Genitive marking of subjects in West Pahari.

Zoller, Claus Peter

1 Introduction

This study grew out of an occupation with the pronominal system of Norwegian Romani (tatersprak) within the frame of a linguistic project on this language. The project is financed by the Norwegian Research Council, to which I want to express my thanks.

The first and second persons singular in Norwegian Romani have the form miro 'I' and diro 'you', and it is generally assumed that the words continue the Indic possessive mera 'my' and tera 'your'. Thus Yaron Matras (2002: 147) says, "[S]candoromani selects the genitive possessive form (miro 'I', diro 'you' < Romani tiro 'your' contaminated with Scandinavian din 'your')." It thus appeared appropriate to look for possible parallels in New Indo-Aryan (NIA). Use of possessive pronouns or nouns marked with a possessive suffix in subject position have so far been known from within the NIA language area only from Bengali, Assamese and Oriya (see for instance Colin P. Masica 1991: 346ff. and Masayuki Onishi 2001b). In these languages the Genitive Subject (1) is not an Agent but, as in the comparable Dative Subject constructions of many other NIA languages, an Experiences. However, coming across these forms in Norwegian Romani reminded me of having occasionally observed Genitive Subjects in the Bangani variety of West Pahari. Even though Guro Flgstad (see the contribution in this volume, pp. 153-168) and I were sceptical from the outset that the usage of a possessive pronoun as subject in all these languages would be due to a common historical origin, it caused us--and especially me--to look more closely at the evidence in Bangani and some other varieties and languages of West Pahari. Our scepticism rested mainly upon the facts that subject marking with a possessive within Romani is limited to Scandinavia; moreover, whereas in Bengali it is a "non-canonical" but not infrequent phenomenon, in Bangani and other West Pahari languages it is not only non canonical but also used rather rarely. This contrasts with the situation in Norwegian Romani where the possessive pronoun has been generalised as the subject marker. On the other hand, Romani nouns in subject position are not marked with a possessive suffix. Since the historical origins for the use of the possessive pronoun in subject position in Norwegian Romani are so unclear or, if the phenomenon is at all explicable, it has developed independently due to local factors in Europe, we decided to write two separate articles.

Thus the main goal of this article is first to draw attention to the fact that within NIA, Genitive Subjects are not only found in Bengali, Assamese and Oriya but also in some varieties of West Pahari. This article analyses their various functions.

The data presented below stem partly from records of an oral epic, the Panduan, which I recorded several times in the 1980s and 1990s, and a mythological story recorded previously in 1983 (see below); partly from a short field research trip in Bangan and surroundings conducted by Flogstad and myself in May 2008, and partly from interviews done with speakers of Bangani and neighbouring Deogari who live in New Delhi. (2) Data for Genitive Subjects in the Koci and Kotgarhi varieties of West Pahari are found already published in Hans Hendriksen 1986, and for the Bhalesi variety of West Pahari in Siddheshwar Varma 1948. It is interesting to see that Bangani, Deogari, Koci and Kotgarhi form one continuous geographical area at the eastern end of West Pahari whereas Bhalesi is located at the western edge of West Pahari in a remote area. The two sources of Bangani oral texts used in this article are:

* A mythological story called "The little old gentleman" (bur[??]-khur[??] sad[??]ru) (abbreviated LOG). I translated and publishedit in 2007. The story consists of 236 sentences, but contains just two clauses with Genitive Subjects.

* The Panduan (abbreviated P) is an oral version of the Mahabharata. The roughly eight-hour-long record (Zoller forthcoming) consists of many thousand sentences. Within this substantial body there are a few dozen sentences with Genitive Subjects.

This shows that Genitive Subjects are used quite rarely in this area. Thus, the question is posed: are they examples of non canonical subject marking? However, instead of answering this question with a yes or no, I will rather begin by looking at the definition of this term as offered in Aikhenwald, Dixon and Onishi (2001), which does not always seem to be useful. Still, many of the data described and analysed in their book do indeed have close parallels in the West Pahari data I am going to present. Thus the authors say (2001: ix): "For example, in a nominative-accusative language, S[ubject] and A[gent] functions may be marked by nominative case for most verbs (the canonical marking) but by dative or genitive case for a small set of verbs (the non-canonical marking)." According to this definition, the non-canonical status of a subject marked with a specific case correlates with its infrequency when compared with the 'standard' case. Moreover, the definition proposes that this infrequency depends directly on the semantics of the predicate. Consequently, split ergativity as a basic grammatical phenomenon does not belong here; however things like the Dative Experiencer constructions, which are usually explained in terms of predicate semantics, do fall under this definition. See, for instance, this issue discussed in NIA languages (Masica 1991: 346ff. and reference to further literature). The problem with the definition, if applied to the West Pahari languages discussed in this article, is that the great majority of their verbs with animate subjects can be both marked with nominative (overtly unmarked) or ergative on the one hand, and with genitive on the other. The same does not hold true for the Dative Experiencer constructions in these West Pahari languages, which indeed seem to depend, as related constructions in other NIA languages, on the predicate semantics. Thus they are fundamentally different from the Genitive subject constructions and therefore not considered here.

A widespread type of Genitive Subject in NIA is constructions with the subject functioning as genitivus possessivus. Here the possessor is the logical subject, while the possessed object is the grammatical subject of an equative sentence. The con struction typically expresses inalienable possession, e.g. Hindi:

1. H. (3)

us=ke do bacce hai

He.OBL=GENPOP-PL-M two children are

'He has two children'

This genitivus possessivus construction is not further discussed here either. On the other hand, the type of Genitive Subject constructions presented below are, to my knowledge, geo graphically restricted within West Pahari to some eastern varieties, namely Bangani, Deogari, Koci and Kotgarhi (and perhaps some more nearby dialects), and to the extreme western variety called Bhalesi. I will not attempt at this stage of analysis to compare these constructions with the (superficially) similar ones in Bengali, Oriya and Assamese.

