.
8
For expressing the main constituents order I will use abbreviation small “x” for adjunct argument, while core arguments 3
inɣa
(2) Kabyle
wərgaz
aqʃiʃ.
3.SG.M.PER.kill man.NOM boy.ACC "A man killed a boy."
Aikhenvald 1995:45, König 2008:182
(3) Middle Egyptian Rdj sn-j
t n nd̰s
give brother-1S POSS bread DAT poor.man ‘My brother gives bread to the poor man’
Bakir 1984:125, Claudi 1994:221
Different from three above illustrated branches of Northern Afroasian, Cushitic and Omotic display generally V-final SOV order, while the word order of Chadic is diverse, but Hausa, one of best known lingua franca in West Africa keeps V-medial SVO ordering. Concerning the immigration of Ethiopian Semitic form Arabian peninsula into the North-East region of Eritrea-Ethiopian high land there has been a lot of discussions about the contact induced language changes, which finally was explained by the term ‘the Ethiopian linguistic area’ as a linguistic convergence, or as a structural accommodation.9 The core point lies on the mixture two opposing types of V-initial and V-final syntax. The latter type of head final ordering (SOV) is best displayed by the neighboring Cushitic language as the following Sidaama.
(4) Sidaama
dagunč-u
gere‘čó
it-Ø-i.
leopart-NOM.M sheep(ACC) eat-3SG.M-S.PRE.3SG.M ‘The/A leopard eats the/a sheep.’
Kawachi 2007:181
Amharic shows the verb final syntax, different form other non-Ethiopian Semitic languages like Classical Arabic, presumably by following the Cush/Omotic SOxV-pattern as the next two intransitive and transitive sentences illustrate. (5) Amharic a.
setɨ.jo-wa-Ø
bə.-fɨ.t’nət hed-ətʃtʃ
are expressed large S, V, O. Since the adjuncts are optional, they are not directly governed by the relevant verb. 9
Crass & Meyer (2008) summarize the research history of this area, since Ferguson (1970) first proposed the term ‘the Ethiopian Linguistic Area’. 4
woman-DEF.F MAN-fast go.PF-3F ‘The woman went quickly.’
b.
setɨjo-wa
ɨssu-n
gəddəl-ətʃtʃ-ɨw
woman.DEF.F 3M-ACC kill.PF-3Fsub-3Mobj ‘The woman killed him.’
Ahland 2009: 687
Virtually the grammatical relations like subject and object in Amharic are not exclusively expressed by the fixed constituent order, but also through the morphological marking on the head verb as a form of subject agreement in person and gender in one hand. And in transitive sentence we can see, in addition, the overt accusative case marker as an object function. Therefore we can conclude Amharic exploit all three basic morphosyntactic strategies of marking grammatical relation. If we consider the level of phrases, one of peculiar constructions appears as the following adpositional phrase. (6) Amharic
a.
bä-t'or PREP-spear
b.
'with a spear'
bä-säw lay PREP-people on
'on people' Leyew 2003, Dimmendaal 2008:281
(6a) shows instrumental case prefix, while (6b) postposition is added to the case marked noun as its complement. So we can say, Amharic displays two different kind of ‘flagging’ for dependent marking (Haspelmath 1918/19), namely one time case marking, and in the other, adposition in one and same adjunct construction. Typologically speaking given language tends to follow a type of one of several possibilities according to the consistent pattern such as prepositional phrase or postpositional phrase as an adposition. Otherwise it can adopt instead case marking strategy. However in Amharic case all three types of expressions are incorporated as one constituents unit. (7a) is another example for the ablative case prefix, which are marked directly on the modifying adjective. The head noun is not immediately adjacent to the modifier, but detached by the definite marker {-u} on the adjective. Hence, there seems to be a discrepancy between immediate constituent ordering of adjective-noun without disturbing through inserting definite marker in-between. Namely the definite suffix should be 5
followed the noun phrase. Ablative case prefix can be interpreted as a proclitic which govern NP, while the definite suffix is hardly match to given NP. This kind of mismatch of constructions would be the result of mixing or conflict between two opposing principles of processing constructions, namely head-initial and head final. (7)
a.
kä-tɨllɨk’-u
bet
from-big-DEF house ‘from the big house’
b.
mist-u-n
bä-mä-gdäl
wife-his-ACC by-NOML-murder ‘by murdering his wife’ (Leslau, 1995:400, Baker & Kramer 2014: 142)
In (7b) the position of preposition or more correctly instrumental case marker is irritating, because through this marker as prefix the constituent of verb and its complement is separated. This kind of peculiar construction seems to be the result of mixing two different constituent ordering, i.e. headinitial Semitic vs. head-final Cushitic.
