§1. Nominal Themes.

Although historically Pāli inherits its inflexion from the older language the original themes on the whole are continually being replaced by later ones mostly derived from oblique cases, a process seen clearly in the consonantal declension. Original vowel stems however are inflected as they are except in the dipthongal and ṛ-declensions. In the former beside a few historical forms (PLS §88) cases formed from a new stem derived from original oblique cases are frequently found: e.g. nāvā M I.134 “ship”, nom. sg. from a stem nāvā- (< Skr. acc. sg. nāvaṃ, inst. Nāvā); gavassa M I.429 “of the cow” dat.-gen. sg., abl. sg. gavā D I.201, loc. gave Sn 310 from a stem gava- (< Skr. inst. sg. gavā etc.); a stem gāva- is also found in gāvī f. nom. sg. A IV.418, Ud 8,49, the long ā being due to the pl. forms (< Skr. nom. pl. gāvaḥ etc.). In the latter (ṛ-declension) a few cases occur from derived stems: e.g. satthārā inst. sg. D I.163 from a stem satthārā- (< Skr. acc. sg. satthāraṃ). In the gāthā literature there is evidence for a stem in -u (from the base exhibited in the Skr. abl., gen., sg.) e.g. satthuno dat.-gen. sg. Sn 547,573; Th 1.131 (PLS §90). But Geiger (loc. cit.) explains it as being due to the fact that in compounds the ṛ of the stem appears as in Pāli. But the consonant stems, especially those identical with roots, are comparatively rare in the Nikāyas due to the phonetic law of the falling off of the final consonant in Pāli as in Prk. (cp. PLS §75). Only a few historical forms survive which point to consonant stems. Of these the new themes are mainly formed in two ways:

1. The final consonant of the Skr. nom. sg. which is either the stem terminal or its phonetic variant is elided and the stem thus vocalized is inflected according to the corresponding [2] vowel declension of that gender, thus: (a) Radical stems: parisā- from Skr. pariṣad f. “assembly” e.g. parisāyaṃ loc. sg. D II.218; parisāsu pl. S II.27; It 64; vijju- from Skr. vidyut f. “lightning” e.g. vijju nom. sg. S I.100; A I.124; (b) an-stems: brahma- from Skr. brahman m. e.g. brahmaṃ acc. sg. M I.2, 328; muddha- from Skr. mūrdhan m. “head” e.g. muddhaṃ acc.sg. Dh 72; Sn 987; D I.95; also neuter stems kamma-, pabba-, etc., from Skr. karman, parvan (PLS §94 for instances); (c) in-stems: seṭṭhi- from Skr. śreṣṭhin m. “treasurer” e.g. seṭṭhissa gen. sg. S I.90; hatthi- from Skr. hastin m. “elephant” e.g. hatthī nom.pl. S I.211; sāmi- from Skr. svāmin m. “lord” e.g. sāmiṃ acc. sg. Sn 83; cakkavattissa M III.176 gen. sg.; (d) s-stems (Skr. -as, -is, -us,): mana- from Skr. manas nt. “mind” e.g. manaṃ acc. sg. S IV.7; manassa dat.-gen. sg. S IV.17; mane loc. sg. A II.158; S I.40, also manasmiṃ S V.171; raja- from Skr. rajas nt. “dust” e.g., rajena inst. sg. M I.25; rajassa dat.-gen. sg. Sn 406; sira-, ura-, teja- from Skr. siras, uras, tejas, e.g. sirasmiṃ M II.75; urasmiṃ A I.141; tejasmiṃ A V.319 loc. sg. cp. tamā tamaṃ Sn.278; tapena Sn 655.

In this declension a nom. sg. in -o is frequently found, with the masculine ending -o of the adjective or participle in agreement, pointing thereby to a change of gender. e.g., tamo vihato M I.22 “the darkness is destroyed”; mano anicco S IV.1 “the mind is impermanent”; mano dukkho That dukkha is used adjectivally is seen from the preceding cakkhuṃ dukkhaṃ etc. S IV.2 “the mind is ill”; tejo pātukato M II.184 “the fire is kindled”; cp. mano supaṇihito Sn 155 “the mind is well directed”. But that this change of gender is a later phenomenon due to the influence of the preceding -o on the adjectival ending as seen from the above examples is shown by the fact that when the adjective precedes the noun the original gender is preserved. e.g., santaṃ tassa manaṃ hoti Dh 96. is-stems: sappi- from sarpis and others (vide PLS §75); us-stems: cakkhu- from Skr. cakṣus nt. “eye” e.g. cakkhuṃ nom. sg. S I.115; M III.136.

2. Themes are derived from the bases exhibited in original oblique cases, particularly the acc. sg. by virtue of its frequent use, thus: (a) From original root stems: vācā- from Skr. vāc [3] (< acc. sg. vācaṃ) f. “speech” e.g., vācāya inst. sg. D I.114; vācā nom. pl. M III.76; D III.18; vācānaṃ gen. pl. Sn 454; pāda- from Skr. pād (< acc. sg. pādaṃ) m. “foot” e.g., pāde loc. sg. A II.144; pāde acc. pl. Sn 573; (b) From an- stems as in: rañña- a sporadic stem from the weakest Skr. base rājñ- m. “king” (cp. Skr. inst. rājñā, loc. sg. rājñi etc.) e.g., raññe loc. sg. D III.83; nāma- from Skr. nāman nt. “name” (cp. acc. sg. nāma) e.g., nāmena inst. sg. D II.154; similarly attena inst. sg. M I.297; II.263; S IV.54; dāmena A III.383 also damena S IV.163, 282; (c) From in- stems: vāsina- from Skr. vāsin m. “dweller” (< acc. sg. vāsinaṃ) e.g., vāsine acc. pl. D II.272; similarly, palokine acc. pl. Th 2.101 from Skr. pralokin; pāṇine acc. pl. Sn 220; verinesu loc. pl. Dh 197; (d) From nt- stems: āyasmanta- from Skr. āyuṣmant m. “venerable one” (< āyuṣmantaṃ acc. sg.) e.g., āyasmante loc. sg. S I.56; III.133; āyasmantānaṃ gen. pl. M I.64; similarly, arahante loc. sg. M I.254; mahantasmiṃ loc. sg. A I.148; bhavantānaṃ gen. pl. M II.148; here there is also a new stem ending in -ata formed from the Skr. weak stem (-at) found in sg. oblique cases; e.g. arahataṃ acc. sg. A II.182 (yatra hi nāma taṃ Bhagavantaṃ arahataṃ sammā-sambuddhaṃ āsādetabbaṃ).