2 Valency reductions

A core feature of most Genitive Subject constructions in West Pahari is valency reduction. Onishi (2001a: 12f.) regards "valency-changing derivations" as closely connected with non canonical marking of subjects, and so he uses the term "deagentivisation". I will use this term at some places below in the sense that the subject loses full or partial agency of an action. One well-known case of valency reduction is passivisation. In West Pahari, passive and Genitive Subject sentences are different from each other. However, they also have some features in common. We shall therefore first have a look at passive sentences.

2.1 Passive

Another word for passive is diathesis. When active sentences are changed into passive ones, the semantic roles of agent and patient are retained, but syntactically they change their functions: patient becomes subject and agent an adjunct. In the passive, only one obligatory Aktant remains and thus there is valency reduction. Instead of using the abstract term 'adjunct', one might also say that in this process the agent moves from a central position into a peripheral one. This is the standard pattern for many NIA languages. However, it is not the only alternative. In case of certain negative passives, the agent, instead of moving to the periphery, can remain in the centre. But he has to pay for it, so to say, with a loss of agency. Peter Gaeffke speaks, with regard to modern Hindi, of ,,Verneinte Passiva zur Bezeichnung von Unvermogen im modernen Hindi" (1967: 78). Masica (1991: 317) says in connection with the historical development of the different NIA passives, "The result is impersonal (or "involitive") verbs, expressing the helplessness or non-volitionality of the erstwhile agent, if any." Thus both authors broach the grammatical category of inability associated with NIA passives. Here first an illustration with an intransitive verb from modern Hindi (R. S. McGregor 1972: 117):

2. H.

mujh=se abhi bazar nah[??] ja-ya ja-e-ga

I--ABLPOP right.now bazaar not go.PP go.PM.FUT-M-SG

'I shan't be able to go to the bazaar just now'

Gaeffke says that the periphrastic jana passive (underlying the above construction) developed early in NIA but was unknown in Middle Indo-Aryan (MIA). (4) Already at an early stage of NIA, this periphrastic passive could express an inability on the part of the subject. It was already used in the Old Bengali Carya songs, in Old Rajasthani and in early Hindi, thus covering a large geographical area. An example from the Ramcaritmanas of Tulsidas (Gaeffke 1967: 53): dekheit jaga nana / dekhata banai Na jai bakhana "Ich sah viele Welten, / die gesehen werden konnen, aber nicht beschrieben werden konnen". Whereas in older Hindi the agent of these constructions was rarely expressed explicitly, this is very common in modern Hindi (where the agent is marked with an instrumental postposition). Gaeffke (1967: 78f.) explains this with a different emphasis on "uber-individuelles Geschehen" in older Hindi against the description of the actions of individuals in modern Hindi.

In the West Pahari languages under discussion a jana passive never developed. Instead they continue an old passive with a suffix -i- (5) added to the verb stem, which developed historically from older i (y)a or i(y)a (Masica 1991: 316). This passive does not express inability, even in negative sentences. There is concord with patient/subject (more on which in 2.2.1). Examples:

3. Deog.

cithi dak[??]=di di-a-i-i

letter post=LOCPOP go.CAUS.PSM.PP-F-SG

'The letter was sent by post' (lit.: 'the letter was caused to go in the post')

4. Deog.

bharat[??] indi bol-i-o [??]

India.OBL=LOCPOP Hindi speak.PSM.PP-N-SG is

'Hindi is spoken in India'

5. Deog.

au n[??] zhang-i-a boiri-k[??]n

I.NOM not kill.PSM.PP-M-SG enemy.INSPOP

'I don't get killed by the enemy'

The next is an example from the Bangani Panduan epic with the predicate consisting of a compound verb with the light verb having a PP form:

6. Bng. P

thakur= k[??] dar[??]= k[??] s[??] de-ni buja-i

master=GENPOP door=GENPOP she.NOM give.PP-F-SG

perform.PSM

'She (the epic) is performed (lit.: 'explained') at the door of the master'

2.1.1 "Absolute Passive"

Siddheshwar Varma (1938: 40) reports from Bhadravalir a remarkable passive construction which he calls "absolute passive":

7. Bhad.

'teskera [??]u m[??]r-o-ta?

he.OBL.ABLPOP I beat.PSM.PPRES

"Am I beaten from him?"

He explains the choice of this term thus: "Because both the agent and the subject [patient] of the action are felt as passive, having absolutely no control over the action." Put in other words, this is the passive of an involitive sentence with a transitive verb. The unintentional agent of the involitive action is marked with the same type of ablative postposition as the above agent of the involitive Hindi jana passive. Active involitive sentences with Genitive Subjects are discussed below under 2.2.2. The Bhadravalir involitive passive construction is also remarkable because the ending of the verb looks like a modern continuation of the old inimite passive ending -iyata-. Already in MIA the ending -ata- was added to the above-mentioned passive suffix -iya- in order to realise "unpersonliche" (impersonal) passives (Gaeffke 1967: 49ff.). Modern NIAs continue to have impersonal passives (see Gaeffke 1967: 80ff.), i.e., passives without an agent. However, the above Bhadravalir construction is slightly different as it is 'personal', but the person acts involuntarily. Here both agent and patient suffer the action, so this might be called 'perferitative mood' (from Latin perfero 'to suffer'). Below (2.5) I will present Bangani and Deogari Genitive Subject constructions using verbs with the meaning 'to be' also in order to realise 'perferitative mood'.

