Pramāṇa · Search · About · API

A History of Sanskrit Literature c29: XXI. Grammar

A History of Sanskrit Literature c29

Language
English

Contents

XXI. Grammar

1. The Beginnings of Grammatical Study

c29.s1.1

IN the Brāhmaṇas of the Vedic period we find sufficient proof¹ that, as in Greece, grammatical study in India began with consideration of such points as pronunciation and euphonic combination, and the discrimination of parts of speech which gives us terms such as *vibhakti*, case termination, *vacana*, number, *kurvant*, present tense. Possibly hence it derived its name Vyākaraṇa, though that is often deduced from the later practice of analysis of forms. We find already in Yāska the terms *nāman*, noun, *sarvanāmam*, pronoun, *ākhyāta*, verb, *upasarga*, preposition, *nipāta*, particle.² The next stage is not represented in the Brāhmaṇas, but is fully in being in Yāska’s time; it consists of the analysis of forms, as opposed to the reckless etymologies of the Brāhmaṇas and Plato; we do not know how this came to be arrived at, though it is a plausible conjecture which finds the motive in the fact that in compounds in Sanskrit the first word appears in its stem form without terminations. From this it was fairly easy to distinguish stem and termination in nouns, and then to advance to distinguish in verbs root, terminations and tense and other affixes, and to arrive at the doctrine of the derivation of nouns from nouns by Taddhita suffixes, of nouns from verbs by Kṛt suffixes. A further step was to declare as did Çākaṭāyana that all nouns are derived from verbs, to which Gārgya objected that if this was so, it followed that every thing should have as many names as it had activities, and every name should apply to everything which had the activity it connoted. But the supporters of Çākaṭāyana carried out their principle, and to this period goes back in substance, not in its present form, the *Uṇādisūtra*, containing words which are derived from verbs by unusual affixes, and which in some form Pāṇini evidently knew.

c29.s1.2

¹ See Wackernagel, *Altind. Gramm.*, i, pp. lix ff.; Oldenberg, *Vorwissensch. Wissenschaft*, pp. 79 f., 238 ff.

c29.s1.3

² See Lakshman Sarup, *The Nighaṇṭu and the Nirukta*, pp. 54 ff. Cf. Prabhat-chandra Chakrabarti, *Linguistic Speculations of the Hindus* (1924–5); S. Varma, JRAS. 1925, pp. 21 ff. (on analysis of meaning).

c29.s1.4

This important period of studies was largely concerned with the preservation and interpretation of the Vedic texts; its work is seen in the preparation of the Padapāṭha of the *Ṛgveda* by Çākalya, who is known to Pāṇini, the similar work done on other Vedic texts, the Prātiçākhyas, which in their original form were probably older than Pāṇini, at any rate as far as concerns those on the *Ṛgveda*, *Taittirīya* and *Vājasaneyi Saṁhitās*,¹ and the Çikṣās, which as we have them are probably later than Pāṇini, but doubtless existed in his time, proving the care taken to secure due correctness of pronunciation of the scriptures. But the grammarians were clearly concerned also with the Bhāṣā, the spoken speech of the day, and it was in connexion with it, especially as it grew more distinct from the sacred texts on the one hand and the speeches of the lower classes on the other, that secular grammar grew up. Pāṇini knew and cites by name many predecessors, including Çākaṭāyana, Āpiçāli, and Çaunaka, as well as minor names, and his allusion to easterners and northerners, if it applies to forms of speech used in these parts, is also testimony to the existence of grammarians to note them, unless we are to assume² that he himself of the north lived in the east and noted the differences for himself, which is implausible in the extreme. What is clear from Pāṇini's own work is that he summarizes the efforts of many previous writers, from whom we may be sure he borrowed his form as well as many facts.

c29.s1.5

¹ Cf. Liebich, *Einführung in die ind. einheim. Sprachwissenschaft*, ii. 35 ff., with Keith, HOS. xviii, pp. xxxix-xli, clxxi.

c29.s1.6

² Franke, GGA. 1891, pp. 957, 975 ff.

2. Pāṇini and his Followers

c29.s2.1

The *Aṣṭādhyāyī*³ of Pāṇini consists of about 4,000 short Sūtras divided into eight books, treating of technical terms and rules of interpretation (i), nouns in composition and case relations (ii); the adding of suffixes to roots (iii) and to nouns (iv, v), accent and changes of sound in word formation (vi, vii) and the word in the sentence (viii). But this scheme is constantly interrupted, rules being interpolated illogically because it was convenient to do, or because space could thus be saved, for the whole book is dominated by the aim to be as brief as possible. The, to us, illogical order and impracticability of learning Sanskrit by the use of the grammar are explained, if we remember that the book was to be learned by heart by those who were already accustomed to use Sanskrit in conversation, and had not to learn how to speak it, but to know what forms were correct, what vulgar. Of the incoherence, however, part is doubtless due to the fact that Pāṇini was only working up a mass of traditional matter, as may be seen not merely from certain irregularities in case usage,¹ but also from the employment to denote a Vedic use of three terms, *chandasi*, *nigame*, and *mantre*, of which the first prevails with his followers. The main object of the grammar is to deal with the Bhāṣā, the living speech of the day; an amount of Vedic matter is incorporated. This part is of uneven value, suggesting that it was based on a number of special studies, imperfectly co-ordinated; thus minute details from the *Kāṭhaka* or *Maitrāyaṇīya Saṁhitās* are noted; but at other times a vague reference is made to Vedic irregularities, Vedic words are cited without analysis, and causeless variations of form are permitted as Vedic.