3. Hausa predicative marker of Western Sudan belt area Grammatical relations like subject or object are manifested in language at two different level of linguistic structure, namely syntactic and morphological level. At the latter level of construction the formal marking of grammatical relations are divided into two canonical coding whether it comes on the head or on the dependent (Nichols 1986, Dimmendaal 2005). These two types of coding are represented as agreement (or indexing) on the head, or as case marker or adposition on the dependent NP (Haspelmath 2018/9). From this typology we can say Swahili as head-marking with its subject and object agreement markers at the verbal stem (8) and the Central Cushitic Kemantey as dependent marking of nominative and accusative on the NPs (9). (8) Swahili
Yule mwindaji a-li-mw-ua
adui wake.
DEM hunter 3.SG-PST-OBJ-kill foe POSS ‘The hunter killed his foe.’
Kwon 2009: 17
6
(9) Kemantney(CC)
nïŋ-ïs
kwan-I
xätïy-aɤ.
ice-NOM house-ACC puncture-PST 'The ice pierced the house.' Leyew 2003: 239, König 2008: 71
As we already saw in Kabyle (2) two opposite strategies can be used simultaneously in a given construction. If the morphological coding is absent, the constituents order will be the criteria to distinguish the object from the other subject argument like following Yoruba ditransitive sentence (S V IO DO). (10) Yoruba a.
Ò jó fún ìyá
ní
owó
Ojo give mother PREP(< say) money 'Ojo gave mother money.' Creissels 2005:66
In the West African Niger river region around a peculiar construction regarding the coding type of grammatical relations which are known as “predicative marker”10 as the following Hausa transitive sentence. (11) Hausa
Abdù yaa
karyà
Kafàr teebùr.
Abdu 3MS.PERF break-I leg-of table 'Abdu broke the table's leg.'
Abdoulaye 2011: 43
The second particle between subject and verb build an independent word and contains only grammatical categories, which belong not only to nominal subject like person, gender and number but also to verbal category aspect. Geographically adjacent, but genealogically different language of Songhai Zarma (classified as NiloSaharan, Greenberg 1963) displays the same ‘predicative marker’ like Chadic Hausa, however it contains only verbal categories like aspect and polarity. Since Zarma has SOV basic order, this 10
This term is called differently according to the languages which are described. For example Kießling (1996, 2000) used preverbal clitic cluster (PCC) for describing Southern Cushitic. Discussing the Cushitic language Mous (2005) calls this particle as “Selector”. Different from Mande languages (like Bambara), where only verbal categories are expressed, Hausa includes person category of the subject noun. Therefore it would be more appropriate to call pronominal predicative marker (PPC). Cf. Kwon 2016. 7
predicative marker are positioned separated from the verb by the direct object (Sm11OV). (12) Zarma
a.
muusaa
na
feejoo
wii
Moussa PERF:POS sheep:DEF kill 'Moussa killed the sheep'
b.
muusaa
mana feejoo
wii
Moussa PF:NEG sheep:DEF kill 'Moussa did not kill the sheep' Oumarou Yaro 1993, Creissels et al. 2008: 103-104
The same construction with PM containing aspect and polarity occurs in imperfective clause of Bambara, the language of Mande. (13) Bambara
a.
séku bɛ́ bòli Sekou IPF.POS run 'Sekou runs.'
b.
séku tɛ́ bòli Sekou IPF.NEG run 'Sekou does not runs.'
Creissels 2007: 5
As Bambara has a verb-final constituents order, the predicate marker of verbal categories separated from the verb as the following Bambara dialect Mandinka ditransitive sentence shows (SmOVx).
(14) Mandinka
mòolu ye
kinoo
dii n na
people PAST food give me to ‘The people gave me food’
Gensler 1994: 3
11
Here I use another small letter “m” for indicating predicate marker. The composition of grammatical categories of this particle is diverse from one language to another. However mood and modality, in my view, seem to be more basic verbal categories among others. 8
Canonical predicative marker as a mixing nominal and verbal categories as in Hausa example occurs in geographically distant East Africa, namely East Cushitic Arbore, which belongs to the Omo-Tana group. Even though there is other Mande language of which PM contains nominal categories, it may be rather traits of genealogical inheritance from Afro-asiatic family.12 (15) Arbore (Omo-Tana) mo
'í-y
ḳor kúure
man DEF.IND-3SG tree cut:3SG.M:PERF 'The man cut the tree.'
Hayward 1984: 110, Mous 305
4. Sandawe verbal categories at NP and verbal plurarity in East African Rift Valley Area One of East African Khoisan, Sandawe is surrounded by all other African language families (Kießling et al. 2008) although it is genetically more close to the Khoe branch formerly called as Central South African Khoisan (Güldemann & Elderkin 2010). Accordingly it has three different click sounds with 5 different modifications (Elderkin 1989). Sadawe is head-marking type with the person, gender, number, polarity and mood indexes on the verb as suffix. As the endo-centric type usually does, the clause can be expressed without co-nominal NP, i.e. it is a type of cross-indexing (Haspelmath 2013). (16) Sandawe a.
c.
thímé-sà
thímé-` -sù˳
b.
