Animacy, postpositions, and the spatial cases in Niya Prakrit

1 Introduction and aims

The grammaticalisation of postpositions as new case markers is a well-known trait of New Indo-Aryan (NIA) languages (Masica 1991: 230–248), but precursors of this development can already be found in Middle Indo-Aryan (MIA) and Old Indo-Aryan (OIA).

Using a corpus of texts in Vedic Sanskrit, Pāli, Apabhraṃśa, and Old Avadhī,^[1] Reinöhl (2016a, 2016b) has, for instance, been able to trace the grammaticalisation process that resulted in two postpositional case markers found in modern Hindi: (i) mẽ ‘in’ < OIA madhye, loc.sg of madhya- ‘middle’, and (ii) par ‘on’ < OIA upari ‘above, on top’.^[2] Contrary to a common assumption found in the literature (e.g. in Bubeník 2006), Reinöhl argues forcefully against a direct continuity between such case markers and the Vedic local adverbs like ádhi ‘above’ (Reinöhl 2016a: 65–83, 2016b). As shown by her, the major source for the NIA postpositional cases are rather “relational noun constructions with a possessor encoded by the genitive”, i.e. constructions such as “in the middle of X” or “on the top of X”. As part of her argument, Reinöhl identifies three grammaticalisation processes: (i) the relational noun construction gradually loses its originally more concrete semantics and comes to be reinterpreted as a postposition encoding grammatical case relations (2016a: 85–117, 2016b: 111–116); (ii) there is a development from free word order to a fixation of the order genitive + postposition, although even in Apabhraṃśa this has not yet become a compulsory rule (2016a: 119–145, 2016b: 116–121); and (iii) a genitive complement is in origin not obligatory with these relational noun constructions but becomes so over time (2016a: 147–173, 2016b: 121–123). Finally, Reinöhl observes that these genitive complements tend to be animate nouns and pronouns, although this preference for animates is not (yet) a fixed rule in the MIA languages studied by her (2016a: 158–162).^[3]

While Reinöhl’s studies certainly are “a bench mark for further studies on Indo-Aryan postpositions” (Verbeke 2018: 123), a drawback to her work is that she did not include data from non-literary sources such as Prakrit inscriptions or remarks by ancient grammarians (Verbeke 2018: 123–124). This exclusive reliance on literary texts stands in contrast to recent work on the development of case marking in both Greek and Arabic, where the importance of a detailed linguistic study of documentary texts written on papyrus is rightly emphasised (cf. e.g. Stolk 2017 for Greek and Kootstra 2022 for Arabic).

The only text corpus in a MIA language that can be compared in content and form to documentary papyri are the so-called Niya documents (3rd–4th century CE), which were drafted in a dialect of Central Asian Gāndhārī known as Niya Prakrit.^[4] In this paper, I would therefore like to take a closer look at the case system of this MIA language, in particular its ablative and locative.^[5] Drawing on Reinöhl’s observations on the role played by animacy in the selection of synthetic versus analytic case marking in Indo-Aryan languages (see above), I aim to demonstrate that (i) the Niya Prakrit synthetic ablative and locative are basically restricted to inanimate nouns, and that (ii) the postposition paride ‘from’ supplies an analytically formed ablative for animate nouns and pronouns, while vaṃti ‘in, at, to’ does the same for the locative. The approach will be mainly synchronic and language-internal, but occasional comparisons with both Indo-Aryan and non-Indo-Aryan languages will allow us to frame the findings of this article within a diachronic and typological perspective.

The article consists of the following parts: a brief introduction to the Niya Prakrit case system (Section 2); an investigation into the postposition paride ‘from’ and the synthetic ablative (Section 3); a parallel investigation into the postposition vaṃti ‘in, at, to’ and the synthetic locative (Section 4); a discussion of the Niya Prakrit personal pronouns from the perspective of animacy (Section 5); a note on the Khotan Prakrit postposition sag̲aj̲i ‘in, at, to’ (Section 6); and some concluding remarks (Section 7). The data on which the findings of this article are based can be found in the Appendix.

2 A brief introduction to the Niya Prakrit case system

To facilitate the discussion on the ablative and locative below, I briefly summarise here the main points about the Niya Prakrit case system (Burrow 1937: §§51–73, pp. 22–30).

In line with other MIA languages, the Niya Prakrit declensional system has been greatly simplified when compared to the rich morphology of OIA. As also happened in Apabhraṃśa, nominative and accusative have been merged in Niya Prakrit to what I shall call the “direct case” (dir) (following Baums 2009: 211 et passim).^[6] While there are traces left of other declensions (Burrow 1937: §§68–72, pp. 27–29), only the a-declension still enjoys large productivity in Niya Prakrit. The paradigm of this declension is exemplified below, using the word bhuma- ‘land’ as a model; not all forms given are actually attested for the word bhuma-, and infrequent variant endings are indicated in between brackets.