c29.s2.2

³ Ed. and trans. O. Böhtlingk, Leipzig, 1887; Śrīśa Chandra Vasu, Allahabad, 1891-8.

c29.s2.3

¹ Cf. Weber, IS. xviii. 508 ff.

c29.s2.4

The principle underlying the grammar is the derivation of nouns from verbs, Pāṇini avoiding dealing with the hard cases by alluding to the Uṇādi list existing in his time. All derivation is done by affixes, and, therefore, when the word agrees with the root form of a verb, or one nominal form is the same as that whence it is derived, it is necessary to assume suffixes which are invisible, e.g. *badara*, fruit of the *badara* tree. Phonetics do not receive investigation save incidentally as changes of words occur in processes of derivation. But in this field Pāṇini, or more correctly his predecessors, achieved very remarkable results, as in the postulate of Guṇa and Vṛddhi changes, of forms with long *r̥* vowel, roots in *ai*, *masj* as the original of *majj*, dive, *s* as the ending of inflexions. The analysis of forms is normally carried out with great acumen; it is very rare to find such a phenomenon as the periphrastic future, e.g. *kartāsmi*, treated as a simple verbal form. In comparison with the work of Greek grammarians Pāṇini is on a totally different plane in this regard. The suggestion that he and his predecessors were creating a language, or that the forms which are not recorded in earlier literature are not to be accepted as prima facie valid, is now definitely disposed of.

c29.s2.5

To secure the brevity aimed at many devices are adopted; the cases are used pregnantly, verbs are omitted, leading rules are understood to govern others which follow; above all algebraic formulae replace real words; the rule that a vowel is changed into the corresponding semi-vowel when a vowel, not itself, follows is denoted by *iko yaṇ aci*; the last Sūtra *a a* denotes that *a* which has been treated in the grammar as an open letter, corresponding to which we have long *ā*, is really a closed letter pronounced like *u* in ‘but’. Older than Pāṇini are probably some technical terms of ungrammatical make-up, such as *parasmaipada*, active, *ātmanepada*, middle, *napuṁsaka*, neuter; others are reduced forms of the original as *it* to denote a letter not pronounced, appended to a word to indicate some feature regarding its treatment, from *iti*, so. The use of such Anubandhas is doubtless before Pāṇini, as the term Uṇādi itself proves.

c29.s2.6

Pāṇini’s date is unhappily uncertain.¹ He was later than Yāska and Çaunaka, probably he came after not only the Brāhmaṇas but also the older Upaniṣads and was alive during the Sūtra period of Vedic literature, but unhappily these facts give us nothing save a relative chronology. We know he was a native of Çalātura near the modern Atak, where Hiuen Tsang saw a statue to his memory; his mother was Dākṣī, and a legend ascribes his death to a lion. His connexion with the north-west is important, when we find in his work Yavanānī, meaning probably Greek (Ionian) writing. We may, of course, scent an interpolation, and, if so, the word is valueless. If not, it leaves us still in doubt, for the assumption that it is a proof that Pāṇini wrote after the invasion of Alexander the Great, though it has been supported² by the occurrence in the *Gaṇapāṭha* of the names Āmbhi and Bhagala, Omphis and Phẹgelas, is clearly unfounded, seeing that India was in contact with Greece as early as the expedition of Xerxes. On the whole, however, it seems needless to carry back Pāṇini beyond the fourth century; if he flourished *c.* 350, then Kātyāyana, who may be placed *c.* 250–200, might easily have found sufficient divergence of speech to justify his corrections. There are, indeed, proofs that language had changed, as we have seen; but to assign Pāṇini to the sixth or seventh century B. C. on that score appears to lack any plausibility.

c29.s2.7

¹ Keith, HOS. xviii, pp. clxviii f.; *Aitareya Āraṇyaka*, pp. 21 ff.; Lüders, SBA. 1919, p. 744; Liebich, *Pāṇini* (1891); Kielhorn, GN. 1885, pp. 185 ff.; Wecker, *Bezz. Beitr.* xxx. 1 ff., 177 ff. A date *c.* 700–600 is claimed by Belvalkar (*Systems of Sanskrit Grammar*, p. 15; cf. Bhandarkar, JBRAS. xvi. 340 f.; Keith, IOC. ii. 242.

