Studies in Language 30:4 (2006), 655–732.
issn 0378–4177 / e-issn 1569–9978 © John Benjamins Publishing Company
Christa König
Frankfurt am Main, Germany
Africa is a continent where grammaticalized case systems are a rare phe- nomenon. But there is one exception: East Africa is a region where there is a relatively high occurrence of case languages (that is, languages with a grammaticalized case system). With regard to the type of case systems which occur in Africa, again, the picture is crosslinguistically unbalanced as there are hardly any ergative languages. In other words, of the two most common case types worldwide, accusative and ergative(/absolutive), essentially only one is represented in Africa, namely the accusative type. From a worldwide perspective, Africa seems to be a continent where case has nothing special to offer. However, in East Africa there are so called marked-nominative languages which seem to be quite unique worldwide. hey are somehow a mixture: On the one hand they share features with prototypical accusative languages, on the other hand they share features with prototypical ergative languages.
In the present paper I will, first, define the typical features of a marked- nominative language. Second, I will give an overview of the languages which have a marked-nominative system. hird, I will deal with the question of whether the distribution of marked-nominative languages is genetically or areally motivated. And fourth, I will speculate on how such unusual systems could have developed.
. Preliminaries
. Marked-nominative languages
In order to define typical features of a marked-nominative language, it is neces- sary to briefly illustrate how prototypical case systems are defined. Case systems are distinguished with regard to the three basic syntactic functions as defined by Dixon (1994:62ff.) and others, namely S, the intransitive subject function, A, the transitive subject function, and O, the transitive object function. In an
accusative system (accusative in short), S and A are treated the same and si- multaneously different than O. In an ergative system, S and O are treated the same and simultaneously different than A. hese patterns are illustrated in Fig- ure 1. he case which covers S and A in an accusative system is called nomina- tive1 and the case covering O accusative. he case which covers A in an ergative system is called ergative and the case covering S and O absolutive. Furthermore the nominative of an accusative system is prototypically morphologically the unmarked form,2 functionally the unmarked form, and the form used in cita- tion. he absolutive of an ergative system on the other hand is prototypically the morphologically unmarked form, functionally the unmarked form, and the form used in citation.
With “morphologically unmarked” I mean zero realization, “morphologi- cally marked” means morphologically non-zero realization. “Functionally un- marked” means being used in a wide range of different functions, which oten is the default form. “Functionally marked” means being used in few functions only, it is not the default form. he morphologically unmarked form is some- times called ‘basic form’ in the literature. he morphologically marked form
S = intransitive subject function A = transitive subject function O = transitive object function
O A
nom acc erg abs
Accusative system Ergative/(Absolutive)-System Nominative = morphologically
unmarked Absolutive = morphologically unmarked
= functionally
unmarked = functionally unmarked
= used in citation = used in citation
A
S
O
S
Figure . Case systems.
is derived3 from the morphologically unmarked form by adding some extra element. he morphologically unmarked form is shorter and/or underived vis- à-vis the morphologically marked form.
Marked-nominative languages are a mixture of both systems, as pointed out by Dixon (1994:64f.): he pattern of A, S and O is the same as in accusative languages, namely A and S are treated the same and simultaneously different than O. hey share this feature with accusative languages. But the accusative in marked-nominative languages is the morphologically unmarked form, at least typically (see below); it is used in citation, and it is also functionally the unmarked form. he nominative on the other hand is the morphologically marked form in a marked-nominative system; A, the transitive subject, there- fore is encoded by the morphologically marked form. his feature of marked- nominative languages is shared with ergative systems.
heoretically one could also argue that accusative and ergative systems are a mixture as they share the feature that S is morphologically unmarked. How- ever this feature is the expected one. his can be seen in Greenberg’s universal 36: “Where there is a case system, the only case which ever has only zero al- lomorphs is the one which includes among its meanings that of the subject of the intransitive verb.” (Greenberg 1966:95)
he oddity of marked-nominative systems can be seen in the fact that they violate Greenberg’s universal 36.
From a typological point of view one may wonder whether there are er- gative languages which show an unusual markedness with a morphologically marked absolutive and a morphologically unmarked ergative. According to Plank (2005) there are a few ergative languages in which the absolutive, encod- ing S and O, is morphologically marked vis-à-vis a morphologically unmarked ergative, encoding A.4 Two types are distinguished: (a) Either there is marking on noun phrases which are not focused or topicalized and are not case-marked at all, as in Nias (Sundic, Western Malayo-Polynesian, Austronesian), or (b) there is marking on noun phrases which are focused, as in Yukaghir (isolate or Uralic). According to (Plank 2005), in Chukchi (Chukchi-Kamchatkan) the absolutive singular, encoding S, is the morphologically most complex noun form, expressed by a variety of suffixes, or by reduplication of the stem (see also Donohue & Brown 1999).
Within accusative languages, the nominative is typically encoded in a mor- phologically unmarked form but there are some languages where both case forms, nominative and accusative, are morphologically marked. Two subtypes therefore need to be distinguished among the accusative languages.
In a similar fashion, with regard to the morphological markedness of nom- inative and accusative, two subtypes among the marked-nominative languages are to be distinguished:
Type 1 (the more common one), in which the accusative is the morpholog- ically unmarked form and the nominative the morphologically marked form, and type 2, in which both case forms, nominative and accusative, are morpho- logically marked. In type 1 of marked-nominative languages, the accusative is morphologically unmarked, functionally unmarked and used in citation. In type 2, the accusative is morphologically marked, functionally unmarked, and used in citation.
Of the three properties used above to define a particular case system, i.e., morphological markedness, functional markedness, and citation form, one is criterial, namely functional markedness. In a prototypical case system all three properties apply. he degree to which a marked nominative system differs from an accusative and an ergative system varies. Type 1 marked nominative languages share more properties with ergative languages than type 2 marked nominative languages.
In sum, marked nominative languages are defined as follows: A marked nominative language is present when at least two cases are distinguished, namely an accusative covering O, and a nominative covering S and A. he ac- cusative must be the functionally unmarked form; it is the default case, that is, the case which is used with the widest range of functions. If one of the two cases is derived from the other, it must be the nominative which is derived from the accusative and never the other way round. Within marked nomina- tives, type 1 shows more marked nominative properties than type 2.
In the following I will classify a language as a marked-nominative language even if it is only defectively so, meaning that there are neutralizations where there no longer is any case distinction. Conditions for neutralization are e.g. definiteness (Wolaitta), person (Datooga), noun phrase structure (Dinka), gender (Cushitic), number (Cushitic), or constituent order (Suri-Chai). If a marked-nominative language shows such a defective system I refer to this as a split language.
he morphologically unmarked form in marked-nominative languages should be systematic or “extra-paradigmatic”, meaning that all nouns of the language appear in the morphologically unmarked form to cover the accusa- tive, and not only a few nouns or noun classes. Extra-paradigmatic is used by Carstairs (1981) for systematically morphologically unmarked forms (such as the nominative or absolutive in Turkish) as opposed to intra-paradigmat- ic, which is used if the forms are only sometimes morphologically unmarked (such as the nominative in Latin).
.2 Terminology
Marked-nominative systems are sometimes also called ‘extended ergative’. Dixon considers both terms ‘extended ergative’ and ‘marked nominative’ to be rather insufficient, but he prefers the latter:
In summary, we have distinguished three kinds of ‘markedness’ among case inflections covering the three core syntactic functions A, O and S. Basically, either of the transitive functions can be marked. If A is marked (by ergative case) then both O and S may be shown by the unmarked ‘absolutive case’, which will again be used for citation. But the marking on A can also be ex- tended to cover S, with the unmarked case being confined to O function and most instances of citation. Strictly speaking, none of the terms ‘nominative’,
‘accusative’, ‘absolutive’ or ‘ergative’ are really appropriate for this third pos- sibility. I will employ ‘marked nominative’ as less potentially confusing than
‘extended ergative.’ (Dixon 1994:66f.)
his quote may illustrate how different the marked-nominative system is in comparison to the two other standard case systems, ergative and accusative.