§2. Archaic Adverbs.

As remarked above there are a few historical forms in Pāli which without exception can be traced back to Vedic (cp. R.O. Franke, Pāli und Sanskrit, p.150 et seq.). Some of these have lost their inflexional value in Pāli and come to be regarded as adverbs or prepositions, and, according to commentators, even as particles or indeclinables (nipāta). This is chiefly characteristic of genuine adverbial cases like the acc. and inst., and to a lesser extent of the abl., gen. and the loc. as well. In the Nikāyas 18 such forms occur mostly as adverbs of time and place and rarely of manner, viz. a.) from original acc.: uddhaṃ “above” D I.23, 153, 251; II.293, 294; III.104; A III.323; V.109; Sn 894; acc. sg. of Vedic ūrdhvá-; ciraṃ “for a long time” Sn 678, 730, from Vedic cira-; alaṃ “rightly” M I.130; S II.18, from Vedic áraṃ acc. sg. of an obsolete stem ára- “sufficient” [4] (VGS §178); nattaṃ “by night” Sn 1070, from Vedic acc. sg. náktaṃ (VGS §178.2); nāma “by name” or “namely” S I.33, 235; Sn 157, 177, from Vedic acc. sg. nma of nāman nt. “name”; raho “secretly” M II.251; III.157, from Vedic ráhas acc. sg. of ráhas; khippaṃ “soon” or “quickly” A II.118; III.164; Sn 413, 682, 998; Dh 65, 137, 236, 289, from Vedic adj. kṣípra (VGS §197.5.b.); sayaṃ “by -self” D I.12; Sn 57, 320, from Vedic svayáṃ originally nom. sg. of svá- (VGS §115.a.). b.) from original inst.: dva “by day” S I.183; M I.125; Dh 387, from Vedic inst. sg. dva; micchā “wrongly” Sn 438, 815 (vide P.T.S. Dict. s.v.), from Vedic inst. sg. mithuy found as mithy in the Brāhmaṇas (VGS §199.6.a.); sahasā “forcibly” Sn 123; A II.209, from Vedic inst. sg. sáhasā (VGS §178.3); musā “falsely” D 1.52 from Vedic inst. sg. mṣā “by or with neglect”.c.) from original abl.: pacchā “after, afterwards” D I.205; Sn 645, 773, 949; Dh 172, 314, 421, from Vedic abl. sg. paśct; ārā “far, far from” Sn 156, 736; Dh 253, from Vedic ārt (VGS §178.5.). d.) from original gen.: cirassa “since long”, “after a long while” e.g., na cirass’ eva D III.11; sucirass’ eva S I.193 also cirassaṃ in same sense D I.179; S I.142, where the final nasal is due to the analogy of the frequent use of acc. ending -aṃ as adverb; divassa used adverbially in compound with divā (see b.) e.g., divā-divassa “at noonday” The sense of “early in the day”, “at sun-rise”, given to this compound by the P.T.S. Dict. (s.v. divā) is obviously erroneous. S I.89; A.V.185; e.) from original loc.; ratto “by night” Sn 223; Th 2.312; Dh 296, from Ved. loc. sg. rtrau. Pāli has two archaic dat. forms cirāya “for long” Dh 342 and svātanāya “for tomorrow” D I.125 which however do not occur as such, that is to say adverbially, in the older dialect. Compare, however, Epic Sanskrit cirāya and the acc. adv. śvastanam (Monier-Williams, Skr.-Eng. Dict., s.v.).

§3. Dual Forms.

The main inflectional peculiarity of Pāli and Prk. as compared with Vedic and Classical Skr. is the loss of the dual. Its place is taken by the plural in all declensions (vide PLS §77.1.). So we find it with all names of things by nature considered in pairs such as eyes, ears, hands, legs etc. e.g., hatthe dhovati [5] M II.138 in place of hastau dhovati in the earlier language; similarly in dvandva-compounds usually expressed by the dual in Skr.: e.g., ime pi candimasuriye evaṃ mahiddhike etc. (acc. pl.) M I.69 and candimasuriyānaṃ (gen. pl.) D I.10. According to Geiger (PLS §77) dve and ubho are the only regular dual forms existing in Pāli. The latter is the normal masculine form (nom. and acc.) corresponding to Skr. ubhau but the former is only the feminine or neuter form, which is due to the fact that the nom. and acc. forms of numerals were used without discrimination for all genders in Middle Indian (cp. Pischel, Prk. Gr. §438). So it is found with masculine nouns. e.g., dve dhammā D II.60; dve pabbajitā D I.57. The form duve occurs only in gāthā literature e.g. Th 1.245 (vide PLS §114). These forms however exist only sporadically for the dual and as a number no more influences Pāli declensions. This is attested by the oblique case-forms of these in use as the loc. ubhosu (tīresu) S III.137; Sn 778; inst. ubhohi (hatthehi); gen. ubhinnaṃ S I.62, which are formed simply by adding the normal plural endings to a stem *ubha- in the first two examples and the gen. appears to be formed on the analogy of other numerical forms (dvinnaṃ, tiṇṇaṃ, catuṇṇaṃ etc.), beside the proper nom. ubho S I.87; A III.48; It 16, 43; Sn 661. There are however three or four other instances where we meet with probably the dual forms of dvandva-compounds. The first of these occurs in kasirena ghāsacchādo labhati A I.107, where the Burmese MS. (Ph) has the v.l. ghāsaccaṃ which is evidently a later ‘correction’. At A III.85 the above reading of the P.T.S. text is repeated, but once at A III.385 the phrase occurs as kasirena ghāsacchādo labbhati, the passive form of the verb showing that ghāsacchādo is here regarded as the masculine nom. sg. of ghāsacchāda-. But the compound consists of two masculine words ghāso “food” = Skr. ghāsaḥ and acchādo “clothing” = Skr. ācchādaḥ, and as such, must be treated either as a dual or collectively as a neuter sg. The proper construction then would be either ghāsacchādaṃ labbhati or ghāsacchādā labbhanti, the pl. being employed for the dual. So the above (A III.385) reading with the passive (sg.) is ruled out, establishing the first reading (A I.107) ghāsacchādo [6] labhati as correct. Consequently the ending -o would represent the older masculine dual in -au, as in ubho (< ubhau). There is however another alternative, that is, we may possibly have here an earlier ghāsacchāde acc. pl., the -e having been later regarded as an eastern form and changed to -o. The other occurs in the phrase natthi hāyanavaḍḍhane natthi ukkaṃsâvakaṃse “there is no high and low, there is no increase and decrease”. These occur at least twice in the Nikāyas viz., at S III.212 and M I.518 and so cannot be misprints. If these coordinative compounds are taken as neuter sg. the form may be the eastern -e, but if, as is quite possible, they are used as plurals then the ending -e represents the dual nt. nom., subject of atthi which can agree with any number. Another instance of a similar doubtful character is āyasmante, voc. M I.474, which probably is an eastern form standing for āyasmanto corresponding to the Skr. dual āyaṣmantau (vide §10). There is greater probability in accepting the pronominal form etc. (not cited by Geiger, PLS §107.1) occurring at Sn. 869, 870, as a neut. dual. acc. since it clearly refers to sātaṃ asātaṃ ca “what is pleasant and what is unpleasant”.

