Show Summary Details

Page of

PRINTED FROM the OXFORD RESEARCH ENCYCLOPEDIA, LINGUISTICS (linguistics.oxfordre.com). (c) Oxford University Press USA, 2016. All Rights Reserved. Personal use only; commercial use is strictly prohibited. Please see applicable Privacy Policy and Legal Notice (for details see Privacy Policy).

date: 27 May 2017

Historical Developments from Middle to Early New Indo-Aryan

Summary and Keywords

While in phonology Middle Indo-Aryan (MIA) dialects preserved the phonological system of Old Indo-Aryan (OIA) virtually intact, their morphosyntax underwent far-reaching changes, which altered fundamentally the synthetic morphology of earlier Prākrits in the direction of the analytic typology of New Indo-Aryan (NIA). Speaking holistically, the “accusative alignment” of OIA (Vedic Sanskrit) was restructured as an “ergative alignment” in Western IA languages, and it is precisely during the Late MIA period (ca. 5th–12th centuries ce) when we can observe these matters in statu nascendi. There is copious literature on the origin of the ergative construction: passive-to-ergative reanalysis; the ergative hypothesis, i.e., that the passive construction of OIA was already ergative; and a compromise stance that neither the former nor the latter approach is fully adequate. In the spirit of the complementary view of these matters, more attention has to be paid to various pathways in which typological changes operated over different kinds of nominal, pronominal and verbal constituents during the crucial MIA period.

(a) We shall start with the restructuring of the nominal case system in terms of the reduction of the number of cases from seven to four. This phonologically motivated process resulted ultimately in the rise of the binary distinction of the “absolutive” versus “oblique” case at the end of the MIA period). (b) The crucial role of animacy in the restructuring of the pronominal system and the rise of the “double-oblique” system in Ardha-Māgadhī and Western Apabhramśa will be explicated. (c) In the verbal system we witness complete remodeling of the aspectual system as a consequence of the loss of earlier synthetic forms expressing the perfective (Aorist) and “retrospective” (Perfect) aspect. Early Prākrits (Pāli) preserved their sigmatic Aorists (and the sigmatic Future) until late MIA centuries, while on the Iranian side the loss of the “sigmatic” aorist was accelerated in Middle Persian by the “weakening” of s > h > Ø. (d) The development and the establishment of “ergative alignment” at the end of the MIA period will be presented as a consequence of the above typological changes: the rise of the “absolutive” vs. “oblique” case system; the loss of the finite morphology of the perfective and retrospective aspect; and the recreation of the aspectual contrast of perfectivity by means of quasinominal (participial) forms. (e) Concurrently with the development toward the analyticity in grammatical aspect, we witness the evolution of lexical aspect (Aktionsart) ushering in the florescence of “serial” verbs in New Indo-Aryan.

On the whole, a contingency view of alignment considers the increase in ergativity as a by-product of the restoration of the OIA aspectual triad: Imperfective–Perfective–Perfect (in morphological terms Present–Aorist–Perfect). The NIA Perfective and Perfect are aligned ergatively, while their finite OIA ancestors (Aorist and Perfect) were aligned accusatively. Detailed linguistic analysis of Middle Indo-Aryan texts offers us a unique opportunity for a deeper comprehension of the formative period of the NIA state of affairs.

Keywords: nominal system, phrasal case, pronominal system, lexical aspect, grammatical aspect, tense, ergative alignment, Middle Indo-Aryan

Introduction

The Indo-European phylum consists of twelve families of languages (Celtic, Italic, Germanic, Baltic, Slavic, Albanian, Greek, Anatolian, Armenian, Indo-Aryan, Iranian and Tocharian). The Ancient language of India is known to us through its old texts called Vedas (Sanskrit veda ‘knowledge’), the sacred writings of the Hindu religion. Its later stage is referred to as Classical Sanskrit, a language of an enormous literary output (religious, scientific, poetry, drama) written during the 1st millennium bce and continued during the medieval period. The Iranian languages are closely related to the Indo-Aryan languages in their grammar and lexicon (their earlier common stage is called Indo-Iranian).The earliest literary documents of the Iranian family came down to us in the form of Avestan, the liturgical language of the religion founded by Zarathustra (ca. 900 bce). Its later stage is known to us through the cuneiform inscriptions of the Achaemenid kings (6th–4th c. bce). The Middle Iranian period is documented by several extinct languages (Parthian, Middle Persian, Sogdian, Khwarezmian). The New Iranian languages – Farsi, Kurdish, Pashto, Balochi and others – arose after the Arab conquests of the Iranian regions (beginning in the 7th c. ce).

1. Middle Indo-Aryan Period

The Middle Indo-Aryan period comprises all the varieties, called Prākrit ‘dialects’/‘languages’ (< prākrta ‘vernacular language’ as opposed to Sanskrit < saṁskrta ‘cultivated language’). They were used in an oral and literary form from the middle of the 1st millennium bc down to the 11th and 12th centuries ce. Its early period is represented by inscriptions and in literature by Pāli (4th century bce–2nd century ce); its middle period by various Prākrits (Māhārāṣṭrī, the language of the lyric poetry; the stage Prākrits of Sanskrit drama; Ardha-Māgadhī, the dialect of the Jain Canon; Jain-Māhārāṣṭrī and Jain-Śaurasenī in post-canonic writings; and Paiśācī, known from the statements by grammarians). Several Late MIA varieties, referred to by the pejorative term Apabhraṁśa (“vulgar dialect”) were in use between the 5th and 12th centuries ce. They are known to us from the Jain poetic works (virtually no prose works came down to us).