c29.s2.8

² Lévi, JA. 1890, i. 234 ff.

c29.s2.9

Kātyāyana probably lived in the third century B. C.,¹ though no strict proof is possible, and this date really depends on the fact that he apparently did not long precede Patañjali; the impression left by Kātyāyana's Vārttikas is certainly that sometimes, not by any means always, he is attacking or correcting Pāṇini on the score of differences in usage which had arisen between the time of the two, while with Patañjali it seems as if he and Kātyāyana were parted by no great interval of time. Kātyāyana was not a captious critic of Pāṇini; he was not the first to call in question his rules; what he did was to examine criticisms, rejecting some, accepting others, and therefore supplementing and limiting Pāṇini's rules. But, while we need not treat him as hostile, he seems not to have been sorry to find Pāṇini in error. Patañjali, whose *Mahābhāṣya* has preserved us Kātyāyana's Vārttikas of some 1,245 Sūtras, takes up Kātyāyana's criticisms, and in many cases defends Pāṇini, but by no means as a matter of course. Moreover, he carries out in great measure his predecessor's work by examining other Sūtras of Pāṇini and correcting or explaining them. It is clear that Patañjali had many criticisms and works before him beside that of Kātyāyana; there are Vārttikas in verse which need not all have been Kātyāyana's, and Kārikās, memorial verses which probably are by various hands, including Patañjali himself; the variety of metres used in these verses is remarkable, including some later quite rare, but complex, metres. Among others Patañjali mentions Vyāḍi, of whose work—the *Saṁgraha*—much has been conjectured but very little is known, Vājapyāyana, Pauṣkarasādi, Goṇikāputra, and Gonardīya, with whom he was formerly erroneously held to be identical.¹

c29.s2.10

¹ Jayaswal's arguments for a period 248–200 (IA. xlvii. 138; xlviii. 12) from Vārttika, ii. 1. 60 are invalid. For his style see V. G. Paranjpe, *Le Vārtika de Kātyāyana* (1922), who claims an earlier date; cf. Smith, EHI. p. 470.

c29.s2.11

¹ Kielhorn, IA. xv. 81 f.; xvi. 101 f.; GN. 1885, pp. 189 ff., who postulates a considerable period between Kātyāyana and Patañjali; *Kātyāyana and Patañjali* (1876).

c29.s2.12

Our information regarding the personality of Kātyāyana and Patañjali is negligible. Kātyāyana, however, either bore the alternative name of Vararuci or was early confused with a person of that name, and to a Vararuci many works are ascribed, including the first extant Prākrit Grammar, *Prākṛtaprakāça*; book iv of the *Kātantra* and the *Liṅgānuçāsana*²; the *Vārarucasaṁgraha*,³ twenty-five Kārikās on case construction, compounds, verbs, and nominal formation; a lexicon; the Vedic *Puṣpasūtra*; and stanzas of poetry. As Patañjali mentions a *Vāraruca kāvya* we may believe in a poet Vararuci of early date, but we need not identify him with the author of the Kārikās. His identity with the author of the *Prākṛtaprakāça* is most implausible, as the Prākrit of that work is very late in character, and we may assume that the other attributions are of no value. Late tradition makes Vararuci a contemporary of Pāṇini, and also a minister of the Nandas of Pāṭaliputra; Kumāralāta⁴ actually confirms this point, but even if this poet existed, it proves nothing for the grammarian, as Kumāralāta speaks only of a poet. Of more value is Patañjali's proof that Kātyāyana was a southerner.

c29.s2.13

² Liebich, *Einführung in die ind. einheim. Sprachwissenschaft*, i. 11. See Winternitz, GIL. iii. 391.

c29.s2.14

³ Ed. TSS. 33, 1913.

c29.s2.15

⁴ *Sūtrālaṁkāra*, trans. E. Huber, p. 88.

c29.s2.16

Patañjali is regarded as an incorporation of the snake Çeṣa, Viṣṇu's resting-place during his slumber, and he is believed to be the author of the *Yoga Sūtra*, a view implausible on grounds of certain grammatical slips by the latter and slight deviations in philosophic terminology, apart from the fact that the tradition is very late and obviously due to likeness of name.⁵ His date⁶ is still disputed. The evidence for it is that statements in his grammar undoubtedly refer to a sacrifice for Puṣyamitra, whose reign began *c.* 185 or 178 B.C., and to a recent attack on Sāketa and Madhyamikā by a Yavana, who is very plausibly identified with the Greek Menander (*c.* 156–153), dates which give *c.* 150–149 for the composition of the work, on the assumption probable, yet not conclusive, that the references are Patañjali's own. Some slight confirmation may be gained from the fact that Kātyāyana, but not Pāṇini, notes the title *devānām priya*, famed in Açoka's inscriptions, suggesting that he fell after 250 B.C., which would not suit ill with 150 B.C. for Patañjali. If this be rejected, we must content ourselves with noting that Kalhaṇa records a revival of the study of the *Mahābhāṣya* in Kashmir under Abhimanyu, whose date, however, we do not know, and that Bhartṛhari (*c.* 650) proves long study of the text before his time.