2.2 Valency reductions in West Pahari

Several of the above examples illustrate the common pattern of the agent being de-centred. We can now turn our attention to Genitive Subject constructions where the subject remains in the centre but loses agency. They are semantically related to the above negative jana passive constructions in that both realise some sort of inabilitative mood, and they are morphologically related to the non-periphrastic passive constructions in that they too employ, at least in a large number of cases, the old -i-passive element. They can be classified thus:

* Inabilitative mood (on this term see Rajesh Bhatt 2006: 159): Subject is unable to realise an action;

* Involitive mood: Subject is unable to control an event;

* 'Perferitative' mood: Subject is unable to prevent an essential change of its own condition and has to suffer it.

These three different moods are realised morphologically in three different ways (see below 2.2.1, 2.2.2 and 2.5): inabilitative in negative sentences, involitive in non-negative sentences, and 'perferitative' only with verbs meaning 'to become'. However, they do not cover all possible West Pahari Genitive Subject constructions. They are also used to realise such different phenomena as reflexivity, a gerundive, and con temporaneity and anteriority in complex sentences (see below 2.2.3 , 2.3 and 2.4).

All three mood constructions use Genitive subjects and add the passive -i- to the verb stem. However, at least in the case of Bangani and Deogari, one needs to further differentiate between two clearly different types of inability mood:

* Temporary inability: constructed with the passive -i- and Genitive Subject;

* General inability: constructed:

* either with normal passive and an obligatory agent marked with an instrumental or ergative postposition;

* or with a nominative subject construction and a modal verb.

Temporary inability means that the subject is, due to any kind of personal or external reason, unable to realise an action as long as the causation persists. In other words, the subject experiences only a temporal reduction of her/his capability which is not an essential trait of her/him. General inability, on the other hand, doesn't mean a permanent reduction of one's capability, but it means that 'no living being X' or 'no one' is in principle able to do action Z (it is infeasible). Thus, the opposition between temporal and general inability also corresponds with the inability of an individual versus infeasibility per se. The first two examples to illustrate this difference are from Deogari: (6)

8. Deog.

m[??]r[??] zhu[??]ni=zhav n[??] ur-i-ndc

I.GEN.OBL moon ALLPOP not fly.PSM.PPRES-M-SG

'I cannot fly to the moon' (for the time being, but later I can)

9. Deog.

au zhu[??]ni=zhav uri (7) n[??] b[??]l-da

LNOM moon ALLPOP fly- not can.PPRES-M-SG

'I cannot fly to the moon' (because I am in principle unable to do this)

I illustrate now normal negative passive (i.e. not the inabilitative) with obligatory agent with two sentences from the Panduan. In the epic, Bhimsena and Arjuna regularly get into tough brawls during which they become so much wedged together that nobody is able to separate them. The second sentence (11) below is a near-repetition of the first one (10). But it is sung, in the version recorded by me, several hours after the first one. However, the first sentence uses an ergative postposition for marking the agent whereas the second uses an instrumental postposition. It is a typical stylistic feature of the Bangani Panduan that the singer may repeat a sentence with slight variations either shortly after the first one or after a long lapse of time:

10. Bng. P

(God Narayana says): "Two combatants are wedged together,

Arjuna and Bhimasena,

y[??] n[??] chura-i-d[??] kuni"

they not separate.PSM.PPRES-M-PL anyone.ERG

they cannot be separated by anyone"

In fact they do get separated, namely by super-strong Hanumana. Now the second brawl:

11. Bng. P

(God Narayana says): "Two brothers are engaged in fighting and dying,

[??] n[??] k[??]si = k[??] chura-i-d[??]"

they not anyone.OBL=INSPOP separate.PSM.PPRES-M-PL

they cannot be separated by anyone"

In fact they do get separated, once again by super-strong Hanumana. So this type of construction realises a general impossibility, and the subject is marked not by the genitive but, for instance, by an instrumental or ergative marker.

2.2.1 Inabilitative mood

In this section I will quote more examples realising the inabilitative mood in order to illustrate the statements made above. With regard to the construction of the arguments, the following needs to be added: The Genitive Subject appears with an invariable (oblique) masculine -[??] ending. Since inabilitative mood is realised with negative sentences, the predicate has the form of a participle (see footnote 6). The participle predicate can be simple or complex. A simple predicate, and most complex ones, add the passive -i- to the (main) verb stem. In very rare cases the latter type of predicate adds a conjunctive participle ending -ui to the main verb stem (see example 13 below). The second component of complex verbs-which carries the inflection--is an auxiliary like 'to be' or 'to stay'. Whereas in the passive constructions there is concord with the subject/agent (see above), in the Genitive Subject constructions there is concord with the object. Here is an illustration for this from Deogari, which has three genders: (8)

12. Deog.

m[??]r[??] b[??]l[??]d n[??] zhang-i-und[a.bar]

I.GEN.OBL ox.M not kill.PSM.AUXPPRES-M-S

'I cannot kill the ox'

'Ox' is masculine, therefore the ending -a. In m[??]r[??] tiria n[??] zharigiundi 'I cannot kill the woman' there is the feminine ending -i; and in m[??]r[??] s[??]ngav n[??] zhangiund[o.bar] 'I cannot kill the snake' it has the neuter ending -o as snakes are understood as being neither male nor female.

The fact that inability expressed with a Genitive Subject is temporary is best shown not with sentences from questionnaires but with sentences found in authentic (oral) texts. But to rule out any misunderstandings: The following sentences from the Panduan epic are all sentences which basically can also be used in everyday language. In one scene Bhimsena elopes with a giantess and sleeps for six months. At the end he needs some time to recover his strength, so he says:

13. Bng. P

m[??]r[??] thad[??] n[??] r[??]-[??] biuz-ui

I.GEN.OBL upright-EMP not stay.PP-M-SG arise.CP

'I cannot get up by myself at all'

For clarification a literal but uncorrect translation into Hindi: mere khra hi na raha uthkar.