There are two comprehensive comparative grammars of MIA. The earlier was published in the Grundriss der indo-arischen Philologie und Altertumskunde by Pischel in 1900 (since 1965 it has been available in Jha’s English translation). The other, A Comparative Grammar of Middle Indo-Aryan, was written and published in India by Sukumar Sen (1960). Both works are valuable for data on phonetics/phonology and morphology of MIA dialects. In 1953, Sen published a long article entitled “Historical Syntax of Middle Indo-Aryan.” In 1934 Bloch’s masterpiece Indo-Aryan from the Vedas to Modern Times appeared. His introduction of the diachronic dimension into Indo-Aryan studies set up an example for the studies of individual New Indo-Aryan languages on historical principles in India. Bloch viewed the MIA period only as a transitional stage between OIA and NIA and did not try to elaborate on MIA dialects systematically in terms of their morphology and syntax. In the meantime, Indian scholars devoted their energy to the editing of Prākrit and Apabhraṁśa manuscripts, whose editions usually contain useful grammatical sketches and lexica. A number of linguistic analyses dealing with MIA phonology, morphology, semantics, and various literary problems were written during the previous decades by both Indian and Western scholars, but, on the whole, syntactic matters of MIA were seriously neglected (Bender’s 1969 bibliography of MIA studies lists 147 titles but only one devoted to syntax). Singh’s Syntax of Apabhraṁśa (1980) contains a wealth of morphological and syntactic data from the final stages of MIA but does not offer any theoretical framework for their evaluation. In 1982 Misra and Misra published a historical grammar of Ardha-Māgadhī, and in 1989 Bhayani published a short study of Apabhraṁśa. In the 1990s, Bubenik attempted to synthesize the previous research in phonology, morphology, and syntax in the monograph The Structural Development of Middle Indo-Aryan Dialects (1996). Another Bubenik monograph, A Historical Syntax of Late Middle Indo-Aryan (1998), exploring the issues of the restructuring of the nominal and pronominal systems, the evolution of the grammatical and lexical aspect, the emergence and development of the ergative construction, and the scope of the causative, was designed to fill in the gap of MIA syntactic studies.

During the past few decades a number of articles addressing the fundamental typological issue of “ergativity” and “alignment” in New Indo-Aryan languages in terms of their synchronic and diachronic perspective have been written, mainly by Western scholars. Most recently, there has been a surge of interest in the origin and development of the ergative alignment in IA, which crystallized in the Late MIA and Early NIA periods (see Bubenik, 2016). The present essay is meant to synthesize my and other scholars’ salient findings in this demanding field of inquiry.

Innovative features during the Late MIA period—measured against the earlier stage of conservative Prākrits—usher in the Early New Indo-Aryan state of affairs. While in phonology Apabhraṁśa preserved the Prākrit phonological system virtually intact, morphology and morphosyntax underwent far-reaching changes that altered fundamentally the synthetic morphology of earlier Prākrits in the direction of the analytic typology of NIA. We will study these changes in the following sections:

  • Section 2: Restructuring of the nominal system and the evolution of the phrasal case

  • Section 3: Restructuring of the pronominal system

  • Section 4: Complete remodeling of the aspectual system as a result of the loss of earlier synthetic forms expressing the past tense; evolution of lexical aspect (Aktionsart) leading to the florescence of “serial” verbs as known from NIA

  • Section 5: The coexistence of the old synthetic and the new analytic passive

  • Section 6: The development and the establishment of the ergative construction during the Late MIA period.

2. Restructuring of the Nominal System and the Evolution of the Phrasal Case

The nominal system of Old Indo-Aryan underwent a gradual erosion of case contrasts during its long MIA period of one and half millennia (5th century bce–10/11th century ce.) At the end of this long period Apbhraṁśa ended up with only one form for earlier Nom vs. Acc, Instr vs. Loc, and Gen vs. Abl. Table 1 puts these matters into the diachronic perspective of Old (Sanskrit) and Middle IA (Old and Middle Prākrits represented by Pāli and Ardha-Māgadhī).

Table 1. The Nominal System (a-Stems) of Old and Middle Indo-Aryan

OIA

Old Prākrits

Middle Prākrits

Late Prākrits

Sanskrit

Pāli

Ardha-Māgadhī

Apbhraṁśa

Sg Nom

putr-aḥ

putt-o

putt-o

putt-u

Acc

-am

-aṁ

-u

Instr

-eṇa

-eṇa/aṁ

-eṇa

-eṁ

Dat

-āya

-assa, -āya

-assa

-aho

Gen

-asya

-assa

-assa

-aho, -ahu

Abl

-āt

-asmā

-ā(o)

-ahe, -ahu

Loc

-e

-asmiṁ

-aṁsi

-i

By the end of the MIA period the seven fusional cases of OIA were reduced to four: Nom/Acc –u, Instr –eṁ, Dat/Gen/Abl –aho/-ahu, and Loc –i/e: in more general terms Direct versus Oblique cases. As a consequence of phonological changes, the polysemous morphology of oblique cases became insufficient to express the notions of spatial removal (Ablative), possession (Genitive), reference (Dative), and location (Locative).

In early MIA the distinct ablative suffix –āt lost its –t, but its homophony with the Nom Pl was prevented by the adoption of the pronominal suffix –asmā in Pāli (< OIA t–asmāt ‘from him’): OIA putr-āt ‘from the boy’ > Pāli putt-asmā. Another replacement for the ablative suffix was the derivational suffix –to ( < OIA –tas), which was continued in the form –āo in AMg. The source of the ablative suffix –he in Apabhraṁśa is uncertain (the OIA feminine pronominal form t-asyai ‘to her’ ended up as tahe ‘from her’ in Ap).

In the possessive construction the nominal or pronominal possessor stands before the possessed (the head of the construction): OIA putr-asya grhaṁ ‘the son’s house’, tasya grhaṁ ‘his house’ > Pāli putt-assa gharaṁ, tassa gharaṁ. In OIA and Middle MIA agreement with the possessed object was not shown on the pronominal possessor. This state of affairs changed fundamentally when Apabhraṁśa introduced two genitival postpositions, kera and tana, in that they display full adjectival agreement with the possessed object:

(1)

Historical Developments from Middle to Early New Indo-Aryan

At that point Apabhraṁśa reached the NIA state of affairs. In Hindi the genitival postposition is=kā bhāī his=GEN/M brother ‘his brother’ displays adjectival agreement with the possessed. In historical retrospective in Hindi the genitival postposition preserved its adjectival characteristics (agreement in gender and number), and the Apabhraṁśa stage shows how it happened. The adjectival form kera (allegedly from the gerundive kārya ‘to be done, made’) was cliticized to the pronoun and underwent some phonological erosion. Precious evidence comes from the westernmost Indo-Aryan variety, European Romani, which preserved essentially the late MIA state of affairs in not showing any phonetic attrition of the original adjectival postposition: les=ker-o phral his=GEN/M brother ‘his brother’. Romani also preserved the distinction of gender on the possessor which was lost in Hindi. Here are the other forms:

(2)

Historical Developments from Middle to Early New Indo-Aryan

Some other IA languages have also preserved distinction of gender on the 3rd Pers personal pronouns (e.g., Marathi tyā-cā mulgā ‘his son’, tī-cā mulgā ‘her son’).

