Pramāṇa · Search · About · API

A History of Sanskrit Literature c7.s3: 3. The Characteristics and Development of Sanskrit in Literature

A History of Sanskrit Literature c7.s3

Language
English

3. The Characteristics and Development of Sanskrit in Literature

c7.s3.1

It is a characteristic feature of Sanskrit, intimately connected with its true vitality, that, unlike Medieval Latin, it undergoes important changes in the course of its prolonged literary existence, which even to-day is far from ended. Moreover, we must note the existence of two streams of movement, the Sanskrit of the Brahmanical schools as summed up in the grammar of Pāṇini, and the less formal language of the ruling class and the Brahmins in their entourage as shown in the epics. The works of Classical Sanskrit literature show the clearest evidence of influence in both directions; the Brahmins, to whom or to whose influence and tradition we owe most of the literature, were schooled in grammar and were anxious to avoid solecisms, but they were also under the literary influence of the epics, and in special of the *Rāmāyaṇa*, and it was not possible for them to avoid assimilating their language in great measure to that of their model.

c7.s3.2

Hence it follows that much of what is taught by Pāṇini and his followers has no representation in the literature. As we have seen, Kātyāyana and Patañjali recognize the disuse of certain verbal forms; there disappear also many idioms,² such as *anvāje-* or *upāje-kṛ*, strengthen, *nīvacane-kṛ*, be silent, *mano-* or *kaṇe-* *han*, fulfil one's longing, *celaknopam vrṣṭaḥ*, 'rained until the clothes were wet'; many words are no longer used, such as *anvavasarga*, allowing one his own way, *niravasita*, excommunicated, *abhividhi*, including, *utsañjana*, throwing up, *abhreṣa*, equitableness. The pronominal base *tya* disappears; in the verb the infinitive *tavai* is lost, many formations such as *jajanti* disappear, and the perfect participle middle in *āna* is disused. The adverbial form in *trā*, as in *devatrā*, and the old word *parut* are lost. Many nominal derivatives are not exemplified, and the use of such phrases as *çuklīsyāt* disappears. Many syntactical rules are obsolete, such as the use of the accusative with adjectives in *uka*; the instrumental with *saṁjñā* or *saṁprayam*; the dative with *çlāgh* and *sthā*; *tṛṇaṁ man* or *çune* or *çvānaṁ man*; the ablative with words denoting far or near; the genitive with verbs of remembering other than *smṛ*, with *nāth*, hope, with *jas* and other verbs denoting injury, and impersonally with expressions of illness, *caurasya rujati*; the instrumental with *prasita* and *utsuka*; *uta* in simple interrogations, and many other usages.

c7.s3.3

² Bhandarkar, JBRAS. xvi. 272; Speijer, *Sansk. Synt.*, pp. 39, 45, 61 f., 65 f., 72, 89 f., 108.

c7.s3.4

It is, however, true that beside this feature we have the deliberate employment by poets of usages, prescribed in the grammar, but so rare as to reveal themselves as purely learned reminiscences. From Açvaghoṣa on, the great authors are fond of displaying their erudition; Kālidāsa has *anugiram*, 'on the mountain', though this is given by Pāṇini¹ merely as an optional form, and *sausnātaka*, 'asking if one has bathed well', from a Vārttika.² Māgha is adept in these niceties; he has *khalu* with the gerund to denote prohibition; *mā jīvan*, 'let him not live'; he distinguishes *vi-ṣvan*, eat noisily, and *vi-svan*, howl; he affects the passive use of the perfect, revives aorist forms and gerunds in *am*, including *vastraknopam*, and uses *klam* as a finite verb. Çrīharṣa, author of the *Naiṣadhīya*, is responsible for the solitary example of the first person periphrastic future middle, *darçayitāhe*, yet cited.³ The case is still more extreme with Bhaṭṭi, whose epic is at once a poem and an illustration of the rules of grammar and rhetoric, and who has imitators in Bhaumaka's *Rāvaṇārjunīya* and Halāyudha's *Kavirahasya* (10th cent.). Even in writers of the folk-tale knowledge of grammar sometimes is exhibited quite unexpectedly in the shape of recondite forms culled from Pāṇini or his successors. So serious a philosopher as Çaṅkara resorts to the use of the negative with finite verbs—which originally must have been merely a comic use—and he is guilty also of the employment of the comparative of a verb, *upapadye-tarām*, a linguistic monstrosity of the worst kind.

c7.s3.5

¹ v. 4. 112 (Senaka).

c7.s3.6

² iv. 4. 1, v. 3.

c7.s3.7

³ Cf. grammatical similes; Walter, *Indica*, iii 38.

c7.s3.8

The influence of the grammarians explains also the free use of the aorist in the writers of elaborate prose · Bāṇa and Daṇḍin, moreover, observe the precise rule for the use of the perfect in narration prescribed by the grammarians. It has been suggested that this may be explained by the derivation of prose from a different tradition than poetry, but the suggestion appears needless.¹ Subandhu ignores the rule as to the perfect, and the simple explanation of the accuracy of the other writers is the desire to display their skill in grammar, which was naturally facilitated by the absence of metrical restrictions. The same liberty explains their practice in postponing the verb to the end of the sentence, unquestionably its traditional resting-place, but one impossible to observe in verse.

c7.s3.9

¹ Speijer, *Sansk. Synt.*, §§ 328 ff.; Renou, *La valeur du parfait*, pp. 86 ff.

c7.s3.10

Very different was the effect on Classical poetry of the influence of the epics.² They show, with special frequency in the case of the *Mahābhārata*,³ the tendency of uncultivated speech to ignore fine distinctions and by analogical formations to simplify grammar. Thus rules of euphonic combination are not rarely ignored; in the noun the distinction of weak and strong case-forms is here and there forgotten; there is confusion of stems in *i* and *in*; by analogy *pūṣāṇam* replaces the older *pūṣaṇam*; there is confusion in the use of cases, especially in the pronoun; in the verb primary and secondary endings are sometimes confused; active and middle are often employed for metrical reasons in place of each other; even the passive is found with active terminations; the delicate rules affecting the use of the intermediate *i* are violated at every turn; the feminine of the present participle active is formed indifferently by *antī* or *atī*; the middle participle of causatives and denominatives is often formed by *āna*, partly doubtless on grounds of metrical convenience; the rule that the gerund is formed by *tvā* in simple, in *ya* in compound, verbs is constantly disregarded; minutiae such as the substitution of *dhāvati* for the present of *sṛ* are habitually neglected. The tendency to prefer *a* bases is seen in the verb and the noun alike, giving such forms as *diçā* and *duhitā*.

c7.s3.11

² For the *Rāmāyaṇa* cf. Böhtlingk, BSGW. 1887, pp. 213 ff.; ZDMG. xliii. 53 ff.; Roussel, *Muséon*, 1911, pp. 89 ff.; 1912, pp. 25 ff., 201 ff.; JA. 1910, i. 1-69; Keith, JRAS. 1910, pp. 468 ff., 1321 ff.

c7.s3.12

³ Holtzmann, *Gramm. aus d. M.* (1884).

c7.s3.13

It was inevitable that so distinguished models as the *Mahābhārata* and the *Rāmāyaṇa* should deeply affect later poets, and Patañjali, in citing an epic fragment containing the irregular term *priyākhya* in lieu of *priyākhyāya*, expressly asserts that poets commit such irregularities (*chandovat kavayaḥ kurvanti*). We find, therefore, occasional errors such as the confusion of *anti* and *atī*, of *tvā* and *ya*, of active and middle, as well as regular disregard of the specific sense of the past tenses as laid down by the grammarians but ignored in the epic. As in the epic, the perfect and imperfect freely interchange as tenses of simple narration without nuance of any kind. Even Kālidāsa permits himself *sarati* and *āsa* for *babhūva*, and Çrīharṣa with the *Rāmāyaṇa* uses *kavāṭa* for the *kapāṭa* of Pāṇini. Lesser poets, especially the poetasters who turned out inscriptions, are naturally greater sinners by far against grammatical rules, especially when they can plead metrical difficulties as excuse.

c7.s3.14

Neither the epic nor the grammarians, however, are responsible for the fundamental change which gradually besets the Kāvya style, in the worst form in prose, but in varying degree even in verse. This is the change from the verbal to the nominal style, as Bhandarkar¹ not inaptly termed it. In the main, Vedic and epic Sanskrit show a form of speech closely akin to Greek and Latin; verbal forms are freely used, and relative clauses and clauses introduced by conjunctions are in regular employment. The essential feature of the new style is the substitution of the use of compounds for the older forms.² In its simplest form, of course, the practice is unobjectionable and tends to conciseness; *hataputra* is less cumbrous than 'whose sons have been slain'. But when new members are added there are soon lost the advantages of an inflective language with its due syntactical union of formed words into sentences; brevity is attained at a fatal cost in clearness. A compound like *jalāntaçcandracapala*, 'fickle as the moon reflected in the water', is comparatively innocuous, but even a stylist like Kālidāsa permits himself such a phrase as *vīcikṣobha-stanitavihagaçreṇikāñcīguṇā*, 'whose girdle-string is a row of birds loquacious through the agitation of the waves'. True, in such a case there is no real doubt as to the sense, but often this is not the case, and in point of fact it is one of the delights of the later poets to compose compounds which contain a double entendre, since they can be read in two ways; of such monstrosities Subandhu is a master. Moreover, the nominal forms of the verb are given a marked preference; the expression of past time is regularly carried out by a past participle passive in form of an intransitive verb, such as *gatas*, he went, or if the verb is active the subject is put into the instrumental and the past participle passive is employed, as in *mṛgeṇoktam*, the deer said. Or an active past participle is created by adding *vant* to the passive participle, *kṛtavān*, he did; a distant parallel in the grammarians has been seen in the sanction by Pāṇini of the use of such forms as *dāçvāṅs* in lieu of a finite verb. Or the use of any save a verb of colourless kind may be avoided by substituting such an expression as *pakvaṁ karoti* for *pacati*, he cooks, or *pakvo bhavati*, it is cooked, for *pacyate*. Similarly the periphrastic future is preferred to the finite verb. Or the verb may wholly disappear as when for *ayaṁ māṅsaṁ bhakṣayati* we have *māṅsabhojako 'yam*, he is a meat eater. In harmony with this is the tendency to lay great stress on case relations as expressing meaning, a practice which in the later style in philosophy, exegesis, and dialectics results in the occurrence of sentences passim with no verb and practically only the nominative and ablative cases of abstract nouns. Frequent, and indeed in some forms of composition, such as the folk tale, tedious in its reiteration, is the use of gerunds in lieu of subordinate clauses.

c7.s3.15

¹ JBRAS. xvi. 266 ff.; cf. Bloch, MSL. xiv. 27 ff.; Renou, *La valeur du parfait*, pp. 90 ff.; Stchoupak, MSL. xxi. 1 ff.; Jacobi, IF. xiv. 236 ff.

c7.s3.16

² Jacobi (*Compositum und Nebensatz*, pp. 25, 91 ff.) points out that they are properly used for ornamental description, not for important qualifications, and also suggests poetic convenience as a cause of popularity; cf. Chap. II, § 4. See also Wackernagel, *Altind. Gramm.*, II. i. 25, 27, 159; Whitney, *Sansk. Gramm.*, § 1246.

c7.s3.17

We are reduced to conjecture as to the cause of this tendency. The desire for brevity is already seen in the style of the Vedic Sūtras, and the grammarians carried it to excess; their works furnish abundant instances of insistence on using cases in a pregnant sense and in affecting compounds; gerunds are frequent in the ritual texts. It has been suggested that the love for participial forms is partly explained by Dravidian influence;¹ the periphrastic future in both Sanskrit and Dravidian uses the auxiliary verb only in the first and second persons; the type *kṛtavān* has a parallel in *çeydavan*; the rule of the order of words in which the governed word precedes and the verb is placed at the end of the sentence is Dravidian. Unhappily, the arguments are inconclusive;² the omission of the auxiliary in the third person is natural, for in that person in any sentence whatever it is commonly omitted as easily understood; the order of words in Sanskrit has parallels in many other languages than Dravidian and rests on general rules of thought.

c7.s3.18

¹ Konow, LSI. iv. 279 ff.; Grierson, BSOS. I. iii. 72; Carnoy, JAOS. xxxix. 117 ff.; Chatterji, i. 174 ff.

c7.s3.19

² Cf. R. Swaminatha Aiyar, POCP. 1919, i, pp. lxxi ff., who legitimately points out that the evidence of Dravidian is very late in date, and these languages probably borrowed from Aryan. K. G. Śaṅkar (JRAS. 1924, pp. 664 ff.) points out that the *Tol-kāppiyam*, the oldest Tamil work, must be after 400 A. D. as it refers to the *Poruḷadhikāramsūtra*, horary astrology, and that the Moriyas of the Sangam are the Mauryas of the Koṅkaṇa, who date after 494 A. D.

c7.s3.20

Beside the correct or comparatively correct Sanskrit of the poetic literature we find, especially in technical and non-Brahmanical works, abundant evidence of a popular Sanskrit or mixed Sanskrit in various forms. Generically it can be regarded as the result of men who were not wont to use Sanskrit trying to write in that language, but there are different aspects. Thus the early Buddhist writers who decided to adapt to the more learned language the Buddhist traditions probably current in Ardhamāgadhī were hampered by the desire not to depart unduly in verse at least from their models, a fact which explains the peculiar forms found especially in Gāthās, but also in prose in such a text as the *Mahāvastu*.³ Traces of this influence persist even in much more polished Buddhist writers such as Açvaghoṣa, and much of it may be seen in the *Divyāvadāna*, though that work marks in part a successful attempt to adapt Sanskrit prose, as known at Mathurā and elsewhere, to Buddhist use. The degree of cultivation of those who endeavoured to write in Sanskrit might vary greatly; thus the Sanskritization of the treatises in the Bower Manuscript, perhaps of the fourth century A.D., is comparatively good in the case of those on medicine, and decidedly poor in those on divination and incantation. In part the deviation from Sanskrit as laid down in the grammars is purely a case of Prācritic forms intruding scarcely disguised into the texts, but in other instances popular influence reveals itself in a Sanskrit which ignores delicate distinctions and confuses forms. The distinction between Prākritisms and careless Sanskrit is not absolute, but it is convenient and legitimate.

c7.s3.21

³ Cf. Senart, i, pp. iv, xiii ff.; Wackernagel, *Altind. Gramm.*, i, p. xxxix. Contrast F. W. Thomas, JRAS. 1904, p. 469, who regards the mixed Sanskrit as representing middle-class speech. Poussin (*Indo-européens*, p. 205) stresses convention as stereotyping usage.

c7.s3.22

Thus we have in the phonology of this popular Sanskrit as seen in the Bower MS. some confusion of *ṛ* and *ri*, of *n* and *ṇ*, of *ç*, *ṣ*, and *s*; metrical lengthening and shortening of vowels is not rare; *ml* becomes *mbl*, and rarely *a* is prefixed as in *alatā*. In Sandhi hiatus and hyper-Sandhi, even to the extent of an elided consonant (*açvibhyānumataḥ*), are known, while *ā* is occasionally elided when initial. In declension we find *īs* and reversely *ū* as feminine nominatives for *ī* and *ūs*; *īs* is often replaced by *yas* as the accusative feminine, and *in* stems are treated as *i* stems, as in *pittīnām* for *pittinām*. In the verb we have simplification in class, as in *lihet* for *lihyāt*, *piṣet* for *piṁṣyāt*; and, as in the epic, very free interchange of active and middle forms; the gerunds in *tvā* and *yā* are confused. Stem formation shows frequently the mixture of bases in *a*, *i*, or *u* for those in *as*, *is*, or *us*, and, rarely, such a base as *hantāra* from the accusative of *hantṛ*; there is confusion in feminine suffixes, as in *ghnā* for *ghnī*, *caturthā* for *caturthī*, while ordinals in composition are sometimes replaced by cardinals. Very characteristic is confusion of gender, especially between masculine and neuter, more rarely between masculine and feminine or feminine and neuter. Case confusion is common, as is non-observation of rules of concord and confusion of numbers, while the interpolation of particles within compounds or sentences, absolute constructions, and very loosely compacted clauses are common.

c7.s3.23

Existing as it did side by side with Prākrit dialects, it was inevitable that there should be frequent borrowings on either side,¹ despite the objections raised from time to time by grammarians and sticklers for purity in the use of the sacred language in sacrificial matters.² Thus, though Classical Sanskrit lost many of the words and roots recorded in the *Gaṇapāṭha* and the *Dhātupāṭha* associated with Pāṇini's grammar, it was enriched by numerous additions, some easy, others difficult, of detection. In many cases the Prākrit forms were taken over with only the necessary changes requisite to make them seem to have terminations allowed in Sanskrit. It appears as if even Pāṇini³ recognized this practice, since he allows eastern place-names to pass as correct though having the Prākrit *e* and *o* for the regular *ai* and *au* which his rules require. In other cases the retention of the Prākrit form was aided by the possibility of regarding the form as genuine Sanskrit; thus the poetic technical term *vicchitti*, really from *vikṣipti*,⁴ in all likelihood seemed to be derivable from *vi-chid*; Kṛṣṇa's epithet Govinda, perhaps Prākrit for *gopendra*, was felt as *go-vinda*, winner of cows; in late texts *bhadanta*, from the phrase of greeting *bhadraṁ te*, is defended as from *bhad* with the suffix *anta*, and *uttṝ* is not recognized as from *avatṝ* through Prākrit *otarati*; *duruttara*, hard to overcome, really from Prākrit *duttara* for *duṣṭara*, was felt as *dur-uttara*. In many cases, doubtless, Prākrit words were correctly rendered into good Sanskrit equivalents, in which case borrowing cannot now be established. In others, however, the process is betrayed by false forms; thus Prākrit *mārisa*, friend, where *s* stands for *ś*, was mechanically made into *māriṣa*; *guccha*, for the lost *gṛpsa*, became *gutsa*, cluster; *masiṇa*, Sanskrit *mṛtsna*, reappeared as *masṛṇa*, soft; *rukkha*,⁵ for *rukṣa* or rather *vṛkṣa*, *rūkṣa*, tree; and *heṭṭhā*, from *adhastāt*, gave by reconstruction *heṣṭa*. A common formation in Jain texts is *vidhyai*, go out, which is based on Prākrit *vijjhai*, from Sanskrit *vikṣai*; similarly *vikurv*, produce by magic, is traced through *viuvvai*, *viuvvae* to *vikṛ*. Later there are borrowings from vernaculars such as Gujarātī or Marāṭhī or Hindi.¹ Often, of course, the Sanskrit version has been ingeniously made to appear valid in itself, as when *pabbhāra* is metamorphosed into *prāgbhāra*, though *prahvāra* is its origin.

c7.s3.24

¹ Zachariae, *Beitr. z. Lexikogr.*, pp. 53 ff.

c7.s3.25

² See Çabarasvāmin and Kumārila on *Mīmāṁsā Sūtra*, i. 3, 24 ff.; *Sarasvatīkaṇṭhābharaṇa*, i. 16; *Mahābhāṣya*, i. 5.

c7.s3.26

³ i. 1. 75.

c7.s3.27

⁴ Zachariae, *B. Beitr.*, xiii. 93; cf. *argala* (IA. xix. 59) through *aggala* for *agralaka*; Kielhorn, GN. 1903, p. 308.

c7.s3.28

⁵ See Hultzsch, CII. i, pp. lxx ff., *contra* Turner, JRAS. 1925, p. 177. I agree with Oldenberg that in RV. vi. 3. 7 *rukṣa* is not = *vṛkṣa*.

c7.s3.29

¹ Cf. Bloomfield, *Festschrift Wackernagel*, pp. 220–30; Hertel, HOS. xii. 29 f.

c7.s3.30

Occasionally we find the process of Sanskritization applied to what was really Sanskrit; probably thus are to be explained *prasabham*, violently, from *pra-sah*; Naghuṣa for the older proper name Nahuṣa, *varṣābhū*, frog, for *varṣāhū*.

c7.s3.31

From foreign sources borrowings also occurred naturally enough in those cases where, as in the Dekhan or Further India, Sanskrit was used side by side with a native speech. Kumārila permits the incorporation of Dravidian terms, provided that they are given Sanskrit terminations, and names especially such as Sāyaṇa were freely thus Sanskritized. The *ḻ* which marks South Indian texts² in lieu of the *ḍ* and *l* of the north is doubtless in part due to Dravidian influence. On the other hand, invasions from the north brought early and late Iranian words such as *lipi*, writing, Old Persian *dipi*,³ *kṣatrapa*, satrap, and perhaps *mudrā*, seal,⁴ or *divira*, scribe, *mihira*, Mithra, *bahādura*, *sāha*, and *sāhi*. The Greek invasions in the north left little trace in the language, but probably later India borrowed *suruṅgā* from syrinx in the technical sense of an underground passage, and a large number of terms of astrology. Many of these they ingeniously altered to seem true Sanskrit, as when for hydrochoos we find *hṛdroga*, or *jāmitra* for diametron. With similar ingenuity the useful camel was metamorphosed into *kramela*,⁵ suggesting connexion with *kram*, go. The Mahomedan invasion brought with it Arabic and Turkish terms, and the European powers have contributed occasional additions to the modern Sanskrit vocabulary, testifying to its capacity of assimilation. The scientific literature in special has shown its willingness to appropriate the terms used by those from whom knowledge has been acquired, together with considerable skill in disguising the loan.

c7.s3.32

² Lüders, *Festschrift Wackernagel*, p. 295.

c7.s3.33

³ Bühler, *Ind. Stud.*, iii. 21 ff.; Hultzsch, CII. i, p. xlii.

c7.s3.34

⁴ Franke, ZDMG. xlvi. 731 ff. Hāla has *vandī*, captive. Cf. Weber, *Monatsber. Berl. Ak.*, 1879, pp. 810 ff.

c7.s3.35

⁵ Lévi (*De Graecis vet. Ind. Mon.*, p. 56) doubts this, but the word is late; *lopāka* (ἀλώπηξ) is different, as *lopāça* is Vedic. Hāla has *kalama* (κάλαμος) and *maragaa* (σμάραγδος).

c7.s3.36

As the passage of time made Sanskrit more and more a language of culture, it reveals in increasing measure a lack of delicate sensibility to idiomatic use of words, such as is engendered by usage in a living speech more closely in touch with ordinary life. The defect, however, is sometimes exaggerated, for it must not be forgotten that poets of all times are apt, through considerations of metre or desire for effect,¹ to adopt unusual senses of words and to strain meanings; Pindar and Propertius illustrate a tendency which is found more or less markedly throughout classical literature, while the Alexandrian Lykophron is guilty of as distinct linguistic monstrosities as any Indian poet. The tendency in their case was accentuated by the growing love for paronomasias, and the tendency to study poetic dictionaries which gave lists of synonyms, ignoring the fact that in reality two terms are practically never really coextensive in sense. The grammatical knowledge of the poets also led them into inventing terms or using terms in senses etymologically unexceptionable but not sanctioned by usage.

c7.s3.37

¹ Catullus' curious compounds in the *Attis* illustrate this theme.

Linked texts (16)