Note: Even though the grammatical head of the above predicate realises grammatical past tense, the whole construction is in present tense (there are numerous parallels of such a construction in the Panduan). Note also that even though the dependent verb has a conjunctive participle suffix, both verbs together form what I have called "combined verbs with "light" main verbs", which means that the conjunctive participle is the semantic head of the predicate (for more examples and a detailed description of their functioning see Zoller 2007: 103ff.).

This sentence 13 is repeated in the epic a little later almost verbatim (for stylistic reasons) but also with a slight change in meaning:

14. Bng. P

m[??]r[??] thad[??]-i n[??] biuz-i-d[??]

I.GEN.OBL upright.EMP not arise.PSM.PPRES.OBL(?)

'I cannot g et up at all'

Again literal Hindi: mere khra hi na ulhta. In the following example Bhimsena is engaged in devouring buns weighing many centners. (9) Seeing this, King Karna says to Bhimsena with regard to himself and to the Kauravas:

15. Bng. P

amar[??] pithi=di bi na gin-i-di, tai pete=di s[??]mai go-i

we.GEN.OBL back.OBL=LOCPOP even not carry.PSM.PPRES-F-PL,

you.ERG stomach.OBL= LOCPOP place-go.PP-F-PL

'We cannot carry (the buns [feminine]) even on the back,

(while) you have placed (them) in (your) stomach'

One may ask why King Karna here uses a Genitive Subject. My guess is he wants to indicate that he and the Kauravas are not weak as such, even though carrying the buns is a challenge they cannot meet right now. The intricate relationship between tem porary inability and basic infeasibility is further illustrated with the following examples. God Narayana instigates Bhimsena into going to a city and trying to cheat a trader. He provokes Bhimsena with the following words:

16. Bng. P

zoik[??] kirari [??]-li, toik[??] th[??]g-i-a tia- s[??], kirar[??] n[??] t[??]r[??] th[??]g-id[??]

where female.trader be.FUT.3.SG, there cheat.ITM.IMP

she.OBL.OBJ, trader not you.GEN.OBL cheat.PSM.PPRES-M-S

'Wherever you meet upon a tradeswoman cheat her thoroughly, (because) you cannot cheat a (male) trader'

God Narayana hasn't yet heard about equal gender treatment, but Bhimsena cheats both at the end, the trader and his wife. Thus the god's (not really serious) expectation that Bhimsena could have a moment of weakness vis-a-vis a male trader was unfounded. Compare this with the following example from the Panduan which describes a general infeasibility and therefore uses a different "peripheral case" (Roman Jakobson) to mark the agent of the passive construction. The bard uses here a similar poetic technique as in sentence 16, namely that of contrasting two opposite facts. The Bhimsena of the Panduan has the gift to adopt various (sometimes bizarre) shapes. In one scene he metamorphoses into a very thin wooden stick and holds a magic iron rod in his hand. The 'stick' and the rod lie on the ground in order to signal to the Kauravas that Bhimsena is no longer himself. To describe this state, the bard uses the following image:

17. Bng. P

du kua-u=kh[??] na tap-[??], ek-i= k[??] na gin-i-[??]

two.OBL crow.OBL=BENPOP not suffice.PRES.3.SG,

one.EMP=INSPOP not carry.PSM.PRES.3.SG

'He doesn't suffice (as food) for two crows, (but on the other hand) he cannot be carried by one (crow)'

In the second sentence part no Genitive Subject is used because the idea is not that there is an individual crow which has lost its energy; the meaning is rather that Bhimsena remains too heavy for everyone. When the Kauravas realise that they indeed cannot lift Bhimsena, they nevertheless insist that this is just an accidental weakness, and they consequently use the Genitive Subject:

18. Bng. P

eik[??] d[??] z[??]l-n[??], Biu= k[??] bar[??] amar[??] na cug-i-d[??]

here give.I1VIP burn.INF.OBL, Bhima=GENPOP load

we.GEN.OBL not lift.PSM.PPRES-M-SG

'Let's abandon (him), (10) we cannot lift the load of Bhima'

However, Bhimsena disagrees with this, and therefore he repeats what the Kauravas said, but with a passive sentence and the agent in the oblique case in order to signal to them that they are wimps. The sentence doesn't contain an overt negative particle, but his ironic question conveys precisely the basic inability of the Kauravas:

19. Bng. P

tum-u bai cug-i-[??] mu=ag[??]=k[??] bar[??], Biu=k[??] bar[??] tum-u bai cug-i-[??]?

you.OBL really lift.PSM.PRES.3.SG LOBL=LOCPOP=GENPOP

load, Bhima=GENPOP load you.OBL really lift.PSM.PRES.3.SG

'Can you really lift the load of me, the load of Bhima, do you really (think you) can lift?'

Suggesting that the Kauravas are wimps doesn't mean for Bhimsena that they are handicapped. He knows how to differ entiate. They are certainly never able to lift him up. However, during a ball game played by the Kauravas and Pandavas, Bhimsena kicks the ball away and then calls upon the Kauravas to search for it. Since he doesn't want to insinuate that the Kauravas are unable to trace anything that has disappeared, he says to them after they have returned from an unsuccessful search:

20. Bng. P

tumar[??] lor-i-[??]?

you.GEN.OBL search.PSM.PRES.3.SG

'You and searching?'

So again a sentence without an overt negative particle, but again the sentence is clearly meant in a negative sense: 'You cannot search (and trace) the ball which I kicked away.' And then Bhimsena points to the ball which is right above them on top of a tree. So this is a singular event, and therefore the use of a Genitive Subject is appropriate.