The notion of reference (tādarthya) could be expressed by a number of postpositions (enumerated by Hemacandra in his Prākrit grammar of the 12th century): kehiṁ, kesiṁ, resi, resiṁ, and taṇeṇa. The form kehiṁ (< Instr M Pl form of ka ‘who’) corresponds to Hindi lie ‘for (the sake of)’

The notion of spatial and temporal location was expressed by the locative in OIA as in grh-e ‘in the house’. Given the phonetic attrition of the case morphology in MIA it became necessary to use the adverb majjhe/i ‘inside’ (< OIA madhy-e ‘in the middle’) with the depending noun in the genitive: ghar-aho majjhe ~ majjhe ghar-aho ‘in the house’; subsequently, the adverb majjhe/i ‘inside’ was grammaticalized as the postposition meṁ in NIA: ghar=meṁ in the house’.

The notions of accompaniment and instrumentality were expressd by the postpositions/prepositions samau/samāṇu (< OIA adverb samānam ‘jointly’) sahũ/sahu/sau (continuing OIA postposition saha ‘with’ < adverb saha ‘jointly’) with the noun in the instrumental:

(3)

Historical Developments from Middle to Early New Indo-Aryan

3. Restructuring of the Pronominal System and the Rise of “Double-Oblique” Alignment

The pronominal subsystem of the third person remained practically unchanged during the whole MIA period, as can be seen in Table 2 (only the singular forms of ‘he, that’ are shown).

Table 2. The Pronominal Subsystem (3rd Sg) of Old and Middle Indo-Aryan

OIA

Pāli

AMg

Ap

Sg Nom

saḥ

so (sa)

so, se

so, su

Acc

tam

taṁ

taṁ

taṁ

Instr

tena

tena

teṇa

teṇa, teṁ, tiṁ

Abl

tasmāt

tasmā, tamhā

tāo

tāsu

Dat/Gen

tasya

tassa

tassa

taho/u

Loc

tasmin

tasmiṁ, tamhi

taṁsi, tammi

tahiṁ

At the end of the MIA period the pronominal subsystem of the first and second person (participants in discourse) reduced the six distinct pronominal forms of earlier MIA to four: Nom vs. Acc/Instr vs. Abl/Gen (in Sg), and Nom/Acc vs. Instr vs. Abl/Gen vs. Loc (in Pl). In semantic terms, both subsystems differentiate between the agent/subject and patient; the notions of spatial removal and appurtenance are syncretized (Abl/Gen); and there was the general locative case. A salient feature of the Western Ap system (as described by Hemacandra in his Prakrit grammar of the 12th century ce) was the existence of the morphological syncretism of the Acc Sg and the Instr Sg mai(ṁ) ‘I’ and pai(ṁ)/tai(ṁ) ‘you’. In merging the locative maiṁ with Acc/Instr, Western Apabhraṁśa created a new typology of direct vs. oblique forms (anticipating the NIA state of affairs). In (3) the state of affairs in early Prakrits (Pāli) is contrasted with that found in Western Ap:

(4)

Historical Developments from Middle to Early New Indo-Aryan

This unusual distribution of pronominal forms in (3), called “double-oblique” alignment (see Harris & Campbell, 1995, p. 240), did not exist in OIA. As far as I can tell it made its first appearance in the Ardha-Māgadhī texts of the 4th–3rd centuries bce. Its appearance in the AMg pronominal system was an extension of the situation that already existed in the system of pronominal clitics in OIA. The conservative Prākrit dialects preserved the OIA accusative forms maṁ and tvaṁ (> taṁ in Pāli), but AMg started using the universal clitic forms me and te instead. Here are some salient AMg examples (after Bubenik, 1998, p. 89):

(5)

Historical Developments from Middle to Early New Indo-Aryan

The earliest pieces of evidence for the double-oblique alignment with orthotonic (non-clitic) pronominal forms in Apabhraṁśa are found in Kālidāsa’s Vikramorvaśīya of the 5th century ce (assuming that the Apabhraṁśa songs in the Fourth Act are genuine). Here the same pronominal form paiṁ encodes both the agent in the passive construction (in the past) and the object in the active construction (in the present):

(6)

Historical Developments from Middle to Early New Indo-Aryan

The double-oblique system with full pronouns is amply documented in the Jain texts composed in Western and Southern Apabhraṁśa during the 10th to 12th centuries.

In the westernmost periphery of India, the double oblique alignment survived in modern Sindhi and Lahnda. In Grierson’s description of Sindhī (1919) there is a set of pronominal suffixes that may express the possessor with nouns (piu=me ‘my father’), the agent (in ergative tenses), and the object (in non-ergative tenses):

(7)

Historical Developments from Middle to Early New Indo-Aryan

Nowadays the use of pronominal clitics of the first and second person has become obsolete (see Bubenik, 2016).

4. Evolution of Grammatical and Lexical Aspect

4.1. Grammatical Aspect

During the MIA period we witness the gradual loss of the synthetic morphology of OIA aspectual and tense forms. Pāli syncretized the aorist and the imperfect into a single past tense form called usually “aorist/preterit.” Mayrhofer (1951, p. 153) presents it as a category consisting of four conjugations which continue the root, thematic, sigmatic, and -aorists of OIA: ahū ‘he was’ (< a-bhūt); agama ‘he went’ (< (a-gama-at); assossi ‘he heard’ (< (a-śrau-ṣ-īt); and agam-is-aṁ ‘I went’. Their ancestral imperfects are seen in forms such as abravi/ī ‘he spoke’ (< a-brav-īt) and agacchi ‘he went’ (< a-gacchat). In Ardha-Māgadhī and Māhārāṣṭrī we witness a complete loss of the finite morphology of the aorist/preterit and its replacement by the passive/ergative construction based on the PP with the agent expressed by the instrumental: Pāli mayā kataṁ ‘it was done by me’ ~ ‘I did (it)’ and Mah mae kaaṁ, Ap maiṁ kiu ‘I did (it)’. The present and sigmatic future survived much longer: Pāli karomi ‘I do’; karissāmi ‘I will do’; Mah karemi, karissaṁ; and Ap karauṁ, kare/isami ~ karihimi. Both forms are still found in NIA: Hindi subjunctive karuṁ continues Apabhraṁśa karauṁ; the sigmatic future predominated in Southern Apabhraṁśa ( > Marathi); in Western Ap both types existed (in NIA the s-future is found in Gujarati and Lahnda, the h-future (< -s-) in Marwari, Braj, Awadhi, and Bundeli).