c29.s2.17

⁵ Cf. Woods, HOS. xvii, pp. xv ff.; Jacobi, GGA. 1919, pp. 14 ff.; DLZ. 1922, p. 271.

c29.s2.18

⁶ Cf. Smith, EHI. pp. 227-9; Winternitz, GIL. iii. 389; Bühler, *Die indischen Inschriften*, p. 72; Keith, IOC. ii. 243 f.

c29.s2.19

The *Mahābhāṣya*¹ is interesting stylistically as giving us a lively picture of the mode of discussion of the day. A question is posed; an Ācāryadeçīya deals with it, not altogether incompetently but not quite satisfactorily, and an Ācārya solves the issue. The style, therefore, is lively, simple, animated, and as in Açoka's inscriptions—possibly a confirmation of the date proposed—not rarely do we find the question 'Wherefore?', 'How?', or 'What?' put and then answered. Proverbial expressions and references to matters of everyday life are introduced and serve both to enliven the discussions and to give us valuable hints of the conditions of life and thought in the time of Patañjali, who thus is a source of information for religious and social history as well as for literature. A good example of his style is afforded by a famous reference² to the Mauryas: Pāṇini has a rule providing for the addition of the suffix *ka* to a name to denote an image of the person, but adds that it is dropped if the image is used to secure a livelihood (*jīvikārthe*) and is not vendible (*apanya*). Patañjali says: *apanya ity ucyate tatredaṁ na sidhyati Çivaḥ Skando Viçākha iti. kiṁ kāraṇam? Mauryair hiraṇyārthibhir arcāḥ prakalpitāḥ. bhavet tāsu na syāt. yās tv etāḥ saṁprati pūjārthās tāsu bhaviṣyati.* 'The difficulty is raised, with regard to Pāṇini's proviso that images are not to be vendible, that on this doctrine the forms Çiva, Skanda, Viçākha, are incorrect. Why is that? Because the Mauryas, in their greed for money, used as means images of the gods (i.e. they bartered them, so that the forms should be Çivaka, &c.). (Final answer.) Very well, granted that the rule for dropping *ka* does not apply to those images of the Mauryas; still as regards images now used for purposes of worship it does apply.’ It will be seen that the amount which must be understood to make such passages intelligible is rather a strain on the reader,¹ and in point of fact the *Mahābhāṣya* evidently gave serious trouble to later students. Bhartṛhari, who died *c.* 651, wrote a commentary on it which is all but lost, and also the *Vākyapadīya*,² in three books of verse, which mainly deals with questions of the philosophy of speech; a difficult work, it contains much evidence of thorough knowledge of contemporary philosophical disputes. Kaiyaṭa’s³ commentary on the *Mahābhāṣya*, which may belong to the twelfth century but which tradition places earlier, borrows largely from Bhartṛhari, and is itself commented on by the voluminous writer Nāgojī Bhaṭṭa (*c.* 1700). Both show that they often had as much trouble to understand Patañjali as have we.

c29.s2.20

¹ Ed. Kielhorn, BSS. 1906 ff.

c29.s2.21

² Bhandarkar, JBRAS. xvi. 206 ff.

c29.s2.22

¹ Cf. B. Geiger, *Mahābhāṣya zu P. vi. 4. 22 und 132* (SWA. 1908).

c29.s2.23

² Ed. with Puṇyarāja’s comm., BenSS. 1887–1907; Kielhorn, IA. xii. 226 ff.; Pathak, JBRAS. xviii. 341 ff.

c29.s2.24

³ Bühler, *Report*, pp. 71 f.; Peterson, *Report*, i, p. 26.

c29.s2.25

Save for Bhartṛhari, Patañjali closes the line of great grammarians. We do not doubt that he drew on the speech of his day; his preface insists on the absurdity of learning words that are not used, and like Kātyāyana he views Pāṇini in the light of a living language. Thereafter use is made of the three great grammarians, efforts are made to explain them, or to re-expound their systems for purposes of more effective exposition, but nothing is done to restate the facts of language with reference to living speech. For reasons which we cannot certainly explain, the authority of Pāṇini and his immediate followers prevailed; deviations from his rules were even in great poets like Kālidāsa deemed to be errors.