cook-3F.SG.RLS
cook-FUT.3F.SG.IRR
‘She cooks/cooked’
‘She will cook’
thímé-ʧhì-sú
d.
thímé-`-sù˳-ts’é
cook-NEG-3F.SG.IRR
cook-FUT-3F.SG.IRR-NEG
‘She did not cook’
‘She does not/will not cook’ Eaton 2002:5
As for the strategies for expressing the grammatical relations, however, Sandawe shows first of all two peculiar patterns which violates the generally assumed principle, that the grammatical categories of noun and verb represented directly to the nominal or verbal stems each. So gender, case, number 12
However we will see the same predicative marker in Kx’a (Ju/hoan) and Khoe-Kwadi (ǂKhomani ~ N||ng) as illustrated in the following section 6. 9
used to be expressed on the noun as affixes or stem alternation. Sometimes it can be a clitic on the NP. On the other hand the verbal categories like TMA, polarity and voice marker occur on the verb or around neighboring auxiliary. As the following transitive sentences show, verbal categories of person, gender, number realized as suffix {-sa} are expressed either at the object noun or/and at the verb.13 (17) Sandawe
a.
iyoo
ǀnining’-sa
ǁaa
mother maize-3FSG
b.
ǁaa-sa
ǀnining’-sa
plant
iyoo
Plant-3FSG maize-3FSG mother ‘Mother planted maize’ Dobashi 2001: 57-8, Dimmendaal. 2008: 284
The basic constituent order of Sandawe transitive clause is SOV, and the subject function is coded at the object noun as suffix, which contains not only the nominal categories like number, person, but also the verbal category of aspect. This mixture of both opposite kind of categories is reminiscent of the above mentioned (pronominal) predicative marker of Hausa-type languages. (18) Sandawe
a.
tʃı́ dìyá-s
màntʃà-a
I egg-S1SG:PERF eat-SG ‘I have eaten an egg.’
b.
tʃı́ dìyá-s
màntʃà-wàa
I egg-S1SG:PERF eat-PL ‘I have eaten eggs.’
Kießling et al. 2008: 209
The second point of the typological peculiarity of Sandawe is the marking of number on the verb, as the (18b) shows. The contrasted singular/plural suffixes {-a/- wàa} distinction on the verb have 13
We will see the same PNG-marker not only at nominal argument but also at the predicative verb in Namibian Khoekhoe (section 6, example 27 in this paper), which is assumed to belong to a same genetic group with Sandawe. 10
nothing to do with the verb per se, but they indicate the number of object nouns. As the object noun as the controller of the number marking on verb, it can be said that there is an object agreement in number on the verb. What is curious thing in this case is, why object agreement is first given without subject agreement if we consider the hierarchy of grammatical functions like ‘Subject > Object’. As the following example indicates there is actually subject agreement too. This is the case of intransitive clause. (19) Sandawe
a.
kû
ǂó̼o-gâ
rope be.cut-S3m:PF 'The rope has been cut.'
b.
kû
ǂó̼o-gâ
rope be.cut-PL:S3m:PF 'The ropes have been cut.'
Kiessling 2000: 9-10
We can conclude as Kiessling (2000) notes that the object agreement of Sandawe follows the ergative pattern, e.g. the single argument of intransitive clause S and the patient-like argument of transitive P are treated same, while agent like argument of transitive A is treated differently.
5.
Päri ergative and marked nominative alignment pattern
Western Nilotic Päri has somewhat unordinary OVS basic word order in transitive (20b), but in intransitive clause the subject comes first, so SV order (20a). That means the treatment of subject function is differentiated by the valence of the verb. Transitivity plays key role in this language for deciding the case marking form of grammatical relation. While the subject of intransitive (S) and the object of transitive clause (P) marked by the same absolutive case, the subject function in transitive (A) marked differently with an ergative case form. The different basic order SV/OVS indicates also the same alignment pattern e.g. ergative system (Andersen 1988). In addition to the constituents order and case marking strategies, Päri makes use of other strategy for grammatical marking e.g. indexing at the verb. In this way it seems to be the most complicated system of marking of grammatical relations across the African continent. In (20), two strategies of word order and case marking are presented. (20) Päri
a.
dháagɔ̀
a-ŋɛ`ɛth-ɔ̀
woman.ABS C-laugh-SUF 'The woman laughed.' 11
b.
dháagɔ̀
á-yàaɲ ùbúrr-ì
woman.ABS C-insult Ubur-ERG 'Ubur insulted the woman.' Andersen 1988: 292, cf. Stirzt 2014: 255.
From discourse-pragmatic functions of topic or focus, subject can be moved to the clause-initial position, then the alternative constituent order SOV appears as the following example (21). The clause initial subject is marked by the unmarked absolutive case instead of ergative in normal word order, as a result which becomes to have the same case forms with object without differentiating the given syntactic function. In this case the third method of cross-referencing is recruited to indicate the subject. So word order, case marking and cross-referencing are complementarily distributed for each other. Indexing can be regarded as a derived strategy.