The rise of analytic formations in both the Imperfective and Perfect(ive) categories was the main morphosyntactic innovation during the MIA period. At its end the Late MIA aspectual system presents periphrastic counterparts to the Imperfective (Present, Imperfect) and Perfect (Perfect, Pluperfect) categories, as shown in Table 3.

Table 3. The Rise of the Late MIA Aspectual System

Imperfective

Perfective

Perfect

OIA

Present

Karoti

Cakāra

Past

akarot (Imperfect)

akārṣat (Aorist)

acakrat (Pluperfect)

Pāli

Karoti

akāsi

tena kataṁ

Ap

Present

kar-antu acch-ai

ki-u

ki-u acch-ai

Past

kar-antu acchi-u

ki-u

ki-u acchi-u ~ ṭhi-iu

In Pāli the imperfective aspect in the present and the past could be realized by combining the present participle of the main verb with the verb tiṣṭh-ati ‘stand’ or vicar-ati ‘walk’:

(7)

Historical Developments from Middle to Early New Indo-Aryan

The continuous action in the future may be expressed by combining the gerundive or the PP with the copula in the future tense: ‘tapo car-itabbo (GERUNDIVE) bhav-issati (FUT) ‘he will be doing penance’. Habituality may be expressed by combining the gerund (absolutive) with the verb vat-ati ‘become’ (< OIA vart-ate ‘turn; remain; exist’);

(8)

Historical Developments from Middle to Early New Indo-Aryan

These rare formations were employed to express finer nuances of the imperfective aspect (continuous/progressive and habitual). During the MIA period they were continued, further developed, and grammaticalized in the following fashion in Apabhraṁśa:

(9)

Historical Developments from Middle to Early New Indo-Aryan

(10)

Historical Developments from Middle to Early New Indo-Aryan

The progressive aspect could also be expressed by the present participle in its gerundial (absolutive) function:

(11)

Historical Developments from Middle to Early New Indo-Aryan

A remarkable innovation of Apabhraṁśa is the imperfective passive participle formed by the suffix –anta (Pres Part) from the passive stem in –ijj, e.g., k-ijj-anta ‘being made’ (replacing OIA kr-iya-māṇa):

(12)

Historical Developments from Middle to Early New Indo-Aryan

The form hamm-antu ‘being killed’ (< han-y-ant) is opaque; the contrast between the active participle ‘killing’ and the imperfective passive participle was apparently not particularly viable (compare han-antu ‘killing’ with hamm-antu ‘being killed’) and the latter form was discontinued in Early NIA.

4.2. Lexical Aspect

Apabhraṁśa data contain examples of verbal compounds expressing several Aktionsart (lexical aspect) categories. As in NIA languages, they are composites of main verbs with one of a small number of “auxiliary” verbs. Cross-linguistically, the lexical meaning of the auxiliary has been grammaticalized toward various actional notions such as inception and completion of the event expressed by the main verb, its continuity, its resumption, its subitaneity (suddenness), and others. New Indo-Aryan languages offer some of the most studied examples of these “serial” verbs (see Hook, 1974; Nespital, 1997). Compared with NIA, languages that compound either the oblique infinitival form (of the Hindi type bol-ne lag-nā speak.INF/OBL be-attached.INF ‘begin to speak’) or the bare verbal roots with the auxiliaries (of the Hindi type ro uṭh-nā cry rise.INF ‘burst into tears’), we find an earlier state of affairs in Ap. Here the main verb may be realized by one of the quasinominal forms: the gerund (-vi, -viṇu, -eppi, -eppiṇu), the infinitive (-aṇahaṁ, -ahuṁ), or the gerundive (-ev(v)ae).

Our Apabhraṁśa data allow us to identify two exponents of the notions of inception and completion, expressed by the Aktionsart auxiliaries, namely lag(g)- ‘begin’ (as a lexical verb lag-nā means ‘seem, appear’, ‘be attached’ in Hindi) and jā- ‘go’ (< OIA yā-). The inceptive auxiliary may combine with any of the quasinominal forms of the main verb, i.e., gerund, infinitive, or gerundive. Apabhraṁśa examples of all the three available combinations (from Svayaṁbhūdeva’s Paumacariu of ca. 700 ce) are provided in (13):

(13)

Historical Developments from Middle to Early New Indo-Aryan

The completive/resultative auxiliary ‘go’ stresses the fact that an action is completed or carried through as a process (see Mc Gregor, 1977, p. 99, regarding Hindi jā-). While Hindi jā- is suppletive in forming the PP gayā (< OIA ga-ta ‘gone’ belonging to another root gam-), the Apabhraṁśa form jā-u contninues the OIA PP yā-ta ‘gone’.

Thus a typical completive event such as ‘he (has) died’, vah mar gayā in Hindi, would be realized as jā-u mu-u in Apabhraṁśa:

(14)

Historical Developments from Middle to Early New Indo-Aryan

The other auxiliary, ga-u go+PP ‘gone’, is used as the marker of the passive voice (ushering the ‘go’-passive of NIA languages):

(15)

Historical Developments from Middle to Early New Indo-Aryan

The PP ga-u is common with intransitive verbs, but there are also instances of the collocations with transitive verbs such as le- take’:

(16)

Historical Developments from Middle to Early New Indo-Aryan

There is some evidence in our Apabhraṁśa data that the notions of version and ablation could be expressed by the auxiliaries le- ‘take’ (also nī- ‘take’) and de- ‘give’. The auxiliary le- ‘take’, heralding the NIA state of affairs, has a general reflexive (selfbenefactive) sense. Functionally, it may be understood as the OIA mediopassive, which was lost in MIA (Pāli kariyati/kayirati and Ap karijjai continue the OIA finite passive kri-ya-te). This loss was compensated for by the grammaticalization of lexical verbs ‘take’ and ‘give’:

(17)

Historical Developments from Middle to Early New Indo-Aryan

These (rare) examples should be distinguished from numerous instances of the compound verbs consisting of de- as a fully meaningful lexical verb ‘give’ combined with nouns such as ‘command’, ‘fight’, etc. to yield compound verbs ‘to command’, ‘to fight’, etc. The most numerous are the compound verbs consisting of kar ‘make’ and nouns. There are quite a few doublets, such as jujjhu kar- ~ jujjhu de- ‘to fight’, rakkhaṇu kar- ~ rakkhanṇu de- ‘to protect’, etc.