Inflexional Terminations

§4. Stems in -a (m.&nt.).

In the singular all the historical endings are retained, with the phonetic changes peculiar to Pāli, except the dat. which has been superseded by the gen. form -ssa. The older ending -āya, however, appears in the Nikāyas quite a number of times but it has there almost completely lost its original significance and in the few instances attested, denotes only aim, direction or purpose (vide §§96,113,106 & 107; cp. PLS §74). The reason for this replacement is to be found in the fact that already in the earlier dialect the dat. by virtue of its syntactical character had come into logical contact with the gen. in many of its proper functions. So, even as early as in the period of the Brāhmaṇas (VGS §202.B.2.a.), the latter had encroached upon many uses of the former and in later Skr. almost ousted it from its legitimate [7] sphere of employment (SS §80). This process is seen also in the Prākṛts where the -āya form as represented by its phonetic developments -āa, āya, -āe etc. (Pischel Prk.Gr. §363) occurs mostly in the artificial dialect of dramatic poetry (SS p.100 f. n.).

§5. Pronominal Endings.

The influence of the pronominal declension is found in the endings -smā and -mhā which exist beside the normal -ā (Skr. -āt) of the abl. sg. and -smiṃ, -mhi beside the historical -e of the loc. sg..

a. An analysis of the Dīgha- and Majjhima Nikāyas has shown that the form -smā occurs only 4 times (leaving aside the repetitions) against some 95 of the -ā form. Of the latter 21 denote cause, -smā being never used in that sense in spite of the causal implication of the pronominal adverbs kasmā “why”, tasmā “therefore” and yasmā “wherefore”. It is also significant that it is always the -ā form that is used in syntactical agreement with the 6 prepositional adverbs ‘governing’ the abl., viz., yāva, aññatra, tiro, uddhaṃ, adho and paraṃ. e.g.,

yāva c’aggā yāva ca mūlā D I.75; M II.170; III.12
“from top to bottom”;

aññatra avusitattā D I.90
“except from imperfection”;

tiro raṭṭhā tiro janapadā M II.167
“across country and province”;

uddhaṃ pādatalā adhokesamatthakā D III.104; M III.90
“above from the soles of the feet and down from the top of the head”;

paraṃ maraṇā M III.101
“after death”.

The forms in -smā and -mhā are confined to one particular syntactical category, viz., the abl. of separation (in the wider sense). So the former (-smā) mainly occurs in connection with the verb pabbajati “sets out”, especially in the stock phrase

agārasmā anagāriyaṃ pabbajati
“he sets out from home to homelessness”

e.g., D I.18, 60, 115, 202; II.16, 230; III.31, 147; M I.200, 267, 345, 459; II.66, 181; III.261. It also occurs in gāthā literature e.g., Sn 1002, 1003, and with another verb of motion at S I.185, i.e. agārasmā anagāriyaṃ nikkhantā; also with the causative pabbājeti “expels” at D I.92, i.e. raṭṭhasmā pabbājesi. The historical form -ā however is the [8] more popular even here, occurring in the Dīgha- and Majjhima Nikāyas over 25 times with about 15 different verbs of motion. The verb pabbajati itself occurs twice with the -ā form, viz., Sakyakulā pabbajito M II.167, and once with its causative: raṭṭhā vā nagarā vā pabbājeyyuṃ D I.90, 91. In all the Nikāyas the -smā form appears only with 4 other verbs all of which signify detachment, release, or aloofness. e.g.,

anissaṭā bhavasmā Ud 33
“not free from becoming”;

gaṇasmā vūpakaṭṭho M III.110; Ud 41; A IV.435
“detached or aloof from the crowd”;

virato methunasmā D II.241 (verse)
“abstaining from copulation”;

na parimuccanti dukkhasmā M I.8, 65
“are not released from sorrow”.

The form -mhā which is its phonetic development is even rarer in the Nikāyas, occurring only 3 times in the Dīgha and Majjhima, viz. Naṅgaramhā pāyāsi M II.119; Rājagahamhā niyyāsi D I.49 and

muñjamhā isīkaṃ pabbāheyya M II.17
“would draw out the reed from the muñja grass.”