(21) Päri
ùbúr
dháagɔ̀
á-yàaɲ`-ɛ`
Ubur.ABS woman.ABS C-insult-3SG 'Ubur insulted the woman.'
Andersen 1988: 293
As Anderson (1988) already noted the so-called ‘marked nominative’14 or ‘extended ergative (from A to S)’ system occurs in the subordinated clauses. This peculiar case assignment system is widely distributed in Northeast-central Africa including Afroasiatic Cush-Omotic and Nilo-Saharan languages (König 2008). Structural characteristic of this type is the marked S and A leaving the P with unmarked with more basic and so more frequent usages. Co-occurrence of the two opposed assignment patterns in a single language like Päri indicates the transitory or intermediate stage of structural changes from marked nominative to ergative. As Andersen (1988) observes, it might be the origin of the African ergative system in Western Nilotic and neighboring Surmic and other languages. The following examples are one of three subordinate clauses given by Anderson (1988) e.g. interrogative form but with the form of { ì} with locative meaning, it constructs like cleft sentence. The subordinated clause of (22a) is intransitive with S marked ergative, which is also the marker of transitive subject (A) in (22b). If we follow the convention of the marked nominative terms, the ergative should be termed as ‘marked nominative’ and the object as absolute (or accusative). 14
For discussions on ‘the marked nominative’, see König 2006, 2008, Dimmendaal 2012, Creissels 2009 among others. 12
(22) Päri a.
pìr
ŋɔ̀ ì
pʌ̂ ʌr
matter what LINK jump
cícʊ̀-ɛ̂. man-ERG
"Why did the man jump?" Andersen 1988:318, König 2008:100
b.
pìr
ŋɔ̀ ì
cʊ̀ɔl yí
ɲìpɔǹd`-ɛ̀.
matter what LINK call 3.SG.O child-ERG "Why did the child call her?" Andersen 1988:319, König 2008:100
6. Kx’a15/Tuu features of VE/MPO with Khoe-Kwadi ACC-marker in Kalahari basin Non-Bantu languages with click sounds in South-Western Africa, which was called ‘South African Khoisan’ by Greenberg (1963), has two typologically contrasted structures, viz. V-medial Kx’a (formerly Northern group) and Tuu (formerly Southern group) in one hand and V-final Khoe-Kwadi (formerly Central group) lineages in another. Previously-called genetic names of ‘Northern’ and ‘Southern’ branches of ‘South African Khoisan’ followed by the new term ‘non-Khoe’, has basically isolating-analytic structure of SmVOx in its simple transitive sentence with adjunct. Regarding the typological peculiarities of this lineage, I would like to consider above all, two phenomena, namely valence-external (VE) argument marker in clause construction and the multi-purpose oblique (MPO) case form for introducing devices for extra argument as an adjunct. In the following Ju/’hoan transitive clause, we can see the predicative marker (IMP) between subject and verb just as same as in Hausa, Zarma, Bambara and Arbore examples above. The object noun comes after verb with VEsuffix {-a} which can be regarded as a kind of indexing for expressing grammatical relation.
(23) Ju/'hoan
ha
kú //ohm-a !aìhn.
CL1.PRO IMP chop-VE tree 15
New genetic group Kx’a was proposed by Heine & Honken (2010) to unite existing Juu Greenberg’s
Northern group) and ǂHõa which has been remained as isolate until then. Ju and Tuu are new terms for Northern and Southern South African Khoisan (Güldemann & Vossen 2008). The disappeared Kwadi in Angola is grouped into Khoe to build new Khoe-Kwadi. (Güldemann & Elderkin 2010)
13
'He was chopping the tree' Dickens n.d.:22, Güldemann & Vossen 2000:110
As the following (24a) indicates, the absence of VE marker means verb-final intransitive. In contrast to (24a), in (24b) new argument is added, however it needs not to be a core argument. So it does not concern the matter of transitivity or the government relationship between verb and its arguments.
(24) Ju/'hoan
a.
vs.
aíá tsí
b.
aíá
tsí-á mí
my.mother come
my.mother come-VE 1S
My mother came.
My mother came to me. Dickens n.d.: 19, 20, Güldemann 2004: 2
In addition to the valence-external argument marker, there is another formative to build an adpositional phrase in Kx’a group. It concerns primarily on the peripheral argument which is composed as prepositional (25a) or postpositional phrase (25b).16 What seems to be interesting is the fact that two opposite forms of pre- and post-position can occur simultaneously, which can therefore be called double adpositions.
(25) Ju|'hoan a.
Ha
kú
//ohm-a
!aìhn kò
g/úí.
CL1.PRO
IMP
chop-VE
tree MPO forest
He was chopping the tree in the forest.
b.