All that has been said above about Bangani also applies to Deogari. I have many more Deogari examples with inabilitative mood, but since they don't add any new insights, it is not necessary to quote them here. Hendriksen (1986: 143) quotes some examples of inabilitative mood constructions from Kotgarhi. However, since they are given without context, it is not certain whether they also realise temporary inability, although it does look so. The following sentences are constructed like the above example 12 from Deogari with a participle of the auxiliary 'to be':

21. Ktg.

t[??]r[??] nei a-u[??]

he.GEN.OBL not come._be.PP-M-SG

"He could not come"

22. Ktg.

t[??]r[??] kich bi nei sun_h-u[??]

he.GEN.OBL something also not hear be.PP-M-SG

"He could not hear anything"

At the other, western end of West Pahari in Khasali and its closely related variant Bhalesi, Genitive Subjects appear (on the basis of very limited data) to be used only in involitive mood constructions (see next section, 2.2.2). For expressing the inabilitative mood the agent takes the ergative case both for intransitive and transitive verbs (the latter show concord with the object). This partially resembles the Bangani general inability mood. The examples from those languages, however, suggest that the construction rather expresses individual (temporary?) inability. The first example is from Bhalesi (Varma 1948: 53) and the second from Khasali (Varma 1938: 41):

23. Bhal.

mei n[??] he's-jou

LERG not laugh.AP (11)

"I could not laugh"

24. Khas.

hij mi 'duijo 'rotti khei jei

yesterday LERG only.two breads eat.AP-F-PL

"Yesterday I could eat only two chupaties"

Despite the lack of an overt negative particle, the last sentence also realises inability, as it says indirectly that 'I could not eat even three chupaties.'

2.2.2 Involitive mood

Involitive mood (12) is the non-negative correspondent to the inabilitative mood. It expresses that the subject is doing something which she/he cannot control. It might be even against his/her will. I could not locate any examples in my Bangani oral text corpus. This doesn't mean that Bangani and Deogari do not use this construction. However, it is certainly much less common there than the inabilitative construction. All following examples are therefore from the other West Pahari languages under consideration, namely Koci and Kotgarhi (Hendriksen 1986: 143), and Bhalesi (Varma 1948: 51 and 53). The first three employ participles of the auxiliary 'to be':

25. Kc.

mere apne cheure katt_hue rossa=matthi

I.GEN.OBL own.OBL wife.OBL cut_be.PP.OBL

anger=LOCPOP

"I happened to cut my wife down in anger"

26. Ktg.

m[??]r[??] hass-u[??]

I.GEN.OBL laugh-be.PP-M-SG

"I burst out laughing"

27. Ktg.

ter[??] mucc hu[??] d[??]r[??]=mar]epsilon]

he.GEN.OBL urinate be.PP-M-SG fear--ABCPOP

"He pissed with fright"

28. Bhal.

'm[??]ra hes's-jou

I.GEN laugh.AP

"I laughed involuntarily"

2.2.3 Deagentivisation with reflexive verbs

We have seen above that the passive marker -i- is added to the verb stem in inabilitative and involitive sentences with Genitive Subject. A variant of 'involuntarily' is reflexivity. In Bangani it can be realised by adding -i- to a transitive verb stem. The subject remains in nominative case. An example from the Panduan:

29. Bng. P

seu tek-[??] g[??]z

he.NOM hold.PRES.3.SG rod

'He holds the rod'

Versus:

30. Bng. P

seu tek-i-[??] g[??]z=par[??]

he.NOM hold.PSM.PRES.3.SG rod=LOCPOP

'He holds (himself) on to the rod'

My corpus does not contain much clear evidence for Bangani reflexive sentences with Genitive subjects. However, the above discussed example 20 tumar[epsilon lori[??]? 'You and searching?' has in my opinion a reflexive meaning aspect. It resembles German reflexive sentences of the type 'such dir doch einen Freund'. Sentences with reflexive meaning using a Genitive Subject are, however, found in Koci and Kotgarhi. Hendriksen (1986: 142f.) discusses under the heading "involitive and reflexive verbs" various sentence types out of which the following are, in my opinion, all reflexive. The form of deagentivisation that takes place here is not one of 'involuntarity' but the subjects are befallen by an event. The sentences all employ the passive marker -i- plus either a present tense ending or the past participle of an auxiliary 'to be' for the past tense:

31. Ktg., kc.

barek git sun-i-a ter[??]

one song hear.PSM.PRES.3.SG he.GEN.OBL

"He unexpectedly hears a song" (better: 'He listens to a song

for himself')

32. Ktg.

k[??] takka kuch sun-i-a

INT you.GEN something hear.PSM.PRES.3.SG

"Can you hear anything?" (better: 'Is anything audible to

you?')

33. Ktg.

j[??] des bitt[??] m[??]r[??] zan-i-a

that area beautiful I.GEN.OBL know.PSM.PRES.3.SG

"I like this place very much" (better: 'The place pleases me a

lot')

2.3 Contemporaneity and anteriority

In this type of construction with Genitive Subjects no deagent ivisation takes place, but different nuances of contemporaneity and anteriority are expressed. In all of them the verb has either the active past participle ending -n[??] (OIA past participle -na-), preceded again by the same passive infix -i-, or the verb has the active past participle ending -iu. (13) In case of -i-n[??] there is the same reflexive meaning as above, e.g., with nominative subject and not expressing anteriority but simple past:

34. Bng.

(s[??]) na-i-n[??] - do-i-n[??]

(they) bathe.PP-M-PL - wash.PP-M-PL

'(They) bathed and washed themselves'

An example from the Panduan with Genitive Subject and -n participle:

35. Bng. P

thik[??] tiur[??] puz-i-n[??], kua "muz[??]r" bi a [??]

Exactly they.GEN.OBL worship.PSM.PP-M-SG, crow "crow"

also come.PP-M-SG

'Exactly (when) they had worshipped (a deity), also the crow (named) "crow" arrived' (That is, the crow arrived when they had just imished their worship.)

Now two examples from the epic with Genitive Subject and -iu participle. The context to sentence 36 is a scene where the Lord of the World tells Kunti that he has long kept a boon for her:

36. Bng. P

tau=kh[??] th[??]-iu mer[??]