The abl.sg ends in -ade (-āde), where the occasional spelling with a long vowel is probably to be taken seriously given the exact cognates in Śaurasenī -ādo and Mahārāṣṭrī -āo;^[7] no abl.pl is attested so far, though one would expect it to take the same form as the ins.pl, i.e. -ehi (cf. Burrow 1937: §63; p. 26). The loc.sg in -a(ṃ)mi derives from the OIA pronominal ending -asmin; the old nominal ending in -e is only rarely attested in Niya Prakrit, and the same holds for the loc.pl in -eṣu (cf. Section 4.2).

3 The postposition paride ‘from’ and the synthetic ablative

In this section, I aim to show that there is a complementary distribution between the Niya Prakrit postposition paride and the synthetic ablative: they are used in the same function, viz. to indicate a source; but the former is used with animate nouns, while the latter is restricted to inanimates. Since the postposition paride ‘from’ has received but scanty attention in the literature (Burrow 1937: 41), I will in addition elaborate on its semantics and etymology.

4 The postposition vaṃti ‘in, at, to’ and the synthetic locative

In this section, it will be shown that animacy does not only affect the marking of ablatival relationships in Niya Prakrit. The synthetic locative is restricted to inanimates as well, whereas for animate nouns, locatival relationships are expressed with the postposition vaṃti ‘in, at, to’.

5 Animacy and the paradigm of the personal pronouns

As discussed in a recent article (Schoubben 2022: 3–7), the paradigm of the first and second person pronouns in Niya Prakrit consists of a threefold system with a direct case form (e.g. dir.sg ahu ‘I’ < aham); a genitive/dative case form (e.g. gen./dat.sg mahi ‘I’ < mahya(m) and gen./dat.sg mama ‘I’ < mama); and some remnants of instrumental case forms (e.g. ins.sg maya ‘I’ < mayā). What has not been discussed in that article is the complete lack of ablatives or locatives in the paradigm of the first and second person pronouns. This absence is noteworthy, though, given that other MIA languages, including Apabhraṃśa, do inflect personal pronouns in the ablative and locative.^[41]

We can now account for the lack of such forms in Niya Prakrit as another instance of animacy governing the case system. Since first and second person pronouns are animate by necessity (cf. e.g. Santazilia 2023: 22–23), they are expected to have lost synthetic forms for the ablative and locative in a language that restricts those to inanimate nouns. Indeed, the postpositional constructions with respectively paride ‘from’ and vaṃti ‘in, at, to’ are attested with pronominal complements in the genitive,^[42] thus supplying an analytically formed ablative and locative for the personal pronouns. Such a distinction between core cases and peripheral cases in the paradigm of personal pronouns is also typologically common. The Australian language Dyirbal, for instance, is another language where pronouns cannot be put in the ablative or locative case because the latter are restricted to inanimates (Aristar 1997: 319, 337; Dixon 2022: 30).

Third person pronouns can be either animate or inanimate, and for these, synthetically formed ablatives and locatives do occasionally occur in Niya Prakrit. Since, however, locatives are rare, and not much can be added to Burrow’s discussion of them (1937: §§80–82, pp. 33–35),^[43] I shall focus here on ablatives.

To begin with, some comments are due on tade, the synthetic abl.sg of the 3rd person pronoun sa-/ta- ‘this’ (Burrow 1937: §80, pp. 33–34). Leaving badly intelligible examples^[44] and cases where tade can just mean ‘then’ (cf. Skt. tataḥ) aside,^[45] tade normally means ‘from it’ or ‘from there’.^[46] Like nouns inflected in the ablative case (Section 3.2), tade is thus inanimate by definition; and when 3rd person pronouns have animate referents, the postposition paride ‘from’ is used to express ablatival relations, e.g. tasa (tas̱a) paride ‘from him’ (CKD 140 cr5; 719 o4) or edeṣa paride ‘from them’ (CKD 189 o2). In one special construction, i.e. when tade is used in a partitive sense dependent on the numeral eka ‘one’ or the adjective avaśiṭha- ‘the remaining’, tade seems indifferent to both gender and number; compare tade … eka ‘one of those’ (CKD 106 uo7), tade eka arivag̱a ‘one guide from those’ (CKD 253 r1), tade avaśiṭhe ‘the rest of those (children)’ (CKD 279 o3), tade eka uṭa ‘one camel of these’ (CKD 359 r1), and tade vaḍ̱avi 1 ‘one mare from them’ (CKD 509 uo3). paride, by contrast, is not found governed by a numeral, which may explain why tade is used for animates (and plurals) in this particular construction.

The usage of imade (also iṃmade), the synthetic abl.sg of the near-deictic pronoun i-/ima- ‘this’ (Burrow 1937: §82, pp. 34–35), is more straightforward: it is always used in an adverbial sense ‘from here’ and thus inanimate by definition.