Ha
kú
CL1.PRO IMP
//ohm-a g/úí
kò
!aìhn.
chop-VE forest MPO tree Dickens n.d.:22, Güldemann & Vossen 2000:110
16
As the glosses indicate the particle {kò} is translated as MPO (multi-purpose oblique), because it appears pre- or postnominal complement NP. Moreover, it can occur before postpositional phrase with multiple functions (26). In the sense that the adjunct are doubly marked with pre- and post-position, it is reminiscent of Amharic double marking adjunct construction which are composed with case prefix and postposition, as we showed in previous section already. Somewhat different but similar construction of double marking postpostions in locative uses are known in Cush/Omotic languages, where nominal possessive construction is grammaticalized into postposion. See for example Maale in Dimmendaal 2008:277. 14
The next ǂKhomani example of MPO comes from !Ui branch of Tuu family, which shows pre- and post-positions. This double adposition construction seem to be the result of grammaticalization of nominal possessive construction of determination into postpositional phrase of government relation in one hand, while preposition {ng} governs again the following NP.
(26) ǂKhomani ~ N||ng
/oe-ke
ke
n//aa
ng
n//ng //a'i
children-P DECL stay MPO hut inside the children are in the house
Güldemann 2004: 2
While Kx’a and Tuu introduce peculiar markers of clause construction VE & MPO, Khoe-Kwadi family shows a suffix {-è} as the following object noun of Khoekhoe transitive sentence (27c) indicates. If we regard the subject argument as zero marked, then we see Khoekhoe as ‘nominative/accusative’ case assignment type. Khoe-Kwadi is famous for their PGN-marking at predicate (27a) and also at argument (27b). (27) Standard Khoekhoe a.
//oo-s
ge
go
die-3F.S.SBJ DECL PST 'She died.'
b.
ti-ta
Güldemann 2006:120
gye
1SG-1SG.SBJ DECL
//ō
tite
die
NEG.FUT
'Ich werde nicht sterben.' [=I won't die.] Dempwolff 1934/5:53, Güldemann 2006:120
c.
≠'ṹu-p
kxa
eat-3M.S.SBJ PQ PST
kè//'ĩ́ i-p-à
//án-'è
3-M.S-DSBJ
meat-3C.S:OBJ
'Did he eat the meat?' Hagman 1977:143, Güldemann 2006:120
The unique marker of accusative in Tuu, however, is not obligatory, e.g. somewhat default for marking the core grammatical relationship. As the following example shows it covers also the function of indirect object in Khwe. 15
(28) Khwe (West Kalahari Khoe) màìci-m̀
à
Matthew-3sg.m acc
ǀ’áò
à
money acc
tí
ǂxàró-á-tà
1sg
give-j-pst
‘I gave money to Matthew.’
Kilian-Hatz 2013: 374
Beside the accusative function of patient role, this particle can also express beneficiary role as grammatical adjunct as in the following Shua examples.
(29) Shua (East Kalahari Khoe) a.
taa
ʔ
pii
tyana-ma
1sg
acc
milk
bring-appl
‘Bring me some milk.’
b.
tse:
ʔ
aka
k’ohu
ngǀ:-a-ma-e-ha
1pl.c
acc
pst
meat
cook-j-appl-pass-pst
‘The meat was cooked for us.’
McGregor 2018
The same particle can be used not only accusative and dative above but also focus marker for intransitive subject argument like following Khwe clause. Thus the functions of the so called ‘ACC’ form are various not only for marking grammatical relation in the clause, but also for designating discourse-pragmatic functions like focus.
(30) Khwe (West Kalahari Khoe) kúcugucugu
à
ǁgèvùu-à-tè
eagle
foc
fly-j-pst
‘An eagle is flying.’
Kilian-Hatz 2013: 370
The optional accusative marker {á} can be used further as copula in intransitive nominal clause. These multi-functional characteristics of the particle suggest the discourse-oriented, not yet completely grammaticalized into the category of valence or transitivity. 16
(31) Khwe (West Kalahari Khoe)
yì
á
ǀéú
tree
cop
big
‘The tree is big.’ Kilian-Hatz 2008: 135 (32) Khwe (West Kalahari Khoe) nǀĩĩ́
dáó
dem15 path
à
ǂ’ó
dáó
à
foc
small
path
cop
‘This path is a small path.’ Kilian-Hatz 2008: 99, 208
7. Bantu conjoint/disjoint distinction for clause structure Among Bantu languages, there are typologically different group of languages, which distinguish the conjoint construction from disjoint one17 as the following Ndebele, one of Nguni (Southern Bantu, with Zulu, Xhosa, Swati) languages. This alternation occurs in the present and recent past tenses only. In present tense conjoint form realizes as zero morpheme, while its counterpart form is {ya}. Actually it is the tense form, but it expresses, in addition, the information about argument frame of given sentence. The conjoint (33a) requires obligatorily the following object NP. The verb governs the object NP as core-argument, while the disjoint construction (33b) does not need the core argument. If it occurs, its status is not anymore a core argument of the verb, but discourse-pragmatically specified participant as disposed apposition.