In East Africa a somewhat confusing terminology has been used: he mor- phologically unmarked form is called absolute or even absolutive, as it is called in ergative languages. What makes it so confusing is the fact that absolute is used to refer either to a nominative in an accusative system or to an accusative in a marked-nominative language. It is even used for a non-zero marked ac- cusative in a marked-nominative language, namely in Haro (an Omotic, East Omotic language, Afroasiatic) (cf. Woldemariam 2003:645). To illustrate this: Tucker and Bryan (1966:14) for instance use the term ‘absolute’ either for the nominative or for the accusative case, and it is always the case in citation:
he Term ‘Absolute’ refers to the form of the Noun or Pronoun, &c., used when cited. In some languages this is in the Nominative Case, in others in the Accusative Case. (Tucker and Bryan 1966:14)
In Cushitic languages such as Bilin and Awngi (called Awia by Tucker and Bry- an 1966), the ‘absolute’ refers to the nominative case and in languages such as Sidamo, Oromo (called Gala by Tucker and Bryan), Somali, and Beja (called Bedauye by Tucker and Bryan), it refers to the accusative case.
In Kemantney, a type 2 accusative language with case being restricted to definite nouns, the term absolutive refers to the morphologically unmarked form, that is, the case-less form of the noun. With indefinite nouns, the absolu- tive is used for S, A and O (see Leyew 2003:237ff. and König forthcoming a).
Outside of Africa the term absolute is used to refer to the unmarked nomi- native, as e.g. by Lewis (1967). He used ‘absolute’ for the Turkish morphologi- cally unmarked nominative case form.
In the general literature about case, there is also a clear opinion on how the terms absolutive and nominative should be used. Blake for instance points out that he strictly wants to restrict the term absolutive for grammatical relations subsuming S + O (= P in his terminology) only and nominative for all other options including S + A:
he term absolutive is used in some circles for a case or case form covering S + P functions. Normally such forms are unmarked. (Blake 1994:26)
My preference is to use nominative for the case that is used to encode S irre- spective of whether this case covers S + A or S + P or indeed whether it is ex- clusive to S or covers S + A + P. his case will normally be unmarked and will be the case used in isolation from constructions. I will reserve the term abso- lutive for a grammatical relation that subsumes S and P. (Blake 1994:187) I will follow Blake and avoid the term absolutive when not covering S and O. I will stick to nominative covering S and A, irrespective of whether it is morpho- logically marked or unmarked, accusative for the case covering O, irrespective of whether it is morphologically marked or unmarked. he term absolute will be avoided completely here since its use may be confusing.
In an article about case in Maasai, Mel’čuk calls the accusative (my termi- nology) the nominative and the nominative (my terminology) either ‘oblique’ or ‘ergative’ or ‘subjective’ (see Mel’čuk 1997:57 & 136). Furthermore he states that Maasai has an ergative structure. His argumentation goes as follows: i. Mel’čuk takes the citation form as the sole basis for the case label. In his
view, the form of the noun used in citation must be called nominative, ir- respective of the syntactic use of this form:
In a language L that has grammatical cases, the case used to NAME objects or situations, i.e. to mark a noun in isolation, must be called the nominative, whichever role it plays in the syntax of L and whichever is its formal exponent (Mel’čuk 1986:71).
ii. Maasai, as will be shown below (see Maa), is a marked-nominative lan- guage where the form of the noun used in citation is identical with the form of the noun used to encode O. If the syntactic use of a case is ignored, the accusative of a marked nominative language (my terminology) would be called nominative in Mel’čuk’s terminology.
iii. Mel’čuk argues that since the structure of Maa is not nominative it is erga- tive (1997:141).6
In the present paper the terms ergative and nominative will be used with regard to the case patterns they cover as introduced in 1.1. herefore I will not fol- low Mel’čuk’s proposal. Accordingly, Maasai is not an ergative language and his nominative will be called here accusative. In the general case literature, Mel’čuk’s view about ergativity has been criticized (cf. Dixon 1994:21–22). In particular Mel’čuk’s statement that Lezgian is an ergative language has been criticized by other authors working on Lezgian languages (cf. Haspelmath 1991).
Sasse (1984:11–112) proposes the term subject case for the nominative when dealing with East Cushitic languages (Gidole, Saho, Konso etc.). In the Berber linguistic tradition the nominative is referred to by the terms annexed state or oblique case. he morphologically unmarked form, the accusative, is referred to by the terms free state or direct case (see Aikhenvald 1990:113).
As this short excursion into the literature might have illustrated, the case phenomenon discussed here is covered by a range of even contradictory terms.
In Table 1 the somewhat confusing terminology is summarized. he ter- minology used here is found in the last line. Some authors are listed twice, such as Tucker & Bryan. his is due to the fact that, as we saw above, they use the term absolute in both a marked-nominative and an accusative language.
Table . Various case labels used in the literature for similar functions.
Author Function
S + A O S + O Citation form Case-less form
Tucker & Bryan, Dimmendaal, Last & Lucassen
Nominative Absolute Absolute
Hayward, Kießling, Wolde- mariam
Absolutive Absolutive
Lewis Absolute
Tucker & Bryan Absolute Accusative Absolute
Sasse Subject case
Mel’čuk Oblique,
Ergative Subjective
Nominative Nominative
Leyew Absolutive
Blake, König, Randal
Nominative Accusative Absolutive
2. Case studies of marked-nominative languages
I will now illustrate the behavior of marked-nominative systems by discuss- ing a variety of marked-nominative languages. To be consistent, I change the terminology used by the authors if necessary. I also change the glosses and explicitly gloss the unmarked case form as accusative in order to avoid misin- terpretations. If needed, I add a gloss.
2. Tennet (South West Tennet, Surmic, Nilo-Saharan)
he following description of case in Tennet is based on Randal (1998:219–272). Tennet is a verb-initial language, with a basic VAO word order, and peripheral participants occur sentence-finally. Tennet is a case language with inflexional case. Four cases are distinguished, namely accusative, nominative, genitive, and oblique. he accusative is the morphologically unmarked form. All other cases are marked by suffixes or tone. he oblique case covers peripheral partici- pants such as time, location and instrument. In a transitive clause with a VAO structure, A is encoded in the nominative case and O in the accusative case. he nominative case is the morphologically marked form, expressed either by the suffix -I, as in the following example, or by the suffix -a, or by tone in other environments. he nominative covers A and the accusative O (see 1).
(1) "¢k"¢t Lowór-I Yomá. VAO PFV.spear Lowor-NOM Yoma.ACC
Lowor speared Yoma. (Randal 1998:231)
In an intransitive clause with VS order, S appears in the nominative case (see 2).
(2) ~¢k mányúdí-I mgI¢n""tI¢. VS go.PFV squirrel-NOM there
Squirrel went there. (Randal 1998:231)
With regard to the case schema illustrated in Figure 1, we are dealing so far with a pattern following the accusative system, as A and S are treated the same, namely both are encoded by the nominative and are simultaneously different than O, as O appears in the accusative.
On the other hand, the case profile of Tennet resembles an ergative pat- tern in so far as A is encoded in the morphologically marked form, namely the nominative case, by means of a suffix or tone.