5. The Old Synthetic and the New Analytic Passive

The MIA passive in –ijja- continues the OIA passive in –ya-, e.g., kri-ya-te ‘is made’ > Pāli kari-ya-ti (with metathesis kayirati) > Ap kar-ijja-i; drś-ya-te ‘is seen’ > Pāli dissai; dī-ya-te ‘is given’ > dī-ya-ti; uh-ya-te ‘is carried’ > vuyhati (root vah-). The passive may be used not only in assertions but also in modalized statements to express demands and deontic modality; it is also used in expressions of ability and cohortation (referring to the first person).

Unlike in OIA (kri-ya-te Present vs. a-kri-ya-ta Imperfect), only the present tense forms are available in Apabhraṁśa. The future tense forms (k-ijji-hi-i ‘it will be done’) are isolated, and their OIA ancestry is doubtful.

(18)

Historical Developments from Middle to Early New Indo-Aryan

The use of the passive in modal contexts of dialogues can be understood as an option left to the speaker of not expressing the addressee as the subject (in the second person), or, positively, as the option of expressing the object of the action as the subject:

(19.a)

Historical Developments from Middle to Early New Indo-Aryan

(19.b)

Historical Developments from Middle to Early New Indo-Aryan

(19.b) can be classified as a demand whose addressee is not necessarily the interlocutor as is the case of the command in (19.a). The potential performer of (19.b) can be any other third person (lit. ‘may the penitence be practiced by whoever’). The generic performer brings us into the domain of deontic modality, whose chief exponent in OIA was the gerundive. With the recategorization of the gerundive as the infinitive in late MIA, it became possible to use the finite passive in its stead:

(20)

Historical Developments from Middle to Early New Indo-Aryan

The form kar-ijja-hu corresponds to the OIA imperative kri-ya-tām (or rather *kri-ya-tu with the suffix of the active imperative). The above deontic statement could be expressed in two ways in Sanskrit either by the gerundive idaṁ karman kar-tavy-am or by the passive “imperative” idaṁ karman kri-ya-tām ‘may this deed be done by whoever’. Because of the recategorization of the gerundive in late MIA, the passive ‘imperative’ in (20) would be ambiguous between a deontic statement and a demand. The latter interpretation became appropriate when the addressee was expressed by the personal pronoun, as in (21):

(21)

Historical Developments from Middle to Early New Indo-Aryan

(Another spelling of the same passage, dijjau, goes back to *dī-ya-tu, with the suffix of the active ‘imperative’ versus the OIA mediopassive form dī-ya-tām).

The irregular active imperatives of Hindi (kīj-ie ‘do’, līj-ie ‘take’, dīj-ie ‘give, pīj-ie ‘drink’) go all the way back to the OIA passive ‘imperatives’ in the third person:

(22)

Historical Developments from Middle to Early New Indo-Aryan

The finite passive may be used in various modal meanings such as abilitative and cohortative:

(23)

Historical Developments from Middle to Early New Indo-Aryan

The form vaṇ-ijja-i goes back to an earlier indicative *varṇa-ya-te ‘it is depicted/described’ used as an impersonal construction (see Latin Quo modo describitur X?)

(24)

Historical Developments from Middle to Early New Indo-Aryan

The source of the form gammau is the 3rd Sg mediopassive imperative gam-ya-tu (see OIA gam-ya-tām) used impersonally (see Russian idët=sya). In Sanskrit it was possible to express the subject by the instrumental form gam-ya-tām asmābhiḥ‘let us go’ (lit. may it be gone by us).

A remarkable innovation of MIA was the rebuilding of the OIA imperfective passive participle kri-ya-māna ‘being made’ by means of the active suffix –ant (see OIA kurv-ant- ‘making’) > k-ijj-anta ‘being made’. Its extensive usage in the absolute constructions in late MIA texts (such as Svayaṁbhūdeva’s Paumacariu of ca. 700 ce) are probably sanskritisms (see Bubenik, 1998, p. 202).

(25)

Historical Developments from Middle to Early New Indo-Aryan

Constructions with the PP and the auxiliary ‘be’ (=mhi ‘I am’, =si ‘you are’, atthi ‘s/he is’, āsi ‘s/he was’) may be interpreted passively or ergatively (i.e., actively). The passive interpretation is available in the context when the agent is either unknown or too obvious to be mentioned:

(26)

Historical Developments from Middle to Early New Indo-Aryan

In dialogues and with verbs of speaking, the ergative interpretation is more likely:

(27)

Historical Developments from Middle to Early New Indo-Aryan

‘We don’t know that we have committed such a sin’ (rather than ‘… that such a sin was committed by us’)

The ambiguity between these two interpretations (passive in the past and the active perfect) is symptomatic of the shift toward ergative typology during the Late MIA period.

6. The Emergence and Development of the Ergative Construction

The subject of the rise of tense/aspect-conditioned split “ergativity” can be approached in “holistic” terms (considering ergativity as a major typological parameter that correlates significantly with a number of other syntactic and semantic features) or in terms of “contingency” (considering the rise of ergativity as a by-product of low-level phonological and morphological innovations, which ultimately introduced changes in the morphosyntactic profile in the language). Both approaches offer certain advantages, especially the latter, which is vital for addressing diachronic issues endemic in the topic.