You.OBL=BENPOP keep.PP I.GEN.OBL

'I have (long-since) kept (a boon) for you'

The context of example 37 is a scene where God Narayana meets two giantesses who are searching for Arjuna and Bhimsena. Since he has seen them just previously, he says:

37. Bng. P

ze ca-i tumu khatirzun bius[??]n, s[??] mer[??] dekh-iu

if want.OPT you.DAT Arjuna Bhimsena, they.NOM I.GEN.OBL

see.PP

'If you want Arjuna (and) Bhimsena, I have already seen

them'

Occasionally -iu takes on the function of a future active participle. In one scene the Lord of the World gets annoyed by a honeybee that keeps on flying around him. But then he says:

38. Bng. P

sun-o (14) dei bel[??] ia-k[??] bol-iu

listen.I1VIP give.I1VIP dear.one she.OBL.GENPOP-M-SG say.PP

'Listen, dear one, (what the bee) is about to say'

Like -i-n[??] also -iu is frequently used with nominative subjects. My impression is that then the construction simply realises past tense. But this is not always easy to determine. Two examples from the epic illustrate this. In the first scene, the Five Gods have set out on a pilgrimage to Lake Manasarovar. However, they imd the holy water polluted because a shoe maker woman has taken a bath before them. When they take her to task, she says that she is innocent because:

39. Bng. P

udr[??] au na-iu, ub[??] tum na-i-a

down LNOM bathe.PP, up you bathe.PSM.IMP

'I have bathed downstream, bathe you upstream!'

In the next example the Pandavas pay a visit to the Kauravas in their capital Hastinapura. They are aghast when they discover that the Kauravas live in caves! So Bhimsena rebukes them with the following words:

40. Bng. P

phet dada teri, z[??]di-k[??] a-iu to th[??] t[??]di-k[??] a-iu au [??]-d[??], t[??]

sunei-sune-ki ban-u th[??][??]thna-z[??]ita

faugh brother your, when.GENPOP-M-SG come.PP you.NOM was

then.GENPOP-M-SG come.PP LNOM be.PPRES-M-SG, then

go1d.EMP-gold.GENPOP-F-SG build.PRES.I.SG was Hastinapura

'O brother, shame on you! If I had come at the time when

you had come (here) then I would have built a Hastinapura

of gold over gold'

2.4 Gerundive

Usually 'gerundive' means a construction with a verbal adjective that expresses necessity. This is the case in the following examples. They seem to come close to Onishi's "modality (irrealis)" feature characterising certain non-canonical subject markers (2001a: 39f.). In Bangani, the gerundive is realised with the passive future participle -n[??] (Masica 1991: 322). Instead of the usual -i- passive marker one finds, not surprisingly, another marker -[??]-. However, this marker has frequently no clearly determinable function and therefore should not interest us further. Tense is optionally indicated by an auxiliary (as in the following example from the epic):

41. Bng. P

thik[??] tiu-k[??] u[??] kholi= k[??] dar[??]= ag[??] p[??]ic-[??]-n[??] ... t[??]ti[??]=kh[??] se

bi p[??]i[c]-[??]

Exactly they.GENPOP-M-SG be.PP-M-SG gate=GENPOP.OBL

door.OBL=LOCPOP arrive.SF.FPP-M-SG ... so.much=BENPOP

they.NOM also arrive.PP-M-PL

'Exactly (when) they had to appear at the door of the gate ...

exactly then (lit. 'so much=for') they in fact arrived'

Again literal Hindi: thik thak unka hua phatak ke dvar par pahucna ... tabhi ve bha pahuce. This sentence is the only clear example of a gerundive with Genitive Subject. There is no doubt that this construction is used very rarely. The two following examples from the epic are not so clear because the word k[??] is in epic Bangani both the oblique form of the postposition k[??] (15) and a local postposition meaning 'with, nearby'. The first sentence appears in a scene where Bhimsena has to cut through the long hair of Draupadi because of the Kaurava Duhsasana holding fast to it. Draupadi advises Bhimsena to tell Duhsasana that he should place the cut hair at the side of his brother Duryodhana whereas he, Bhimsena, would place a bride at his side. Upon this Bhimsena answers:

42. Bng. P

p[??]r tes- k[??] [??]lkh[??] cetai-n[??] (16)

but he.OBL.GENPOP(?) light let.feel.FPP-M-SG

'But that he must feel as a light (punishment)'

The next example comes from a scene where the Pandavas encounter a hostile water mill (!). They wonder why the mill has become their enemy because Bhimsena had built it and:

43. Bng. P

Kuta Mata- k[??] pis-n[??]-k[??] th[??]

Kunti mother.GENPOP(?) grind.FPP.OBL.GENPOP-M-SG was

'(The mill) was (to be used) for grinding (flour) by mother

Kunti'

Hendriksen (1986: 106) quotes the following short sentence from Kotgarhi as an example of a gerundive:

44. Ktg.

yaks-[??] hamm[??] kha-n[??]

demon.GEN.OBL we.NOM eat.FPP.OBL

"'The troll will eat us' (lit. "to-the-troll we (are) to-be

eaten")"

But this sentence also looks suspicious to me because the sentence could also reflect an ordinary construction like Hindi ham raksas ka khana hai 'we are the food for the demon'. In the overwhelming number of cases, Bangani and Deogari mark the subject in gerundive constructions with the ergative. So the question is why there are at least a few examples with Genitive Subjects. However, I fear that for the time being I will have to owe an answer to the reader.