In copula constructions, the nominal predicate appears in the accusative form, as in 3a. In ‘I am a teacher’ the copula is clause-initial, the subject ‘I’ ap- pears in the nominative and the nominal predicate ‘a teacher’ appears in the accusative clause-finally.7 here is a corresponding construction without any copula present as in 3b. 3b is used for instance as the response to the question
‘who is the teacher’. In 3b both participants, S and the nominal predicate, occur in the accusative.8
(3) a. k-eénj "nn"¢ deméz-zóh-t. COP S N.PRED 1-be 1SG-NOM teach-AGEN.ACC-SG NOM ACC I’m a teacher. (Randal 1998:233 & 2000:72)
b. anét deméz-zóh-t. S N.PRED
1SG.ACC teach-AGNM-SG.ACC ACC ACC I’m the teacher. (Randal 2000:71)
Peripheral participants introduced by the applicative are also encoded in the accusative form. his may be illustrated by 4a and 4b. In both clauses the va- lency of the verb is increased by the applicative in order to include the recipient of the action. he recipient occurs always in the accusative, whether placed clause initially between A and O (as in 4a) or clause finally (as in 4b). In 4a and 4b the accusative occurs twice, for O and the recipient.
(4) a. i-ttón-êk Lokóri-I Yomá k"vIy"¥k. V A PP O PFV-send-APL Lokor-NOM Yoma.ACC news.ACC
Lokori sent news to Yoma. (Randal 1998:244)
b. i-ttón-êk Lokóri-I k"vIy"¥k Yomá. V A O PP PFV-send-APL Lokor-NOM news.ACC Yoma.ACC
Lokori sent news to Yoma. (Randal 1998:244)
Nominal possession can be expressed by two different constructions. In the first construction the possessee precedes the possessor, the possessor is encoded in the genitive, and a special particle is standing between the two, called associa- tive9 by Randal (see 5a). Alternatively, nominal possession can be encoded in a possessor-possessee order, but only expressed by juxtaposition. No genitive case is needed; instead, the possessor occurs in the accusative case, as ‘fish’ in 5b, followed by the possessee. 5a and 5b are variants expressing the same con- tent. Alternatively, 5b can be analyzed as an instance of possessor raising. (5) a. k-7-t7¢7¢d-" ff cI¢ ulúg-t-ó. PEE AM POR
1-PRF-cut-1SG head.ACC AM fish-SG-GEN I cut the head of the fish. (Randal 1998:241)
b. k-7-t7¢7¢d-" ulúg-t ff. POR PEE 1-PFV-cut-1SG fish-SG.ACC head.ACC
I cut the head of the fish. (Randal 1998:241)
he citation form of nouns is the accusative, as illustrated in 6a. (6) a. k"vIy"¥k.
news.ACC News.
b. Lfk~´lI cí á-r~´h Lfh"¥m. AVO Lokuli.ACC AM IPV-beat Loham.ACC
It is Lokuli who is beating Loham. (Randal 1998:261).
At least to some extent, S and A appear in Tennet in the accusative if placed before the verb. his holds for some focus constructions, as in 6b; in others however this does not apply (see further König forthcoming a).
Figure 2 gives an overview of the functions covered by the nominative and the accusative in Tennet. To summarize, the nominative covers S and A, the accusative covers O, peripheral participants introduced by verbal derivation such as the applicative, citation, nominal predicate in copula clauses, and pos- sessor in juxtaposed possession. In other words, one could say the accusative is the default case to encode dependent nouns if no special case is required. he accusative has a much wider range of uses; therefore it is also functionally unmarked.
nom subject (S & A) ater the verb
(a) citation form (b) O
acc (c) nominal predication
(d) subject (S & A) before the verb (sometimes) (e) possessor in juxtaposed possession (f) peripheral participants introduced by head- marking devices (verbal derivation) such as the applicative
Figure 2. Functions covered by the nominative and accusative case in Tennet.
2.2 Maa (East Nilotic, Nilo-Saharan)
Maa is a VAO language. Two cases are distinguished, nominative and accusa- tive (called absolute by Tucker and Mpaayei 1955).10 Case inflexion is marked by tone, meaning that accusative is the basic form, and the nominative can be derived from the basic accusative form by rules. he basic form is morphologi- cally and functionally unmarked. Note that the language has complex rules, not all nouns follow the same rules in order to derive the nominative from the accusative form; instead, Maa has according to Tucker and Mpaayei (1955) four basic different noun classes, each having subclasses which follow different rules. In the following examples, high tone is marked by an acute accent, and low tone is let unmarked. To illustrate how the rules work by which the nomi- native form is derived from the accusative form, there is a list of four nouns in Table 2 all following the same rule, all belonging to class I. he tone rule for the case inflexion goes as follows: “Leading stem syllables are depressed in the Nominative… all syllables have lowered tone except the last” (Tucker and Mpaayei 1955:177). Maa is a language with vowel harmony. Generally speak- ing, within a word all vowels are either [+ATR] or [−ATR].11 Basically, the harmony is triggered by the dominant lexical root. his means that grammati- cal affixes or function words occur always in both [ATR]-qualities, such as the preposition t7 or te (cf. 18 and 20; note that all allomorphs of a certain element are not always enumerated).
Table 2. Case inflexion in Maa of nouns belonging to class I (Tucker and Mpaayei 1955:177).
Accusative Nominative Meaning
7l~¢k~¢nyá 7l~k~nyá ‘head’
7mányátá 7manyatá ‘warriors’ village’
7mbártá 7mbartá ‘horse’
entíto entitó ‘girl’
he following Maa examples are either from Maasai, a Southern Maa dialect, or from Camus and Samburu, closely related Northern Maa dialects. Unfortu- nately some of Tucker and Mpaayei’s data on Maasai (1955) is not marked for tone. herefore, whenever in the following examples there is no tone marking this is due to this fact.
In a VAO structure, which is the basic constituent order, A occurs in the nominative, as fl-kI¢t7`ŋ ‘ox’, and O occurs in the accusative, as en-kóítóí ‘road’ in 7 (called absolute by Tucker and Mpaayei 1955).
(7) 7¢-df¢l-I¢ta fl-kI¢t7`ŋ en-kóítóí. V A O 3.SG-see-PROG M.SG-ox.NOM F.SG-road.ACC
he ox sees the road. (Tucker and Mpaayei 1955)
In an intransitive clause S appears in the nominative as fl-kI¢t7`ŋ ‘ox’ in example 8.
(8) e-kuet-ita fl-kI¢t7`ŋ. V S
3.SG-run-PROG M.SG-ox.NOM
he ox is running. (Tucker and Mpaayei 1955:7) In citation, the accusative is used, as in 9.
(9) fl-t~ŋánì M.SG-person.ACC
person (Tucker and Mpaayei 1955:175)
In copula clauses, the nominal predicate appears in the accusative case, as
‘Sironka’ does in 10. (10) á-rá Sirónkà.
1.SG-be Sironka.ACC
I am Sironka. (Tucker and Mpaayei 1955:175)
he language has two prepositions, one (f¢ or ó) takes the nouns in the accu- sative, the other (t7) in the nominative case. hus, in 11, f¢ is the preposition which takes the following noun in the accusative.
(11) á-df¢l In-k7¢rá ó n-kitúààk 1.SG-see F.PL-child.ACC PREP F.PL-woman.ACC
I see the children and the women. (Tucker and Mpaayei 1955:215) he possessor ater the genitive particle12 is encoded in the accusative, as ol- ayíónì ‘boy’ is in 12, irrespective of whether the possessee is encoded.