There is copious literature on the origin of the ergative construction in NIA languages: (a) passive-to-ergative reanalysis (Anderson, 1977); (b) the ergative hypothesis, i.e., that the passive construction of OIA was already ergative (Klaiman, 1978); and a compromise stance that neither (a) nor (b) is fully adequate (Andersen, 1986; Hock, 1986,). All these theories explain the state of affairs in NIA from the oldest recorded stages without paying sufficient attention to the crucial MIA stage, which was exactly the period when the typological passive-to-ergative shift was running its course. The crucial form in all discussions of ergativity in IA languages is the ta-form of OIA ( < Proto-Indo-European form *-to), classifiable as a verbal ‘noun/adjective’. In IE languages it is documented as a past passive participle (PP). In PIE the *to- form could be used in equational predication with intransitive verbs (‘he is gone’ *so gwm-tós) and with transitive verbs in the construction *toi (INSTR) or tosyo (GEN) wīr-os gwhn-tós ‘by him / of him the man was killed’. In IA the latter construction should be interpreted as passive in the absence of the finite passive in the perfective aspect (mediopassive forms were available in the imperfective aspect: Present kri-ya-te ‘it is being made’, Imperfect a-kri-ya-ta, but there was also the medio-passive Aorist form a-kār-i). To express the active voice ‘he killed the man’ one could use the finite form of the aorist *é-gwhen (> *a-han, the undocumented form of the root aorist was replaced by the sigmatic aorist a-hāniș-at) or the perfect *gwe-gwhon-e (Vedic ja-ghān-a). Classical Sanskrit and contemporary MIA dialects show an increase in the use of the construction of the type tena krtam ‘by him done’ but as long as the finite Aorist and Perfect were available, there was no need to interpret the construction tena krtam as ergative. To stay with ‘killing’ an example such as mayā naro hatah could be interperetd alternatively as ‘a/the man was killed by me’ ~ ‘I killed the man’ where the active or passive interpretation can be argued for by the postulates of the functional sentence perspective. To express unambiguously ‘I killed the man’ one could use the finite active Imperfect (naram ahanam) or the sigmatic Aorist (naram a-hāniṣ-am) or the Perfect (naram jaghana), i.e., there is not much point in characterizing OIA and early MIA as an ergative-absolutive language (especially in view of their nominative–accusative morphology). In OIA the ta-form and the gerundive could be accompanied by the agentive phrase in the instrumental or in the genitive (the latter case is found verbs of perception, consumption, and ‘enjoyment’). In MIA the agentive phrase with the ta-form could be expressed by either the Instr or the Gen. Statistics for the inscriptions by the King Aśoka (3rd century bce) were elaborated by Andersen (1986), and they may be interpreted in the sense that verbs denoting activities favor the instrumental agent in the passive construction, whereas the “ingestive” verbs favor the genitival agents. The situation is complicated by the neutralization of the contrast between the Instr and the Gen in pronominal clitics in favor of the latter case (Pāli mayā ‘by me’ mama/mayhaṁ ‘of me/to me’ vs. clitic me Gen/Dat). Thus it is debatable whether the following example featuring the pronominal agent in the clitic form should be given an active or passive interpretation:

(28)

Historical Developments from Middle to Early New Indo-Aryan

But on the whole the ergative interpretation of mayā krtam ‘by me done’ > ‘I did (it)’ (i.e., passive-to-ergative shift) became readily available with the gradual attrition and final loss of the finite morphology of Imperfect, Aorist, and Perfect during the Late MIA period (see Table 3). At the same time we witness the appearance of the absolutive case (see Table 1). Contrast Apabhraṁśa absolutive marking—mae (INSTR) nar-u (ABS) mār-i(y)a(u) (PP) ‘I killed the man’ and naru (ABS) calai the man walks’—with OIA nominative marking—mayā naro (NOM) hataḥ and nar-aḥ (NOM) carati.

It could be pointed out that the seeds of this state of affairs were already present in the OIA stage in the case of neuters that do not distinguish formally between the Nom and the Acc. Thus a fictive Sanskrit sentence such as

(29)

Historical Developments from Middle to Early New Indo-Aryan

can be given either interpretation. Under the passive interpretation of (29), the NP mitram would be considered to be a grammatical subject, and the sentence would be translated “the friend was killed by me.” Under the ergative interpretation the NP mitram could be taken for the absolutive case (neuters display same suffix –am for both Nom and Acc), which expresses the patient, and the sentence would be translated “I killed the friend.” However, in OIA there were not too many animate neuters (kalatram ‘wife, female’ is another one), whereas in Late MIA this morphosyntactic ambiguity spread to all nouns as a result of the Nom/Acc syncretism; that is, the same suffix –u would be used with the intransitive subject, the subject of the passive sentence, and the object of the active sentence. The preference for the passive or active interpretation would ultimately be dictated by the postulates of FSP (functional sentence perspective), which is not easy to demonstrate in our earlier MIA texts exhibiting “free” word order. We stand more chance of demonstrating the pragmatic effects (in terms of topic and focus) with pronominal arguments. To judge by the behavior of pronominal clitics in contemporary languages, it could be that the contrast of the type clitic form of the personal pronoun versus its orthotonic form was exploited in context-dependent sentences to implement pragmatic contrasts such as that of the unmarked statement versus a statement with the subject in focus: ‘I heard it’ versus ‘I (not YOU) heard it’. Other options of putting the subject in focus are the passive construction ‘It was heard by ME’ or (in English) the clefted contruction ‘It was I who heard it’.

(30)

Historical Developments from Middle to Early New Indo-Aryan

In the latter example the full pronominal form could carry the sentential stress. (Clearly, coming up with minimal pairs of this nature in our written texts is extremely difficult; see Bubenik, 1998, pp. 140–141.)

In Section 2 we studied the emergence of the double-oblique system in the pronominal forms of the 1st and 2nd Sg [examples from Ardha-Māgadhī and Apabhraṁśa were given in (3)–(5)]. The case marking with the verbs of speaking is of a particular interest from the point of view of later developments in NIA. In Apabhraṁśa verbs of speaking pattern both ergatively (express their speaker by the instrumental case) and nominatively (express their speaker by the absolutive case). Some of them allow both patterns (volliu ‘said, spoke’, pajaṁpiu ‘said’ (< OIA pra-jalp-), pabhaṇiu ‘addressed’), and some are documented only in the ergative construction (kahiu ‘told’, pucchiu ‘asked’); see Hindi, where kahnā and pūchnā are always used with the ergative postposition =ne but bolnā (< Ap voll-) only rarely so. Their addressee may be expressed by the absolutive or the ‘oblique’ case (Gen/Dat):

(31)