(33) Ndebele
a.
u-Ø -dl-a *(uku-dla). 1S-TNS.CJ-eat-ASP 15-food 4 ‘S/he eats (the) food.’
b.
u-ya-dl-a (uku-dla) 1S-TNS.DJ-eat-ASP 15-food ‘S/he is eating / eats (the) food.’
17
Conjoint vs. disjoint alternation was called variably in the literature as long vs. short form, or sentence final vs. non-final form. Since the paper of Van der Spuy (1993) there were a lot of discussions on this peculiar construction. See for example Creissels (1996, 2017), Voeltz (2004), Buell (2005, 2006), Van der Wal (2011), Halpert (2015), Zeller et al. 2017, van der Wal & Hyman (2017) among others. 17
The same is true in the case of recent past tense, while the conjoint and disjoint forms are replaced by {-é} and {-ile}.
(34) Ndebele
a.
u-Kuthula u-Ø -dl-é
*(isi-tshwala).
1a-Kuthula 1S-TNS-eat-ASP.C 7-polenta ‘Kuthula ate (the) polenta.’
b.
u-Kuthula u-Ø -dl-ile
(isi-tshwala).
1a-Kuthula 1S-TNS-eat-ASP.D 7-polenta ‘Kuthula ate (the) polenta.’ Abloiu & Avery 2009: 166
The grammatical relations are expressed in Bantu, first of all, by the constituents order, e.g. SVO. In the case of subject, it should be marked obligatorily in declarative, indicative sentence. But the way of marking object varies from one Bantu to another. Object marking as index at the head verb is another strategy of coding grammatical relation in addition to the constituent order. In Ndebele as other conjoint/disjoint Bantu languages like Zulu, Makhuwa, Tswana the object marking at the verb requires disjoint construction normally without following object NP. In this sense, it is the type of ‘pro-indexing’ according to Haspelmath (2013)’s three ways-classification of indexing.
(35) Ndebele
a.
u-Phita
u-Ø -yi-khab-ile/-*e
in-ja.
1a-Peter
1S-TNS.-9OM-kick-ASP.DJ
9-dog
‘Peter kicked the dog.’
b.
u-ya/-*Ø -ku-dl-a
uku-dla
1S-TNS.DJ-15OM-eat-ASP 15-food ‘S/he is eating/eats the food.’
CJ/DJ-distinction in Zulu are realized almost same as in Ndebele example above. The optional occurrence of object NP in DJ construction is interpreted as sentence external element. The more typical examples are following sentences as the traditional grammars call CJ as “short”, or sentence 18
non-final form, while DJ is called as “long” and clause-final form.
(36) Zulu a.
Ngi-ø-fund-a
i-n-cwadi.
1S-CJ-read-FV AUG-9-book ‘'I’'m reading a book.’'
b.
Ngi-ya-fund-a. 1S-DJ-read-FV ‘'I’'m reading.’'
Zeller et al. 2015 [2017]
Peculiarity lies in the fact that this alternation belong to the indexing, in other word a type of headmarking, which do not correspond to the agreement with other co-occurring argument nouns, but it just show the pattern of argument frame of the clause. Through this alternation the speaker choose the boundary of core arguments which are governed directly by the verb, while the core argument can be disposed to the right outside of the governing boundaries. In this way the dislocated NP now functions as focus or as resumptive information unit of already announced by object marker.
8. Concluding remarks
In this paper I have tried to introduce six typologically peculiar constructions across the genealogical boundaries. The languages which are treated here are located also very broadly throughout the continent. One of basic assumption of the comparison of the constructions is that the peculiar constructions are the outcome of linguistic contact between the diverse languages genetically and typologically. These structures are sometimes hard to assign into already well established certain type of language, so in some cases it is almost impossible to describe adequately in terms of syntactic terminologies for clause description. At that point we call the more widely applicable concepts of topic of focus, which are not be easily understandable without considering the communicative situations like speech act and behavior. Now we extend out view from the sentence structure to more wide discourse-pragmatic functions for explaining the complex linguistic phenomena.
