The grammaticalization of verbal adjectives (or gerunds) in past, future, modals and infinitive.

Annie Montaut, Inalco, Paris (UMR 8202 INALCO/CNRS/IRD, Labex 083)Submitted (Andrej Malcukov & Martin Haspelmath eds)

Introduction

Whereas the grammaticalization of the Sanskrit passive past participle into ergative alignment in many Indo-Aryan languages has been profusely studied, its grammaticalization into definite past (or preterit) has raised less interest, as well as the nominative realignment in Eastern languages. The parallel realignments of the passive future verbal adjective of obligation or gerund has similarly raised only a very limited (obligation patterns) and recent interest, and the grammaticalization of the form into infinitive is still practically unstudied. The aim of the paper, in continuation of Montaut (in press), is to account in a unified manner for the various grammaticalizations of both forms and try to understand why different paths can be traced in different languages with different outcomes, by considering the whole context of constructions and other existing forms, with a comparison with similar grammaticalizations in Latin and Romance languages.

1. The grammaticalization of the passive participle into finite past and “ergative alignments” in Indo-Aryan

As is well-known since Kellogg (1875) and Grierson (1903-28), what is today called the ergative construction or alignment, and was in the 19th and early 20th centuries mentioned as a “kind of passive construction”, stems in Indo-Aryan out of a particular type of the Sanskrit nominal sentence, generalized in classical language since the stories of the Vampire display 1115 nominal forms for past against only 38 finite tense forms (Bloch 1906:60)1. The past passive participle (hereby glossed PPP) or verbal adjective behaved as the predicate, replacing the numerous Vedic synthetic past tense, and agreed in gender and number with the patient, the agent appearing in the instrumental, a case later renewed with a postpositional marker, the so-called ergative case marker, specific only in certain IA languages.

The following example in classical Sanskrit (1) leads to the modern ergative construction in Hindi (2a) in contrast with present nominative sentences (2b), a construction extensively studied since the eighties for its syntactic as well as pragmatic properties (for Hindi among others Kachru 1987, Montaut 2004, Mahajan 1997, Davidson 2006, who all notice that tha agent has all the control properties and the patient only few discourse properties.

(1) mayā / mama tat krtam (SK)1SG.INS / 1SG.GEN DEM NOM.N.SG do.PPP.NOM.N.SG ‘I did/have done that (lit. ‘by me/ of me this done’)’2

(2) mai.ne yah / apnā kām kiyā (HINDI)1SG.ERG DEM.M.SGREFL work.M.SG do.M.SG‘I did this /my work’

1 Phonological erosion of the –ita ending gave –iya then –a endings, to which were added the modern gender/number endings in MSH and most Hindi dialects: the morphology of the predicate in the definite past is still a nominal form, varying only in gender and number, the reason why in most grammars of IA till the late 20th

c. such tenses were considered as “participle tenses”, in contrast with the “radical tenses”.2 The genitive marking of the agent is restricted to pronouns, instrumental being by the most current marker.

1

1.1. Early NIA data: alignment shift and acquisition of temporal meaning

Equally well-known is the fact that this split ergativity (for definite past and derived tenses such as present perfect, pluperfect and compound forms with the past participle in other moods) is now restricted to the Western part of the IA speaking zone, if we except Assamese in the North-East. What has been less commented is the fact that it was prevailing throughout Indo-Aryan up to the middle stages of New Indo-Aryan (14-16th centuries), depending on the regional varieties. Here are early examples of the extension, both in Western and Eastern languages, right from the Prakritic stage of Old Indo Aryan; in (3) from Ashoka’s first edict which displays both Western (3a: Girnar) and Eastern (3b: Jaugada) dialectal varieties, and (4), from the play writer Kalidasa, which shows the contrast between instrumental agent and nominal predicate in the past, and nominative agent and finite predicate in the present:

(3) a. iyam dhammalipī devānampriyena priyadassina ranna lekhapita (3) b.iyam dhammalipi devānampiyena piyadassina [lajina] lekhita

this law-scripture of-gods-friend friendly-looking king inscribed NOM.F.SG NOM.F.SG INST.M.SG INST.M.SG INST.M.SG NOM.F.SG

‘The friendly looking king beloved of gods has (made) engraved this law-edict’ (PRK)

(4) hau pai pucchimi … ditthī pia pai sāmuha jantī 1SG.NOM 2.OBL ask.PRS.1SG seen.F.SGloved.F.SG 2.OBL in.front passing.NOM.F.SG‘I ask you… Did you see (my) beloved passing in front (of you)?’ 3 (PRK)

All Western and Central languages, now ergative, displayed at an older stage contrasts similar to (4), with personal endings on the finite verb in the present whereas in the past the verb retains a nominal morphology, gender number agreement with the patient, and the agent is case marked in the oblique, as a result of the usual syncretism abl/dat/loc in the area. The series goes from Westernmost languages to Central, and one of the oldest literary texts in Sant Basha from Kabir (13th c.), the literary koine which is considered the cradle of Hindi/Urdu:

(5) a. guri dānu ditta (OLD PUNJABI:guru.OBL/LOC gift.M.SG given.M.SG

‘The guru gave the gift’ Guru Granth Sahib)(5) b. sundari.nai Bharat.hai rākhī (OLD RAJ/GUJ)

beautiful.lady.F.SG.ACC Bharath.OBL.M.SG given.F.SG ‘Bharata kept the beautiful lady” (Tessitori 1914: 167)

(5) c. mai nahi mākhan khāyau (OLD Braj)1.SG NEG butter.M.SG eat.M.SG ‘I did not eat the butter’(Surdas 25.1 mid 16th c.)

(5) d. krpā kel.i tumhi ai se pāhi.l.e myā Old MARATHI pity.F.SG do.l.F.SG 2.OBL DEM.S.SG see.l.N.SG 1SG.OBL/INS‘You have had pity’ (Jnaneśvari 11.255, in Bloch 1965: 261)

(5) e gur.i diyā palītā (SANT BHASHA)guru.LOC/OBL give.M.SG stick.M.SG

‘The guru gave the stick’ (Kabir 8.3)

3 The form pai for 2nd person is already used as a syncretic marker for several oblique cases.

2

(5) f. Virrsigh Joysi.ya bhās pāi ‘ (Old central Pahari)Virsingh Joshi.OBL proclamation/bond.F.SG get . F.SG‘Virsingh Joshi (the king) received the bond’ (from Pant 2009 in Stronski 2014: 283)

Already some variations appear such as the agreement with a marked object in Old Marwari (5b), a marking itself rather recent, a –l- suffixation on the predicative participle in Old Marathi (5d), variations in the oblique case marking (hai, -hi /-i, iya ending. But the basic structure is the same.

It was also maintained unchanged in the Eastern languages, with a similar contrast between present, nominative agent and finite form agreeing with agent on one side, and non finite participial forms in the past with oblique agent (6a): still in the 16th century the Bhojpuri first person pronoun had an overtly nominative form inherited from the Sanskrit nominative aham (hau manus “I [am] a man”) whereas it displayed an oblique form overtly inherited from the Sanskrit instrumental in the past (mai pāi “I obtained”) (Tiwari 1966: 158).Series (6) illustrates a few of these Eastern varieties in the early NIA stage, starting with examples from the oldest buddhist poems or caryas (Chatterji) at the origin of middle Maithili, Oriya and Bengali, which contrasts intransitive construction (nominative subject) and past transitive (ins/obl agent), then older stages of various language:

(6) a.hau acchilo (?e), mai bujhila, dila OLD BENGALI1SG come.il.1SG 1SG.INS understand-ila give.ila ‘I came, I understood, gave…’ (caryu 35, SKC 964)

(6) b. bhala na kala mane dela bisavāsa (middle MAITHILI)good.M.SG NEG do.la 1.SG.OBL give.la trust.M.SG‘I did not [the] good, [that] I gave trust’4

(6) c. Tirahuti leli jānhi (KH 20) Tirahuti.F.SG take.li.F.SG REL.OBL‘By whom (the city) Tirahuti was taken (in Jha 1958)

(6)d. eka rāta sapnā mai dekhā (OLD AWADHI) one night dream.M.SG 1SG.OBL see.M.SG

‘One night I saw a dream’ (Nur Mohammed 4)

Here again light differences start emerging, such as the erosion of gender in all Eastern varieties stemming from the Magadhean Prakrits (as opposed to those stemming from the Sauraseni Prakrits in the West), so that agreement loses clarity. But again, the structure is essentially maintained, almost everywhere with the –l- suffix.

As indicated by the translations of both series, mostly borrowed from the author from whom examples are taken,5 the temporal meaning of the form extends from present perfect and plu-perfect to the mere representation of anterior events, as a preterit As soon as the old passive participle started grammaticizing as the only expression for past, it assumed both a stative / resultative meaning (perfect) and an anterior meaning, in what Nespital (1989) identified as a “Proto-Aktiv Satz” right from the Pali stage in Milindapanha. This change in meaning has been well documented in Peterson (1998:190) for Pali and Breunis (1990) for 4 But moe ghenili hāveri-mālā 1sg.obl enrich.l.f.sg bone-garland.f.sg “I have enriched the bone garland’ with different oblique form for the first person pronoun, although also distinct like in ex () from the direct form (hau) check5 Except when they were translated by passive sentences. As for the gloss, never given, I adopted the gloss ins in case the pronoun has a distinctively instrument form.