c29.s2.26

One commentary of Pāṇini deserves praise for its extent of information, its comparative clearness, and its evidence of changes in Pāṇini’s text, the *Kāçikā Vṛtti*⁴ of Jayāditya and Vāmana, which was written before I-tsing visited India, when he found it regularly used by Chinese in order to study Sanskrit grammar, and when he records its school use by boys for five years after attaining fifteen years of age. Books i–v seem to have been Jayāditya’s; presumably Vāmana finished it by reason of his death. A comment on it was written *c.* 700 by the Buddhist Jinendrabuddhi, and the *Nyāsa* referred to by Māgha¹ appears to be this book. Another Buddhist, Çaraṇadeva, wrote in 1172 under the supervision of Sarvarakṣita a *Durghaṭavṛtti*,² dealing with the difficult passages of Pāṇini’s text. Among his many citations are three verses of the *Jāmbavatīvijaya* of a Pāṇini, whose identity with the grammarian we may safely dismiss as unproved. But Pāṇini could not teach Sanskrit, for which end his grammar was not written. For such purpose re-writing and re-arrangement were essential, giving us (*c.* 1400) Rāmacandra’s *Prakriyā-kaumudī*,³ based on which is Bhaṭṭoji Dīkṣita’s well-known and not unsatisfactory *Siddhāntakaumudī*,⁴ on which he wrote a comment, the *Prauḍhamanoramā*. From it come two school grammars of Varadarāja, *Madhyasiddhāntakaumudī* and *Laghu-kaumudī*.⁵

c29.s2.27

⁴ Ed. Benares, 1898; B. Liebich, *Zwei Kapitel der Kāçikā* (1892); on Pāṇini’s text, Kielhorn, IA. xvi. 178 ff.

c29.s2.28

¹ ii. 112. Ed. by Srish Chandra Chakravarti (Rājshahi, 1914 ff.), see i. 47, 48 on the authorship of the *Kāçikā*. On it is based Puruṣottamadeva’s *Bhāṣāvṛtti* (*c.* 1150); ed. 1918.

c29.s2.29

² Ed. TSS. 6, 1909.

c29.s2.30

³ S. C. Vidyabhusana, JPASB. 1908, pp. 593 ff.

c29.s2.31

⁴ Ed. Bombay, 1882. Date seventeenth century.

c29.s2.32

⁵ Ed. and trans. J. R. Ballantyne, Benares, 1867.

c29.s2.33

As we have seen, Pāṇini presupposes an *Uṇādisūtra*⁶; our extant text contains late words like *dīnāra* or *mihira* and omits some, e. g. *pāntha* mentioned by Patañjali; Çākaṭāyana or Vararuci is given as author. The *Dhātupāṭha* goes back in substance to Pāṇini; it gives the roots according to classes, with indicatory letters containing information regarding their formation; on it are based the *Dhātupradīpa* of Maitreyarakṣita, the *Daiva* of Deva, and the *Puruṣakāra*, a joke on the name, by Kṛṣṇalīlāçuka,⁷ who is later than Hemacandra, and the *Mādhavīya Dhātuvṛtti*⁸ ascribed to Sāyaṇa’s brother Mādhava in the fourteenth century. The *Gaṇapāṭha* has been interpolated, and Vardhamāna’s *Gaṇaratnamahodadhi*⁹ (1140) is not based on it but on some other grammar. Rules on accent, Vedic and classical, are dealt with in the *Phiṭsūtra*¹⁰ of Çāntanava, who is later than Patañjali. The rules of interpretation which govern the construction of the grammar must have been early formulated, if not explicitly laid down by Pāṇini himself; of several collections, that commented on by Nāgojī Bhaṭṭa in his *Paribhāṣenduçekhara*¹ is best known.

c29.s2.34

⁶ Ed. Böhtlingk, St. Petersburg, 1844; Ujjvaladatta’s comm., ed. London, 1859.

c29.s2.35

⁷ Ed. TSS. 1, 1905.

c29.s2.36

⁸ Ed. *Pandit*, iv–viii, xvii–xix.

c29.s2.37

⁹ Ed. J. Eggeling, London, 1879.

c29.s2.38

¹⁰ Ed. F. Kielhorn, AKM. iv. 2, 1866.

c29.s2.39

¹ Ed. and trans. Kielhorn, BSS. 1868; ed. ĀnSS. 72.

3. The Later Schools

c29.s3.1

The later schools present no features of essential interest and may be reviewed briefly. The oldest was probably the *Kātantra*,² ‘little treatise,’ called also *Kaumāra* or *Kālāpa*, the latter names indicating acceptance of the legend that its author Çarvavarman wrote under Çiva’s special favour. The legend which brings him into contact with Sātavāhana has been noted and its worth questioned.³ What is certain is that in Kashmir and Bengal the work had much influence, and that it affected deeply the Pāli grammar of Kaccāyana and the Dravidian grammarians. Originally of four books, it appears with supplements both in the Tibetan translation and in Durgasiṅha’s commentary; fragments have been found in Central Asia,⁴ and the *Dhātupāṭha* is extant only in the Tibetan version. In addition to Durgasiṅha’s *Vṛtti* on which he himself wrote a *Ṭīkā*, a sort of commentary is provided in Ugrabhūti’s *Çiṣyahitānyāsa* (1000).⁵ Tibetan tradition ascribes to Çarvavarman the use of the grammar of Indragomin, and this work seems to have been popular among the Buddhists of Nepal, but it is lost, though the reality of its author’s existence is certain.