(12) é-ípot fl-cf¢r7 l-f¢ [o]l-ayíónì 3.SG-call M.SG-friend.NOM M.PEE-M.SG.POR M.SG-boy.ACC he friend of the boy called him. (Tucker and Mpaayei 1955:213)
Maa has a rich verbal derivation; some of the categories are valency-increasing devices. Maa has predominantly a head-marking strategy to encode peripheral participants by verbal derivation. Among the valency-increasing devices there are applicative, causative, and instrumental. hey all follow the same rule with regard to participant marking. he participant introduced by the verbal deriva- tion always occurs in the accusative form. hus, in 13, the beneficiary ‘father’
occurs in the accusative. It is introduced by the verbal applicative suffix -ókì (with the allomorphs -akI, -oki in the imperfective and -aka, -oko in the perfec- tive). In 14, the instrumental participant ‘knife’ is encoded in the accusative; it is introduced by the verbal instrumental derivation, marked by the suffix -íé. (13) á-ból-ókì papá f¢l-b7¢n7.
1.SG-open-APL father.ACC M.SG-basket.ACC
I open the basket for father. (Tucker and Mpaayei 1955:129) (14) á-dúŋ-íé 7nk-ál7¢m.
1.SG-cut-INST F.SG-knife.ACC
I cut it with a knife. (Tucker and Mpaayei 1955:142)
In 15, the verb is derived by the causative, and the agent 7n-k7¢ráí ‘child’ is en- coded in the accusative.
(15) 7-Ita-nak 7n-kitok 7n-k7rai. 3-CAUS-suck F.SG-woman.NOM F.SG-child.ACC
he women suckles the child. (Tucker and Mpaayei 1955:147)
Whenever the subject, S or A, occurs before the verb, as in focus constructions for instance, it no longer appears in the nominative but in the accusative, as in 16. In 16 there is an AVO order.
(16) en-tító na-df¢l nIny7¢. F.SG-girl.ACC REL.F.SG-see 3.SG.ACC
It is the girl who sees him. (Tucker and Mpaayei 1955)
Further functions covered by the nominative, besides encoding S and A ater the verb, are as follows: (a) he nominative occurs ater the vocative particle lf¢ as in 17.
(17) lf¢ t~ŋani. VOC person.NOM
O person! (Tucker and Mpaayei 1955:176)
(b) Ater the second preposition of the language, t7,13 all nouns have to occur in the nominative case. t7 is a preposition which introduces all kinds of peripheral participants; locative (18), beneficiary (19), manner (20), time (20) and instru- ment (21a) are just some of them.
(18) Camus (North Maa)
e-tó-wwán-á t7 n-káji. locative
3-PFV-stay-PFV PREP F-house.NOM
He stayed in the house. (Heine and Claudi 1986:102)
(19) Samburu (North Maa)
k-é-sídáí t7 n-k7rá-í beneficiary
k-3-good PREP F-child-SG.NOM
It is good for the child. (Heine and Claudi 1986:102) (20) Maasai
ti aŋ ‘at home’ locative to layioni ‘by the boy’ beneficiary te keraisho ‘in childhood’ time
ti obor ‘in peace’ manner (Tucker and Mpaayei 1955:42) Sometimes t7 provides an alternative strategy to the verbal derivations that were mentioned above: he following noun always occurs in the nominative, that is, instrumental is encoded in the nominative ater the preposition t7, as in 21a; in 21b the applicative is used instead, with the instrument appearing in the accusative.
(21) a. Camus
a-í-ŋI¢c-a t7 n-kal7¢m. instrument
1-PFV-cut-PFV PREP F-knife.NOM I cut it with a knife.
b. a-í-ŋI¢c-ie n-kál7¢m. 1-PFV-cut-APL F-knife.ACC
I cut it with a knife. (Heine and Claudi 1986:102)
here are also valency decreasing devices, such as passive,14 expressed by a suffix -I¢. Surprisingly, S in passive clauses is encoded in the accusative, such as Sirónkà in 22a. herefore, subjects of passive clauses, as 22 illustrates, are encoded in the accusative. As Heine and Claudi (1986:79–84) convincingly show, the passive suffix goes back to a third person plural pronoun *ki: PASS í <
*ki 3.PL. Literally, the passive construction as in 22a goes back to the structure
‘hey praise Sironka’. he verb still shows traces of the source construction in various ways: First, the subject is cross-referenced as an object (O) and not as a subject (S&A). his can be seen in cases where the subject refers to first or second person. Maa has complex portmanteau pronouns prefixed to the verb, which cross-reference both subject (S&A) and object (O). In 22b the subject refers to first person singular. he pronoun used on the verb however encodes third person subject (S or A) and first person object. Second, Maa has some suppletive verbs sensitive to number agreement. In passive constructions it is always the plural form which is used — even if the subject has singular refer- ence (22c).
With regard to the expression of the agent, the picture is not consistent. According to Heine and Claudi (1986), only some speakers allow the notion of an agent, others don’t. If the agent can be expressed it always occurs consis- tently in the nominative (see 22d–e). In 22d, S (patient) ‘house’ is encoded in the accusative and the agent ‘by the child’ (as a peripheral participant) is en- coded in the nominative. According to Heine and Claudi (1986), the agent in passive clauses is pronounced with a little preceding pause and agents are not allowed to be placed before the subjects. he preceding pause may be a trace of the structure by which the agent was initially introduced into passive clauses: either as a kind of aterthought, or originally introduced by the preposition t7, which always takes the nouns in the nominative. he passive construction is a hybrid, the patient is promoted, but not to the extent that it contains all prop- erties of S found elsewhere. Alternatively one could argue that constructions like 22 are not passives but subjectless clauses. Whatever position one favors, the construction remains complex. For our purpose it is of importance that in constructions of this kind (subsumed under the label passive or passive- like), the only core participant occurs in the accusative and not in the nomina- tive. his holds for the passive constructions in other Nilotic languages as well (see below).
(22) a. Maasai
é-I¢sI¢s-I¢ Sirónkà. 3.SG-praise-PASS Sironka.ACC
Sironka is praised. (Tucker and Mpaayei 1955:175) (Lit.: hey praise Sironka.)
b. Samburu (North Maa) k-áa-ipot-o-kI¢. k-3.1-call-PFV-PASS
I have been called. (Heine and Claudi 1986:80) c. áa-puonunú-i áa-iŋuraa.
3.1-come.PL-PASS INF.PL-look
I’ll come to be seen. (Heine and Claudi 1986:80)
d. k-7¢-jIŋ-I¢ n-kají, n-k7rá-I¢. V S PP (Agent) k-3-enter-PASS F-house.ACC F-child.SG-NOM
he house will be enterd by the child. (Heine and Claudi 1986:80) e. Maasai
e-rik-i [i]nk-ishu aainei [I]l-m~rran. V S PP (Agent) 3-lead-PASS F.PL-cow.ACC 1.SG.F.PL.POSS M.PL-warrior.NOM
My cows will be led by the young men. (Tucker and Mpaayei 1955:81)
A summary of the different functions covered by the accusative and the nomi- native is given in Figure 3, which provides a profile for the nominative and ac- cusative in Maa. As can be seen, the accusative is the case with the wider range of functions. his is so even if the nominative encodes more than just S and A. herefore in Maa as well, the accusative is the case which is not only morpho- logically but also functionally the unmarked one.
(a) subject (S & A) ater the verb
nom (b) agent of passive
(c) vocative
(d) peripheral participants introduced by the preposition t� (a) citation form
(b) O
(c) nominal predicate
acc (d) subject (S & A) before the verb (e) possessor
(f) peripheral participants introduced by head-marking devices (verbal derivation)
(g) patient (S) of passive (e) ater prepositon ��
Figure 3. Functions covered by the nominative and accusative cases in Maa.
2.3 Datooga (South Nilotic, Nilo-Saharan)
Datooga is, like most of the other South and East Nilotic languages, a verb-ini- tial language. It has ATR-vowel harmony.