Much the same is true about the adverbially used adehi (also atehi) ‘from there’ (Burrow 1937: §91, pp. 40–41), which supplies a far-deictic equivalent to imade. Etymologically, adehi consists of ade < ataḥ, abl.sg to the pronoun a- ‘that’, plus an element -hi, thus parallelling (Jaina-)Māhārāṣṭrī iohi ‘from here’ < itaḥ + -hi.^[47]

6 A note on Khotan Prakrit sag̲aj̲i ‘in, at, to’

As already noted in Section 1 (fn. 4), Niya Prakrit is not the only known dialect of Central Asian Gāndhārī; in addition, we have Kuča and Khotan Prakrit. In the former, the postpositions discussed in this article are attested as well, but neither has so far popped up in Khotan Prakrit.^[48] Because of the fairly limited corpus available for this dialect (CKD 661; 843), this may be due to chance. With Noble (1931: 453), I would, however, like to observe that in (21) the Khotan Prakrit postposition sag̲aj̲i /saγāźi/ occupies the same syntactic slot as vaṃti does in Niya (and Kuča) Prakrit. The noun phrase governed by sag̱aj̱i encodes the recipient (∼ a dative) of the prs.1sg vikrināmi ‘I sell’, just like in (13) a noun phrase governed by vaṃti was used to express the recipient of pst.3pl vikridati ‘they sold’.

Another parallel between Khotan Pkt. sag̱aj̱i and Niya Pkt. vaṃti lies in their etymology. As discussed in Section 4.1, vaṃti derives from the OIA relational noun upānte ‘in the proximity of’, loc.sg of upānta-; similarly, sag̱aj̱i goes back to the OIA relational noun sakāśe, loc.sg of sakāśa-, also meaning ‘proximity’. Compare further the Pāli postpositions santike ‘in the presence of > with, before’ and samīpe ‘in the proximity of > near’, cited in Section 4.1.

While dependent on the discovery of additional Khotan Prakrit materials, it thus seems possible that the expression of the locative case for animate nouns is an isogloss separating Niya from Khotan Prakrit (cf. also Schoubben 2024: 402–403; 523–526).

7 Concluding remarks

To summarise, the Niya Prakrit postposition paride ‘from’ has been grammaticalised as a marker of ablatival relationships and is always accompanied by a genitive complement right in front of it. This complement being animate by default, paride fulfils the role of an animate ablative in the language, whereas the synthetic abl.sg in -ade (-āde)^[50] has come to be restricted to inanimate nouns. In the same way, the postposition vaṃti ‘in, at, to’ should be characterised as a marker of animate locatives, while the synthetic loc.sg in -aṃmi and the loc.pl in -eṣu typically select inanimate complements. As a result, the paradigm of the Niya Prakrit a-declension given in Section 2 only applies to inanimate nouns such as bhuma- ‘land’ and not normally to animate nouns such as maṃnuśa- ‘man’ (< OIA manuṣya-), whose ideal paradigm would rather be as follows:

Consequently, one can say that the direct, the genitive, and the instrumental are the three core cases in Niya Prakrit, while the ablative and the locative seem to be peripheral.

The pervasive role animacy plays in the nominal morphology of Niya Prakrit makes this language an outlier in the spectrum of MIA languages we know of. As briefly hinted at in Section 3.1, Reinöhl found that the outcomes of madhye ‘in the middle of’ and upari ‘above, on top’ only inconsistently select animate complements in Pāli and Apabhraṃśa, and that the grammaticalisation process towards case markers was fully completed only in 16th-century Old Avadhī. Thus, we seem to have here another instance where Niya Prakrit parallels developments found in other Indo-Aryan languages, “but at a much earlier period” (Jamison 2000: 65).

Sometimes, grammatical differences between Niya Prakrit and other MIA languages can be attributed to contact with neighbouring languages (cf. Schoubben 2024: Ch. 5). Yet, in the present case, no convincing source seems to be at hand. Animacy plays a certain role in the case system of Tocharian A and B, where there is differential object marking (DOM) in that the obl.sg of thematic nouns only ends in -ṃ when the referent is [+human] (e.g. TA oṅkaṃ ‘man’; TB eṅkweṃ ‘id.’).^[51] This, however, is hardly to be compared to the Niya Prakrit ablative and locative.^[52] In Bactrian, too, an exact comparandum is lacking; in Kuṣāṇa Bactrian, plural nouns governed by prepositions take the obl.pl ending in -ανο /ānə/ for animates and the dir.pl ending in -ε /e/ for inanimates (Sims-Williams 2004 [2008]: 66), but this is again too different from the Niya Prakrit pattern to have been its source.^[53]

If not language contact, which other factor may account for the Niya Prakrit case system behaving differently from that of other MIA languages? I submit that it is the difference in text genre. Niya Prakrit is known from administrative documents, whose language is “remarkably varied and flexible” (Jamison 2000: 64).^[54] Other MIA languages are mainly attested via literary compositions – texts whose language does not normally reflect the spoken vernacular (cf. Masica 1991: 51), and where influence from prestige languages like Sanskrit and Māhārāṣṭrī often obscures linguistic developments (cf. Bubeník 1998: 16–19).^[55] If, then, postpositions are not yet fully integrated into the case system of these MIA languages, including Apabhraṃśa, and if animacy plays a less decisive role than in Niya Prakrit and in NIA, this may tell us more about what authors wanted the language of their literary texts to look like than about the actually spoken vernacular.^[56] In other words, postpositional case marking could have come into being at a (much?) earlier point in the history of Indo-Aryan than the evidence from literary Prakrits would suggest.