3

classical Sanskrit. I have accounted for this shift (Montaut 1996, 2007) on the lines of Bybee, Perkins & Pagliuca (1994) within the general process of grammaticalisation of the new periphrastic form: as long as the nominal form, initially a marked innovation, competed with the old tensed forms, it retained its original restricted meaning (stative). When the old forms disappeared, the new form, no longer stylistically expressive, occupied the whole space of past and acquired what Bybee calls an open meaning. This meaning conveyed the values of preterit or anterior, resultant, stative-resultant and stative : as stated by Bloch (1906: 60), “du contexte et du sens de la racine dependent la valeur active ou passive et la nuance temporelle et modale du participe. Il est donc le substitut de toutes les formes verbales du passé à tous les modes et à toutes les voix”. When a copula came into use, first, in Sanskrit, in order to prevent ambiguities when overt pronouns in the first two persons were omitted (Bloch 1906), then to signal a restricted meaning stative-resultant as an expression of stylistic emphasis (Breunis 1990:141), the simple form started to contrast with the new copula one as the unmarked form with the marked one, and restricted its meaning to what was not expressed by the copula form: anterior event, namely a preterit.6

1.2. Further divergent grammaticalizations in modern New Indo-Aryan: shift from passive to active or “possessive perfect”?

Modern Eastern IA languages shifted back to a nominative alignment in the past between the 14th and 16th centuries, and for most of them added personal endings to the participle, making it a finite tensed form as in Bengali:

(7) a. āmi boi.ta por.l.ām b. tumi boi.ta por.l.e (BENGALI)1SG book.DEF read.PST.1SG 2 book.DEF read.PST.2‘I read the book’ ‘You read the book’

As noted by S. K. Chatterji, the new personal endings in the past (1 -ām, 2-i, -e) are distinct from the inherited personal endings of present (1 -i, 2 -ish, o) and come from pronominal stems. As for the -l-, which is now analyzed as a past tense marker, it originates from an adjectival suffix, the same as Hindi –il- (rang.il.ā “coloured”), a further evidence of adjectival nature of the predicate in the pre-ergative alignments (Chatterji 1926: 928, Tessitori 1914). Its re-analysis as a past (PST) tense marker corresponds to the renewal of the pre-ergative alignment into a nominative alignment.7 Chatterji in 1926, following the then usual interpretation, considers this evolution as a shift from passive to active, but he gives all the elements for a proper understanding (regarding the nominal feature of the old form, hence the stative rather than passive meaning of the original pattern)8.

Similarly, Eastern Hindi, which also displays the same suffix -l- in the definite past, is systematically interpreted as a shift from passive to active in Saxena (1937: 247sq) for Bhojpuri, Jha for Maithili (1985 [1958]: 492 sq) and Tiwari (1966: 171) for Awadhi: “when the original passive construction was lost in Bhojpuri as in other Magadhean dialects, the Prakritic constructions with the passive participle became a regular verb in Bhojpuri, and it began to be conjugated by adding personal terminations which came from the radical tense as well as from the s/h future” (Tiwari 1966: 171).

6 The topicalizing fronting of the agent seems to be acquired in the mid MIA (Breunis 1990, chapter 6 on word order) but Bubenik & Paranjape 1996 (116-7) date the linguistic perception of the oblique noun as a semantic subject later (end MIA).7 The fact that the same suffix may also occur at other tenses (certain persons of the future in Bhojpuri for instance) shows that this recent re-analysis as a tense marker is limited to Bengali.8 An “archaic” remnant of the old system survived in the classical language with the –e ending for transitive past.

4

Again similarly, other Indo-European languages such as Persian (Cardona 1970) and Latin then Romance languages went through the cycle finite past tense > nominal sentence with predicative participle agreeing with the patient and oblique agent > restructuration with nominative agent and verb agreeing with agent, as stated in Kurylowicz (1931, 1965). The Persian data considered are (8a) for old Persian and (8b) for modern Persian:(8) a mana tyâ karta.m

1SG.GEN DEM NOM.N.SG do.PPP.NOM.N.SG ‘I have done that’

(8)b man in kar.d.am1SG.NOM this do.PST.1SG ‘I did that’

The corresponding Latin data involves a periphrastic past with participle and dative of the agent (dativus auctoris), restructured with the have auxiliary and a nominative subject with still today a vestige of the old pre-ergative pattern in the form of the agreement of the participle with the direct object when preposed (les lettres que j’ai écrites, je les ai écrites)

(9) a mihi id factum (est) (LATIN)1SG.DAT DEM NOM.N.SG do.PPP.NOM.N.SG (be.PRS.3S)‘I did/have done that’ (lit. ‘by me/ of me this done’)

(9) b. ego id factum habeo (LATIN)1SG.NOM DEM NOM.N.SG do.PPP.NOM.N.SG have.PRS.1SG‘I have done that/it’ (lit. I have this done)

(9)c j’ai fait ceci (FR) io ho fatto questo (IT) yo he hecho esto (SP) (ROMANCE LG)

As stated by Kurylowicz in his paper on the formation of tenses in Romance languages, further developed in his study on the evolution of grammatical categories (1965): “ in the evolution that we consider, the decisive step is the replacement of the dative + esse [be] + nominative by nominative + habere [have] + accusative. The passive construction has been transformed into an active one” (1931: 107).

Against this classical analysis, reflected by the Indian scholarly literature mentioned above on similar evolutions, Benveniste, in a pioneering paper of 1952 about the meaning of the perfect, proposed his own view of the perfect as basically possessive: the case marker in such periphrastic expressions with participle was in no way an agent marker but a possessor marker, since classical Latin uses the dative and verb “be” to represent possession9, a construction later on restructured with verb “have” and nominative subject in vulgar Latin:

(10) a . mihi est filius / pecunia (CLASSICAL LATIN)1SG.DAT be.3SG son.NOM.M.SG/ money.NOM.F.SG

(10) b. b ego filium/pecuniam habeo (Late LATIN)1SG.NOM son ACC.M.SG money.ACC.F.SG have.PRS.1SG‘I have a son/ money’

Consequently the evolution of the perfect has nothing to do with a shift from passive to active, according to Benveniste, but is a mere reversion (“renversement, retournement”) of the old possessive pattern. Similar conclusions were presented later by Pirejko (1979) and by Trask (1979: 397) who associates the possessive origin with “the incorporation into the inflectional paradigm of a nominal form” with a genitive agentive complement.

9 Whereas the standard case marker for agents in passive clause is a + ablative, and similarly in old Persian hacama + ablative and not genitive.

5

1.3. Re-interpretation of the IA grammaticalization path: localizing meaning

This very famous analysis of Benveniste of the Indo-European perfect raises however a problem regarding the Indo-Aryan data, since it is exclusively (and polemically) based on case marking. As already mentioned, the agent, particularly when afull NP, is standardly in the instrumental in classical Sanskrit (see for details Montaut in press) and not in a possessive case, which argues in favour of a passive origin. Yet, further developments of the pre-ergative pattern into a fully ergative alignment do not emphasize a passive origin and the new marker used to reinforce the old syncretic oblique case are very commonly derived from location nouns, not to mention the fact that, till inflected forms were in use, they also served for locative (the –i ending in Sant basha). The most wide-spread of the now ergative case markers is ne (ni, nai, ne, ne), found in Hindi/Urdu, Panjabi, Marathi, Gujarati, and it is derived from the locative (* karnasmin, a renewal of the classical form karne by analogy with the most current paradigm) of the noun karna “ear” (Tessitori 1914: 65sq ; more details and examples in Montaut in Press). Then from karnasmin, kannahi, kannhai, kannai, nai, nai, ne, ne. Tessitori, the fist scholar who identified the correct origin of ne, and was followed by Trump (1972: 401) in assigning to it an original locative meaning, by Tiwari (1961), Saxena (1937), Chatterji (1926) as well as Chatak (1966) for Garhwali and all reliable scholars throughout the 19th c. early 20th c. Tessitori (1914-6: 68-70) gives examples such as (11a) with a clearly locative meaning and (11b) with an agentive meaning:

(11) a. mithyādrsthi loka kanhai sravai vasirau nahifalse.look people LOC hermit.M.SG dwell.PRS.3M.SG NEG‘A shravaka (hermit) should [does] not live near heretics’

(11) b. adiśvara naï diksā lidhi (OLD RAJASTHANI)Adishwara LOC/ERG consecration.F.SG take.F.SG‘The Adishvara took the consecration’

The second most widespread ergative marker is le (lai, al), found in Kumauni, Garhwali, early Nepali, and it is derived from the verbal root lag “to be in contact, touch” (lagi/lāgi > laï, lai, le) originally meaning “having come in touch with”, “up to”, “for the sake of” (Turner). Reflexes of both markers are significantly used for dative (nai/ne/nū in Rajasthani, Panjabi, Gujarati; lā in Marathi), to a far larger extent than they are for instrument (ne in Marathi, , -an/le in Garhwali/Kumaoni). These connections should however not be over emphasized, given the extremely weak semantic content of the initial etymon, and similarly the fact that the same ending –i was used in the Sant Basha of Kabir both for agents, such as (5e) above, and locative complement such as (12):

(12) ābari dīsai ketā tārā sky.LOC be.seen.3M.PL. how.manystar.M.PL‘How many stars we have seen in the sky!’ (146.1)

Last evidence, the Gujarati ergative marker presents an alternation ne/e, the first form to be related to the above mentioned ne whereas the –e ending is also used as a locative case marker (Cardona & Suthar 2003: 678) and seems to have been the initial marking, since we find it in both functions in the Jain Gurjar Kavyo in the 14-15th c. (Desai: 1926)

(13) Sita.e kagal vacyo MOD GUJARATISita.ERG letter.M.SG read. M.SG‘Sita read the letter’

6

(14) a. jamunajī.n.e tat.e (OLD GUJARATI)Yamuna.HON.GEN.LOC bank.LOC‘On the bank of the Yamuna river’10 (630)

(14) b. ame yamuna gayā(…) rokyā nandanā nānhadī.e1PL Yamuna go.M.PL stop.M.PL Nanda.GEN small.boy.LOC/ERG

‘We have been to the Yamuna (…), Nanda’s boy stopped [us] ([we] were stopped by N)

The specialization of a given case marker for a given function is clearly recent (and in some languages still very weak: Montaut 2015, under press, Stronski 2010), which explains such contradictory associations (INS/ERG, vs DAT/ERG or LOC/ERG)11. Parallel to the originally weak semantic load, the construction itself can be considered a predication of localisation in an abstract sense: the process or the result is merely located in relation with a source, not represented as the direct source as in the transitive model. Coming back to Benveniste and the Latin restructuration with “have” auxiliary, he clearly states that “have” is but “an inverted be”, that is a stative verb. This notion was developed in another pioneering paper on “être” et “avoir” dans leurs fonctions syntaxiques), particularly its use in French and Romance languages for expressing transient states (avoir faim, peur, mal “be hungry, cold, feeling pain”). It is in this abstract sense that I suggested that ergative pattern as well as dative subject sentences widespread in IA for non action verbs (physiological or psychological processes) should be considered as predications of localisation (Montaut 2004).

The existence of a similarly “pre-ergative” pattern for modality and future in various early NIA languages confirms that the so-called ergative alignment in IA is not such a well-delimitated (and, within IE, exotic) innovation for transitive verbs in the past.

Besides, the variations in the so-called ergative alignment are so huge that one may question the very notion of ergativity in IA. Whereas it is a quite clearly ergative pattern in Hindi/Urdu, despite the default agreement with marked object (TAM split), and in Panjabi, Gujarati and Marathi (TAM and person split), “ergativity” in Nepali has only the agent marker for claim since the verb agrees with the agent (ex?) and similarly in Assamese, where the –e marking for transitive subjects is now considered as an overt nominative (Goswami & tamuli 2003: 432); in Rajasthani what remains of it is only the agreement with the object even marked, but the agent is unmarked; ergative marker, again with the exception of Hindi/Urdu/Panjabi, are rarely specific, etc.. Such variations are too complex and important to be treated here, being outside the main purpose of the paper (for a better approach see the excellent synthesis of Stronski 2010), but invite to question the treatment of the ergative alignment as a privileged one, and only alternative to the nominative alignment.

2, The grammaticalization of the modal verbal adjective into future and potential

The past passive participle or verbal adjective is not the only form which gave rise to non-nominative alignments via the nominal sentence with oblique agent, further re-aligned with nominative subjects and tense verb forms. Eastern IA languages are known to have developed their future tense from the Sanskrit verbal adjective of obligation (glossed OVA) in –tavya, “to be V.ed”. The parallel with the evolution of past has long been noticed by the historians of 10 Like in Hindi, the genitive postposition is an adjectiving suffix and the noun in the genitive inflects for gender, number and case in agreement with the head noun.11 Bangaru (Panjab-Haryana) is well-known for displaying the same case-marker, nai in certain dialects, si in others, for ERG, DAT/ACC and INS (for examples see Montaut 2007, 2015. Kului, a Western Pahari (Himachal, North of Panjab) also presents this peculiarity (example in Stronski 2014). Quote examples? Maithili too displays “contradictory” uses of the postposition so, INS/ABL and DAT/ACC (Jha 1958: 30).

7

those languages but not considered as a highly relevant fact before Montaut (1996, 2007). In languages which went through the whole cycle nominal sentence with instrumental agent and passive obligative participle agreeing with the patient (or, if no patient in the neuter) > future meaning with oblique agent > “active” shift with nominative subject and personal endings on the predicate, the parallel with the evolution of past is perfect and confirms the analogy emphasized by Kurylowicz regarding the evolution of tense in Romance languages. Let us first see this type of evolution before turning to other developments in Western languages.

2.1. The evolution from Sanskrit to Bengali: identical grammaticalization of future and past

In classical Sanskrit the periphrastic construction involving a verbal adjective (sometimes called gerund or gerundive) was the standard way to express obligation (15) in a pattern strictly parallel to the past one (1), but which is found both with transitive (15) and intransitive verbs (16); example (17) gives a typical illustration of the frequency of both past and modal nominal sentences:in the narrative register

(15) mayā tat kartavyam (SK)1SG.INS DEM NOM.N.SG do.OVA.NOM.N.SG ‘I have to/should do that (lit. ‘by me this to-be-done’)’

(16) yamayor apramattayā tvayā bhavitavyam (SK: Vetala 35)constraint.LOC.DUAL attentive.INS.F.SG 2.INS be.OVA.N.SG

‘You should be attentive (non-distracted) regarding the observances”(17) mantriputrenoktam (mantriputrena-uktam) “adya tvayā gantavyam”;

minister.son.INS.M.SG say.PPP.N.SG now 2.INS go.OVA.N.SG tayoktam (= tayā.uktam) “gantavyam”3.INS.F.SG say.PPP.N.SG go.OVA.N.SG

“Now you should go”; she said: “should go” (Vetala 9.8.13)

This pattern remains unchanged in the various Prakrits of Middle Indo-Aryan such as the Magadhean variety in Ashoka (18), the direct ancestor of Bengali:(18) hida no kimci jive alabhitu pajohitavye no pi ca samāje kattavye

here no some living kill sacrifice. no and assembly doNOM.N.SG CV OVA.NOM.N.SG NOM.M.SG OVA.NOM.M.SG‘One should not sacrifice by killing a living creature nor hold a meeting’(it should not be sacrificed by killing a living being nor a meeting should be held)

It seems to have started including a temporal meaning as soon as Ashoka’s time in the Eastern region, since according to Chatterji (1926: 966) the rock edit of Sarnath (19) has “a vague mandatory sense, with an express future implication”:(19) iyam sāsane vinapayitavye

DEM NOM.M.SG principle.M.SG make.known.OVA.NOM.M.SG ‘This principle should/will be taken to knowledge (made to be known)

Old Bengali displays a very similar construction, both for intransitive and transitive verbs with the characteristic –b- suffix as an eroded remnant of the old –tavya, and agreement with the:patient of transitive verbs, before the loss of gender marking and an invariable –ba after it:(20) a.toe sāma kariba maï sānga OLD BENGALI

2.OBL with do.ba(M.SG?) 1SG.OBL company.M.SG

8

‘I shall accompany you (do company with you)’ (carya 10, Chatterji 1926: 967)(20) b. maï dibi piricha (OLD BENGALI)

1SG.INS give.b.F question.F.PL‘I will ask questions’ (carya 15 Ibid)

(20) c . tabe to.ka rakhiba kona jāne (OLD BENGALI)then 2.ACC protect.ba which person.OBL

‘Then who will protect you?’

(20) d mai jaivo (=jaiba?) govinda saha khelana1SG.INS go.b Govida with .play.INF

‘I shall go to sport with Govinda’ (Chatterji: III, 30) Manasollasa, 1129)

Around the 15th century, this construction was transformed into a nominative one, with nominative (unmarked) subjects and tensed verb agreeing with the subject. The verbal form consists in the old –ba- form now suffixed with person endings, and these suffixed person endings are the same as the ones used for past, both distinct from the present endings (for the matter –ish and –o respectively for second familiar and second respectful person)

(21) . tu boi.ta por.bi b. tumi boi.ta por.b.e (MOD BENGALI)2SG book.DEF read.F.2SG 2H book.DEF read.F.2H‘You (familiar) will read the book’ ‘You (respectful) will read the book’

The verb morphology is now analyzed as Base+future marker –b- + person ending, in the same way as the past is now analyzed as Base + past marker –l- + person ending, both processes of reanalysis occurring at the same time. With a different formulation, the above mentioned grammarians of the evolution of Eastern IA languages acknowledged this parallel as a common “active transformation”: Chatterji for Bengali (1926: 987) indicates that “the affixes are exactly on the lines of the past”12, in contrast with those for present (for 2nd person intimate –ish, and 2nd person neutral- o), and also that the shift in alignment (“construction”) occurred at the same time for future and past.

Other Eastern languages also went through this “active transformation” but its outcome is not so clear as in Bengali since in Awadhi for instance the sigmatic inherited future resisted and is still prevailing at certain persons (and similarly in Bhopjpuri), and Maithili which now consistently displays a –b- future has acquired a complex agreement system indexing several participants. But in older times the construction was maintained with gender agreement (now lost) still visible:

(22) a. sumarabi mori name … prema sumaraba Old MAITHILIremember.b.F myF name..F love.M remember.b.M‘You will remember my name, you will remember my love’ (Vidyapati: 9, Jha 1958)

(22) b. hamahu nāgari sabe sikhaūbi (V 52, Jha1958: 495)1.OBL lady.F all(=pl) teach.b.F‘I shall instruct the ladies’

(22) c. mane ki bolaba sakhi apana genāna (Vidyapati 24, Jha 1958: 495)1SG.OBL how speak.ba friend REFL experience.M.PL‘O friend, how shall I speak out any experiences’13

12 First person now in -o shows less clear analogy, but the vowel is labial like the consonant for past (-am) and the form was previously more visibly stemming from the same affix.13 In example (22b) hamahu is the oblique form of the first person (singular and plural, today hama), whereas (22c) displays an oblique case of the base for first singular person, now gone out of use, with a case marking unglossed by the author, but evoking the palatal oblique forms for first person (Hindi mujh, mujhe).

9

Awadhi and Bhojpuri, which have now a complex paradigm in the future, also displayed the same morpho-syntactic pattern in their early stages before the “active transformation” which created “the future affixes for the first, second and third persons masculine and feminine singular and plural […] in a line with those of simple past” notes Tiwari for Bhojpuri (1966: 161). It is remarkable that in Awadhi, the –ba predicate with oblique agent of the older pattern prevailing till the 16th century had a meaning still conveying necessity, whereas it is no longer the case for the Bengali or Maithili equivalent. The resistance of the old sigmatic future for certain persons may explain why the –b- form took longer to grammaticize into a future14:

(23) a.kathā bhāśā-baddha karabi mai so-saba-hetu kahaba mai gāīstory.F.SG language-ridden do.b.F 1SG.INS this all-because say.b. 1SG.INS sing.CV‘That story is to be composed by me in the vernacular, for this reason it is to narrated

by me by singing’ (or: I shall compose, I shall tell by singing) (Tulsidas 18)15 AWADHI

(23) b.ghara kaisai paithaba mai chūche, kaunu utara debau tinha pūchehouse how enter.ba 1.SG.INS deprived which answer.M.SG give.b.M.SG 3PL.OBLask‘Empty how shall I enter the house, when they ask what answer shall I give’

(Jayasi p. 121, in Saxena 1937: 261)(23) c.prāna-priya siya jānibi, nija kinkarī-kari mānibi

life-dear Sita.F.SG know.b.F.SG refl slave consider.b.F.SG‘Sita is to/will be regarded as beloved like life, she is to/will be accepted as thy slave’

(in Chatterji III: 96 his translation as obligative)

2.2. Kurylowicz and the theory of the (cognitive) equivalence of perfect and future

This parallel evolution of past and future systems was also noticed by Kurylowicz for Romance languages, with totally different, but equally convincing morpho-syntactic evidence. The future in modern Romance languages is indeed a very particular innovation, involving the verb “have” constructed with the infinitive. The same auxiliary is used as in the perfect, simply the form is now fused in the future whereas the auxiliary is kept unbound in the perfect (cf. above). Here are the forms in French and Spanish, with similar formation in Italian:

1st person I will sing 2nd person You will sing 3rd person He will singFrench Je chanter.ai Tu chanter.as Il chanter.aSpanish Yo cantar.é Tu cantar.as Ello cantar.a

1SG sing.INF-1SG 2SG sing.INF-2SG 3SG sing.INF-3SGThe endings are either identical to (French) or derived from (Spanish) the present paradigm of “have” verb (j’ai, tu as, il a). French is the clearer since in the past too the verbal form is the same (j’ai chanté, tu as chanté, il a chanté), whereas it displays an initial aspirate in Spanish as a separate auxiliary (he cantado, has cantado, ha cantado).

14 Depending on the dialects: only first person have the –b- ending in Western dialects of Awadhi, first and second person in Eastern dialects, the h form (inherited sigmatic future s > s > h) prevailing at the third person, but in early Awadhi –b- was found in all persons, although not systematically (Saxena 264-6). 15 Same example translated in Saxena (1937: 260) by a future : bhasa-badha karabi mai ”I shall render it in popular language”.

10

Historically, the “have” periphrastic future acquired its temporal meaning from an original meaning of necessity in Vulgar Latin, with the infinitive shifting from passive to active.

(24) a. ego id cant.ari habeo > cantare habeo1SG thissing.INF.PASS have.1SG sing.INF.ACT-1SG‘I have this to be sung, to sing > I will sing’

The whole demonstration of Kurylowicz, who concludes to an essential analogy between perfect and future (in contrast with the present), relies on the origin of the habere periphrastic future, an initially modal construction which he traces back to the Latin verbal adjective of obligation (sometimes called gerund or gerundive in –nd-. This –nd- verbal adjective was used with an agent in the dative case and agreed in gender and number with the patient in the nominative, like the verbal –tavya sentence, but unlike it, the copula was present (agreeing with the nominative term). The well-known following examples were still in use as classical proverbs not long ago:

(24) a mihi/nobis virtus colenda est1SG.DAT/1PL.DAT virtue.F.SG cultivate.nd.F.SG be.PRS.3SG‘I/We should cultivate virtue’ (lit. virtue is to be cultivated to me/us)

(24) b. Carthago delenda estCarthago.F.SG destroy.nd.F.SG be.PRS.3SG‘Carthago should be destroyed/[We] must destroy Carthago’

From these examples, which can be superimposed to the Latin evolution of perfect in the following table, itself comparable to the IA data, Kurylowicz in a very illuminating paper on the evolution of grammatical categories (1965), developed the hypothesis of the deep similarity between these two tenses: perfect and future are both basically non active because they do not aim at depicting an action, but to represent viewpoints on action, viewpoints from the present (1965)

alignment perfect futureNominative: synthetic verb form, NOM subject, person agreement Latin

(ego) feci/cantavi (ego) faciam /cantabo

Non-nominative : participial verb, DAT agent Late Latin, no person agreement

mihi factum est mihi faciendum est

Nominative: V +have, NOM subject, Person agreement (Later Latin)

(ego) factum habeo (ego) fieri/cantari habeo

Nominative: V+¨have, NOM subject, person agreement (French)

j’ai fait/ chanté je fer.ai/chanter.ai

Although Benveniste also reached similar conclusions regarding the symmetry of past and future (both “axial” relatively to the present, both representing views on action rather than processes), he radically rejected Kurylowicz interpretation of the evolution from Latin, arguing that the Latin construction in habere never had a obligative meaning and was used as a “future of predestination” in Christian predicators, in the meaning “deemed to happen”. This argument has later on been proved wrong, as well as the accusation of mistaking a passive infinitive for an active infinitive, but his major reason for rejecting the symmetry of both evolution might have had something to do with his own thesis of the “possessive perfect”, since it seems more embarrassing to make the meaning of the future possessive (more details in Montaut 1997).

11

The ambiguity between the original meaning of obligation (or potential) and the new meaning of future started between the 2nd and 3rd centuries in grammarians like Tertullian and Pompeius, and the future meaning was later on found mainly in Christian writers and grammarians, with no particular connotations of predestination, almost only in learned texts, and mainly in Africa (Bourova 2005: 304), its earliest, still ambiguous, instance in a colloquial register coming from letters of Wâdi Fawâki (Adams 2011, 2013: 659).16 No study to my knowledge points to a direct transformation of the dative alignment with –nd- gerund into the nominative alignment with habere but the situation is the same for the equivalent transformation regarding the perfect. The striking event in the history of Romance languages is the total extinction of the Latin synthetic future and the subsequent formation of new futures, many of them out of an obligative periphrasis (habere): the grammaticalization of obligation into future is nowhere questioned (Adams 1991, 2011, Bourova & Tasmowski 2007). In this respect it parallels the Eastern IA data, which on their side display an observable continuous history from non-nominative alignment to nominative alignment.

2.3. Divergent evolutions in Western languagesHowever this evolution from modality to future is not pan-Indian, and this is a strong difference with the past in the general evolution of Indo-Aryan. Whereas the –ta form, being the normal expression of past, rapidly became the substitute for all the verbal forms of past , the –tavya form never became the normal expression of future because the old synthetic future was maintained in many regions and prevented the new periphrasis from extending to the field of future.

In Western languages the old verbal adjective of obligation was maintained as a potential or obligative finite form in few modern languages only, whereas it prevailed in all, up to the middle stage of NIA: The ergative-like pattern of Ashoka (early MIA) presented as the origin of the Eastern –b- future (18) has a Western equivalent since the same edict is found in Girnar in the West (Girnar, now Pakistan):

(25) hida na kimci jivam arābhitpā prajuhitavyam na casamājo kattavyohere no some living kill sacrifice. no and assembly doNOM.N.SG CV OVA.NOM.N.SG NOM.M.SG OVA.NOM.M.SG

‘One should not sacrifice by killing a living creature nor hold a meeting’(it should not be sacrificed by killing a living being nor a meeting should be held)

The same pattern is preserved till late MIA (Apabhramsha stage, turn of the millennium) in the well known Western Jain text of Paumacariu, without or with oblique agent, clearly patterning like the past sentences (26c):

(26) a. annu na nam.ev.auother.M.SG NEG respect.v.M.SG

‘No other is to be respected’ (Paumacariu 26.3.2)(26) b. navara ekku vau mai pālevau

only one vow.M.SG 1SG.INS keep.ev.M.SG‘I shall/should observe only one vow’ (from Bunenik 1998: 194)

(26) c. tā keumaie hau gharaho nīya

16 Ambiguous (future/alethic) example from Tertulian quoted in Adams (1991: 148): si enim sustuleris istam tertiam, remanere habent duae “for if you take away the third [syllable) two will (have to) remain” (GL 129.6). The letter of Wâdi Fawâki on Ostraka to Rustius Barbarus is also ambiguous between future and deontic modality: adferre habes “you have to bring” or “you will bring” (Adams 2013: 659).

12

‘then Ketumati.INS/OBL 1SG home.LOC conduct.PPP.M.SG‘Then I was taken home by Ketumati’/ ‘Then K took me home’ (Bubenik 1986: 148)

The NIA languages inheriting this state of affairs, Gujarati, Marathi, Rajasthani, which for most of them also retained the sigmatic future, maintain the –tavya pattern with an obligative meaning, contrary to the Eastern languages which started shifting to a future meaning during their early NIA stage. Gujarati for instance shows agreement with patient and instrumental agent (-im), and the obligation meaning, in the 16th c. text of Upadeśamālā:

(27) a. isī upamā jānivi OLD GUJARATIsuch.F.SG comparison.F.SG know.v.F.SG‘Such a comparison should be known’ (Dave 64-5, )

(27) b. sisyii te kārya tatkāla ācarivaupupil.INS DEM.M.SG work.M.SG immediately do-v-M.SG‘The pupil should immediately do that work’

(27) c. te pāpiu jāni.v.au3M.SG.INS know.v.M.SG ‘He should be considered as a sinner’ (Dave: 54)

Already in the 14th c. language of the sadāvaśyaka, oldest written testimony of old Gujarati, the “gerund” used as a predicate had according to Pandit (1976: 23) an “imperative [= obligative] sense”: rakhivau in the masculine “is to be saved”, karivau “is to be done”, vyavasthāpivi in the feminine “is to be founded”).