According to Kießling (2001),15 case is expressed by tone. Two sets of bound pronouns encode subjects, S/A, and objects, O. he bound pronouns show an accusative pattern. he subject set is prefixed, the object set is suffixed to the verbal root. Similar to other Nilotic languages, there is verbal derivation which increases or decreases the valency of the verb.
Unlike in other Nilotic marked-nominative languages where case is ex- pressed by tone, three cases are distinguished: accusative, nominative and, as Kießling calls it, anti-genitive, or anti-genitival (2001:11). he morphologi- cally unmarked form, accusative, is opposed to the nominative and the anti- genitive. Both are derived from the accusative (called absolutive by Kießling 2001) basically by changing the accusative tone pattern into initial and final high for the nominative (cf. Table 3, nominative column), and into final high for the anti-genitive (cf. Table 3, anti-genitive column). he claim that the ac- cusative is basic and the nominative and anti-genitive are the derived cases
is corroborated by the following observations: First, in the accusative, nouns occur in various tone patterns (cf. Table 3, accusative column), whereas the nominative consistently shows an initial and final high, and the anti-genitive a final high. Second, there is a tonal rule according to which declaratives always end in a final low tone, accusatives are resistant to this rule, nominatives not. For Kießling (2001:10) this is a further hint, he argues, that the accusative is not affected by the rule, because it reflects the basic tone pattern of the nouns whereas the nominative is affected since its tone pattern is a surface realization. he tonal shape of the noun might vary on whether it occurs in a prepausal position or not. In prepausal position the nominative loses its final high (cf. Kießling 2001:15) (cf. Table 3, prepausal nominative column). Table 3 gives an overview of the different tonal forms which occur with nouns belonging to dif- ferent tone structures in their accusative form, listed in the let column. Table 3. Case forms in different contexts in Datooga (Kießling 2001:15). Tone
patternAccusa-tive Prepausal accusative Nomina-tive Anti-genitive Prepossessive
nominative Prepausal nominative Meaning
HL wéedà wéeda® wéedá wéedá wéedá wèeda® ‘moon’
LL gèetà gèeta® géetá gèetá géetá gèeta® ‘tree’
HHL éenígà éenîga® éenígá éenígá éenìgá éenìga® ‘rivers’ LHL dìibígà dìibîga® díibígá dìibígá díibìbá bíibìga® ‘children’ HLL búunèedà búnnèeda® búunéedá búunèedá búunèedá búnnèeda® ‘people’
LLL ùhùudà ùhùuda® úhúudá ùhùudá úhùudá úhùuda® ‘head’
According to Kießling (2001:17), the basic constituent order is, as mentioned above, VAO (cf. 25a).16 For pragmatic reasons, other constituent orders are possible, such as VOA for contrastive focus of O (25b), or a topicalized subject, A or S, placed before the verb (AV or SV, cf. 29b).
he accusative covers the following functions: It is used as the citation form of nouns (cf. 23a) and for nominal predicates (cf. 23b). O (cf. 25a), and even IO (cf. 31) appear in the accusative. Topicalized subjects, A/S, placed before the verb, occur in the accusative (cf. 29b). In nominal possession, the possessor always takes the accusative (cf. 26a). he possessee precedes the possessor. he noun ater a preposition (cf. 28 and 29) is encoded in the accusative. According to Kießling (2001:16) all prepositions take the noun in the accusative, examples are presented for the prepositions áa ‘and, together with’ (cf. 28), and jèedá ‘in’ (cf. 30a & 30b). Nominal modifiers, such as adjectives and numerals, also ap- pear in the accusative (cf. 27). Additional participants can be included by ver- bal derivation, such as an instrumental suffix -an (or -a), and a recipient/goal -s
(Kießling 2001:20). he additional participant always appears in the accusative (cf. 35a and 35b) if encoded by the head-marking device of verbal derivation.
he nominative is used to encode S and A ater the verb only (cf. 24 and 25a–25b).
he anti-genitive is used in nominal possession in order to encode the pos- sessee. As mentioned above, the possessor is let unmarked and occurs in the accusative. he possessor follows the possessee (cf. 26a). he anti-genitive not only encodes the head of nominal possession. It is also used with the head in modified noun phrases, such as nouns modified by an adjective (cf. 27), or numerals. If a modified noun phrase fills the slot where the syntax requires the accusative, the head noun appears in the anti-genitive (cf. 27). Concerning the nominative, it remains unclear whether nouns in the nominative also take the anti-genitive if used as heads in a modified noun phrase or as possessee (26b–i). As anti-genitive and nominative both occur with final high tone, it cannot be decided whether final high is triggered jointly by the anti-genitive and the nominative in these contexts or by the nominative alone (cf. Table 3, columns anti-genitive and nominative). Within the NP, the anti-genitive is a head-marking device. If the anti-genitive would be blocked by the nominative and would only occur instead of the accusative it also would have a dimen- sion which goes beyond the NP-level. In the latter case the anti-genitive would be used on the clause level, in complementary distribution with the accusa- tive. Anti-genitive is not only used with complex NPs, and in possessive con- structions, it is also used for head nouns of relative clauses, if otherwise, in a simple NP, it would occur in the accusative (cf. 34a). In 34a, IO occurs in the anti-genitive introducing a relative clause: he accusative of the object ‘elder’ is gwáargwèedà.
he following can be said about case marking within the noun phrase: With regard to the anti-genitive, only the head noun is case marked, the modi- fiers occur in the accusative. With the head in the nominative it is not entirely clear: here are hints that the final high of the nominative at least to some extent spreads to the modifiers; nevertheless, Kießling (2001:1617) says that the modifiers do not show any agreement. In 33, the quantifier sèen ‘all’ appears with a high-low tone, basically with a low tone pattern. he final high of the head is spread to the initial of the modifier (cf. Kießling 2001:14): búunèe sèen
‘people.NOM all.ACC’ > búunèe sêen. In relative clauses, the spread of the final high occurs as well if the head appears in the nominative. In the latter case, the final high of the head noun of the relative clause is replaced by a low tone; the final high spreads to the relative marker; thus, ‘who’ occurs as jáa instead of jàa, its basic form (cf. 34b compared to 34a). he spread of the final high with
the nominative could suggest that the nominative has been encoded originally by a clitic at the end of the noun phrase, possibly the nominative that would have been a phrase marker. In this respect the nominative and the anti-genitive differ, since with the modifiers of the anti-genitive there is no spread of the final high (cf. 27). As the spread of the final high only occurs with the head in the nominative and not with the head in the anti-genitive, this might suggest that nominatives really block the occurrence of the anti-genitive.
In Datooga, case is a split system; it either shows a marked-nominative system or no case distinction at all. Case only appears with nouns and third person pronouns.20 First and second person pronouns do not occur with the nominative. In general, selfstanding pronouns are rather rare, Datooga is a pro-drop language. Bound pronouns on the verb encode subjects and objects in first and second person. In the third person the system is defective, O is not encoded at all, S/A for third person plural always, for third person singular sometimes. he occurrence of selfstanding pronouns is restricted to pragmati- cally marked constructions, e.g. when topicalized. In the latter case, pronouns appear preverbally and therefore in the accusative only. he third person pro- noun mostly appears in the nominative in cases of switch reference in dis- course (see Kießling 2001:19) (cf. 32).
Datooga (23) a. màydà.
calf.ACC
Calf. (Kießling 2001:10)
b. sàawà màanàŋóódìgà gîl 3.PL.NOM wealthy.people.ACC COP
hey are wealthy people. (Roland Kießling p.c.)
(24) qwàyèet bàasta®. V S
burn fire.NOM
he fire burns. (Kießling 2001:9)
Table 4. he cross-reference system of Datooga (Kiessling 2001:19).