These examples show that -mhā too is used only for the notion of separation, particularly with verbs of motion denoting that from which there is a movement. It is evident therefore that the pronominal endings were confined, in the case of the abl., to its function of signifying separation, while the historical ending -ā was still in the Nikāyas the popular form for all its varied uses in general.

b. The pronominal form of the loc. sg. in -smiṃ and its phonetic variant -mhi are not so rare in the Nikāyas as the corresponding abl. forms. The former is by far the more frequent of the two and occurs in almost every syntactical function of that case. e.g.,

lokasmiṃ viharati D I.23
“he lives in the world”;

veyyākaraṇasmiṃ bhaññamāne D I.46
“while the explanation was being declared”;

sīlasmiṃ hoti D I.65, 66, 67
“is part of (his) virtue”.

The latter -mhi however is less frequent and is mostly used in verse, decidedly metri causa. e.g., vanamhi jhāyato Sn 221; setamhi chatte anuhīramāne D II.15 (verse) and the intermediate phonetical stage is also found in verse. e.g.,

antalikkhasmi S I.67
“in the intermediate space or sky”.

The v.1. -asmiṃ which appears in one text is not in keeping with the metre. [9]

§6. Inst. sg. in -ā.

With regard to the inst. sg. it has to be remarked that beside the usual form in -ena a form in -ā corresponding to Vedic -ā (inst. sg. of a-nouns, m. & f.) occurs many times in the Nikāyas (cp.PLS §78). Franke has conclusively shown that such forms represent the inst. and not the original abl. sg. in -āt (Z.D.M.G. 1892, pp.313-315). It occurs with both masculine and neuter nouns, especially in the frequent phrase

sahatthā santappesi
“served or fed with his own hand”.

e.g., M I.393; II.50; A I.274; D I.109; Sn p.107 etc. which the Comy. glosses in most places by sahatthena (e.g., ‘sahatthâti sahatthena’ Manorathapūraṇī II.372; Sumaṅgalavilāsinī I.277). It occurs but once in the Nikāyas outside this context i.e. in

na sahatthā paṭhaviṃ khaṇati M II.51
“he does not dig the ground with his hand”.

Here too the Comy. has asahatthena (Sum. III.814). This ending however is not restricted to the above word. In

mā sokā pahato bhava Th 1.82
“do not be overcome by grief”,

sokā is definitely the inst. sg. denoting means (cp. PLS §78). Since the abl. in -ā (< Skr. -āt) does not occur in this function, that is, to signify means in general or instrument, in the Nikāyas, we may regard the following as representing original inst. in -ā of neuter verbal nouns: dassanā pahātabbā M I.7 et. seq.; bhāvanā pahātabbā M I.12;

vinodanā pahātabbā
“should be got rid of by ...”

M I.12. In all these examples the verb pahātabbā would require an inst. of means (by which) rather than an abl. of cause (through which), since effort on the part of the agent is implied. Similarly in viriyā nimmathitaṃ padhānâbhinibbattaṃ M II.130 the sense prompts us to regard the -ā as inst. sg., Chalmers translating it correctly as “kindled by effort and fired by striving”, the preposition by implying means and not cause. In sahatthā referred to above also the inst. denotes means and is not due to a preposition saha which Franke (loc. cit.) thought is here contracted to sa-. On the other hand the compound stands for *svahastā, sa- being the reflexive pronominal adjective Vedic or Skr. sva. When this inst. occurs with saha, the preposition meaning with or together with, the sense implied is simultaneity or association (vide [10] Sociative Inst. §64). e.g.,

saha parinibbānā D II.156; S I.159
“simultaneously with the passing away”;

saha vacanā Ud 16
“simultaneously with the word”

i.e. “as he spoke” (cp. Geiger, PLS §78.1. “zugleich mit dem Wort, im Augenblick, wo er es sagte”).

§7. The Ending of the Acc. Pl.

In the plural of the a- declension Pāli differs from Skr. in the acc. and dat. The historical ending -ān of the acc. is lost due to the fact that, since phonetically it becomes -ā by the falling off of the terminal consonant, it is liable to be confused with the nom. pl. in -ā. The form in -e which is the regular acc. ending in Pāli and Prākṛt is borrowed, as Geiger suggests (PLS §78.3.), from the pronominal declension, where the original masculine acc. pl. -tān took the form of the nom. -te because it had lost its accusative character through the dropping of -n and in order to distinguish it from the feminine -tāḥ which too would give in Pāli -tā. But Geiger has drawn attention to one solitary survival of the -ān form appearing in gāthā literature, viz., in the phrase vehāsān-upasaṅkamiṃ Th 1.564.

§8. The Inst. Pl. in -e.

In the inst. pl. the regular form is -ehi, the phonetic development of Skr. -ebhis. It has been shown that the aspirate bh in Pāli is retained when it is in the body of the word but is generally reduced to -h- in inflexional endings (cp. R.L.Turner, The Phonetic Weakness of Terminal Elements in Indo-Āryan, J.R.A.S. 1927, p.277). Nevertheless the intermediate form -ebhi also occurs, though not frequently, mostly in archaic instances (cp. Geiger PLS §79). e.g., ariyebhi Dh 162; Ud 6. The same is found in other declensions. e.g., jhāyibhi jhānasīlibhi M III.13. The inst. pl. in -ais which is in fact the older of the two in Old Indian has come to be gradually lost even in Vedic (vide Macdonell, VGS §78.f.n.). It has left no trace either in Pāli or in Prk. owing to the confusion with acc. pl. in -e, except for one [11] solitary instance. The form dhīro occurring at Dh 207, it has been suggested by V. Lesný (A new reading of the Dhammapada 207, J. p.T.S. 1928), stands for dhīre, the older inst. pl. In all the MSS. of the Pāli version of the Dhammapada the reading is:

bālasaṅgatacārī hi dīgham addhāna socati,
dukkho bālehi saṃvaso amitten’ eva sabbadā,
dhīro ca sukhasaṃvāso ñātīnaṃ va samāgamo.