2.5 Essential state and non-control

Both notions characterise quite well the constructions discussed in this section. Onishi (2001a: 38f.) mentions in the paragraph titled "stativity" two types of deagentive derivations in Bengali (with two different auxiliaries), one resulting in "non-control" (i.e., again deagentivisation) and the other in "resultative 'state"'. My use of the two notions, however, differs in some ways from the situation in Bengali. Both 'essential state' and 'non-control' appear in constructions that express the change from one state into another only in the specfic construction types below. Both Bangani and Deogari use the same constructions (Hendriksen provides no evidence for Kotgarhi and Koci). What we may tentatively conclude from the not very broad database is that Bangani seems to stress more the aspect of 'non-control' and Deogari that of 'essential state', but this might need to be checked again. 'Non-control' here means that the expressed change from one state to another was caused by an external or personal factor not under the control of the logical subject. So this is different to some extent from the other two moods of inability and involuntariness, and I suggest calling this perferitative mood, that is the mood where the subject suffers an event. And 'essential state' means that the resulting state is regarded as having an essential and not just a superficial quality characterising the subject. The following constructions differ from all the above sentence types in that they can only be constructed with verbs meaning 'to become'. Moreover, the Genitive Subject doesn't appear in the oblique case but is in concord with the complement. The first example is from the Bangani story of the little old gentleman, followed by sentences from language consultants. The background of the first example is the regionally famous story of the advent of God Mahasu in Bangan. There was a man-eating giant who spread fear and terror in the region until God Mahasu together with his guardian deities arrived from Kashmir. The guardian deities killed the giant; however his heart stayed alive and later became a demon god:

45. Bng. LOG (sentence 24)

t[??] de-[??] tipu=par[??], t[??] tetk[??] b[??]n-[??] tesr[??] jibalu

Then go.PP-M-SG Tipu=LOCPOP, then there become.PRES.3.SG

he.GEN-M-SG Jibalu

'Then it (namely the heart of the killed demon) went over to

(the village of) Tipu, then it became (the demon-deity called) Jibalu'

46. Bng.

tesr[??] b[??]n- cur[??]

he.GEN-M-SG become.PP-M-SG powder

'He became powder', i.e., 'he was beaten up very badly' (of

course against his will and in a decisive way)

A slightly different way of idiomatic expression but with basically the same meaning:

47. Bng.

mer[??] b[??]n- [??] pint[??]

I.GEN-M-SG become.PP-M-SG ball

'I became a ball' (17)

An idiomatic expression:

48. Bng.

tesr[??] go-[??] jangu

it.GEN-M-SG go.PP-M-SG carrying-basket

'It went (became) a carrying-basket'

This is said when something has turned into a mess. Note that go[??] 'went' is used here in the sense of 'became'.

The following are examples from Deogari. In the first sentence pair, two almost identical facts are expressed, however, in (49) suggesting superficiality and in (50) essentiality:

49. Deog.

se b[??]n-i durp[??]ti

she.NOM become.PP-F-SG Draupadi

'She became (the ancient heroine) Draupadi' (e.g. by putting on appropriate clothes)

Versus:

50. Deog.

teski b[??]n-i durp[??]ti

she.GEN-F-SG become.PP-F-SG Draupadi

'She became (the ancient heroine) Draupadi' (because Draupadi's spirit entered her)

And while it is possible to say in Deogari:

51. Deog.

seu b[??]n-a pradhan

he.NOM become.PP-M-SG mayor

'He became mayor'

--It is wrong to say:

52. Deog.

* teska b[??]n-a pradhan

he.GEN-M-SG become.PP-M-SG mayor

because being a mayor means holding an office. This is not an essential quality of a person. One final example to illustrate this. In Deogari it is possible to say both:

53. Deog.

se b[??]n-i birali

she.NOM become.PP-F-SG cat

'She changed into a cat'

And:

54. Deog.

teski b[??]n-i birali

she.GEN-F-SG become.PP-F-SG cat

'She changed into a cat'

In the latter case it is understood that the woman is a witch, and witches metamorphose regularly into cats in order to harass victims. In the former case it is understood that the woman had been reborn as a (normal) cat in her past life, because now she is essentially a woman and not a witch (and thus cannot meta morphose into a cat). The above examples from Deogari suggest that there is in the area an underlying system of essential vs. non essential character features which of course cannot be in vestigated on the basis of a limited number of sentences, but which determines the correct or wrong forms of these sentences. Obviously, 'well-formedness' is neither determined here by abstract syntax nor by the semantic properties of the predicates.

3 Conclusions

The above data from the different languages of West Pahari show remarkable similarities with features pointed out by Aikhenvald et al. for non-canonical markings of subjects, especially their so-called deagentivisation. Still, a major theoretical gap remains. Whereas the approach favoured in the above-mentioned book rests, as I understand it, on the assumption that the choice for non-canonical subjects is a matter of verb semantics (an obvious case for this are the Dative Experiencer constructions), the data from West Pahari rather promote a grammatical basis for Genitive Subjects. E.g., an inabilitative sentence with Genitive Subject can have as predicate any verb. These constructions are used to express inabilitative, involitive and perferitative mood, they are used to express contemporaneity and anteriority, and they are used to express necessity. This is very different from the experiencer subjecthood of Dative Subject constructions.

Are there antecedents in older forms of Indo-Aryan out of which the modern Genitive Subject constructions might have developed? I am not aware of comparable constructions in the older stages of NIA. However, there are the absolute constructions in OIA and MIA. Of interest here might be the genitivus absolutus. However, apart from the fact that it was used in subordinate sentences to express anteriority and contemporaneity (see Bubenil 1998: 197ff.), which makes them comparable with the above constructions in section 2.3, there is no evidence how these absolute constructions could have further developed into the modern Genitive Subject constructions.