Person & number S and A O18
1.SG aa-/da- -aan
2.SG í- -i\i- ~ -ey
3.SG (G)a-/Ø19 Ø
1.PL èe/ si- 77sa
2.PL oo- 77gwa
3.PL ée-/si- Ø
(25) a. gàyqábár qáarèemáŋgá bùŋêeda®. V A O 3P-dig youths.NOM grave.ACC21
he youths will dig the grave. (Kießling 2001:8)
b. gàyqábár bùŋêedà qáarèemáŋgá®. V O A 3P-dig grave.ACC youths.NOM
he youths will dig the grave. (Kießling 2001:8)
(26) a. qábár màydá dêeda®. V O = PEE POR
beat calf.AGEN cow.ACC
He beats the cow’s calf. (Kießling 2001:11) b. qábár máydá dêeda®.
beat calf.NOM cow.ACC
(i) he calf beat the cow. V A O
(ii) he cow’s calf beat him. (Kießling 2001:11) PEE POR
(27) qwàat gàtmòodá hàw. V O (N ADJ)
take wife.AGEN big.ACC
He takes the big (= first) wife. (Kießling 2001:13)
(28) gáwày qòh qáar7`7máŋgá áa háw7¥7ga®. V S PREP 3.PL.go ?? youth and girls.ACC
he young men and the girls went home. (Kießling 2001:16). (29) a. gàjáanùwâa háadígá jèedá húlándá\àwà. V S LOC
must.go men.NOM in men’s tent.ACC
he men must have gone into the men’s tent. (Kießling 2001:17) b. háadígà gàjáanùwâa jèedá húlándá\àwà. S V LOC men.ACC must.go in men’s tent.ACC
As for the men, they must have gone into the men’s tent. (Kießling 2001:17)
(30) a. gwándà gádéemgá jèedá dûgwa®. COP S LOC be woman.NOM in cattle.ACC
he women were among the cattle.
b. gàdéemgà gwándà jèedá dûgwa®. S COP LOC woman.ACC be in cattle.ACC
As for the women, they were among the cattle. (Kießling 2001:17) (31) qòodâw \áawúudá gùdêeda®. V A IO
give cat.NOM dog.ACC
he cat gave it to the dog. (Kießling 2001:7)
(32) gwàéewà áséetà: “áŋI`Iŋí dúgwáa-gú gágùt nàahà” say sun.NOM you.ACC cattle.ACC-you.PL.POSS kill what he sun said: “You, what has killed your cattle?”
gwàèewà ní\: “gágùt ásèeta” V S reply 3.SG.NOM kill sun.NOM
And he replied: “he sun killed them.” (Kießling 2001:19) (33) qàkà búunèe sêen
come people.NOM all
All people came. (Kießling 2001:14)
(34) a. ákòodâw gwáargwèedá [jàa qóonâal bálláandà qàmûŋga®]. give elder.AGEN REL teach boy.ACC drink.beer? …and he gives it to the elder who will teach the boy how to drink
beer. (Kießling 2001:14).
b. qwàyìin gwáargwèedà [jáa qóonâal bálláandà qàmûŋga®]. put elder.NOM REL teach boy.ACC drink.beer? gìléeŋgà ákwàyìin ùhùudá bállâanda
beads? put? head.AGEN boy.ACC
he elder who will teach the boy how to drink beer puts the beads, puts them on top of the boy’s head. (Kießling 2001:14).
(35) a. gát-à bálláandá gàacéedà dáráwéetà. V A INST O shoot-INST boy.NOM arrow.ACC swala.ACC
he boy shoots the swala22 with an arrow. (Kießling 2001:20) b. gágwàl-s-á gísàmjáŋgá mùránéedà dúgà. V A BEN O select-GOAL Gisamjanga warrior.ACC cattle.ACC
he Gisamjanga selected cattle for the warrior […]. (Kießling 2001:20)
here is no data available on whether Datooga has a passive construction and, if there is one, how case is encoded.
In sum, Datooga is a marked-nominative language with a split system: It either shows a marked-nominative system or no distinction at all. he split is triggered by person; in the first and second person the case distinction is neutralized. All core participants occur in the accusative. he accusative has the widest range of uses, it covers functions such as: citation form, nominal predicate, O, IO, participants introduced by verbal derivation, A/S before the verb, first and second person S and A in all positions, possessor, dependents in an NP, nouns ater prepositions. Unlike in other Nilotic marked-nominative languages, there is a third case, called anti-genitive by Kießling (2001). he lat- ter is something like a construction case; nouns which are modified in a noun
phrase have to appear in the anti-genitive. Interestingly, it interferes with case on the clause level as well, first, because it might be used only when the noun otherwise takes the accusative. he nominative however probably does not take an anti-genitive inflexion if modified. he accusative has the broadest range of functions compared to other accusatives in marked-nominative languages. At least ten different functions are covered by the accusative. Furthermore, the anti-genitive covers nearly as many functions as the accusative; except for the citation form, the anti-genitive occurs in all remaining nine functions.
3. Accusative in marked-nominative languages
How broad the range of functions of the accusative in a particular language in a marked-nominative system can be may be seen in the following statement about Zayse (Ometo, East Omotic) by Hayward. Note that Hayward uses the term absolutive instead of accusative.
[…] it would be true to say that this is the form [absolutive23] which occurs in every syntactic function except that of subject. (Hayward 1990:242)
Ater having illustrated the general behavior of marked-nominative languages by presenting three languages, namely Tennet, Maa, and Datooga, I would like to make some general remarks especially on the profile of the accusative in marked-nominative languages.
Table 5 gives an overview of functions which may be covered by the accu- sative and by the nominative in marked-nominative languages. For the accusa- tive it is more of a maximum list in that it lists all the functions being covered by the accusative according to the literature; an accusative of a single language may only cover part of the functions listed in Table 5. I will now discuss each of
NOM subject (S & A) ater the verb, third person (a) citation form
(b) O
(c) nominal predicate
ACC (d) subject (S & A) before the verb (e) possessor
(f) peripheral participants introduced by head-marking devices such as verbal derivation
(g) IO
(h) ater prepositions
(i) S and A, irst and second person, all positions (j) nominal modiiers in NP
Figure 4. Functions covered by the nominative and accusative cases in Datooga.
these functions in turn. he database for some languages was not exhaustive. If a language is not mentioned under a certain function this may mean either that this function is not covered by the accusative or nominative, but it may also mean that there was no data available.
Table 5. List of functions covered by the accusative and nominative in marked-nomi- native languages.
Case Function Language
ACC a Citation form all, except Arbore type 2 nouns only definite in Haro
b Object function all, only definite in Haro c Nominal predicate in copula
clauses all, except Arbore; only definite in Haro d S & A before the verb (in basic
verb-initial languages either topi- calized or focused)
Maa, Datooga, Teso, Turkana, Päri, Jur- Luwo, Dinka, Pokot, Nandi, Sebei, Kipsigis, Datooga, Omotik, Majang, Didinga, Murle, Tennet, Baale, Chai, Berber
e Possessor Maale, Datooga, Tennet & Maa only if juxtaposed
f Indirect Objects Arbore, Datooga (always), Maale (mostly DAT, but sometimes ACC)
g Participants introduced by head- marking devices such as verbal derivation (applicative, causative, instrumental)
Tennet, Maa, Datooga
h Ater prepositions Datooga, Maa (f´), Dinka neº, Berber (only few languages)
i Basis for case doubling Maale, Zayse, Wolaitta, Alaaba, K’aabena for ablative.24
Exceptions: Genitive in K’aabena (for other cases), Haro
j Patient of passive (S) Maa, Päri, Nandi, Turkana k Further particitants Alaaba: Time
Case Function Language
NOM a S & A all; only ater the verb in all Vi and Vm languages, that is, non-focused or non-topi- calized subjects
b Prepositions Berber, Dinka, Maa (t7)
c Agent of passives Dinka, Maa
d Basic for case doubling Berber
e Possessee Berber, Dinka
f Possessor Dinka (ater the preposition èº)
a. Citation form
his function is covered by the accusative in all marked-nominative languages belonging to type 1. In languages belonging to type 2 the picture is different: In languages where both forms (accusative and nominative) are derived, the citation may differ from the accusative. Arbore, a Cushitic language belonging to the Afroasiatic phylum, is mainly type 1, meaning that the accusative is typi- cally the morphologically unmarked form and the nominative the morpho- logically marked form, encoded either by tone or by a suffix, e.g. -é. here are however a few nouns in the language where both cases, nominative and accu- sative, are morphologically non-zero. Arbore therefore is a language belonging mainly to type 1 but also to type 2. With nouns following type 2, the accusative is derived from a so called ‘basic form’ by tone. Among the latter, it is not the accusative that is the form used in citation but the basic form. In 36, nouns of both types are listed: se‘ ‘cow’ belongs to type 1, accordingly the citation form is identical with the accusative form, but kšoró ‘wood’ follows type 2 and therefore the basic form is used in citation and not the accusative.
(36) Arbore (Omo-Tana, East Cushitic, Afroasiatic) se‘ ‘cow’ = cow.ACC
kšoró ‘wood’ (kŸoro ‘wood.ACC’) (Hayward 1984)
In all other type-2 languages where case marking is not dependent on definite- ness, the citation form is always identical with the accusative.
In languages where the case system is restricted to definite nouns only, the default citation form is the caseless indefinite form (as e.g. in Haro see Section 4). Nevertheless, the functional unmarkedness of the accusative can be seen in the fact that definite nouns occur in the accusative as their citation form (cf. Woldemariam 2003:63).
b. Object
Objects are marked in the accusative in all positions irrespective of whether expressed ater or before the verb, and irrespective of whether the language belongs to type 1 or 2.
c. Nominal predicate in copula clauses
Nominal predicates are always encoded by the accusative, except where the language has developed a case just covering this function. Nandi, an East Ni- lotic marked-nominative language, may demonstrate the typical behavior. In 37, the nominal predicate ‘child’ is encoded in the accusative in copula clauses. Note that in 37 the subject Kibet also appears in the accusative as it follows the
rule ‘no case before the verb’, which in Nandi includes the copula element kò (see (d) below).
(37) Nandi (East Nilotic, Nilo-Saharan)
kipe:t kò la:kwét. S PRED.N
Kibet.ACC ?? child.ACC
Kibet is a child. (Creider & Creider 1989:125)
Arbore is an exception again. Arbore has developed a copulative (called pred- icative by Hayward) which is used in this function (cf. 38). he copulative en- codes focused participants and nominal predicates.
(38) Arbore
w[aºBáº]25tto ‘[oºBóº]l-a. DEM donkey-COP
hat is a donkey. (/‘ohól/) (Hayward 1984)
d. S & A before the verb (either topicalized or focused or just the default word order)
his feature does not apply to verb-final languages — for obvious reasons, oth- erwise they wouldn’t be case languages — illustrated once more by Arbore, a verb-final language. In Arbore, S and A are encoded in the nominative before the verb, which reflects the basic word order (cf. 39 and 40).
Arbore
(39) néekŸ ‘í-y y-eečč-e. S V
lion.NOM PVS26-3MS 3MS-come-PER A lion came. (Hayward 1984)
(40) n[eŸ‘ eŸ]-t-é ý móh t-[oºBoº]m-e. A O V lioness.F-NOM PVS.3MF man.ACC 3FS-eat-PER
he lioness ate the man. (Hayward 1984:157)
But the feature ‘S and A before the verb are encoded in the accusative’ applies to all verb-initial and verb-medial languages. Feature (d) follows an irregularity which goes beyond marked-nominative languages but applies to all case lan- guages: Among all East African case languages there is one irregularity, which I referred to by the slogan ‘no case before the verb’, more appropriately ‘no case distinction before the verb’27 (König 2002b). I mentioned this above with regard to example 16 in Maa: Whenever a core participant is frontshited be- fore the verb, this participant occurs always in the morphologically unmarked form, irrespective of whether it is S, A or O. he feature ‘no case before the verb’ does — as has been shown — not occur in verb-final languages; other than
that, nearly all case types and word orders are covered, such as verb-initial and verb-medial languages, including marked-nominative and ergative languages (such as Päri). How substantial this feature is can be seen in the fact that even in languages which (at least) today have an AVO order, the feature holds true. I will illustrate this with examples form Suri-Chai, another Surmic language. Suri-Chai has a basic AVO constituent order.28 It is a marked-nominative lan- guage with an unmarked accusative (called absolute by Last & Lucassen) and a nominative, derived from the accusative by a suffix -o. Further cases are geni- tive, encoded by a suffix -í\, and locative, encoded by a suffix -ó.29 Suri-Chai also follows the rule: no case distinction before the verb. Accordingly, in the basic orders SV and AVO both core participants occur in the unmarked ac- cusative. In 41, S (‘my children’30) occurs in the accusative in an SV-order and in 42a, A (‘Bume’) occurs in the accusative in an AVO order. Ater the verb however, S and A appear in the nominative. In 42b the same content as in 42a is presented in a different word order, namely an OVA order. Hereby A (‘Bume’) occurs in the nominative case. 42b reflects the pragmatically marked structure, 42a reflects the basic order. In 42b the object is topicalized. S in a VS order also appears in the nominative, as in 43 ‘I’ and ‘you’ in ‘I know’ and ‘you are ignorant’.31
(41) Suri-Chai (Southeast Surmic, Nilo-Saharan)
7¢rró-á-gà\ú ílá:gásè gó:ré nà rògònó r7séò. S V children.ACC-my32 be.sick.IPV.3PL very NAR tomorrow die.IPV.3.PL My children are very sick and tomorrow they will die. (Last & Lucassen
1998:406)
(42) a. bume haŋae ŋakogine. A V O
Bume.ACC chase.PFV.3PL.O3SG Ngakogine.ACC
he Bume chased Ngakogine. (Last & Lucassen 1998:408)
b. ŋakogine haŋae bume-o. O V A
Ngakogine.ACC chase.PFV.3PL.O3SG Bume-NOM
he Bume chased Ngakogine. (Last & Lucassen 1998:407)33
(43) kagai a\-o: cí7¥: imàgí í\-ó: V S V S
know.1SG I-NOM but be.ignorant.2SG you-NOM I know, but you don’t know. (Last & Lucassen 1998:397)
In verb-initial languages, the frontshiting of A and S is due to pragmatic func- tions such as topic or focus, as e.g. in Maa and Berber; this does not hold for languages with a basic verb-middle order, such as Suri-Chai.
e. Possessor
Some languages have developed a genitive case to encode the possessor, but even then it may be that there is an alternative, juxtaposed, possessive con- struction in which the preceding possessee occurs in the accusative. his holds true for Tennet, as has been shown above (examples 5a and 5b), and for Maale. Maale, an Omotic language spoken in southern Ethiopia, is also a marked- nominative language. It has a basic verb-final order. According to Amha (2001) ten cases are distinguished, either by tone or by suffixes.34 Among the ten cases there are three different locative cases. Case inflexion differs on whether the noun is indefinite or definite. In the definite use, the accusative and nomina- tive are both derived forms (type 2), both encoded by suffixes. With indefinite nouns, the case distinction of accusative and nominative is encoded by tone, with the nominative being derived from the accusative. Maale has also two dif- ferent structures to encode nominal possession, where the possessor is encod- ed in two ways: Both constructions share the possessor-possessee order, but in the first strategy the juxtaposition possessor-possessee is enough. No genitive case is involved; the possessor appears in the accusative case (cf. 44a). In the second strategy, the possessee appears in the genitive, which again is suffixed to the accusative form of the noun, as in 44b. here is a meaning difference between the two constructions: he use of the genitive, as in 44b, implies that the rest of the body is not short.