“Verily he who walks in the company of fools suffers for a long time; living with fools is always painful as with an enemy; living with the pleasant is wise, like meeting with kinsfolk”. The italicized words give the literal rendering of the phrase dhīro ca sukhasaṃvāso and the Comy. (Dhammapadaṭṭhakathā P.T.S.Vol.III. p.272) too follows the same reading but does not comment on the form dhīro. However, as has been pointed out so cleverly by Lesný such a translation does not indeed make good sense, though grammatically there can be no objection to it. It is evident from the parallelism with the first part of the second line, viz. dukkho bālehi saṃvāso, that the reading should be either sukho ca dhīrasaṃvāso, as Max Müller suggested, or more likely dhīre ca sukhasaṃvāso, as Lesný takes it. The latter is supported by the Kharoṣṭhī version which attests to the fact that the second part is -sukhasaṃvāso ( ... suhavasa ñātihi va samakamo, 39, Les fragments Dutreuil de Rhins, par Emile Senart. Journal Asiatique 1898, p.297), and not dhīrasaṃvāso, and also by the Skr. text which fills in the lacuna in the Kharoṣṭhī version by the inst. pl. dhīrais (dhīrais tu sukhasaṃvāso, XXX.26, L. de la Vallée Poussin Documents sanscrits de la seconde collection M.A. Stein J.R.A.S. 1912, p.369). Geiger (PLS §79) has instanced another place where the inst. pl. -e is authentic (Buddhavaṃsa 2.32, guṇe dasah’ upāgataṃ) but the -e forms given by E. Müller in his Simplified Grammar of the Pāli Language as inst. pl. used with the sense of the dat. (such as yācake etc.) are not however instrumentals but only the loc. sg. (-e) denoting the person to whom something is given or offered. [12]

§9. The Dat. Pl. Ending.

The dat. pl. in Skr. is the same as the abl. pl. ending in -(e)bhyas which is retained in Pāli for the latter (abl. pl.) of a- stems as -ehi, which has been shown to be a phonetic development of *ebhio (< *ebhiyo), the aspirate being reduced to - h - as described above (§8). Pandit S. Majumdar Sastri in a monograph entitled ‘The Dative Plural in Pāli’, on the evidence of some survivals of the old dat. pl. in -ehi in the Asokan dialect, suggested the possibility of a few of these forms remaining in Pāli where the form normally used is the ending -ānaṃ of the gen. which as pointed out above (§4) is due to the syntactical displacement of the dat., in the pl. as in the sg., by the gen. But a close investigation of the Nikāyas shows that no certain vestiges of an historical dat. pl. exists in Pāli. There are however some instances of the -ehi form the sense of which seem to be bordering on that of the dat. (or the abl.). In the frequent phrase yāvadeva manussehi suppakāsitaṃ D II.113,114,219; III.122 etc., which Rhys Davids (Dialogues 11.113) translated as “until in a word it shall have been well proclaimed among men”, manussehi can be syntactically the dat. denoting the persons to whom something is proclaimed (vide §93 b.c.d.). There is also the reading Yāva devamanussehi Ud 64 which is supported by the Comy, on D III.122 ‘deva-lokato yāva manussa-lokā suppakāsitaṃ’ and also by the Buddhist Skr. parallel at Divyâvadāna 201 ‘yāvad-deva manuṣyebhyaḥ’. Whatever the reading may be it is an open question whether the ending -ehi here represents an older dat. (pl. -ebhyas) agreeing with the verb ‘suppakāsitaṃ’ or an abl. construed with yāva taken as a preposition. But if the latter be the case the rendering would be “proclaimed up to or as far as (gods and) men” which however does not make good sense. On the other hand if yāva is taken merely as the adverb meaning “completely” (cp. Rhys Davids, “in a word”) or “just” as found in alaṃ vo taṃ yāvadeva sītassa paṭighātāya D III.130 “just enough to stand the cold”, the phrase makes satisfactory sense. Accordingly it is quite probable that what we have here is an old dat. pl. We are confronted with a similar difficulty in the [13] case of -ehi in

ayaṃ bhikkhave uppatti asādhāraṇā puthujjanehi A II.128
“this birth, monks, is not common to worldings”.

The adj. sādhāraṇa in Pāli as well as in Skr. is capable of being construed with either the gen., dat. or inst. (vide Monier-Williams’, Dict, s.v.) but with the inst. its sense is usually “equal” because here a comparison is implied. When, however, the sense is “common to” as in the above example the dat. appears syntactically the more suitable construction. The -ehi ending therefore may here possibly stand for the older dat. pl. rather than the inst., preserved because of the option in the construction.

§10. Eastern Forms.

Among the sporadic forms of the above (a-) declension we may group the so-called Māgadhisms under which Geiger (PLS §80) includes the nom. and voc. sg. in -e both masculine and neuter. In the Dīgha- and Majjhima Nikāyas there are 6 such nom. forms of masculine nouns and 8 of neuters. The existence of these eastern forms can be justified on the ground that all these are put into the mouth of one or the other of the six leaders of heretical schools whose dialect was naturally some kind of eastern Prākṛt. Moreover it is significant that they are clustered together in passages of philosophic importance reported to have been said by them. These statements occur in the Sandaka Sutta of the Majjhima- and in the Sāmaññaphala Sutta of the Dīgha Nikāya. They are: bāle ca paṇḍite ca kāyassa bhedā ucchijjanti D I.55; M I.515,518; doṇamite sukhadukkhe, pariyantakaṭe saṃsāre natthi hāyanavaḍḍhane natthi ukkaṃsâvakaṃse M I.518; D I.54; ājīvasate, paribbājasate, nāgâvāsasate, vīse indriyasate, tiṃse nirayasate M I.517-518; D I.53; sattaguḷe khitte nibbeṭhiyamānaṃ eva paleti M I.518; kamme ca aḍḍhakamme ca M I.517; sukhe dukkhe jīve satt’ ime M I.517 with the less accurate reading sukhe dukkhe jīvasattame D I.56. These statements are repeated in a discourse by the Buddha at S III.211 (§§5,6&7). Also in the Sunakkhatta Sutta of the Majjhima Nikāya the -e form is 5 times used by the Buddha in a talk with Sunakkhatta, the Licchavi. Here too they occur in a passage of [14] philosophic importance, viz., ye lokāmisasaṃyojane se pavutte M II.254; ye anañjasaṃyojane se bhinne M II.255; ye ākiñcaññâyatanasaṃyojane se vante M II.255; ye nevasaññānâsaññāyatanasaṃyojane se ucchinne ucchinnamūle tālavatthukate anabhāvakate āyatiṃ anuppādadhamme M II.255. There are three other instances of the -e form outside the above context, viz., ke ca chave sigāle, ke pana sīhanāde ti? D III.24, where probably it is due to the fact that the phrase is borrowed from popular speech as an exclamatory metaphor conveying a sense of disparagement; ye āyatane veditabbe S IV.98, which the Comy. takes as nom. sg. (‘tasmā ye āyatane veditabbe ti taṃ kāraṇaṃ jānitabbaṃ ti attho’ Sāratthapakāsinī 391, v.l. veditabbo); idha pana bhikkhave bhikkhu ... tasmiṃ ca sukhe anadhimuchite (for -to) hoti M II.223 “here, monks, a bhikkhu is not infatuated in the matter of that happiness”.

Geiger has instanced the voc. sg. (in -e) in ehi tvaṃ samma Bhesike D I.225 which he regards with Pischel (Prk. Gr. §366.b) as a nom. used in address as voc. In fact Pāli like Prk. has sometimes the actual nom. sg. instead of the voc. (-a) in addressing. e.g., kin nu kho āvuso bho Gotamo taṃ jīvaṃ taṃ sarīraṃ udāhu aññaṃ jīvaṃ aññaṃ sarīraṃ D I.157 (cp. Ardha-Māgadhī voc. sg. putto, Prk. Gr. §363). We have also the reverse case where the form in short -a is used for the nom. sg. in -o. e.g., Kahan nu kho bho Nāgita etarahi so bhavaṃ Gotama viharati ... ? D I.150, which may be either due to eastern influence (cp. Ardha-Māgadhī nom. sg. Buddha-putta for Buddha-putto, Prk. Gr. §364) or the sandhi form of the original Gotamaḥ with the dropping of the visarga. In the voc. pl. of āyasmā beside the regular āyasmantā and āyasmanto we find a form -ante used in addressing two persons. e.g., āyasmante (voc. pl. or dual) M I.474. If this be a dual form corresponding to Skr. -antau, standing for āyasmanto, the -e can be regarded as being due to eastern influence. Such influence is positively seen in the archaic nom. pl. ending -āse (Geiger PLS §79.4, for examples) which is the eastern form for Pāli -āso from Vedic -āsas. There are a few eastern forms in the pronominal declension also (vide §16). [15]

§11. Sporadic Forms of the a- Declension.

In the above paragraph we have referred to the archaic ending -āse of the nom. pl. masculine which represents the eastern derivative of the Vedic double ending -āsas, both feminine and masculine. In Pāli however this ending is never found with feminine nouns (PLS §79).

In the neuter of the a- declension there are a few remnants of the older Vedic plural of the nom. in -ā, beside the regular -āni. e.g., rūpā Th 1.455; D I.245; sotā Sn 345; nettā Th 2.257 etc. On the analogy of the masculine inflexion a neuter acc. pl. -e is formed, (m. nom. pl. -ā: m. acc. pl. -e = nt. nom. pl. -ā: nt acc. pl. e). e.g., rūpe passituṃ Ud 30; rūpe ca pajānāti M I.61; rūpe paṭicca S IV.18. This is also found with the verbal nouns in -naṃ, all being used in the plural thus removing the possibility of their being Māgadhī nt. sg. acc. in -e. e.g. nīvaraṇe pahāya D I.73; Sn 17, beside nīvaraṇāni (vide P.T.S. Dict. s.v.). But sometimes masculine adjectives are found used with them. e.g., cattāro satipaṭṭhāne bhāventi M II.11; showing that the identity of forms had later on given rise to change of gender.

§12. Feminine in -ā.

In the feminine ā- declension the older historical endings of the inst., dat., abl., and gen. have been replaced by -āya which is also used for the loc. beside the normal -āyaṃ. The ending -āya seems to be a later phonetic development of the Skr. abl. -gen. -āyas, the -ḥ- being dropped owing to the phonetic law already mentioned (vide §1.) and the shortening of the final vowel being due to the general phonetic weakness of terminal elements in Middle Indian as referred to (§8.). The replacement of the dat. both in the sg. and in the pl. by the gen. form is due to the same syntactical phenomenon as discussed in the case of the masculine declension (§4.). The older inst. -ayā is also lost being replaced by the abl. sg. -āya due to similar syntactical reasons (vide §§62,116 & 118). But a considerable number of inst. fem, in -ā, as in the masculine and neuter declensions, is [16] found in the Nikāyas. e.g., saddhā pabbajitvā M I.16,123 “leaving (home) through or by faith”, beside saddhāya gharā nikkhamma Sn 337; tassā issā na supati Sn 110 “he does not sleep through jealousy for her”; assavanatā dhammassa D II.38; M I.168 (Comy. ‘assavanatā ti assavanatāya dhammassa’ Sum.II.467); vyārosanā paṭighasaññā Sn 148 “through anger and hatred”; ekapuggalassa bhikkhave kālakiriyā bahuno janassa anutappā hoti A I.22 “owing to the death of one person there is worry for many people”. It is however difficult to say whether this ending -ā corresponds to the older Vedic inst. in -ā of feminine nouns as in doṣ barháṇā etc., or is a phonetic contraction of -āya (cp. Prk. -āa). Geiger (PLS §§27.2, 81) is inclined to favour the latter possibility though Franke thought it was definitely the Vedic ending -ā of feminine inst. sg. (vide: Inst, auf -ā von a-stämmen im Pāli, Z.D.M.G. 1892 pp.313 et seq.). Even the form -ā in abhiññā sacchikatvā D II.92,153, beside abhiññāya desitā D II.119 can be inst. sg. of means, though Geiger (§27.2) regards it as a contraction of the gerund in -āya after the Comys. (e.g., ‘tad abhiññā ti tad abhijānitvā’ Sum. I. p.59).

§13. The Vowel Declension (-hetu).

Of the sporadic forms belonging to this declension the form hetu (abl. sg. m.) is interesting owing to its peculiar syntactical function. It is evidently a phonetic development from Skr. hetos, abl.-gen. sg. of hetus m. “reason or cause”, the dropping of the final - h -, and the reducing of the vowel -o to -u being due to the phonetic peculiarities of such terminal elements as described in the foregoing paragraph. As to the weakening of the vowel we may compare sajju (<*sajjo) corresponding to Skr. sadyas (vide Geiger, PLS §§22&23). It is mostly used as a postposition denoting cause in which case it appears as a periphrase for the inst. or the abl. of cause. e.g.,

attahetu parahetu dhanahetu Sn 122
“because of oneself, others or wealth”;

na kho, Udāyi, etassa sacchikiriyāhetu bhikkhū mayi brahmacariyaṃ caranti M II.37
“it is not, Udāyi, due to [17] (the intention i.e. for the purpose of) realizing this ... that monks live the Holy Life under me”;

kāyassa pīṇanahetu M II.191
“for (lit. because of) the pleasing of the body”.

As seen from the rendering of the latter examples -hetu implies not only cause but purpose as well. It may not appear, however, always as postposition in a compound. There are many instances where it is used as a separate word agreeing with a gen. of the noun or pronoun which denotes the material cause implied. e.g.,

puttadārassa hetu M II.187
“lit. from the cause of son and wife”

i.e. “due to or for the purpose of son and wife”; yesaṃ hetu labhāmase Kh 6 “owing to whom, lit., because of whom, we acquire ... (cp. Comy. ‘ye nissāya yesaṃ kāraṇā’ Paramatthajotikā II. p.210). From these it is evident that what we have in the stock phrases taṃ kissa hetu D I.14; M I.1; A II.31, “why is it?”, lit., “because of what is it? and kissa hetu A III.303, IV.393; Sn 1131, is an abl. sg. hetu and a gen. of the pronoun (kissa, cp. kissa nirodhā taṇhā nirodhôti D II.33, where too kissa is gen. sg. “of what” and nirodhā is abl. similar to hetu). The suggestion that -hetu may be an elliptical form of the acc. sg. hetuṃ (vide P.T.S. Dict. s.v.) is therefore unwarranted, cp. SS. 193. f.n. l. where he argues that inspite of Pāṇinī’s rule śaṣṭīhetuprayoge (2.3.26) a comparison with I.E. idiom shows that hetoḥ in the phrase ‘kasya hetoḥ’ is abl. & not gen. and that kasya is gen.

§14. The Consonantal Declension (parisatiṃ).

In the feminine parisā-, originally belonging to the consonantal declension (< Skr. pariṣad), the historical form parisati corresponding to Skr. pariṣadi occurs quite a number of times. e.g., D III.18; A II.180. Here the replacement of - d - by - t - is probably due to the influence of other original consonantal stems like sarit- (e.g., acc. sg. saritaṃ Sn 3) where in Pāli beside a nom. ending in a vowel (cp. sarī parallel to parisā oblique cases are found with a - t - . This is however not a sporadic phonetic change peculiar to Pāli as Geiger suggests (PLS §39.4). In the examples adduced by him viz. kusīta, mutiṅga and pātu- (Skr. kusīda, mṛdaṅga and prādur) it is not quite certain which [18] form is the earlier. The first two are most probably loan-words in Indo-Aryan and the etymology of the last is uncertain. This form which is the loc. sg. is sometimes found with a final anusvāra as parisatiṃ. e.g.,

parisatiṃ dhammaṃ deseti M II.140
“he preaches the doctrine in (or to) the assembly”;

so Rājagahe parisatiṃ evaṃ vācaṃ bhāsati A I.185
“At Rājagaha he tells these words to the (or in the) crowd.”

At another place it occurs with the masculine pronoun. e.g.,

sādhu te pañca dhamme imasmiṃ parisatiṃ bhāsassūti M II.199
“well, declare to (or in) this assembly the five dhammas”.

Here we have a v.l. imissaṃ parisati. The appearance of the anusvāra is probably due to the syntactical fact that verbs of speaking sometimes agree with an acc. of the person to whom the words are addressed (vide §§36.b,58.c.). The proper loc. significance of the historical form parisati being lost due to its archaic nature the construction was replaced by the more popular idiom, viz., the acc. with verbs of speaking. So the acc. ending -ṃ is added to a theoretical stem parisati- (f.). The fact that the loc. form, whether historical or later, is preserved when there is no actual verb but only the participle also strengthens the validity of our surmise. e.g.,

bhāsitā kho pana te es’ avuso Pāṭika-putta Vesāliyaṃ parisati vācā D III.18;
“Were these words spoken by you, friend Pāṭika-putta, at Vesāli among the rabble?”;

parisāyaṃ bhāsato D II.218,
“speaking in the assembly”.

For it is to be generally observed in Pāli concinnity that the loc. appears in such adnominal instances in place of an acc. which is the more usual in the adverbal construction.

The Pronominal Declension Supplementary to Geiger’s presentation (PLS §§104-112).

§15. The Enclitic Forms.

Whereas in Vedic and Classical Sanskrit the enclitic forms me, te sg. are found only for the dat. and gen., no, vo pl. are found only in the acc., dat. and gen., Pāli like Prākṛt has extended their use to other cases as well. (vide Pischel, Prk. Gr. §420; acc. sg. me; inst. sg. me; acc. pl. ṇo, ṇe; inst. pl. ṇe). Though [19] not infrequently, the forms me and te occur as accusatives in Pāli. e.g.,

te ekena khaṇena ekena muhuttena ekamaṃsakhalaṃ ekamaṃsapuñjaṃ karissāmi M I.377
“In a flash, in a moment, I shall reduce you to one mash, one mass of flesh”.