Abbreviations

Languages and texts

Ap. Apabhramsa

bhad. the Bhadrawahi variety of West Pahari

bhal. the Bhalesi variety of West Pahari

bng. the Banger variety of West Pahari

deog. the Deogai3 variety of West Pahari

H. Hindi

khas. the Khasali variety of West Pahari

koc. the Koci variety of West Pahari

ktg. the Kotgarlfi variety of West Pahari

LOG "The little old gentleman", an oral narration from Bangan

MIA Middle Indo-Aryan

NIA New Indo-Aryan

OIA Old Indo-Aryan

P Panduan, an oral Mahabharata epic from Bangan

Grammatical abbreviations

ABC ablativus causae ABL ablative ALL abative AP absolute passive AUX auxiliary BEN benefactive CAUS causative CP conjunctive participle DAT dative EMP emphatic particle ERG ergative F feminine FIL filler word FPP future passive participle FUT future GEN genitive IMP imperative INF infinitive INS instrumental INT interrogative word ITM the iterative marker -i- LOC locative M masculine N neuter NOM nominative OBJ object OBL oblique OPT optative PL plural PM person marker POP postposition PP past participle PPRES present participle PRES present tense PSM the passive marker -i SF suffix SG singular

References

Aikhenvald, Alexandra Y., Dixon, R.M.W. and Masayuki Onishi (eds.) 2001, Non-Canonical Marking of Subjects and Objects. Amsterdam and Philadelphia: John Benjamins.

Bhaskararao, Peri and Karumuri Venkata Subbarao (eds.) 2004, Non-Nominative Subjects. Two volumes. Amsterdam and Philadelphia: John Benjamins.

Bhatt, Rajesh 2007, Covert Modality in Non-Finite Contexts. Berlin and New York: Mouton de Gruyter.

Bubenik, Vit. 1998, A Historical Syntax of Late Middle Indo Aryan (Apabhramsa). Current Issues in Linguistic Theory. Amsterdam and Philadelphia: John Benjamins.

Gaeffke, Peter 1967, Untersuchungen zur Syntax des Hindi. The Hague: Mouton & Co.

Hendriksen, Hans 1986, Himachali Studies; III. Grammar. Det Kongelige Danske Videnskabernes Selskab, Historisk-filosofiske Meddelelser 48,3. Kbenhavn: Munksgaard.

Masica, Colin P. 1991, The Indo-Aryan Languages. Cambridge: Cambridge University Press.

Matras, Yaron 2002, Romania A Linguistic Introduction. Cambridge: Cambridge University Press.

McGregor, R. C. 1972, Outline of Hindi Grammar: With Exercises. Delhi: Oxford University Press.

Onishi, Masayuki 2001a, Non-canonically marked subjects and objects: Parameters and properties. In Aikhenvald, Alexandra Y., Dixon, R.M.W. and Masayuki Onishi, pp. 1-51.--2001b, Non-canonically marked S/A in Bengali. In Aikhenvald, Alexandra Y., Dixon, R.M.W. and Masayuki Onishi, pp. 113-147.

Varma, Siddheshwar 1938, The dialects of the Khasali group. Journal of the Royal Asiatic Society of Bengal 4: 1-65.--1948, The Bhalesi Dialect. Royal Asiatic Society of Bengal Monograph Series Vol. IV: 1-64. Calcutta: Royal Asiatic Society of Bengal.

Zoller, Claus Peter 2007, Is Bangani a V2 language? European Bulletin of Himalayan Research 31, 83-141.

Zoller, Claus Peter (forthcoming). Die Panduan. Ein mundliches Mahabharata-Epos aus dem Garhwal-Himalaya.

Claus Peter Zoller

Oslo

(1) I am aware that the term Genitive refers strictly speaking to the syntactic relationship within a clause constituent. However, this term is used in the literature, and I therefore follow it.

(2) My main language consultants for Bangani have been Mr. Gabar Singh Chauhan and Mr. Trilok Singh Chauhan (both living in New Delhi) and for Deogai3 (spoken south of Bangani) Mr. Shamsher Singh Chauhan and Mrs. Savita Singh Chauhan (both also living in New Delhi).

(3) Already published transliterated sentences have been adjusted whenever necessary to the transliteration and abbreviation standards followed by me in this article. All the West Pahari languages treated in this article are tone languages. However, the tonemes are not shown in my transcriptions.

(4) This is perhaps not quite right as Vit Bubemlc (1998: 125f.) quotes a few examples from Apabhramsa. However, he stresses that "Examples of the innovative go-passive are extremely rare in our Apabhramsa texts."

(5) Tlie element -i- has in fact adopted a range of other functions, as can be seen in some examples in this paper. For instance, it also expresses iteration.

(6) Bangani, Deogai3, Koci and Kotgarhi continue in some of their positive present verb endings inherited OIA forms. The negative present, however, is constructed with a participle -da which is preceded by an -n- in case the verb stem ends in a vowel. Alternatively, the preceding vowel is nasalised.

(7) Tlie final -i is probably originally the same passive vowel. However, here it has no passive function. For Bangani and Deogai3 complex predicates there is the rule that if the main verb consists of the bare stem, then -i always has to be added.

(8) Also visible in the above examples 3-5. Bangani has two genders.

(9) Centner: a measure of weight equivalent to approximately 100 pounds; a hundredweight.

(10) Literally: 'Let (him) burn!'

(11) The abbreviation AP stands for Varma's notion of "absolute passive" and has been explained under 2.1.1.

(12) Some of the following examples of this section don't look (in the translations) like moods but rather like aktionsarten. However, they do realise mood and not aktionsarten because they do not specify the details of an event, but rather the attitude or assessment of the subject vis-a-vis the nature of the reality of the event.

(13) Whether -iu was originally -i-u with passive marker is not clear to me.

(14) The -o is interference from the Hindi imperative.

(15) It is used only in epic language. Elsewhere one uses r[??].

(16) The verb is a causative extension of cetn[??] 'to feel' plus reflexive -i-. The German translation of cetaina[??] would be 'sich anfuhlen lassen'.

(17) Implying that the subject was badly beaten up.