19
Abbreviations 1, 2, 3
first, second, third person
ABS
AUG
augment
ACC, acc
Appl
applicative
ASP
aspect
ATR
advance tongue root
C
completive, common
CJ, C
conjoint
CL
class
Cop
copular
DAT
dative
DECL
declarative
DEF
definite
Dem
demonstrative
DJ, D
disjoint
ERG
ergative
F
feminine
Foc
focus
FUT
future
FV
final vowel
IMP
imperfective
IND
indicative
IRR
irrealis
j
juncture
LINK
linker
M, m
masculine
m
predicative marker
MAN
manner
MPO
multi-purpose oblique marker
NEG
negative
NOM
nominative
NOML nominalizer
NP
noun phrase
O, obj, OBJ
OM
object marker
object
absolutive accusative
Pass
passive
PER, PF, PERF perfect
PL
plural
PNG
person, number, gender
POS
positive
POSS
possessive
PP
prepositional phrase
PQ
question particle
PRE
present
PREP
preposition
PRO
pronoun
PST, PAST, pst past
RLS
realis
S, sub
subject
SG, sg singular
SUF
suffix
TNS
tense
V
verb
VE
valence external argument marker x
adjunct
20
Abdoulaye, Mahamane L. 2011. “Relative clauses in Hausa: a grammaticalization perspective” in JALL 32: 1-41. Ahland, Michael. 2009. "From topic to subject, grammatical changes in the Amharic possessive construction" in Studies in Language 33, 3: 685-717. Aoun, Joseph, E. Benmamoun & L. Choueiri. 2010. The syntax of Arabic. Cambridge University Press. Baker, M. & R. Kramer. 2014. "Rethinking Amharic prepositions as case markers inserted at PF" Lingua 145: 141-172. Bendor-Samuel J. (ed.). 1989. The Niger-Congo languages. Lanham, MD: University Press of America. Buell, Leston Chandler. 2005. Issues in Zulu verbal morphosyntax. diss. University of California Los Angeles. _______. 2006. “The Zulu conjoint/disjoint verb alternation: focus or constituency?” ZAS Paper in Linguistics 43: 9-30. Claudi, Ulrike. 1994. "Word order change as category change", In W. Pagliuca (ed.), Perspectives on grammaticalization, 191-231. Amsterdam: John Benjamins. Clements, N. & A. Rialland. 2008. “Africa as a phonological area” in Heine & Nurse (eds.) 2008: 3687. Crass, J. & R. Meyer. 2008. “Ethiopia” in Heine & Nurse (eds.) 228-250. Creissels, D. 1996. “Conjunctive and disjunctive verb forms in Tswana” South African Journal of African Languages 16: 109-115. _______. 2007. A sketch of” Bambara argument structure. ms. ______. 2017. “The conjoint/disjoint distinction in the tonal morphology of Tswana” in van der Wal & L. Hyman (eds.) 2017. Creissels, D., G. Dimmendaal, Z. Fraizyngier, & C. König. 2008. "Africa as a morphosyntactic area" in Heine, B. & D. Nurse (eds.) 2008: 86-150. Dimmendaal, Gerrit. 2005. “Head marking, dependent marking and constituent order in the Nilotic area” in Voeltz(ed.) 71-92. Dimmendaal, Gerrit. 2008. "Africa's verb-final languages" in Heine & Nurse(eds.) 2008:272-308. 21
Dimmendaal, Gerrit. 2011. Historical Linguistics and the Comparative Study of African Languages. John Benjamins Publishing Company. Amsterdam/Philadelphia. ________. 2012. "Marked nominative systems in Eastern Sudanic and their historical origin" Afrikanistik Online. pp. 23. Eaton, Helen. 2002. “focus as a key to the grammar of Sandawe” ms. Elderkin, Edward D. (1989). The significance and origin of the use of pitch in Sandawe. Unpublished D. Phil dissertation. University of York, Heslington, York. Ferguson, Ch. 1970. “The Ethiopian language area” Journal of Ethiopian Studies 8,2: 67-80. Gensler, O. 1994. On reconstructing syntagm S-Aux-O-V-Other to Proto-Niger-Congo. Proceedings of the Twentieth Annual Meeting of the Berkeley Linguistics Society: Special Session on Historical Issues in African Linguistics. pp. 1-20 Greenberg, J. 1959. Africa as a linguistic area. In Bascom & M. Herskovitz (eds.) Continuity and change in African cultures. Chicago: Chicago University Press. 15-27. ______. 1963. The Languages of Africa. The Hague: Mouton. _____. 1966. "Some universals of grammar with particular reference to the order of meaningful elements" Greenberg(ed.) Universals of Language. Cambridge, Massachusetts: M.I.T. Press. pp. 73-113. _______. 1983. Some characteristics of African languages. In Dihoff I. (ed.) Current approaches to African linguistics 1. Dordrecht: Foris. 3-22. Güldemann, Tom. 1998. The Kalahari basin as an object of areal typology - a first approach, in Schladt (ed.) 137-69. _______. 1999. "Head-initial meets head-final: nominal suffixes in eastern and southern Bantu from a historical perspective, SAL 28(1):P 49-91. _______. 2004. "Linear order as a basic morphosyntactic factor in Non-Khoe Khoisan", Syntax of the world's languages, Keipzig, 8th. Auguist 2004. ms. _______. 2006. "Structural isoglosses between Khoekhoe and Tuu: the Cape as a Linguistic Area" in Matras et al. 2006: 99-134. _______. 2007. Linguistic areas without evidence of contact. Paper presented at the symposium “Language contact and the dynamics of language”, Leipzig. ______. 2008. The Macro-Sudan belt: towards identifying a linguistic area in northern sub-Saharan Africa. In Heine & Nurse (eds.) 151-185. _______. 2010. Sprachraum and geography: linguistic macro-areas in Africa. In Lameli et al. (eds.) 22
Language and space: an international handbook of linguistic variation, vol. 2. Language mapping. Berlin: Mouton de Gruyter. 561-585. Güldemann, T. and E. D. Elderkin. 2010. On external genealogical relationships of the Khoe family. In Brenzinger, Matthias and Christa König (eds.), Khoisan languages and linguistics: proceedings of the 1st International Symposium January 4-8, 2003, Riezlern/Kleinwalsertal. Quellen zur Khoisan-Forschung 24. Köln: Rüdiger Köppe, 15-52. Güldemann T. & R. Vossen. 2000. Khoisan, in Heine & Nurse (eds.) 99-122. Gunnink, H., B. Sands, B. Pakendorf & K. Bostoen. 2015 “Prehistoric language contact in the Kavango-Zambezi transfrontier area: Khoisan influence on southwestern Bantu languages”. Journal of African Languages and Linguistics. 36, 2: 193-232. Halpert, Claire. 2015. “Prosody/syntax mismatches in the Zulu conjoint/disjoint alternation” ms. Haspelmath, Martin. 2013. "Argument indexing: a conceptual framework for the syntactic status of bound person forms" in Bakker & Haspelmath(eds.) to appear. Language across boundaries: Studies in memory of Anna Siewirska. Haspelmath, Martin. 2018/19. "Indexing and flagging, and the head and dependent marking" ms. to appear in Te Reo Hayward, R. 2000. Afroasiatic. In Heine & Nurse (eds.) 74-98. Heine, Bernd. 1976. A typology of African languages, based on the order of meaningful elements. Berlin: Dietrich Reimer. Heine, B. & D. Nurse (eds.) 2000. African Languages: An introduction. Cambridge: Cambridge University Press. _________. 2008. A linguistic geography of Africa. Cambridge University Press. Heine, B. & H. Honken. 2010. The Kx'a family. A new Khoisan genealogy, JAAS 79: 5-36. Kawachi, Kazuhiro. 2007. A Grammar of Sidaama (Sidamo), a Cushitic language of Ethiopia. ph.D thesis, University at Buffalo, State University of New York. Kilian-Hatz, Christa. 2008. A grammar of modern Khwe (Central Khoisan). Cologne: Rüdiger Köppe. _______. 2013. Syntax: Kxoe subgroup: Khwe. in R. Vossen (ed.) The Khoesan languages. 356-378. London: Routledge. Kieβling, Roland. 1996. Verbal inflectional suffixes in the West Rift group of Southern Cushitic. in Griefenow-Mewis & Voigt(eds.) Cushitic and Omotic Languages, Proceedings of the 3rd International symposium, Berlin, March 17-19, 1994. 59-70. ______. 2000. Some salient features of Southern Cushitic (Common West Rift), Lingua Posnaniensis 23
42: 69-89. Kiessling, R. 2000. "Verbal plurality in Sandawe" HAAPI (2000) ms. Kiessling, R. M. Mous & D. Nurse. 2008. "The Tananian Rift Valley area" in Heine & Nurse 2008: 186-227. König, C. 2006. “Marked nominative in Africa” in Studies in Language 30,4: 655-732. _______. 2008. Case in Africa. Oxford: OUP. Kwon Myong-Shik. 2009. Problems of establishing universal semantic roles; with special reference to the relation of Kiswahili predicate and its participants [in Korean], Journal of the Korean Association of African Studies 30: 3-30. ________. 2016. Typological characteristic of marking syntactic relation of Cushitic, with particular reference to Sidaama, [in Korean] presented at the Annual Conference of Korean Association of African Studies. 1st June. 2016. Maddieson, Ian. 1984. Patterns of sounds. Cambridge: Cambridge University Press. McGregor, William B. 2018(?) "Emergence of optional accusative marking in Khoe languages" in Seržant, I. & A. Witzlack-Makerevich (eds.) Diachrony of differential argument marking. 243-279. Mous, Maarten. 2005. "Selectors in Cushitic" Voeltz(ed.) 303-326. Nichols, Johanna. 1986. 'Head-marking and dependent marking grammar', in Language 62: 56-119. ______. 1992. Linguistic Diversity in space and time. Chicago IL: University of Chicago Press. Van der Spuy, A. 1993. “Dislocated noun phrases in Nguni”. Lingua 90, 335-355. Van der Wal, Jenneke. 2011. “Focus excluding alternatives: conjoint/disjoint marking in Makhuwa” Lingua 212, 11: 1734-1750. Van der Wal & L. Hyman. 2017. The conjoint/disjoint alternation in Bantu. Berlin, Mouton de Gruyter. Voeltz, E. 2004. Long and short verb forms in Zulu. Ms., University of Cologne. Voeltz, Erherd F. (ed.) 2005. Studies in African linguistic Typology. John Benjamins Publishing Company, Amsterdam/Philadelphia.
24