Historical Developments from Middle to Early New Indo-Aryan

vuttu ‘said’ (< OIA uktam), volliu ‘said, spoke’ and kahiu ‘told’ appear also in in the finite passive form vuccai, volijjai, and kahijjai with the speaker in the instrumental (in Svayaṁbhūdeva’s Paumacariu this construction was actually more common than its non-finite counterpart tena V+PP):

(32)

Historical Developments from Middle to Early New Indo-Aryan

Summarily, the OIA system was aligned accusatively. The construction tena krtam ‘it was done by him’ was interpreted as the perfective passive (its imperfective counterpart was provided by the finite passive tena kriyate ‘it is being done by him’). In the Early MIA stage (represented by Pāli) we witness the spread of the construction tena kata ‘(it) was done by him’ ~ he did (it)’ into the domain of earlier finite perfective forms but the choice of expressing the perfective aspect by the finite form (akāsi < OIA aorist akārṣat ‘he did’) was still readily available, and the case system was still solidly of the Nom-Acc typology (see Table 1). In the Late MIA period we witness the crucial step to the establishment of the ergative alignment in the rise of the absolutive case. The construction tena kiu (<tena kata < tena krtam) replaced completely earlier finite forms of the ‘preterit/aorist’ but it remained ambiguous between ergative and passive interpretation (‘(it) was done by him’ ~ he did (it)’). Complicated matters in the following centuries cannot be addressed here; but ultimately in NIA the construction us=ne kiyā he.OBL=ERG do.PP is consistently interpreted as an ergative construction ‘he did (it)’. The innovative passive (in Hindi as of 17th century ce) is marked by a different “instrumental” postposition, =se ‘by, with’ (its other meanings are ‘from’ and ‘to’), and the PP has to be reinforced by the auxiliary gayā ‘gone’ (from OIA ga-ta go.PP): us=se kitāb likh-ī ga-ī he.OBL=INS book write.PP/F go.PP/F. Over the centuries before NIA the Vedic agentive adjunct in tena krtam ‘done by him‘ was replaced by the postposition, whose source is not quite clear (Braj and Marwari = nai, Nepali = lāī (< the participial form lāgi ‘stick, adhere’ ?), and the quasi-nominal participial form krta ‘done’ underwent a functional shift toward the quasi-verbal semifinite perfective form kiyā ‘did’. In Marwari and Nepali these are dative postpositions. In Old Nepali le and lai were used interchangeably—now it is only le that is used as an ergative postposition. In Hindi, unlike the present forms, which are fully finite (person, number, tense), the perfective form kiyā is only ‘semifinite’ (marked for number and gender, but NOT for person). The OIA exponents of the perfective aspect (finite forms of the aorist) marked morphologically for person and number are contrasted in (33) with their NIA counterparts, where the person has to be expressed by the free personal pronoun (hosting the ergative postposition):

(33)

Historical Developments from Middle to Early New Indo-Aryan

7. Desiderata for Further Research

Different alignments arose in different subdomains of the grammar of Middle Indo-Aryan dialects as a result of independent factors. The syncretism of case leading to the appearance of the absolutive case in the nominal system was caused by phonological change. On the other hand, the way case marking alignment was distributed over the pronominal system can primarily be attributed to the effects of animacy. However, the Apabhraṁśa “double-oblique” pronominal system differs from the usual predictions in that the 1st and 2nd Pers Sg should be aligned accusatively. Table 3 shows the “intrusion” of the non-finite construction te/iṁ ka(d)a (later kiu) ‘done by him’ into the domain of the erstwhile finite forms expressing the notion of perfectivity; we observed its coexistence with the finite ‘preterit’/’aorist’ in Early MIA (Pāli) and its complete demise during the Late MIA period. The inherited finite and the new analytic passive were shown to be coexistent. The gerundive (modal passive participle) was recategorized as the (polite) active imperative in the context of the passive-to-ergative shift. In Section 6 we saw that all of the typological changes analyzed in Sections 2, 3, and 4 were instrumental in the establishment of the ergative construction during the Late MIA period. Various “inconsistencies” in case marking with the verbs of speaking were of particular interest from the point of view of later parallel developments in NIA.

The detailed diachronic narrative of the “contingency” view of alignment may be summarized in holistic terms (see example 34) as the restoration of the OIA aspectual triad Present-Aorist-Perfect on the basis of two participles: imperfective and perfective (the PP form). The NIA Perfective and (analytic) Perfect categories are aligned ergatively, while their monolectal OIA ancestors (Aorist and Perfect) were aligned accusatively. The imperfective forms of NIA (Present vah kar-tā hai ‘he does’ and Imperfect vah kar-tā thā ‘he did/was doing’) are built on the grammaticalized nominal form kar-tā (< OIA kar-tr -‘doer’), which replaced the Late MIA participial form kar-antu ‘doing’ (< OIA kar-ant-). The perfective (simple Past) and perfect forms (Perfect, Pluperfect, and Future Perfect) are built on the OIA PP form kr-ta ‘done’ (> NIA semifinite ki-yā ‘did’). The present tense auxiliary h-ai ‘is’ goes all the way back to OIA bhav-ati ‘becomes’, and the past tense auxiliary continues the PP form sthi-ta (> thā) of the root sthā- ‘stand’

(34)

Historical Developments from Middle to Early New Indo-Aryan

Further progress in our comprehension of these matters during the early NIA centuries (12th–16th centuries) can only be achieved by close cooperation among historical linguists and connoisseurs of Old, Middle, and New Indo-Aryan languages and dialects on the basis of Early NIA literary works. Dahl and Stroński’s Indo-Aryan Ergativity in Typological and Diachronic Perspective (2016) represents a step in this direction.