In Old Rajasthani similar obligative constructions are still found in the 16th century, the time when the language is supposed to have diverged from Old Gujarati, both with intransitive (28c) and intransitive predicates, without (28a) and with agent (18b):

(28) a.hisā na kar.av.īviolence.F.SG NEG do.AV.F.SG‘Injury is not to be done’ (Tessitori 1915: 120)

(28) b. anere vidya lete vinay karivum other.PL.INS knowledge taking humility.M.SG do.iv.M.SG‘Humility should be observed by others acquiring knowledge’ (RG 16:)

(28) c. tai na jāi.vu2.INS NEG go.v.M.SG‘It should not be gone by you’ (RG 37 in Khokhlova 2013: 101)

Such constructions came to be replaced by the obligative infinitive in –an/-ana plus copula and dative subjects with the nai postposition, when the so-called gerund in –v- was replaced by the infinitive, on the now most widespread pattern for obligation (see infra).

Marathi is the only language which retained up to the modern stage the morpho-syntax of the –tavya predication, initially in the meaning of obligation, further opening to potential: Old Marathi has for obligation, according to Bloch (1970: 264), a “syntax, with the logical subject in the instrumental, (…) very similar to that of the form for past” which continued up to Modern Marathi (29a), with an optional alternation with nominative alignment and agreement with subject 29(b):

(29) a. tyāne ghari yāve (MARATHI)3M.SG.INS/ERG home.LOC come.SBJV.N.SG‘He should come home’

13

(29) b. to ghari yāvā3M.SG.NOM home.LOC come.SBJV.3M.SG‘He should come home’

This “active transformation” is however considered as associated to the optative meaning (“he may”) whereas the ergative marking of the agent and agreement with patient are associated to obligation in Wali (2004: 228)17. A meaning questioned by Pandharipande (1997: 290; 2003: 711) who associates the ergative pattern (glossed Agent) with potential meaning.Significantly an older stage of the language displayed a dative/ergative case alternation as shown in examples (30) from Joshi (1900: 468):

(30) a. majhyāne /malā câlavle (OLD MARATHI)1SG.INS / 1SG.DAT go.POT.PST.NS (Bloch 1909: 265)“I could/was able to go”

(30) b. majhyāne / malā dhadā sikhavlā1SG.INS / 1SG.DAT lesson.M.SG learn.POT.PST.M.SG “I was able to learn the lesson”

What is clear from these various evolutions is that wherever a different form for future was available (the inherited sigmatic future or a new form (-l-) as in Marathi, the –tavya verbal adjective retained its modal value (potential/obligation in Marathi) or disappeared from the TAM paradigm, including obligation. Whereas the –tavya > av / ab form with instrumental agent was up to the 14th century preserved as the standard expression of deontic modality throughout Indo-Aryan, it gave way to periphrastic expressions build with infinitive + copula (or other auxiliaries: cāhiye “must”, from a verbal root “look at”, par “fall”> “incumb”), and a dative agent, the predicate (verb + infinitive) agreeing with the patient if the verb is transitive, a quasi-ergative pattern:

(31) mujhe/ āp- ko citthī bhej-n-ī hai (modern HINDI)1SG;DAT 2PL DAT letter.F.SG send-INF-F.SG be.3SG‘I have to send the letter’

Panjabi Delhi is known to present an alternation dative/ergative for agents in certain obligation constructions (Khokhlova 2013), which evokes the Old Marathi alternation for potential. This suggests that the general shift for dative in obligation sentences might be related both to the old non-nominative pattern with predicative participle which prevailed till mid NIA and to the new dative pattern on the rise after the 15th century for predicates involving an experiencer and transient states in what can be considered semantic alignments (Montaut 2013). This type of alignment is now the preferred argument structure of a considerable part of the predicative lexicon of NIA and has probably attracted to the dative pattern the obligative sentences with infinitive, whereas obligative sentence retained an optional ergative marking in the one language where there is a finite predicate inherited from the old –tavya form (Marathi).

17 Even unmarked pronouns (1st and 2nd person never take the ergative marker in Modern Marathi) require the neutral agreement of the verb in the obligation meaning as they do in the ergative pattern with neuter object, whereas they control person agreement in the potential meaning : tyāne /tu hansāve [3m.sg.erg /2 nom laugh.n.sg] “he/you should laugh”, vs tū hansāvā [2.nom laugh.2.m.sg] “you may laugh” (Wali 2004: 238).

14

3. The grammaticalization of the verbal adjective of obligation into infinitive and verbal noun

The shift towards tense or various modal meanings along with the renewal of alignments in the form of pre-ergative, ergative or back to nominative alignments is not the only outcome of the –tavya grammaticalisation. It also lead to the renewal of the infinitive, a category almost disappeared in its original form between Late Sanskrit and NIA.

A fair number of IA languages display an infinitive form derived from the Sk verbal adjective in –tavya. Gujarati is the best known example: let us first look at it as an exemplary illustration since the language is also a major language historically well documented and profusely described.

3.1. Gujarati infinitive: verbal noun, adjective

The peculiarity of Gujarati is that this –v- form does not alternate with any other form for infinitive and verbal noun such as the widespread –an/-ana. This –ana form, itself derived with a suffix used for nouns of action, is indeed credited to have renewed the category infinitive in new Indo-Aryan (Bloch 1965: 280), after the loss of the inherited Sk infinitive in –tum (maintained only in Marathi). The single form of infinitive in Gujarati is -vu, which is uncontroversially related to the old verbal adjective in –tavya (Dave 1935: 64; Chatterji 1926: 966, who too derives the Gujarati verbal noun karvū from the verbal adjective or passive obligative participle kartavyam and its enlarged form *kartavyakam). The verbal noun agrees in case like the –an-ana form in all IA languages: jovu “to see”, jovā lāyak “worth seeing”.

(32) a khāvā lāyak vastu cheDEM eat.v.OBL worth thing be.3SG

‘This is a thing worth eating’

In periphrastic permissive constructions, where it completes the verb de “give” and inceptive constructions, where it completes the verb lag “touch/start”, it similarly displays the oblique form –ā: karvā de “permit to do”, khāvā lag “start to eat”18. And it agrees in gender and number like the –an infinitive forms of other NIA languages in obligational constructions with a transitive infinitive. But with the Gujarati –v- infinitival form, used with auxiliary “be” or “need”, not only obligation, wish too can be conveyed. The meaning of obligation is conveyed by an enlarged form of the infinitive (-v-an-), with the relational suffix –an (Cardona & Suthar 2003: 677), whereas the desiderative meaning requires the short form (-v-), both constructed with the copula:

(33) a. mar.e caupdī vanc.van.ī che / lekh lakhvan.o che1S.FG.AG book.F.SG read.van.F.SG be.PRS.3 article.M.SG write.van.M.SG be PRS.3S‘I have to read a book / to write an article’

(33) b. tam.ne kyā javanu che2.DAT where go.INF.N.SG PRS.3SG‘Where do you have to go?’ (Cardona & Suthar 2003)

(34) a. mar.e caupdī vanc.v.i che lekh lakh.vo che

18 Cardona & Suthar 688; Dave 1935: 52sq). In these two constructions, languages with –an/na infinitives also display the oblique form of the verbal noun (Hindi karne do [do.inf.obl give.imper] ‘allow [x] to do/let X do’, khāne lagā [eat.inf.obl start.m.sg] ‘(he) started to eat’.

15

1SG.AG book.F.SG read.v.F.SG be.PRS.3 article.M.SG write.v.M.SG be.PRS.3SG‘I want to read a book, to write an article’

(34) b tam ne kyā ja.v.u che2 DAT where go.INF.N be.PRS.3‘Where do you want to go?’ (Cardona & Suthar 2003)

(34) c. mār.e gujrātī bhāsā bol.v.ī che1SG.ERG Gujarati language.F.SG speak.v.F.SG be.PRS.3SG‘I want to speak Gujarati’

It should be noted that, although the meanings (in obligative constructions) are very close to the Marathi obligative or potential sentences with finite forms in –v-, the morphology here is that of a verbal noun exactly comparable to the –an formations of Hindi such as example (31). Interestingly, case marking is not limited to the dative as in all other Western Indo-Aryan languages but presents an alternation dative/ergative (glossed agent in Cardona & Suthar). Case alternation here seems to be associated to the personal pronoun, since the second person is more often in the dative than the first one (agent=erg), yet, there is in other examples no clear explanation of distribution (personal communication Cardona).

In Old Gujarati, obligation was expressed by the predicative use of the verbal in –tavya (cf. supra), but the use of a copula + infinitive already occurs (a) and verbal noun also derived from this –v- origin, is commonly used with the meaning “worth of V”, and in nominal use (āsana nau le.v.au “the taking of the seat”, vinasi.v.ā nai-kāji “in-order to be destroyed” Dave 1935: 54).

(35) a. jīnaī jīvii jīhā jāi.v.au chaiREL.INS person.INS there go.v.M.SG be.3M.SG

‘The person who is designated to go there (in a particular place)’ (Dave: v. 262)(35) b. pāsachā siu anamila.v.au bhalau

depraved.OBL with not.mix.v.M.SG good.M.SG ‘It is good (the good is) not to mix with the depraved’ (Dave 64, v. 223)

(35) c . pāsachā siu boli.v.u, ekai upāśrayi rahi.v.u (v. 224)depraved.OBL with speak.v.N.SG one.LOC hostel.LOC stay.v.N.SG

‘To speak with the depraved, to stay at the same hostel’

(35a) may be the link between the obligative predicative construction as (27) and the new infinitive construction with copula (33). But the infinitive is present right from the end of MIA (Pischel 1900: 388) in 11th century Jain Digambara texts (Tagare 1948: 322), parallel to the obligative predication with a –v- predicate.

Similar verbal nouns are also found in other Western languages such as Braj (māribau, māribau “to strike”), Rajasthani (mārabo), and Kanauji further North (māribo), a long tradition documented for the older stages of Rajasthani Gujarati in Tessitori (1915: 121):

(36) jīp.ava vamchaiwin.ava want.PST.3SG‘He wishes to conquer’ (Tessitori 1915: 121, Chaya to Yogasāstra III-134)

Modern Braj has two forms for infinitive, the –n- form and the the –b-/-v- one, the –n- form particularly frequent in goal functions and complement of verbs according to examples given by Kellogg (1875: 289 § 480):

16

(37) a. sadā kahu sau rahivau nāhi BRAJalways somebody with stay.iv.M.SG NEG‘There is no always remaining with anybody’

(37) b. mere putrani kau pandit karive jog haimy son.PL ACC pandit do.iv.OBL worth be.SG‘You are competent for making my sons wise men’

(37) c. tum sau kahan kau āyau hau 2 SOC say.an DAT come PFT.1SG‚’I have come to tell you’

(37) d. rājā kahani lāgyauking say.an start.M.SG‘The king began to say’

Snell (1991::) also suggests the the -na- form occurs with semi auxiliarized verbs such as de- (permissive), pā (ability) and lag (inception) and alternates with the –b- form in goal functions (19).Bundeli, a South-western language sometimes considered as a Hindi dialect, also displays two types of infinitives, the –na form used in permissive and inceptive constructions (morā kho khānā do “let the boy eat” (Jaiswal 1962: 132) whereas the –b- form is preferred in typically nominal uses, including with the goal DAT/ACC postposition kho:20

(38) a. daur.b.o ūke cala.b.e barābar hai MOD BUNDELIrun.b.M.SG 3.SG.GEN walk.b.OBL same be.3.SG

‘Your running is equal to his walking’ (Jaiswal 1962: 133)(38) b. tumāe karabe kho ite bohota hai

2.OBL do.b.OBL DAT here much be.3SG‘There is much for you to do here’ (Jaiswal 1962: 133)

Only Marathi has not, and never had, a –v- infinitive. Is it because it soon developed an obligative/potential predicative constructions with the old –tavya verbal adjective, which was maintained till now (29), or rather because of the (unique) surviving of the –tu infinitive? Or both? One could think that languages which lost the obligative predicate in –av- during the early second half of the millennium, may-be because a further development into future was blocked by the preservation of the old synthetic future, made it into a verbal noun, when no other form was concurrent as in Marathi, a western feature then21. But such is not the case.

3.2 Infinitive in Eastern languages

But the –b-/-v- infinitives are not a specific feature of Western languages (usually with either inherited sigmatic or a periphrastic -gā future) and we also find them in Eastern languages, which retained the –b- predicative form in a future meaning.

This is the case in modern Maithili, which has three forms of infinitive (-ana/āna, -al/ala, -aba/ab), the two latter forms alternately used as complement: dekh.lā me sunnar or dekh.bā me sunnar “beautiful to look at” (Jha 1958: 519). Similar examples were already attested in

19 As complement of movement verbs for instance : moko pakaran ko āyo “he came to touch me”, moko spars karive kau doryau “he ran to touch me” (VV 227 in Snell).20 Alternation with agent suffix -baro (karna baro « doer », khabe baro « eater ») whereas the -b- form only is allowed with the –aiya suffix (dekhabaiya « seer »).21 This however does not explain the absence of both in MSH/U and Panjabi, the ‘central’ languages.

17

Middle Maithili with the same –ba/-va form, such as kopahū kara.vā joga “fit for showing anger” (Vidyapati 50, in Jha 1958: 519), as well as verbal noun in inceptive and volitive constructions:

(39) a.kamal.āsana kichu kahavā lāgu MIDDLE MAITHILIlotus.seat something say.v.INF;OBL start.M.SG‘The lotus-seated god began to say something’ (KJ1 in Jha: 611)

(39) b. nāgaripana kichu kahabā cāhocitizenship some say.b.INF.OBL want.PRS.1SG‘I wish to describe (his) citizenship’ (R 62, in Jha: 611)

Assamese (North Eastern state) also has an infinitival form in –iba, which is included into the “worth to” suffix (i)balogya (sa.balogyia “worth to see”), and is required as a nominal formative and in concatenation with the ‘be able’ modal (Goswami & Tamuli 2003: 425)

(40) a.za.b.âr xâm.ât ASSAMESEleave.b.GEN time.LOC‘Time of leaving’22

(40) b. xi saikel sâla.bo par.e3M.SG bicycle ride.bo be.able.3M.SG‘He can ride a bicycle’

Similarly the standard Oriya infinitive ends in –ibā and is used as a noun before postpositions (ās.ibā ku “(in order) to come”). Although modern standard Bengali does not display –b-/-v- infinitives, the variety spoken in Assam (Tunga 1995) has similar forms used for completing modal verbs (karibar paro “I am able to do”), and in the early 20 th century Grierson mentions Bengali constructions such as ja.b.ār somoy.i [go.b.GEN time.LOC] “at the time of leaving”, korbā lāgil “he began to do”, continuing Middle Bengali lage balibar “he began to say” (Chatterji 1926: 1008). Besides, standard Bengali still construes the dative verbal complements with the –b.GEN formation (jabar somoi “at the time of coming”; as.(i)b.ar janye “for coming”). As for mid-Eastern languages such as Bhojpuri, and Awadhi, both in their modern and ancient stages, has –be/-bu verbal nouns alternating with the –ana forms (Saxena 282-5). Useless to multiply examples: it seems that, apart from Marathi which maintained the old Sanskrit infinitive, only standard Hindi/Urdu, and Panjabi, the so-called central IA languages, do not have it, but it should be remembered that the three of them are recent under such names, sharing as common ancestors Braj, Avadhi, and, mostly, the popular literary koine of the mystic predicators, the Sant Bhasha23. They retained only the –an/-ana infinitive, which is used in obligative constructions as well as a verbal noun, and developed a new periphrastic –gā future, so that the old verbal adjective has left no trace in the standard modern stage of these languages. But on the whole, more languages still display infinitive than any other tense or modal form derived from the old –tvaya verbal adjective: could not it be that the basic meaning of the form was rather nominal than verbal (modal)?

3.3. Original meaning of the so-called obligative verbal adjective 22 Exactly on the Sanskrit model na ayam vaktavya.sya kālah “it is not the time of speaking” (from Panchatantra, in Bloch 1965 : 278f).23 Panjabi’s founding texts (Guru Granth Sahib) displays a mixture of Sant Bhasha texts, many from Kabir and early Panjabi still not much diffferienciated from the other dialects of the will-be Hindi language. Although attributed to Fariduddin Ganjshakar (13th c.), the emergence of Panjabi as distinct from neighbouring various ‘old Hindi’ dialects is later. As for the history of Hindi/Urdu, see…

18

3.3.1. The Indo-Aryan dataA last evidence for the remarkable persistence of the nominal meaning attached to the –tavya form comes from Romani, a language separated from West-central IA in the beginning of the 2nd millennium. In all Romani dialects, infinitive is notoriously absent (Boretzky 1996). Whereas an infinitive form has been created out of contact with the local languages, the old –tavya form was however maintained as a noun of action in –iben, which is the only suffix for deverbal and de-adjectival abstract nouns in the South Balkanic dialects. For instance the noun referring to a beverage, that is, something worth drinking, is zspiben, from the verbal IA root PI “drink” (Beniśek 2010). Similarly, in most Rom dialects, deverbal abstracts end in -iben,24 a formation parallel to the other gerundival suffix -nya “able to”, “worthy of” which also produces deverbal abstracts such as pāj (common Romani).

So diverse nominal evolutions, from infinitive to deverbal abstract nouns, raise the question of the original meaning of the –tavya form, the most clear evidence being the Romani case, where the only surviving form is used as a nominalizer (for deverbal abstracts), without any trace of the potential or obligative meaning, not to mention a development into future. More precisely, did the nominal and future meanings evolved “side by side” from an originally obligative form as suggested by Chatterjee (1926: 966): “The simple future notion evolved gradually; side by side with it, the old notion of an action to be done continued, and was modified into simply the notion of an act”? Or was not it reversely, that both obligative (the “notion of an action to be done”), then potential or future, meanings evolved from the “simple notion of an act”, which side by side with modal and temporal evolutions, itself evolved into verbal nouns and infinitives?

Going back to the origin of the tavya form in Old Indo-Aryan, it is itself derived from the verbal noun in –tu, the same which in the accusative was used throughout classical Sanskrit as an infinitive (-tum) and maintained in Marathi with the -un infinitive ending (-idum > Maharashtri -ium > Mod. Marathi -un). In the dative, -tave, the form was also used as infinitive (> MIA –tae, > Ardhamagadhi –ttae). To this –tave verbal noun could also be added the gerundival suffix –ya (Debrunner 1954: 612-15), which only later specialized in necessity constructions while the other gerundive on the –ana nouns of action (-anlya) tended to be lexicalized according to Bubenik (1998:190). Also worth mentioning, the nominalising (deadjectival) suffix –pan, still used in NIAs, is also supposed to derive from a gerundive ending –tva (Khatre 1966: 153, Pischel), suffixed with –an (-tvana > -pan), and is very productive (either directly suffixed to the base, such as in bac.pan “childhood” < baccā “child” or to the oblique or direct form of the adjective such as in akele.pan or akelā.pan “solitude” < akelā “alone, lonely”).