(44) a. Maale (Ometo, North Omotic, Afroasiatic) gudúri tóki k’amítsi-ke.
hyena.ACC foot.NOM short-BE35.A.DCL A hyena’s leg is short.
b. gudúri-ko tóki k’amítsi-ke. hyena.ACC-GEN foot.NOM short-BE.A.DCL A hyena’s leg is short. (Amha 2001:63)
In Maa, the possessee occurs always in the accusative form, namely ater the genitive particle (cf. 12). But there are also marked-nominative languages where the possessee occurs in the nominative; Dinka and Berber are examples. Dinka, a West Nilotic language, has four cases: A nominative (called oblique by Andersen 2002) marked by tone, an accusative (zero marked),36 two loca- tives, called allative and inessive-ablative, marked by overlong vowel and tone change. Andersen calls the latter ‘external cases’, which encode the relationship of participants on the clause level, distinguished from internal cases which op- erate within complex noun phrases. All external case inflexions are marked by non-linear tools such as root-internal change. Modified nouns appear in what
Andersen calls ‘construction case’ (see Andersen 2002:29). he possessor oc- curs in the nominative ater a preposition èº. he same preposition is also used as a copula, but when used as a copula the nominal predicate occurs in the accusative (cf. 46). 45 and 46 are nearly identical clauses; the only difference is that in 45 ‘chief’ occurs in the nominative and in 46 in the accusative. he ele- ment èº obviously functions as a copula in 46 but as a preposition in 45. (45) Dinka (West Nilotic, Nilo-Saharan)
máµn¯ èº bàº\ POR = NOM
child.CS1 PREP chief.NOM
the chief’s child (Andersen 2002:15)
(46) máµn¯ èº báº\ N.PRED = ACC
child.CS1 be chief.ACC
a child who is (going to be) a chief (Andersen 2002:15)
In Berber it is also the nominative rather than the accusative which is used as the basic form of the genitive case. Possessors in nominal possession are marked by the genitive case. he genitive case itself, marked by a clitic i-, is prefixed to the nominative case form of the word and not to the accusative one. he use of the nominative, being the form employed in possessive constructions, may be the result of the fact that, in general, nouns ater prepositions occur mostly in the nominative. he genitive itself has developed out of a definite clitic which may have triggered the nominative (see Aikhenvald 1992:43 & Section 8). f. Indirect objects
Indirect objects37 are oten encoded in the accusative but, again, not if the language has developed a dative case covering this function, as for instance in Maale. But even in Maale it may happen that two participants within one clause are encoded in the accusative without any verbal derivation or postposi- tion (cf. 47). With ditransitive verbs such as ‘to feed’, O and IO appear in the accusative case.
(47) Maale
‘ind-a na‘‘-ó ‘awki muuzz-é-ne. mother-NOM child-ACC meat.ACC feed-PFV-A.DCL he mother fed the child (with) meat. (Amha 2001:207)
In Arbore, where there is no grammaticalized dative, the accusative is taken as well to encode the indirect object, once again without any further verbal derivation. he accusative encodes the indirect object (IO) and the direct ob- ject (O) with ditransitive verbs. Both orders O IO V or IO O V are possible (as
exemplified in 48a and 48b). he indirect object is really encoded by the accu- sative and not by the basic form, which can be seen in nouns which have a dif- ferent basic form. In 49, the indirect object is encoded by gˇ[aŸ‘aŸ]r-te, the accusa- tive form of ‘old woman’, and not the basic form, which would be gˇ[aŸ‘aŸ]r-té. (48) a. Arbore
‘í-n¿ biče náag s[iºBiº]s-e. O IO V PVS-1S water.ACC boy.ACC give.1.SG-PER
I gave water to the boy. (Hayward 1984:111)
b. náag biče ‘í siss-e. IO O V
boy.ACC water.ACC PVS.2S give.2.SG-PER You gave water to the boy. (Hayward 1984:159)
(49) gˇ[aŸ‘aŸ]r-te dafar-á-n¿ s[iºBiº]s-a. IO COP V old.woman-F.ACC cloth-COP-1.S give.1.SG-PER
It’s clothes that I’ll give to the old woman. (Hayward 1984:159)
he same holds, as shown above, for Datooga: IOs are always encoded in the accusative (cf. 31).
Other languages, such as Tennet or Maa, do not have three place verbs. hey always use the head-marking device, an applicative, whereby the seman- tic concept corresponding to IO in other languages is encoded in the accusa- tive, as has been shown above.
g. Participants introduced by head-marking devices such as verbal derivation (applicative, causative, instrument)
As I have illustrated above, Tennet, Maa, and Datooga encode the participant introduced by verbal derivation always in the accusative. he same is true for Nandi. In 50, the beneficiary ‘the child’ is encoded in the accusative. he verb has a suffix -ì ‘cut’, which looks like a verbal derivation, so that the dependent participant introduced by verbal derivation is encoded in the accusative. (50) Nandi
kí-ka:c-ì kípe:t la:kwé:t ce:kà. PAST-give-cut Kibet.NOM child.ACC milk.ACC
Kibet gave the child milk. (Creider & Creider 1989:124) i. Case doubling
he term ‘case doubling’ is used for the morphological occurrence of more than one case ending on one noun. It refers to a special building pattern: he case marker is not suffixed to the morphologically unmarked form, but to a form which itself is already a derived case form. he base for case doubling is a stem
which itself is a derived case form (see Maale, 52a and 52b). Note that these case doubling forms are not instances of double case markings, where different syntactic functions are marked on the same noun. In short, a noun with several case endings is an instance of double case if the case endings express more than one function (as in Kanuri38); an instance of case doubling is present if two or more case endings express only one function (as in Maale).
(51) Kanuri (Saharan, Nilo-Saharan) táta kámú-vè-gà rúskә¢nà. son woman-GEN-ACC I.see(?) I have seen the wife of the son.
As mentioned above, Maale has ten different cases; some are suffixed to the accusative form. Case doubling occurs with peripheral cases. he accusative is always the basic form for peripheral cases such as the instrumental, as with
‘snake’ in 52a. Some case forms in Maale are even triple-marked, and again, the basis it the accusative, as in 52b.
(52) a. Maale
‘ízí wóówi-na 1a‘-ínt-é-ne. 3MS.NOM snake.ACC-INST bite-PASS-PF-A.DCL He is bitten by a snake. (Amha 2001:39) b. máár-ó-ídda.ppa
house-ACC-LOC.ABL
from the house (Amha 2001:69)
herefore case doubling forms are not instances of double case markings. In Wolaitta (Ometo, North Omotic) and Alaaba (Highland East Cushitic) there are also case doubling, and the accusative is always the basis for the case doubling forms (Azeb Amha, p.c., and Gertrud Schneider-Blum, p.c.).
here are however other marked-nominative languages which take a dif- ferent basis for case doubling, namely the genitive or the nominative: he geni- tive occurs as the basis for case doubling in particular to some extent K’abeena (Highland East Cushitic), and the Omotic language Haro (see Section 4). he nominative occurs as the basis for case doubling in Berber (see (e) and 85). j. Subjects in passive-like constructions are encoded by the accusative In most marked-nominative languages, S in passive constructions is not en- coded like O, but is encoded like A, namely in the nominative, as 53 from Maale may illustrate.