c29.s3.2

² Ed., with Durgasiṅha’s comm., J. Eggeling, BI. 1874–8. See B. Liebich, *Einführung in die ind. einheim. Sprachwissenschaft* (Heidelberg, 1919), who dismisses the work of Indragomin, now lost; cf. Kielhorn, IA. xv. 181 f.

c29.s3.3

³ Winternitz (GIL. iii. 379) suggests the third century A. D.

c29.s3.4

⁴ Cf. L. Finot, *Muséon*, 1911, p. 192.

c29.s3.5

⁵ Sachau, *Alberuni*, i. 135; *Bodleian Catal.* ii. 129.

c29.s3.6

Use is made in the *Kāçikā Vṛtti*, without acknowledgement, of the *Cāndra Vyākaraṇa*,⁶ the grammar of Candra, which was popular in the Buddhist countries, Kashmir, Tibet, and Nepal, and which reached Ceylon. The date is uncertain, for Bhartṛhari and Kalhaṇa ascribe to Candra study of the *Mahābhāṣya*, while south Indian tradition connects him with Vararuci and makes him condemn the *Mahābhāṣya* as much talk with few ideas. He alludes in his grammar to a victory of a Jarta over the Hūṇas, which points to A. D. 470 as an earliest date, and A. D. 600 seems at least as likely if we may trust Chinese sources. He wrote a comment on his own grammar, and has a distinct terminology from that of Pāṇini, though he is essentially dependent on him. A *Dhātupāṭha*, *Gaṇapāṭha*, *Uṇādisūtra*, and *Paribhāṣāsūtra* belong to the text, and *c.* 1200 the monk Kāçyapa wrote a *Bālāvabodhana* which became popular in Ceylon.

c29.s3.7

⁶ Ed. B. Liebich, Leipzig, 1902; comm., 1918.

c29.s3.8

The Jains, in their turn, had grammars of their own. The *Jainendra Vyākaraṇa*,¹ ascribed to the Jinendra, really written by Pūjyapāda Devanandin, perhaps was composed *c.* 678. The *Çākaṭāyana Vyākaraṇa*² belongs to the reign of Amoghavarṣa (814–77), when Çākaṭāyana compiled it, using the terminology of Pāṇini, of Candra, and also of the *Jainendra*. The grammar has besides a full commentary, abridged by Yakṣavarman in his *Cintāmaṇi*, works on *Dhātu*, *Gaṇa*, *Uṇādi*, *Paribhāṣā*, and a *Liṅgānuçāsana*. Based on it is the *Siddhahemacandra* or *Haima Vyākaraṇa*,³ written for Jayasiṁha Siddharāja who had eight older works brought from Kashmir for his use; the work is practical in arrangement and terminology, which is mainly that of the *Kātantra*, and omits, of course, Vedic grammar and accent. Hemacandra wrote two commentaries, an *Uṇādigaṇasūtra* and a *Dhātupāṭha*.⁴

c29.s3.9

¹ Ed. *Pandit*, N.S. xxxi–xxxiv.

c29.s3.10

² Ed. London, 1913. Cf. Pathak, ABI. i. 7 ff.

c29.s3.11

³ Kielhorn, *WZKM.* ii. 18 ff.

c29.s3.12

⁴ Ed. J. Kirste, Vienna, 1895–9.

c29.s3.13

Other grammars won local acceptance, most at a late date. The *Saṁkṣiptasāra*⁵ of Kramadīçvara, its commentary revised by Jūmaranandin, deals in seven chapters with Sanskrit, in an eighth with Prākrit grammar; its popularity was in western Bengal, and it was written after 1150. Vopadeva's *Mugdhabodha*⁶ and *Kavikalpadruma*, on roots, won greatest popularity in Bengal and were written after 1250 under Mahādeva of Devagiri. Eastern Bengal favoured Padmanābhadatta's *Supadmavyākaraṇa* (1375), Bihar and Benares the *Sārasvatī Prakriyā* with commentary by Anubhūtisvarūpa.

c29.s3.14

⁵ See Zachariae, *Bezz. Beitr.*, v. 22 ff.; IOC. i. 218 ff.; ii. 278.

c29.s3.15

⁶ Ed. Böhtlingk, St. Petersburg, 1847; IOC. i. 230 ff.

c29.s3.16

Of grammatical and lexical importance are the *Liṅgānuçāsanas*,⁷ treatises on gender, similar to those appended to the lexica.

c29.s3.17

⁷ Franke, *Die indischen Genusregeln* (Kiel, 1890).

c29.s3.18

That ascribed to Pāṇini cannot be so old; that in Āryā verses ascribed to Vararuci is known to the *Liṅgānuçāsana* of Harṣadeva (606–47) and Vāmana (*c.* 800). We have also texts ascribed to Çākaṭāyana and Hemacandra.¹

c29.s3.19

¹ Ed. Göttingen, 1886.

4. Grammars of Prākrit

c29.s4.1

It is clear that the Prākrit grammars² which we have were written under the direct influence of Sanskrit grammars. The tradition which ascribes to Pāṇini a Prākrit grammar is doubtless a mere invention to honour Prākrit, and the further contention that Katyāyana produced the *Prākṛtaprakāça*³ of Vararuci is equally absurd. That grammar handles Māhārāṣṭrī very fully, in nine chapters, then gives one each to Paiçācī, Māgadhī, and Çaurasenī, treating Māhārāṣṭrī as the highest form, but finding Sanskrit as the original of all Prākrits. The forms of these Prākrits are clearly later than those of Açvaghoṣa and reflect perhaps the works of the third century A. D. at earliest. What is certain is that Vararuci, if we give this name to the author, was commented on by Bhāmaha, the writer on poetics (*c.* 700). The relative age of Vararuci and Caṇḍa, author of the *Prākṛtalakṣaṇa*⁴ is disputed; unhappily that work has come down in the two recensions, and, even if one is older than Vararuci, the other is younger, while its original views on many important points, as to retention of inter-vocalic consonants, depend on dubious readings; it may give us a stage not otherwise represented of Ārṣa or Ardhamāgadhī, the Prākrit of the Jains, it gives one Sūtra only each to Apabhraṁça, Paiçācikī, and Māgadhikā. As we have seen, another testimony as to a Prākrit is given by the chapter on Prākrits in the *Nāṭyaçāstra*.

c29.s4.2

² Pischel, *Grammatik der Prakrit-Sprachen* (1900).

c29.s4.3

³ Ed. and trans. E. B. Cowell, Hertford, 1854.

c29.s4.4

⁴ Ed. A. F. R. Hoernle, BI. 1880.

c29.s4.5

Vararuci seems to have belonged to the east—which is another sign of distinction from Kātyāyana of the south. His tradition is alleged to have been followed by a mysterious Laṅkeçvara or Rāvaṇa,⁵ who wrote a *Prākṛtakāmadhenu* on which is based Rāma Tarkavāgīça’s *Prākṛtakalpataru*, a work of the seventeenth century, which again was used by Mārkaṇḍeya’s *Prākṛtasarvasva* in that century, while Kramadīçvara’s section on Prākrit grammar represents the same school. Laṅkeçvara is also credited with a comment on Çeṣanāga’s *Prākṛtavyākaraṇa-sūtra*—not, as stated by Grierson, on his own work—and, as the manuscripts alleged to contain his works have vanished, his existence is decidedly in the air.

c29.s4.6

⁵ Grierson, AMJV. III. i. 120 ff.; Mitra, *Notices*, ix, nos. 3157, 3158; these notices are quite inadequate foundations for any theory.

c29.s4.7

The western school is held to be represented by the *Vālmīki-sūtras* which are lost in their original form, and are only preserved in a much expanded late version. In Hemacandra’s grammar,¹ book viii deals with Prākrit; he adds Ārṣa to the list as a good Jain must, and takes note of Jain Māhārāṣṭrī as well as the ordinary Māhārāṣṭrī of the poets; besides Paiçācī he places Cūlikāpaiçācika, and handles Apabhraṅça, giving stanzas of unknown source. For Māhārāṣṭrī he cites Hāla and the *Setubandha*, for Paiçācī, it seems likely, the *Bṛhatkathā*, probably, however, not in the original version. He is followed by Trivikrama in his *Prākṛtaçabdānuçāsana*,² who uses the *Vālmīki Sūtra* terminology, by Siṅharāja (14th cent.) in the *Prākṛtarūpāvatāra*,³ by Lakṣmīdhara (16th cent.) in the *Ṣaḍbhāṣācandrikā* and others.

c29.s4.8

¹ Ed. Pischel, Halle, 1877–80.

c29.s4.9

² T. Laddu, *Prolegomena zu Trivikrama's Prākrit-Grammatik* (1912).

c29.s4.10

³ Ed. E. Hultzsch, London, 1909. Cf. Keith, IOC. ii. 299.