3.3.2 Latin correspondencesThe Latin data leads to similar questions regarding the original meaning of the verbal adjective or gerund of necessity in –nd-. Whereas the well-known construction with copula and possible agent in the dative as mentioned in 2.x has always a necessity reading (Carthago must be destroyed), conveying both passive voice and obligation, the form itself seems to be more generally related to the mere verbal idea. This conclusion was first formulated by Ernout & Thomas (1953: 285): “il exprime simplement l’idée verbale”. It was then developed in the more recent Latin syntax by Touratier (1994: 164 sq) who insists that the construction, not the form, conveys the meaning of necessity, in predications with the verb “be” , with an optional dative agent (41a). Used as an attribute the –nd- verbal adjective gets a vaguely abilitative meaning like the –able suffix (orator legendus “an orator worth reading” “a readable orator”),

24 Deadjectival abstracts usually end in –ipen, a suffix inherited from the IA form –pan, and many dialects present a contamination of both forms.

19

but it is mostly used since Tite Live as a mere verbal noun and behaves “just as an infinitive”, devoid of any necessity meaning (41b):

(41) a. consola-nd-us hic mih-ist (= mihi ist) (Plaute) LATINconsole-nd-NOM.M.SG this.NOM.M.SG 1SG.DAT-be.3M.SG

‘I have to console him’ (= he is to be consoled to me)(41) b. de consilio relique-nd-i Italiam

about decision.ABL leave-nd-GEN Italy.ACC‘Regarding my project of leaving Italy’

In non-predicative constructions, the -nd- gerund is then devoid of the meaning obligation, which grammaticized only in predicative constructions with the verb “be”. Countless formations attest of this nominal behaviour, all in complement constructions, from the well-known Ars Amandi (“art of love”) to the three libidines (desires) stigmatized by Saint Augustinus (libido dominandi, sciendi, fruendi “desire of domination, knowledge, enjoyment). The difference with the Indo-Aryan evolutions is that a further development into standard infinitive never took place, because of the preservation throughout classical and vulgar Latin of the inherited infinitive in –re, still present in modern Romance languages, so that the form disappeared – if not the structure, which can be traced at the origin (?) of the “have” reconstruction. Yet the initial meaning of expressing a mere verbal notion, common to both OIA and Latin accounts for the various grammaticalization paths of this “gerund”, richly illustrated in NIA and aborted in Romance modern languages, but present in Latin.

As for the hypothesis of Southworth (2005), according to whom languages exhibiting this type of verbal noun and futures, along with –l- past forms, would correspond to the “outer circle” of the Indo-Aryan immigration, whereas languages devoid of them would correspond to the kernel of IA (initial immigration flow), it is not consistent with the historical data since all of them, at one stage, displayed similar verbal nouns.25

Conclusions

The evolutions above analysed are not symmetric. Let us sum up the main asymmetries, regarding past participle in IA and Romance languages, future in IA (and romance languages).

The past participle was reanalized everywhere as a past tense form (accomplished aspect). But whenever a (adjectival) suffix happened to be amalgamated to the participial, then finite tense form (Bengali and Eastern IA, Marathi), it is this suffix which was re-analyzed as past, a clear example of grammaticalization of a category (derivational affix) into a different category (tense). All over South Asia, whether or not re-aligned into nominative patterns, the participial form acquired a predicative function with an initially open meaning which came to be restricted to the meaning of definite past, whatever the label, anterior, preterit, aorist, yet certainly distinct from perfective (for arguments against this current terminology, see Montaut 2015, Caudal, in press). This restriction is at striking discrepancy with the Romance data, where it grammaticalized into a perfect, later on re-aligned with “have” restructuration. In Latin the copula was always present, whereas it was optional (for the sake of avoiding ambiguities regarding person in IA), and more important, the old aorist (or preterit) never disappeared and is still more (Spanish, Italian) or less (French: written only) maintained, 25 This renews an old hypothesis by Grierson viewing the arrival of Indo-Aryan speaking tribes in two steps, the first flow (more conservative since reflecting more purely Indo-Aryan speech) occupied a central space in the Gangetic plain, and the second flow settled around the already established population (hence the « theory of circles »). Arguments include of course more features, including phonological ones, which have been widely discussed in the early 20th century and consequently the hypothesis abandoned.

20

whereas in IA it was lost quite soon everywhere, and the open meaning was first restricted when a copula started being used for present perfect, then in the past, for pluperfect. A given form, obviously, occupies all the empty space left by the absence of a competing form.

The same holds true for the development of future out of obligative periphrastic constructions. Here the striking discrepancy is the radical loss of future tense in late Latin (hence the dominance of “have” futures in Romance languages (or in some languages other periphrastic constructions), whereas in Indo-Aryan the old sigmatic future was maintained in the Western and Central regions, hence the dominance of ex obligative constructions in the East.

The development of the so-called obligative gerund or verbal adjective into infinitive raises also another row of problems: before stating that infinitives in IA were largely recreated as a category out of an obligative verbal adjective, and propagating as a plausible grammaticalization path the path necessity > verbal noun (??), one should further inquire into the contextual meaning of the given form in its own history. For the matter the old gerund, only obligative in specific constructions, and initially referring to the verbal notion, logically evolved into a verbal noun, an evolution which was blocked in Romance languages by the persistence of the inherited infinitive, disappeared almost everywhere, whereas it proliferated (except in Marathi) in Indo-Aryan.

Empty space on the semantic map: a condition for grammmaticalization of a non tense form into tense? (although not a prerequisite condition in other languages (Harris & Campbell 1995, Heine & Kuteva 2005).

As for the syntactic constructions involved, which are highly relevant here (see section 2 and 3 for future and infinitive, from the same original form), they have often been interpreted in terms of alignment. The modern writing of the history of the old (past passive) participle in –(i)ta has focused on alignment since the history of ergative construction in Indo-Aryan has become a typological topic of interest. If a similar interest was not raised by the parallel history of the other passive participle in –tavya, the reason is most probably that its present forms in IA are not associated with an ergative alignment. Both evolutions were however strikingly parallel, including the renewed case markers in those languages which maintained obligative/potential predicates directly inherited from verbal adjective (or gerundive). The unprecedent interest raised by the present ergative alignment in Western IA – itself a cover tag for extremely diverse and complex alignments cf. section 1.3 – has lead to consider the MIA and early NIA patterns as pre-ergative, as well as to re-interpret the Iranian data as ergative or pre-ergative depending on the languages considered (Haig 2008; Haig, in press, with critical analysis), whereas the same concept never extended to the same patterns in the same languages during the same stages of evolution for future and obligation/potential. This is more than a question of terminology and gloss, although glosses condition the reading of such accounts in broader typology or general linguistics – why –ne glossed sometimes agent, sometimes ergative and sometimes instrument in the same grammar? do we gloss the function or the form, don’t we superimpose modern functions, or interpretations of them, when glossing an old oblique syncretic case by ergative?

This raises a number of questions, among which: what is an ergative alignment? If the case marker is the most relevant feature, then not Marwari (Magier 1983) which is considered ergative, but Assamese (which is not), if patient agreement is the most relevant, then not Nepali, but all the experiential constructions in all NIA languages (and South Asian non ergative languages), etc. If controlled properties are relevant then no IA language, which have all been attaching subject properties to the agent throughout the “pre-ergative” stages from classical Sanskrit to the modern full fledged ergative sentences in modern Western languages. The fact that all, now ergative and now non ergative languages, where all at a time displaying pre-ergative alignments for obligation and past, and are now displaying syntactically similar

21

alignments for a particular semantic class of predicates with a dative subject, which we may consider semantic alignments, only sheds light, not on how a given language is “aligned” but on how relative, and historically conditioned, is the way we consider alignment.

Equally questionable is the exclusive focusing on only one construction at the origin of a grammaticalization path. Since the construction is an intrinsic part of the grammaticalization process of a form (DeLancey 2011), whether a grammatical word or an affix (-l- past), other constructions of the given form too should be considered in order to check the specificity of each outcome. Finally, in the particular case of Indo-Aryan, the proliferation of divergent paths and outcomes should not be under-estimated under the pressure of better-know and “more major” languages such as Hindi/Urdu. The total lack of any posterity for the –tavya form and constructions, in contrast with the most straightforward preservation and re-inforcing of the –ita form and construction, is still difficult to account for, even if one can invoke the influence of Persian (itself devoid of such posterity) on the cultivated registers during the 17-19th centuries.

Adams, James Noel. 2013. Social Variations and the Latin Language. Cambridge: Cambridge University Press.Adams, James Noel. 2011. Late Latin (Chapter 16). in Clakson, James. A Companion to the Latin Language. London: Wiley-Blackwell. Beames, John. 1970 [1871]. A Comparative Grammar of the Modern Aryan Languages of India. Delhi: Munshiram Manoharlal.Benveniste, Emile. 1966. Problèmes de linguistique générale 2. Paris : Gallimard  ([1952]. La construction passive du parfait transitif. Vol.1 :176-86, [1960]. Etre et avoir dans leurs fonctions linguistiques. Vol.1 : 187-207, [1965]. Les transformations des catégories grammaticales. Vol. 2 :127-136).Bloch, Jules. 1906. La Phrase nominale en sanscrit. Paris: Champion.Bloch, Jules. 1970 [1920]. The Formation of the Marathi Language. Delhi: Motilal Banarsidass.Boretzky, Norbert. 1996. The 'New' Infinitive in Romani. Journal of the Gypsy Lore Society. Fifth Series. 6(1): 1-51.Bourova, Viara. 2005. A la recherché du conditionnel latin: les constructions Infinitif + forme de habere examinées à partir d’un corpus électronique. In Kabatek, Johannes, Pusch, Claus & Raible, Wolfgang (eds.). Romanistische Korpuslinguistik II, 303-16. Tübingen : Narr.Bourova, Viara and Tasmowski, Liliane. 2007. La préhistoire des futurs romans — ordre des constituants et sémantique, Cahiers Chronos 19: 25–41. Breunis, André. 1990. The Nominal Sentence in Sanskrit and Middle Indo-Aryan. Leiden: Brill.Bubenik, Vit. & Ch. Paranjape. 1996. Development of Pronominal Systems from Apabhramsha to New Indo-Aryan. Indo-Iranian Journal 39 (11-32).Bubenik, Viit; 1998. A Historical Syntax of Late Indo-Aryan (Apabhramśa). Benjamins.Bybee, Joan., R.D. Perkins & W. Pagliuca (eds.). 1994. The Evolution of Grammar, Tense, Aspect and Modality in the Languages of the World. Chicago-London: Chicago Univ.Pr.Cardona, George. & B. Suthar. 2003. Gujarati. In The Indo-Aryan Languages. In Cardona, G. & D. Jain (eds.). London: Routledge (658-97).Cardona, G.. 1970. The Indo-Iranian Construction mana (mama) krtam. Language 46 (1-12).Chatak, Govind. 1966. Madhyapahari ka bhashashastrîa adhyayan. Delhi: Radhakrishna PrChatterji, Sunita Kumar. 1986 [1926]. The Evolution of Bengali Language. Delhi: Rupa..Caudal, Patrick. In press.

22

Dave, Trimbaklal, N. 1935. A Study of the Gujarāti language in the 16th century (v.s.) : with special reference to the Bālāvabodha to Upadeśmālā (Suri). London: Royal Asiatic Society.Davison, Alice. 2002. Agreement features and projections of Tense and Aspect, in R.Singh (ed.) Yearbook of South Asian Languages and Linguistics. Delhi: Sage (27-57).Debrunner, Albert. 1954. Altindische Grammatik, II 2. Göttingen: Vandenhoek und Ruprecht.DeLancey, Scott. 2011. Grammaticalization and syntax : A Functional view. In Heine & Narogg, 365-77.Desai, Mohan. 1926. Jain Gurjar Kavyo. Bombay: Jain Svetambar Conference Office.Harris, Alice & Campbell, Lyle. 1995. Historical Syntax in Cross-linguistic Perspective. CUPHeine, Bernd & Kuteva, Tania. 2005. Language Contact and Grammatical Change. CUPGoswami & Tamuli. 2003. Assamese. In Cardona & Jain (eds.): 411-43.Grierson, G.A. 1903-28. Linguistic Survey of India, Delhi (reprint): Motilal Banarsidass.Haig, Geoffrey. 2008. The emergence of ergativity in Iranian: extension or reanalysis? In: Karimi, Simin, Don Stilo & Vida Samiian (eds.). Aspects of Iranian linguistics. Newcastle: Cambridge Scholars Publishing, 113-127. Haig, Geoffrey. In press. Deconstructing Iranian ergativity. In: Coon, Jessica, Lisa Travis & Diane Massam (eds.). The Oxford handbook of ergativity. Oxford: Oxford University Press.Hock, Hans Heinrich. 1992. Studies in Sanskrit Syntax. Delhi: Motilal Banarsidass.Jaiswal, Mahesh Prasad. 1962. A Linguistic Study of Bundeli. Leiden : Brill. EFEOB GRAM BUND 2Jha, Subhadra.1985 [1958]. The Formation of the Maithili Language. Delhi: Munshiram Manoharlal.Juyal, Govind. 1976. Madhya Pahari Bhasha (Garhvali Kumaoni) ka anushilan aur uska hindi se sambandh. Lucknow: Navyug Granthagar.Kachru, Yamuna. 1987. Ergativity, Subjecthood and Topicality. Lingua 71 (223-38).Kellogg, Rev. 1972 [1875]. A Grammar of the Hindi Language. Delhi: Oriental Book Reprints.Kokhlova, Ludmila. 2013. Obligational Constructions in New Indo-Aryan Languages of Western India. Lingua Posnianiensis LV-2: 91-107.Kuryłowicz, Jerzy. 1960. Esquisses Linguistiques, Kraków: Polska Akademia Nauk ([1953]. Aspect et Temps dans l’histoire du persan: 109-118, [1931]. Les Temps composés du roman: 104-108).Khatre, DM. 1966. The Formation of Konkani. Pune: Deccan College Publ.Kuryłowicz, Jerzy. 1965. The Evolution of Grammatical Categories. Diogène 51 (51-71).Mahajan, Anup. 1997. Universal Grammar and the Typology of Ergative Languages. In Studies on Universal Grammar and Typological Variation. Alexiadou A. & T. A. Hall (eds). Amsterdam: Benjamins (35–57).Masica, Colin. 1990. The Indo-Aryan Languages. Cambridge: Cambridge University Press.Montaut, Annie. 2004. Oblique main arguments in Hindi as localizing predications. In Non nominative Subjects (eds. Bhaskararao & Subbarao). Amsterdam: Benjamins (33-56).Montaut, A. 2007. The evolution of the tense-aspect system in Hindi/Urdu: the status of the ergative alignment. Proceedings of the LFG06 Conference (M. Butt and T. Holloway King (Eds.). CSLI online Publications.Montaut, Annie. 1996. L’évolution des systèmes perfectal et futur en indo-aryen occidental et oriental. Journal Asiatique 284.2 : 325-360.Montaut, Annie. 1997. Benveniste et Kurylowicz: deux méthodes, deux trouvailles sur le système aspecto-temporel. Linx N° spécial 9 (Emile Benveniste vingt ans après. C. Normand ed.): 337-53. (available on line: http://linx.revues.org/1080)

23

Montaut, Annie. 2013. The rise of Non-canonical Subjects and semantic alignments in Hindi. in Serzant, Ilja & Kulikov, Leonid. The diachronic typology of Non-Canonical Subjects, 92-117. Amsterdam: Benjamins.Montaut, Annie. 2015. The Verbal form V-â in Hindi/Urdu: An aorist with “aoristic meanings”. In Guentcéva (ed.). Aspectuality and Temporality. Empiric and Theoretical Issues. Amsterdam: Benjamins.Montaut, Annie. In press. Why ergative case marker, etc.Nespital, Helmut. 1986. Zum Verhältnis von Genus Verbi, Nominativ- und Ergativ- Konstruktionen im Hindoarischen. Müinchener Studien zur Spachwissenschaft 47 (127-58).Pandit, Prabodh Bechardas. 1976. A Study of the Gujarati language in the 14th century with special reference to a critical edition of Ṣaḍāvaśyakabālāvabodhavṛtti of Tarunaprabha. Bombay : Bharatiya Vidya Bhavan (EFEO ttp://www.sudoc.fr/107923815)Peterson, John. 1998. Grammatical relations in Pāli and the emergence of ergativity in Indo-Aryan. München: Lincom Europa.Pirejko, L.A., 1979. On The Genesis of Ergative Construction in Indo-Iranian. In Plank, F. Ergativity; Towards a Theory of Grmmatical Relations. London: Academic Press (481-88)Renou, Louis. 1952. Grammaire de la langue védique, p. 309. Lyon : IAC.Saxena, Ram Baburam. 1937. Evolution of Awadhi. Delhi [1971 reprint]. Delhi: Motilal Banarsidass.Schwarzschild, L. A. 1955. Notes on the history of the infinitive in Middle Indo-Aryan. Indian Linguistics 16: 29-34Snell, Rupert. 1991. The Hindi Classical Tradition: A Braj Bhasha Reader. London: SOAS. Southworth, Franklin. 2005. Linguistic Archeology of South Asia [the Grierson hypothesis revisited]. London : Routledge.Strnad, J, 2012. A note on the Locative/Instrumental/Ergative in Old Hindi. Archiv Orientalni 30 (41-64).Stronksi, Kryzstof. 2010. Non-nominative Subjects in Rajasthani and Central Pahari. The Status of the Ergative and Obligative Constructions. Lingua Posnaniensis LII-1 (81-97).Stronski, Kryzstof. 2010. Variations of ergativity patterns in Indo-Aryan. Poznan Studies in Contemporary Linguistics 46(2):237-253.Stronski, Kryzstof. 2014. On the Syntax and Semantics of the past participle and gerundive in early New-Aryan. Evidence from Pahari. Folia Linguistica Historica 35: 275-305.Tagare, G.V. 1948. Historical Grammar of Apabhramsha. Poona: Deccan College.Tessitori Luigi. 1914-16. Notes on the Grammar of the Old Western Rajasthani, with Special Reference to Apabhramsha and to Gujarati and Marwari. Indian Antiquary 42-44. Tessitori, Luigi. 1914. On the Origin of the Perfect Participles in 1 in the Neo-Indian Vernaculars. Indian Antiquary: 225-36.Tiwari, Udayan N. 1961. Hindi Bhasha ka udgam aur uska vikas. Prayag: Bharati Bhandar.Tiwari, Udayan N. 1966. The Development of Bhojpuri. Calcutta: The Asiatic Society vol. X.Touratier, Christian. 1994. Syntaxe latine. Louvain: Peeters.Trask, R. L. 1979. On the Origins of Ergativity. In Plank, F. (385-404).Trumpp, E. 1872. Grammar of the Sindhi Language. London-Leipzig.Tunga, Sudamsu Shekhara. 1995. Bengali and other dialects of South Assam.Wali, Kashi. 2004. Non-nominative Subjects in Marathi. In Non-nominative Subjects (eds. Bhaskararao & Subbarao),223-52. Amsterdam: Benjamins.

24