(cp. Ardha-Māgadhī and Śaurasenī acc. sg. te, Prk. Gr. §421). Franke has also given as acc. sg. in Pāli the forms me and te, (vide, Pāli und Sanskrit, p.152). This employ seems to have originated in the contact between the uses of the acc. and the inst. as in the following causative construction where the causative verb pāpetu can take either the acc. or the inst. of its primitive subject that which would have been its subject in the original non-causative state (vide §59). e.g.,

sādhu me bhante Bhagavā tapojigucchāya aggaṃ yeva pāpetu sāraṃ yeva pāpetûti D III.48
“may the Blessed One make me attain to the summit, to the essence of disgust-for-asceticism”

where me can be either the acc. or the inst. In the pl. no and vo are similarly found for the acc. e.g.,

upāsakā no bhavaṃ Gotamo dhāretu M I.413
“may the venerable Gotama take us as disciples”;

pahāya vo gamissāmi D II.120
“I shall go leaving you”;

āmantayāmi vo D II.156
“I address you”

(cp. Māgadhī and Śaurasenī acc. pl. vo, Prk. Gr. §422). In the inst. there are definite examples where the form me is used for the agent. e.g.,

maggaṃ kho me gacchantena A IV.334
“by me going the way”;

kammaṃ pana me karontena A IV.334
“by me doing an action”;

akatena me ettha kataṃ M I.515 lit.
“by me not doing it is done, i.e., without my doing any task is done”;

mūḷhena me evaṃ kataṃ M II.248
“thus done by my deluded self”.

It is also found agreeing with feminine nouns. e.g.,

suto yeva me ... upanaccantiyā D II.268
“was heard by me ... (while) ... dancing”.

Buddhaghosa regards me in the stock phrase evaṃ me sutaṃ D I.1 as standing either for the inst. or the gen. of agency (vide §154). He says: “me saddo tīsu atthesu dissati: Tathā hi ’ssa gāthâbhigītaṃ me abhojaneyyaṃ ti ādisu mayā ti attho (i.e. inst.). Sādhu me bhante Bhagavā saṇkhittena dhammaṃ desetûti ādisu mayhaṃ attho (i.e. dat.). Dhammadāyādā me bhikkhave bhavathāti ādisu mama ti attho (i.e. gen.). Idha pana mayā sutanti mama sutanti ca atthadvaye yujjati”. (Sum. I . p.28, Papañcasūdanī I. p.4.). There is no doubt therefore that the form [20] me was from very early times used as an inst. sg.. This extension of its use seems to have originated in the agent-use of the gen. forms me and te, which is a frequent construction in Pāli. e.g., api ca m’ettha puggalavemattatā viditā D II.152; Sn p.102 (Comy. ‘api ca mayā ...’ Pj.II.2.436); taṃ kiṃ maññasi gahapati, sutaṃ te ... evaṃ bhante sutaṃ me M I.378; Samaṇo me Gotamo nimantito Sn p.104; te ca me evaṃ puttha D I.192;III.28; etaṃ me abhipatthitaṃ D II.266 (Comy. ‘etaṃ mayā abhipatthitaṃ’ Sum.III. p.702); bhāsitā me esā vācā D III.54; kicchena me adhigataṃ D II.36; yan te karaṇīyaṃ Sn p.39; punar āyu ca me laddho D II.285(V.); sahitaṃ me asahitaṃ te ... āropito te vādo M II.3; taṃ me idaṃ bhante Bhagavā sakkhi diṭṭho M I.370. In all these examples the gen. is as permissible to denote the agent as the inst.. It is probable that the me here originally stood for the gen. but later on came to be regarded as the inst. of agency. Finally, in two instances no and vo appear in the role of nom. plurals. e.g.,

yaṃ no Bhagavā dhammaṃ bhāsissati taṃ no sossāmati M II.5
“what doctrine the Blessed One will preach to us, that we shall hear”;

mā vo muñcittha koci naṃ D II.262
“Ye let not one escape, whoever it be”.

In both these examples there is the alternate possibility of the enclitic being used merely as emphatic particle, but in taṃ no sossāmâti the emphasis, if no implies such, is not needed according to the context, though vo in the other instance may have an emphatic sense.

§16. Sporadic Forms.

Among the sporadic forms of the pronominal declension we may place the nom. pl. amhā formed on the analogy of the a- declension (nominal acc. pl. -e: pronominal acc. pl. -e: nominal nom. pl. -ā: pronominal nom. pl. X). e.g.,

tena ca amhā attamanā M II.132,177
“thereby we were pleased”.

Similarly on the analogy of the a- declension the relative pronoun yo has a dat. sg. masculine yāya. e.g., yāya eva kho pana atthāya D I.90, beside the usual yassa, which is the gen. form used as dat. e.g., yassa atthāya M I.392. In the inflexion of the demonstrative pronoun a gen. pl. sānaṃ f. “of those (women)” occurs [21] beside the regular tāsaṃ or tāsānaṃ. e.g., bāḷhā me dukkhā vedanā abhikkamanti no paṭikkamanti abhikkamo sānaṃ paññāyati S V.80,345. A few archaic forms of the reflexive pronoun sa, (Skr. sva) “one’s own”, occur mostly in the gāthā literature. e.g.,

saṃ ñātiṃ atimaññati Sn 104
“he disparages his own relatives”;

nihīno sena mānena Sn 132
“devoid of his own pride”;

samhi āsane D II.225
“in his own seat”;

sehi dārehi Sn 108; sehi dhammehi Sn 298.

Finally, it may be mentioned that a few eastern forms have crept into the pronominal declension as into the nominal. e.g., ye for yad or yaṃ, and se for tad or tam. These are also found in the passages already referred to (§10). e.g., Tattha yañce savitakkaṃ savicāraṃ, ye avitakke avicāre se paṇītatare D II.278; evam eva kho, Sunakkhatta, sammā-nibbānâdhimuttassa purisa puggalassa ye nevasaññānâsaññâyatana-saṃyojane se ucchinnamūle ... M II.256 (cp. Prk. Gr. §423). The same form se occurs also in the frequent adverbs seyyathā D I.145 for tad + yathā “just as, such as” and seyyathîdaṃ D I.89;II.91; S V.421; It 99 “as follows” for tad + yatha + idaṃ. The eastern form ye for yaṃ is found also in compound yebhuyyena D I.17;II.139, which is made up of Skr. yad and bhūyas. The interrogative ke for ko m. sg. also occurs in one of the above-mentioned passages (vide §10).