c29.s4.11

This account, however, of the schools is based on inadequate grounds, for Rāvaṇa is merely a name to us. In the case of the *Vālmīki Sūtra* we have more evidence, for the Sūtras are recognized in different forms by Trivikrama, Siṅharāja, and Lakṣmīdhara. But the question arises whether it is correct to assume that they are older than Hemacandra. Trivedi,⁴ from a comparison in detail with the Sūtra of that author, holds that the *Vālmīki Sūtra* is an improved version of the work of Hemacandra, basing his conclusion on the fact that the Sūtras are sometimes better expressed, sometimes abridged, Hemacandra. Against this has been set by Hultzsch⁵ the fact that, as both the text of the *Vālmīki Sūtra*, when preserved alone, and Lakṣmīdhara, to whom we may add the *Çambhurahasya*, ascribe the Sūtra to the author of the *Rāmāyaṇa*, the work can hardly have come into being after the date of Hemacandra. This, however, is decidedly conjectural, for we have no certainty of Trivikrama’s date. All that is certain is that he wrote after Hemacandra and before Lakṣmīdhara and the *Ratnāpaṇa* of Kumārasvāmin, who was a son of Mallinātha. Now Lakṣmīdhara was, it seems, the protégé of Tirumalarāja¹ of the third dynasty of Vijayanagara, who flourished about the middle of the sixteenth century A. D., and Kumārasvāmin’s father Mallinātha lived before A. D. 1532, when a verse of his appears in an inscription,² and after A. D. 1400, since he quotes the *Vasantarājīya* (*c.* 1400).³ Kumārasvāmin also knows Lakṣmīdhara, so that there can be no real doubt of the latter’s date. We, therefore, can hardly say that the *Vālmīki Sūtra* could not have been composed after Hemacandra, and at present this seems the most probable conclusion. Siṁharāja’s date is also quite dubious; the reference⁴ to the fourteenth century is conjectural, and it is possible that he is really later than Bhaṭṭoji Dīkṣita. Lakṣmīdhara and he agree in treating of Māhārāṣṭrī as the Prākrit *par excellence*, and then shortly giving the particular features of Çaurasenī, Māgadhī, Paiçācī, Cūlikāpaiçācī, and Apabhraṁça. Of other grammarians Mārkaṇḍeya is noteworthy for his treatment of a large number of forms of Prākrit as a result of his revising the tradition of varieties of dramatic Prākrit; he treats of five principal divisions, Māhārāṣṭrī, Çaurasenī, Prācyā, Āvantī with Bāhlīkī, and Māgadhī with Ardhamāgadhī, and also of Çākārī, Cāṇḍālī, Çābarī with Ābhīrī and Auḍhrī, Ṭākkī, Nāgara and Upanāgara Apabhraṁça, and Paiçācī. It would be interesting to know how far he based his work on earlier writers, how far on the study of texts; that the latter was the case in the seventeenth century, when Prākrit was far more of a dead language than Sanskrit, must, despite Grierson’s assumption⁵ that Mārkaṇḍeya was a predecessor of Pischel, be regarded as wholly implausible, and the fact that Rāma Tarkavāgīça by no means agrees with his statements suggests that both were more or less intelligent compilers, not original investigators.

c29.s4.12

⁴ *Ṣaḍbhāshāchandrikā*, pp. 6 ff.

c29.s4.13

⁵ *Prākṛtarūpāvatāra*, p. vii.

c29.s4.14

¹ Hultzsch, *Report III*, p. viii; EI. iii. 238.

c29.s4.15

² JA. v. 20 *n*.

c29.s4.16

³ EI. iv. 327; Hultzsch, p. iv, n. 4.

c29.s4.17

⁴ Winternitz, GIL. iii. 406, n. 2.

c29.s4.18

⁵ AMJV. III. i. 123.

c29.s4.19

The value of the Prākrit grammarians has been strongly depreciated by Bloch¹ and Gawroński,² while it has been defended by Pischel³ among others. On the whole they do not make a very favourable impression; their rules are often obviously much too wide, a fault shared by them with the Sanskrit grammarians; moreover, they have clearly often generalized, while for many difficulties they afford no aid. On the other hand, recent investigations regarding Apabhraṁça⁴ have proved that they had often real grounds for forms which they give, and it must be remembered that they are often poorly preserved and inadequately explained.

c29.s4.20

¹ *Vararuci und Hemacandra* (1893), pp. 30 ff.

c29.s4.21

² *KZ.* xliv. 247 ff.

c29.s4.22

³ *Gramm. der Prakrit-Sprachen*, pp. 45 f.

c29.s4.23

⁴ Jacobi, *Sanatkumāracarita*, pp. xxiv ff.

c29.s4.24

The Pāli grammarians, though they are deeply influenced by Sanskrit grammar, yet do not assume that Sanskrit is the source whence Pāli is derived, and they write in Pāli, not Sanskrit. Like the Prākrit grammarians, they draw from literature for their models, not from spoken languages. Kaccāyana, the most famous, is later than Buddhaghosa, not perhaps before the eleventh century; he uses freely the *Kātantra* as well as Pāṇini; Moggallāna, in the twelfth century, who started a rival grammar, shows the influence of Candra also. The Burmese monk Aggavaṁsa's *Saddanīti* (1154), which has won fame in Burma and even in Ceylon, depends on Kaccāyana.⁵

c29.s4.25

⁵ Franke, *Zur Geschichte und Kritik der einheimischen Pāli-Grammatik und Lexikographie*, and Geiger, *Pāli*. Sanskrit is used for a Kanarese grammar in Bhaṭṭākalaṅkadeva's *Karṇāṭakaçabdānuçāsana* (ed. Bangalore, 1923), written *c.* 1600 A. D.

Linked texts (34)