Further Reading

Andersen, P. K. (1986). The genitive agent in Rigvedic passive constructions. In Collectanea linguistica in honorem Adami Heinz (pp. 9–13). Wrocław, Poland: Wydawnictwo Polskiej Akademii Nauk.Find this resource:

Anderson, S. R. (1977). On mechanisms by which languages become ergative. In C. N. Li (Ed.), Mechanisms of syntactic change (pp. 317–363). Austin: University of Texas.Find this resource:

Bubenik, V. (2016). On the establishment of ergative alignment during the Late Middle Indo-Aryan period. In E. Dahl & K. Stroński (Eds.), Indo-Aryan ergativity in typological and diachronic perspective (pp. 111–133). Amsterdam: Benjamins.Find this resource:

Bubenik, V., & Paranjape, Ch. (1996). Development of pronominal systems from Apabhraṃśa to New Indo-Aryan. Indo-Iranian Journal, 39, 111–132.Find this resource:

Haig, G. (2008). Alignment change in Iranian languages. A construction grammar approach. Berlin: Mouton de Gruyter.Find this resource:

Hock, H. H. (1986). P-oriented constructions in Sanskrit. In B. Krishnamurti, C. P. Masica, & A. K. Sinha (Eds.), South Asian languages: Structure, convergnece and diglossia (pp. 15–26). Delhi: Motilal Banarsidass.Find this resource:

Khokhlova, L. V. (2001). Ergativity attrition in the history of Western New Indo-Aryan languages (Punjabi, Gujarati and Rajasthani). The Yearbook of South Asian Languages and Linguistics, pp. 159–184.Find this resource:

Klaiman, M. H. (1978). Arguments against a passive origin of the ergative. Chicago Linguistic Society: Papers from the 14th Regional Meeting, pp. 204–216.Find this resource:

Sukumar, S. (1953). Historical syntax of Middle Indo-Aryan. Indian Linguistics, 13, 355–473.Find this resource:

Sukumar, S. (1960). A comparative grammar of Middle Indo-Aryan. Pune, India: Deccan College.Find this resource:

References

Andersen, P. K. (1986). The genitive agent in Rigvedic passive constructions. In Collectanea linguistica in honorem Adami Heinz (pp. 9–13). Wrocław, Poland: Wydawnictwo Polskiej Akademii Nauk.Find this resource:

Anderson, S. R. (1977). On mechanisms by which languages become ergative. In C. N. Li (Ed.), Mechanisms of syntactic change (pp. 317–363). Austin: University of Texas.Find this resource:

Bender, E. (1969). Middle Indo-Aryan. In T. E. Sebeok (Ed.), Current trends in linguistics (Vol. 5, pp. 46–54). The Hague: Mouton.Find this resource:

Bhayani, H. C. (1989). Apabhramśa language and literature. Delhi: B. L. Institute of Indology.Find this resource:

Bloch, J. (1933). L’indo-aryen du véda au temps moderne. Paris: Adrien Maisonneuve.Find this resource:

Bubenik, V. (1989). An interpretation of split ergativity in Indo-Iranian languages. Diachronica, 6(2), 181–212.Find this resource:

Bubenik, V. (1996). The structure and development of Middle Indo-Aryan dialects. Delhi: Motilal Banarsidass.Find this resource:

Bubenik, V. (1998). A historical syntax of Late Middle Indo-Aryan (Apabhraṁśa). Amsterdam: Benjamins.Find this resource:

Bubenik, V. (2003). Prākrits and Apabhraṁśa. In G. Cardona & Dh. Jain (Eds.), The Indo-Aryan languages (pp. 204–249). London: Routledge.Find this resource:

Bubenik, V. (2016). On the establishment of ergative alignment during the Late Middle Indo-Aryan period. In E. Dahl & K. Stroński (Eds.), Indo-Aryan ergativity in typological and diachronic perspective (pp. 111–133). Amsterdam: Benjamins.Find this resource:

Bubenik, V. (forthcoming). Development of tense and aspect from Old to Middle Iranian. Fausböll, V. 1877–1896. The Jātaka together with its commentary. London: Trübner.Find this resource:

Dahl, E., & Stroński, K. (2016). Indo-Aryan ergativity in typological and diachronic perspective. Amsterdam: Benjamins.Find this resource:

Grierson, G. A. (1903–1928/1968). Linguistic survey of India. Calcutta: Motilal Banarsidass.Find this resource:

Haig, G. (2008). Alignment change in Iranian languages. A construction grammar approach. Berlin: Mouton de Gruyter.Find this resource:

Harris, A. C. & Campbell, L. (1995). Historical syntax in cross-linguistic perspective. Cambridge: Cambridge University Press.Find this resource:

Hock, H. H. (1986). P-oriented constructions in Sanskrit. In B. Krishnamurti, C. P. Masica, & A. K. Sinha (Eds.), South Asian languages: Structure, convergnece and diglossia (pp. 15–26). Delhi: Motilal Banarsidass.Find this resource:

Hook, P. E. (1974). The compound verb in Hindi. University of Michigan.Find this resource:

Jain, B. D. (1923). Ardha-Māgadhī reader. Lahore, Pakistan: Panjab University Oriental Publications.Find this resource:

Karmarkar, R. D. (1932). Vikramorvaśīya of Kālidāsa. Pune, India: Karmarkar.Find this resource:

Klaiman, M. H. (1978). Arguments against a passive origin of the ergative. Chicago Linguistic Society: Papers from the 14th Regional Meeting (pp. 204–216)Find this resource:

Mayrhofer, M. (1951). Handbuch des Pāli. Heidelberg, Germany: Winter.Find this resource:

Mc Gregor, R. S. (1977). Outline of Hindi grammar. Delhi: Oxford University Press.Find this resource:

Misra, S. S., & Misra, H. (1982). A historical grammar of Ardhamāgadhī. Varanasi, India: Ashutosh Prakashan Sansthan.Find this resource:

Nespital, H. (1997). Dictionary of Hindi verbs. Allahabad, India: Lokabharti Prakashan.Find this resource:

Pischel, R. (1900). Grammatik der Prakrit-Sprachen. Strassburg, Austria: Trübner (Delhi: Motilal Banarsidass, 1965; English translation by S. Jha.).Find this resource:

Rhys Davids, T. W., & Carpenter, J. E. (1890–1911). Dīgha-Nikāya. London: Pali Text Society.Find this resource:

Sen, S. (1953). Historical syntax of Middle Indo-Aryan. Indian Linguistics, 13, 355–473.Find this resource:

Sukumar, S. (1953). Historical syntax of Middle Indo-Aryan. Indian Linguistics, 13, 355–473.Find this resource:

Sukumar, S. (1960). A comparative grammar of Middle Indo-Aryan. Pune, India: Deccan College.Find this resource:

Singh, R. A. (1980). Syntax of Apabhraṁśa. Calcutta: Simant Publications India.Find this resource:

Vaidya, P. C. (1958/1980). Prakrit grammar of Hemacandra. Pune, India: Bhandarkar Oriental Research Institute.Find this resource: