9. CHAPTER SIX THEORY AND CRITICISM IN THE DHVANYALOKA

CHAPTER SIX THEORY AND CRITICISM IN THE DHVANYALOKA

Anandavardhana's need to justify his departures from

Alamkārasāstra as it had so far been practiced, and

particularly to provide evidence for the previously

unrecognized predominance of rasa in poetry, leads him to

adopt a more empirical approach to literary analysis, quoting

and discussing the works of earlier poets far more frequently

and thoroughly than any of his predecessors had done. Because

his teleological mode of literary analysis requires that

particular features of a poem be evaluated not simply in terms

of their intrinsic beauty but in terms of their ability to

contribute to the intended effect of the entire composition of

which they form a part, his attention is directed, to an

unprecedented degree, toward the specific contexts in which

particular poetic elements are employed; that is to say,

toward specific literary works.

This shift toward a more empirical and more context-

dependent mode of literary analysis gives issues of literary

criticism a relevance they would not have had for earlier

Ālamkärikas, whose approach gave little or no importance to

context. The goal of these Alamkarikas was to identify,

define and categorize those discrete and isolable elements,

phonetic or semantic, which lend beauty to a verse (any verse)

in which they are employed. These elements could be

illustrated equally well with verses composed specially for

this purpose or with verses drawn from existing poems. A

258

259

verse illustrating a particular figure of speech could be

found in some well-known poetic work, but the source from

which it was drawn could have no bearing on its aesthetic

worth. If an illustration meets the defining criteria for the

figure in question it will be beautiful and give pleasure to

an audience, without regard to the context in which it is

employed. For these theorists alamkäras and other desirable

elements of poetry are valued for their intrinsic beauty, not

for their ability to contribute to some unifying aesthetic

goal within a work. Therefore, specific works of poetry are

of little interest to them serving, at most, as a fund of

potential examples

For Anandavardhana, however, the aesthetic value of

particular formal elements depends not on their intrinsic

beauty, but on their capacity to subserve the overriding

aesthetic purpose of the particular work in which they are

employed. Context therefore assumes a new importance in

Anandavardhana's literary analysis. The criteria by which the

isolable and definable elements of poetry are judged are no

longer internal to them; any phonetic or semantic feature can

be properly evaluated only with regard to the purpose of the

whole composition of which it forms a part. Thus for

Anandavardhana it becomes useful to examine and discuss

specific literary works, in a way in which it was not useful

for his predecessors, for whom the main object of

Alamkārasāstra was the analysis of abstracted, repeatable

260

elements which could add beauty to any and every composition.

Literary Criticism in the Dhvanyaloka

Although Anandavardhana's critical remarks on existing

literary works are relatively few in number, given the total

absence of such remarks in earlier Alamkarasastra, they

represent an important innovation and are worthy of serious

attention. For Anandavardhana, as his poetic teleology would

lead us to expect, the primary, indeed the sole, basis on

which elements of particular compositions can be judged is the

extent to which they heighten or diminish the primary rasas of

these compositions.

This establishment of a single, overriding criterion of

aesthetic value requires a re-evaluation of the status of

those rules of composition formulated by earlier aesthetic

theorists, both in Alamkarasastra and in the related but

separate field of Natyasastra (dramaturgy). It can no longer

be uncritically assumed that adherence to such rules is

automatically necessary or desirable. It is always with

reference to the standard of rasa-communication that an

author's adherence to such principles, or failure to adhere to

them, must be judged. This issue is confronted most directly

in Anandavardhana's discussion of the Venīsamhāra of

Bhattanārayana (c. 600-650 AD) .1

The Venisamhara is a drama in six acts which presents the

story of the victory of the Pandavas over the Kauravas in the

1Dhvanyāloka, pp. 336-40, 363.

261

Mahabharata war. Its primary rasa is vira ("heroism"), but in

the second act of the play there is an episode of srīgāra-rasa

: the Kaurava king Duryodhana overhears his wife describing to

her servants an inauspicious dream in which she finds herself

attracted to a mongoose (nakula -- also the name of one of the

Pandavas) who leads her to a secluded place and undresses her.

Hearing this, and not realizing that it is only a dream being

described, Duryodhana flies into a jealous rage, believing his

wife to be involved with the Pandava Nakula. Before he has a

chance to confront his wife, her servants' remarks reveal his

mistake. Relieved, he approaches his wife and calms her fears

regarding the inauspicious dream, and an affectionate episode

ensues between them.

Anandavardhana regards the inclusion of this episode as

a flaw which detracts from our appreciation of the vira-rasa

which predominates elsewhere in the work, and sees this flaw

as the result of blind adherence to the rules of plot

construction laid out by Bharata in the Natyasastra, even when

these contravene the requirements of rasa. According to

Bharata, the plot of a nataka2 must contain five sandhis

("junctures") : mukha ("opening"), pratimukha ("unfolding"),

garbha ("sprouting"), avamarsa ("interruption"), and nirvahana

2The most elaborate and fully developed of the ten varieties of drama Bharata describes (cf. Natyasastra 18.10ff [Parimal Sanskrit Series 4, Vol. 2, pp. 302ff]), and the genre to which the VenIsamhara belongs.

262

("conclusion") .3 For each sandhi, he gives a list of

sandhyangas ("parts [or subordinate elements] of the sandhis")

which should be incorporated within it.4 Among the

sandhyangas to be introduced during the "unfolding", Bharata

(19.59) lists vilasa, which he defines (19.76) as "desire for

the enjoyment of love".

As Ānandavardhana sees it, Bhattanarayana has introduced

an erotic episode in the second act of the VenIsamhara solely

to conform to the rule given by Bharata, without regard for

the primary rasa of his work.

This is another primary condition required for a work to be suggestive of rasa etc .: that the arrangement of the sandhis -- namely mukha, pratimukha, garbha, avamarsa, and nirvahana -- and their angas -- upaksepa ["commencement"] etc. -- should be made with a view toward the suggestion of rasa, as in the RatnavalI,5 and not merely out of a desire to fulfil the requirements of sästra, as in the

3cf. Nātyaśāstra 19.37-43 (ibid., Vol. 3, pp. 23-29) .

4Nātyasāstra 19.50ff (ibid., pp. 31ff) .

5A drama composed by the Harsa, king of Kanauj (r. 606-47 AD) and ruler of one of the largest empires in the history of South Asia. Anandavardhana does not demonstrate specifically the conformity of the sandhis and sandhyangas in this drama to the requirements of its primary rasa, and even the normally thorough Abhinavagupta remarks that to do so in detail would require an inordinately lengthy discussion, but the basic point is clear: the srgära-rasa which predominates is kept in view throughout the course of the play, and there is no irrelevant and digressive development of other rasas, as in the Venisamhara. We have already seen the RatnavalI singled out for particular praise for its insistent focus on the development of its primary rasa (cf. above, Chapter Four, p. 180): In the verse uddāmotkalikam ... (RatnavalI 2.4; "Having blossoms/longing which are/is unrestrained ... "), the poet's punning comparison of a blooming creeper with a beautiful woman is praised because it foreshadows the hero's love for another woman and his wife's jealous anger, thus furthering the development of the primary rasa.

263

Venīsamhara, where the sandhyanga of the "unfolding" called vilasa is introduced in the second act, even though it is not conformable to the development of the contextually appropriate rasa.6

The generally accepted rules of plot composition are not to be

obeyed simply because they are the rules. Whenever it is

necessary for the full and unobstructed development of the

primary rasa, they must be abandoned.

Even Kalidasa, a poet revered by Anandavardhana and

nearly all other Ālamkārikas, is not spared from criticism

where his practice is seen to deviate from what is required

for the fullest possible development of rasa. Anandavardhana

regards Kalidasa's explicit description of the lovemaking of

Siva and ParvatI in the eighth sarga of his Kumarasambhava as

vulgar and therefore ultimately detrimental to the work's

6Dhvanyaloka, pp. 336-340. Abhinavagupta, apparently disturbed by Anandavardhana's suggestion that Bharata's rules should simply be ignored, explains the problem differently. Bharata defines vilasa as "desire for the enjoyment of love" (samīha ratibhogartha). According to Abhinavagupta "the words 'enjoyment of love' are used so as to imply those aesthetic factors suggestive of the basic emotion [sthāyibhava] belonging to the rasa appropriate to the context." (Locana, pp. 340-341). Rati ("love") is the sthayibhāva of srīgāra- rasa, the first rasa described in the Natyasastra and, in practice, the most common; Abhinavagupta understands it as meant to refer, metonymically, to any contextually appropriate sthayibhava. The rasa appropriate to the Venīsamhāra is vīra, and to fill the need for the sandhyanga vilasa, the author ought to have constructed an episode to excite the basic emotion appropriate to this rasa, rather than one appropriate to śrīgara-rasa, as he has done. His fault is thus not that he follows Bharata's rule where he should not, but rather that he misunderstands and misapplies the rule. Through this interpretation Abhinavagupta the Bharata's rules; however, there is nothing to indicate that upholds authority of

Anandavardana (or Bharata, for that matter) understood vilāsa in this way.

264

capacity to evoke rasa.

Änandavardhana's observations on this flaw arise out of

a more general discussion of character type and the

appropriateness of particular actions and emotional states to

particular types of character. He regards such

appropriateness as an important condition for a work's ability

to convey rasa. For Anandavardhana character type (prakrti)

is divisible into three broad classes -- high, middle, and low

(uttama, madhyama, and adhama),7 and further into human and

divine. It is important that characters be consistently

treated as of one type or another.

A sthayibhava constructed so as to consistently conform to this [character type] will become appropriate. Otherwise, if the determination [utsaha] etc. [i.e. the sthäyibhavas] of a divine hero are portrayed by relying on purely human actions, or if those of a merely human hero are portrayed by relying on purely divine actions, they will become inappropriate. And thus, in the description of a purely human hero such as a king, if actions such as leaping across even if they the seven oceans are portrayed, are beautiful, they will certainly be devoid of rasa; and the reason for this is their inappropriateness.

The portrayal of deeds of heroism and the of the determination

and energy out of which they arise is the means by which vira-

rasa is conveyed, but if these deeds are not appropriately

matched with the type of character who performs them they will

7Here he is following Bharata, who discusses these three classes of character in the twenty-fourth chapter of Nātyasāstra (op. cit., Vol. 3, pp. 241ff) . The hierarchy of high, middle and low carries both moral and class connotations.

8Dhvanyaloka, op. cit., p. 330.

265

not convey this rasa effectively. The description of a purely

human character leaping over oceans is so implausible as to

impair the vira-rasa it is meant to enhance.

The problem with the Kumarasambhava is, in

Anandavardhana's view, basically similar.9 It is not that the

explicit description of lovemaking per se is objectionable,

but rather that the description of divine characters engaged

in such actions is inappropriate and hence injurious to rasa.

A potential objection is raised, however, to the

extension of the dichotomy between human and divine character

type from the realm of heroism to that of love.

If [as you sayl an examination is somehow to be made of appropriateness to human or divine character types in the description of heroic determination [utsäha] etc., then let it be made; but in the case of love [rati] what can be the use of such an examination? For love, even for the gods, can be described only by means of conduct which is appropriate to Bhäratavarsa [i.e. South Asia] .10

In the case of heroism, there is a clear distinction between

the actions of which gods and mere humans are capable, and

this distinction makes possible inappropriateness of the kind

Ãnandavardhana has described. In the case of love, there is

no such apparent distinction between human and divine modes of

action; as we have no knowledge of any specifically divine

'Ānandavardhana does not mention the Kumarasambhava by name in this passage, but in an earlier portion of the text (vrtti on 3.6; Dhvanyaloka, op. cit., pp. 317-318) he refers to the impropriety of the description of Parvati's lovemaking in the Kumarasambhava and says he will explain this impropriety later -- clearly referring to this passage.

10Ibid., p. 332.

266

mode of conduct in such matters, one has no recourse but to

describe the love-play of the gods in the same way one would

that of humans.

Ānandavardhana concedes that there is no special mode of

action appropriate to the description of divine lovemaking;

this does not, however, mean that appropriateness and

inappropriateness have no place in the consideration of such

description. One must, indeed, describe the love-play of

divine characters as one would that of human ones; in doing

so, however, one should be careful to treat these divine

characters as if they were human characters of the highest

type (uttama-prakrti).

We do not that there is some different appropriateness relating to divine characters. What say

then? The description of srngara belonging even to divine characters will be beautiful if made as one would do in the case of the highest sort of heroes -- kings and so on -- in Bharatavarsa. Common, vulgar description of srīgara, in the case of kings and so on, is not found in drama; thus, it should be avoided in the case of the gods. If you say that it is only to be avoided in such cases because drama is to be enacted [abhineya] [rather than merely recited], and because enactment which has love-in-union [sambhoga-śrīgāra] vulgar, we reject this: if an enactment having such a as its subject is

subject is vulgar, what prevents a [recited] poem having such a subject from being vulgar? Thus in any poem, whether it is to be enacted or not, the vulgar description of the sexual union of heroes of the highest type, such as kings, with heroines of the highest type, will necessarily be vulgar, just as the description of the sexual union of one's parents would be vulgar. The case is just the same for [description of sexual union] having the highest gods as its subjects.11

Thus, Kalidasa's description of the lovemaking of Siva and

llIbid., p. 332.

267

ParvatI is vulgar, inappropriate for characters of the highest

type, and, ultimately, detrimental to the work's capacity to

convey rasa. That the Kumarasambhava is a purely verbal

product, rather than a drama to be presented visually before

an audience, in no way mitigates this vulgarity. So,

appropriateness of action and emotion to character type is

necessary in works conveying srīgära-rasa, just as in those

conveying vira-rasa.

Even when the poet relates an existing story, rather than

inventing his own, his construction of plot must be guided by

the requirements of rasa, not simply by unthinking

faithfulness to the course of the story as it is known. As a

general rule, Anandavardhana argues that poets, when drawing

on histories or other existing bodies of narrative for their

plots, ought not to tamper with their sources. He quotes with

approval from the prologue of Yasovarman's Ramabhyudaya the

dramatist's claim that in his play there is "not even a slight

deviation in the course of the story. "12 The epics and other

historical sources are themselves rich with rasa, and a poet

should select from these sources episodes or stories which

will convey a desired rasa

However, whenever any element in such a story conflicts

with the rasa the poet wishes to convey, it must be removed,

12Dhvanyaloka, p. 335. The play itself is not extant. Abhinavagupta notes the source of the quotation, and the full verse is quoted by Bhoja in his Śrngaraprakasa (cf. Raghavan, Bhoja's Srīgaraprakasa [Madras: Punarvasu, third ed., 1978], p. 393).

268

and newly invented elements must, if necessary, be added to

the familiar story so as to heighten this rasa.

Anandavardhana offers several examples of poems in which known

stories have been altered in accordance with the needs of

rasa: the works of Kalidasa, Sarvasena's Harivijaya, and

Anandavardhana's own Arjunacarita.13

Kalidāsa frequently takes liberties with his (usually

epic) sources, and Anandavardhana does not specifically

indicate what changes he has in mind. Abhinavagupta,

commenting on the passage, mentions Kalidāsa's description, in

the Raghuvamsa, of the marriage Rama's grandfather Aja and of

various other exploits of Rama's predecessors and successors

which are not described in the epics. This deals only with

Kālidāsa's additions to and not alterations of the stories he

retells, and overlooks what is surely his most notable

departure from his epic sources: the replacement, in his

Abhijñanasākuntala, of Dusyanta's deliberate rejection of

Sakuntalā (whom he has secretly married, and who carries his

unborn child), with the story of a curse which causes him to

forget her. 14

The latter two works mentioned by Anandavardhana are not

extant, and as he does not specifically explain the ways in

13Dhvanyaloka, op. cit., pp. 335-336.

14Kuntaka, in his Vakroktijivita, discusses in detail this transformation of the epic story of Sakuntala and Kalidasa's reasons for introducing it (cf. VakroktijIvita, Dharwar ed., pp. 249-251, and below, Chapter Eight, pp. 410-411) .

269

which these poems' plots deviate from their sources, we are

dependent on Abhinavagupta's comments for information on these

works. "In the Harivijaya, " we are told, "acts such as the

theft of [Indra's] Parijata-tree are described as part of

[Krsna's] placating of his beloved, although not found in the

histories. Likewise, in the Arjunacarita, acts such as

Arjuna's conquest of the netherworld are described, although

not known from the histories. "15

Seen in the light of Anandavardhana's rasa-centered

poetic teleology, faithfulness to one's sources can have no

independent value. Where the plot of one's source narrative

effectively evokes the desired rasa one should adhere to it;

where an addition or alteration will heighten rasa one should

change it without compunction.

The poet, when composing a poem, dependent [paratantra] on rasa. should be totally If he sees in his story any element not suitable for this rasa then he should eliminate this and independently devise narrative element which is suitable for this rasa. For, some other

there is no use in a poet's merely rehearsing a historical plot, since this is accomplished by the

15Ibid., pp. 335-336. As Ingalls notes (The Dhvanyaloka of Anandavardhana, p. 436, n. 2) the story of Krsna's theft of the Parijata is found in most manuscripts of the Harivamsa, but is absent from the Sarada manuscripts of Kashmir as well as several of the Malayalam-script manuscripts (and is, on these grounds, excluded from the Critical Edition of the text). Abhinavagupta may be mistaken in regarding this as an innovation on the part of Sarvasena, who may have known a version of the Harivamsa containing this story, or, as Ingalls suggests, Sarvasena's innovation may have crept back into some versions of the Harivamsa; in any case, as versions of the Harivamsa containing this story seem to have been unknown in Kashmir, Abhinavagupta is presumably correct in taking this to be the alteration to which Anandavardhana refers.

271

This new, empirical orientation, while it allows

Anandavardhana to showcase the unprecedented analytical and

critical possibilities which his teleological, rasa-centered

model of poetry allows for, also makes apparent a major

problem with this model as applied to actually existing s a

literature: that there exists a significant body of poetry in

which the communication of rasa simply is not the primary

object of the poet. The great strength of Anandavardhana's

model of poetry is its ability to provide a general criterion

for aesthetic value in literature, in terms of which

particular prescriptions regarding composition may be

explained and justified and existing compositions may be

evaluated. As we shall see, in the case of poetry which does

not have rasa as its telos, this explanatory and evaluative

power simply disappears.

Gunībhūtavyangya- and Citra-kāvya

Ānandavardhana has made two major innovations in poetic

theory which provide the basis for his model of poetry: he has

argued for the existence in poetry of a heretofore

unrecognized level of suggested (vyangya) meaning distinct

from its directly expressed (vacya) meaning, and he has

introduced a Mimamsa-derived, teleological method of text

analysis which requires that any coherent text be directed

toward a single, predominant element, to which all of its

parts are ultimately subordinate. Dhvani, the type of poetry

with which Anandavardhana is primarily concerned, is defined

272

as poetry in which suggested meaning both exists and is

predominant. But the two basic postulates of his poetic

theory allow for three logically possible classes of poetry

(taking the status of suggested meaning as the basis for

classification). Poetry may or may not convey a suggested

meaning in addition to its expressed sense; if it does so,

this suggested sense may be subordinated to the expressed

sense or may be predominant with respect to it.

Änandavardhana accepts that these three classes are not

only theoretically possible but actually represented in

existing literature. The first type, in which suggested

meaning both exists and is predominant, is, of course, dhvani.

The second, in which suggested meaning is present but is

subordinated to the expressed meaning, is termed

gunībhutavyangya-kavya ("poetry in which the suggested meaning

is subordinate [gunibhuta] "). The third, in which there is no

suggested sense whatsoever, he calls citra-kavya ("flashy

poetry" or "picture-poetry"). As we have seen (above, Chapter

Four, p. 132), Anandavardhana strives to show that rasa has

played a central role in poetry from its very origins; for

him, much if not most great poetry has the communication of

rasa as its primary object, and hence may be classified as

dhvani-kāvya. He does not, however, go so far as to contend

that all poetry has this as its object. He acknowledges that

both gunībhūtavyangya- and citra-kavya form a part of the

poetic tradition as it is known to him, and is ultimately

273

unwilling, despite his professed preference for dhvani-kāvya,

to unqualifiedly dismiss either of these other types as

necessarily deficient (though, as we shall see, he comes close

to doing so in the case of citra).

Gunībhūtavyangya, unlike dhvani, is a negatively defined,

catch-all category. A verse containing some suggested meaning

may for any of a variety of reasons fail to meet

Anandavardhana's criteria for a true instance of dhvani, and

whenever it does so it is to be considered a case of

gunībhutavyangya. The primary condition for the subordination

of a suggested meaning, the one mentioned by Anandavardhana in

his definition of gunībhutavyangya, is that the directly

expressed meaning should be more beautiful than it.18 A

verse conveying a suggested meaning should also be regarded as

gunībhūtavyangya if this suggested meaning is unclear.19

Likewise, if the suggested meaning of a verse is explicitly

referred to or mentioned, as in the following example:

Knowing that her lover was considering what time they should meet, the clever woman closed her play lotus, giving him a hint with laughing eyes.20

18Dhvanyaloka, p. 458. The standard of "beauty" is, of course, thoroughly subjective, and its usefulness as a criterion for distinguishing between dhvani- and gunībhutavyangya-kavya is challenged in later Alamkarasastra, particularly by Mahimabhatta (cf. Vyaktiviveka, pp. 160-168 and below, Chapter Ten, pp. 488ff).

19Dhvanyaloka, op. cit., pp. 281ff.

20Ibid., pp. 250, 461. samketakālamanasam vitam jñātvā vidagdhayā/ hasannetrārpitākūtam iīlāpadmam nimīlitam//

274

Here, the closing of the play lotus suggests that the woman

intends to meet her lover at sunset, when the lotuses close;

because the poet explicitly states that this action is meant

to convey such a hint to her lover, the verse must be regarded

as gunībhutavyangya, rather than as a true case of dhvani.

Similarly, where sọme particular intonation (kāku)

provides a cue which helps an audience to understand the

suggested meaning of a verse, the verse should be regarded as

gunībhutavyangya rather than dhvani. For example:

Having struck at our lives and our masses of wealth by the burning of the lacquer house, by poisoned food, by entering the gambling-assembly, and having pulled at the clothes and the hair of the wife of the Pandavas, do the sons of Dhrtarastra live confidently, while I am alive?21

The verse is spoken by Bhimasena, second eldest of the

Pandavas, angered by his older brother Yudhisthira's efforts

to make peace with the sons of Dhrtarastra and by hearing an

actor speaking a verse he believes to be a blessing for them.

There are no explicit verbal cues in the verse which indicate

that the verse is intended ironically, but the context makes

it perfectly clear that this is so, and in performance an

actor reciting the verse would ordinarily convey this ironic

21Tbid., pp. 477-8. lāksāgrhānalavisānnasabhāpravesaih prānesu vittanicayesu ca nah prahrtya/ akrsya pandavavadhuparidhanakesān svastha bhavanti mayi jīvati dhārtarāstrāh// (Bhattanārāyana, Venīsamhāra 1.8) Anandavardhana quotes only the final pada of the verse ("Do the sons of Dhrtarastra ...? ), the portion in which suggestive intonation would be employed.

275

intent by means of intonation. Thus the ironic intent of the

verse, although it is suggested by the context and the

character of the speaker, is, in performance, more explicitly

indicated by the actor's use of a particular tone of voice.

Just as when there is an explicit verbal reference to the

suggested sense of a verse, when the intonation used in

recitation indicates a suggested sense in this way, it must be

regarded as an instance of gunībhūtavyangya.

Thus there are several essentially unrelated criteria by

which a verse containing a suggested meaning may fail to

attain the status of dhvani and consequently be classified as

gunībhutavyangya.

Gunībhutavyangya as Anandavardhana defines it is a very

broad category -- broad enough to include within it virtually

all of the alamkāras whose definition and classification

absorbed most of the attention of earlier Alamkārikas. As we

have already seen (cf. above, Chapter Four, pp. 157ff),

Anandavardhana concedes that previous Alamkarikas have, in the

course of examining various figures of speech, dealt with all

three possible varieties of suggested meaning -- vastu (in

figures such as samāsokti, paryāyokta, etc.), alamkāra (in

figures such as rupaka and sasamdeha, which suggest an

unstated simile), and rasa (in figures such as rasavat,

preyasvat, etc.) -- albeit without recognizing these as

instances of the general phenomenon of suggestion. He argues

nevertheless that dhvani lies entirely outside the scope of

276

their poetic analyses. It is his contention that in the

instances of suggestion they have dealt with the suggested

meaning is invariably subordinated to the directly expressed

meaning. Therefore, all such figures can be subsumed within

the category of gunībhutavyangya.

The relegation of these alamkaras to the realm of

gunībhūtavyangya grew out of Anandavardhana's need to

establish the existence of dhvani as a type of poetry which

could not be adequately explained within the existing

categories of Alamkarasāstra, and which consequently demanded

the formulation of a new mode of poetic analysis. When,

toward the end of the third uddyota of the Dhvanyaloka,

Anandavardhana turns to examine gunībhutavyangya in more

detail, he carries the argument further. It is not simply

that there are a few selected alamkaras which can be included

within the scope of gunibhūtavyangya; virtually all alamkāras

can be shown, in most cases, to convey some suggested meaning,

and hence to be so included.

Like dīpaka, samāsokti, etc., other alamkāras also can be seen, for the most part, to touch on some suggested meaning, either another alamkāra or a vastu; since, first of all, all alamkāras can be made to contain the [suggested] alamkāra atisayokti ("hyperbole") .22

Ideally, all figures ought to contain some measure of

hyperbole or overstatement -- it is on their deviation from the

ordinary that the beauty of such modes of expression depends.

This assertion of the central role of hyperbole in figurative

22Ibid., pp. 464-5.

277 usage is not new to Anandavardhana, and he cites Bhamaha's remarks on atisayokti in support of his position. "All oblique [atisayokti]; meaning is intensified [vibhāvyate] by expression [vakrokti23] is just this this. A poet should make a great effort with respect to this -- what alamkara can exist without it?" -- Here the meaning that should be understood is this: an alamkāra in which atisayokti resides, due to the imaginative power [pratibhal of the poet, will possess a high degree of beauty; any other will be mere [alamkāramātra]. ornament Thus, because of its suitablity for incorporation in all alamkaras24, this [ [atiśayokti] is said, by figurative non-differentiation [abhedopacāra], to be identical with all alamkāras. 25 All alamkaras, when employed by an imaginative poet, can and should incorporate some element of hyperbole. This hyperbole may be explicitly stated, but will ordinarily be understood via suggestion. Whenever this is so, the alamkara in question will be included within the scope of gunībhūtavyangya (if the suggested hyperbole were not subordinate to the expressed figure which suggests it, this figure could not properly be considered an alamkära at all, but would rather be regarded as an instance of dhvani).

23Here referring not to a particular alamkāra (as in many later authors [eg. Vamana, Kavyalamkarasūtra 4.3.8; Rudrata, Kāvyalamkara 2.14-17; Mammata, Kavyaprakasa 9.1]), but to that indirect or out-of-the-ordinary character which is, or ought to be, common to all poetic usage. 24It would seem that these and the following remarks on the suggestion of atisayokti by other alamkaras ought to be restricted to arthalamkāras (meaning-based figures) alone -- it is hard to see how sabdālamkāras (sound-based figures) such as alliteration (anuprasa) or rhyme (yamaka) could ever be said to suggest the figure atisayckti -- but neither Anandavardhana nor Bhamaha explicitly so restricts it. 25Ibid., pp. 467-468.

278

Having explained the pervasive presence of suggested

hyperbole in alamkaras generally, Anandavardhana goes on to

list a wide variety of specific alamkäras which typically or

necessarily convey some subordinate suggested sense, and which

depend on this suggested sense for their beauty. Subordinated

suggestion, on Anandavardhana's view, plays so central a role

in the operation of the alamkaras that the analysis of

gunībhutavyangya-kavya in general can effectively substitute

for the analysis of particular alamkāras as practiced by

earlier Ālamkārikas.

Alamkāras such as rūpaka [i.e. the arthalamkaras], which acquire a high degree of beauty only when they touch on some suggested element, all fall within the scope of gunībhūtavyangya. And having a subordinate suggested meaning is s the general characteristic of all such [alamkaras], both those already explained [by previous Ālamkarikas] and those not yet explained. If this [gunībhutavyangya] is defined, then they are all adequately defined. They cannot be accurately known by the explanation of the particular form of each without any general infinite; just as definition, because h one they are

through word by word study [pratipadapātha] .26 words cannot be accurately known For the permutations of language are infinite, and the alamkāras are varieties of these. 27

The attempt to exhaustively define and categorize all

figures of speech, which has heretofore served as the primary

object of Alamkarasāstra, is, on this view, ultimately futile.

26An allusion to the paspasahnika of the Mahabhasya (Vrajajīvan Pracyabhāratī Granthamāla 23, Vol. 1, pp. 53-56); here Patañjali explains that the proper forms of words, because they are infinite, cannot be learnt through word by word study, and grammatical instruction must therefore be made by means of generally applicable rules (utsarga) exceptions to these rules (apavāda) . and

27Dhvanyaloka, pp. 472-3.

279

There is no limit to the possible permutations of figurative

usage, and thus no hope of achieving a complete taxonomy of

such usage by the description of specific figures of speech.

The role of subordinated suggestion in the figures is so

pervasive, however, that it may serve as the defining element

which underlies figurative usage in general. The

characterization and analysis of this underlying, defining

element can effectively take the place of the interminable and

ultimately hopeless task of defining and classifying the

figures one by one.

Anandavardhana began his work by sharply demarcating the

object of his own literary analysis, suggested meaning, from

the object of analysis for earlier Alamkarikas, directly

expressed meaning in the form of particular, expressed,

figures of speech (cf. Dhvanyaloka, 1.3-42ª). However, as we

have just seen, further analysis of the role and scope of

suggested meaning in poetry blurs and finally overwhelms this

demarcation: the theory of suggested meaning not only reveals

and explains the existence of a type of poetry, dhvani, which

had gone unrecognized by earlier literary analysts, but also

makes clear, in the form of subordinated suggestion, the

general, underlying principle on which figuration is based and

thus produces a more satisfactory account of the figures, the

primary focus of prior literary analysis, than previous

Ālamkärikas themselves were able to provide.

28Ibid., pp. 47-49.

280

Although Anandavardhana consistently refers to dhvani and

gunībhūtavyangya as "types" of poetry (kāvya-prakāra or kāvya-

visesa), it should be evident from the foregoing discussion,

and from the examination of the subordination of alamkäras to

rasa made earlier (cf. above Chapter Four, pp. 174ff), that

these two "types" are not mutually exclusive and may co-occur

in the same composition or even the same verse. The various

alamkaras, which are all or virtually all to be regarded as

subtypes of gunībhutavyangya-kavya, can and generally should

be employed so as to heighten the capacity of a verse or a

larger composition to suggest, as its primary meaning, a

particular rasa (making it, thereby, an instance of dhvani-

kāvya) .

Änandavardhana explicitly points out this overlapping of

"types" in his discussion of gunībhutavyangya. He is, as

always, anxious to emphasize the prevalence of dhvani,

particularly rasa-dhvani, in the great works of Sanskrit and

Prakrit poetry, and is therefore useful for him to show how

even verses and expressions conveying a subordinated suggested

sense may co-occur with and assist in the process of rasa-

dhvani.

Even this gunibhutavyangya-type may again take on the form of dhvani, when one considers its ultimate intent -- namely rasa etc.29

This can be seen clearly in the examples of gunībhūtavyangya

quoted above. The verse "Knowing that her lover was

29Ibid., p. 483.

281

considering what time they should meet ... " (samketakāla-

mānasam .. .-- cf. above, p. 273) conveys the subordinate

suggested meaning that the heroine wishes to meet her lover at

sundown (and thus may be considered an instance of

gunībhūtavyangya), but the ultimate object of the verse is to

suggest srngara-rasa. Likewise, in the example "Having struck

at our lives ... " (laksagrhanala ...-- cf. above, p. 274) the

words of the verse, aided, in performance, by a certain

intonation on the actor's part, convey, as a subordinated

suggestion, that the verse is intended ironically; the

ultimate purport of the verse, however, is to suggest vira-

rasa ("heroism") (or perhaps raudra-rasa ["fury"] ).

In this way it can be shown that a great many examples of

gunībhutavyangya-kavya are also cases of rasa-dhvani and,

having the suggestion of rasa as their primary and overriding

goal, such examples can easily.be explained and evaluated via

Anandavardhana's teleological mode of literary analysis.

Anandavardhana concedes, however, that gunibhūtavyangya need

not be, and is not invariably, directed toward rasa-dhvani.

But when an expression does not have rasa, etc. as its ultimate purport [tätparya], even when it is conveyed by words containing subordinate subordinated suggestion will a property of the whole.30 suggested meanings,

Even if the individual words or phrases of which a verse is

composed contain subordinate suggested meanings and are

therefore to be regarded as gunībhutavyangya, the verse as a

30Ibid., p. 485.

282

whole may have rasa as its primary object and so be considered

a case of dhvani; but, when such a verse does not have rasa as

its primary object, the entire verse will be regarded as

gunībhūtavyangya. There is then some domain for

gunībhutavyangya-kavya independent from that of dhvani; many

or most instances of gunibhūtavyagya may be rasa-directed and

hence ultimately included within the scope of dhvani, but, in

at least some cases, gunībhūtavyangya occurs without

culminating in any rasa-dhvani. As we shall see, the

existence of such poetry poses a serious problem for

Anandavardhana's poetic teleology, as, in the absence of rasa,

there is no meaningful criterion in terms of which the

aesthetic value of particular phonetic and semantic elements

may be evaluated.

Apart from dhvani-kavya and from gunībhūtavyangya-kavya

(in which rasa and other suggested meanings occupy a

subordinate role), Anandavardhana accepts the existence of a

third type of poetry, citra-kavya, in which suggested meaning

plays no role whatsoever. Such poetry depends solely on the

beauty of its sounds and its expressed meaning for its

capacity to give pleasure. Given Anandavardhana's argument

for the pervasive role of subordinated suggestion in the

figures of speech dealt with by earlier Ālamkärikas, the range

of phonetic and semantic elements which may be employed to

render such poetry beautiful is quite limited; there are,

however, at least a few figurative elements which are totally

283

devoid of suggested meaning.

There [in citra-kāvya] there is a certain element of variegation of sound [śabda-citra] in the form of difficult rhyming [duskara-yamaka31] etc .; apart from this variegation of sound, there is variegation of expressed meaning [väcya-citra] in the form of alamkāras such as "imagining" [utpreksa32] when they are established as the primary meaning of an expression, lacking any rasa-directedness and devoid of connection with any suggested sense.33

Anandavardhana takes a disparaging attitude toward such

poetry, which depends solely on the beauty of its sounds and

surface-meanings, and describes it as "an imitation of poetry

[kāvyānukāra] . 134

Despite his disparaging attitude toward citra-kavya,

however, Anandavardhana insists that citra-kāvya forms a real

and significant part of poetic literature as it is known to

him. He considers and decisively rejects an objection against

the very possibility of poetry totally devoid of suggested

meaning. The objector argues that, while it may be possible

to compose poetry in which there are no suggested vastus or

31Yamaka ("rhyme") may extend well beyond a single word or a few syllables -- it may extend to the juxtaposition of two or more quarter-, half-, or even whole verses which are completely homophonous but convey different meanings; those particular varieties of yamaka which are designated as duskara ("difficult") are illustrated in (sometimes known as Kavyalaksana) 3.38ff. (cf. Kavyalaksana of Dandin' s Kavyadarsa

Dandin, Anantlal Thakur ed., Darbhanga: Mithila Institute, 1957, pp. 214ff) .

32A figure which consists in the attribution to anything of properties or functions which could not possibly belong to it.

33Dhvanyaloka, op. cit., p. 495.

34Ibid., p. 495.

284

alamkäras, there can exist no poetry which does not, in one

way or another, suggest some rasa.

There can be no poem which does not refer to some entity [vastu]. And every entity in the world must certainly be subservient [anga] to some rasa [bhava], if only as a vibhāva. 35 or emotional state For rasas etc. are particular states of mind, and there is no entity which [when apprehended] does not produce some state of mind -- and, if it did not produce any state of mind, such an entity would not be a proper topic for poets in any case.36

All poetry must certainly be about something, and the things

described or referred to in a poem will necessarily generate

some sort of emotional response. Any such emotional response

will fall within the scope of one or another of the rasas or

bhavas (even indifference may be considered an emotional

response -- one which would give rise to santa-rasa

["tranquility"]). Given this, it would seem that any poem

whatsoever would have to suggest some rasa or rasa-like state,

and would therefore have to be classified as either dhvani or

gunībhūtavyangya, depending on the predominance or

subordination of this rasa with respect to the expressed

meanings which suggest it. No third category is possible, as

rasa can never be entirely absent.

In responding to this objection Anandavardhana turns once

again to the tools of Mimamsa discourse-analysis. The

35A factor which prompts or heightens an emotional response -- one of the three factors (along with anubhavas and vyabhicaribhavas) required for the suggestion of rasa (cf. above, Chapter Four, p. 130).

36Dhvanyaloka, pp. 495-6.

285

existence of citra-kävya is justified with reference to the

Mīmāmsā notion of vivaksa ("intent to express").

To this argument the following answer is made: True, there is no such type of poetry in which there is no awareness of rasa, etc. But, when a poet who is devoid of any intent to express rasa, bhava, etc. composes sound- or meaning-based alamkäras then, in dependence on that expressive intent, the meaning [of the poem] is determined to be devoid of rasa etc. For the meaning of words in a poem is only that which becomes the object of expressive intent. And if, in such a case, even in the absence of any expressive intent on the part of the poet, any awareness of rasa etc. should arise by virtue of the implicative capacity [samarthya] of the expressed meaning, this awareness will be extremely weak -- having determined absence of rasa in this way, the scope of citra-kāvya may be established.37

Any poem (indeed any comprehensible utterance whatsoever) will

refer to entities of some sort (whether real or fictitious),

the entities referred to will invariably provoke some sort

emotional response, and the poem in question may thus be said

to be suggestive of some rasa, bhava, or other rasa-like

experience; however, if a poet is intent solely on the

construction of alamkaras and composes totally without regard

for any emotional response his composition may generate, any

rasa or rasa-like experience suggested by this poem will be

incidental to the poet's expressive intent and its effect on

the hearer's understanding of the poem will be negligible. No

poetry can ever be totally devoid of rasa, but such poetry, in

which any suggestion of rasa is incidental and, as it were,

accidental, may legitimately be classified as citra-kāvya.

While he generally looks down on citra as a lesser

37Dhvanyaloka, pp. 496-7.

286

variety of poetry, Anandavardhana is ultimately unwilling to

exclude it entirely from the realm of literature. He does all

he can, however to restrict the scope of citra and emphasize

its marginality to literature as a whole. He does this, first

of all, by drawing a sharp distinction between the descriptive

and prescriptive imperatives of literary analysis:

This citra is considered by us because poets, speech is unconstrained [visrnkhalagirah], have been seen whose

to compose poetry totally without regard for any ultimate purport [tätparya] in the form of rasa etc .; but for present-day poets, now that a proper establishment of the principles of poetry is being made, there should be no mode of poetry apart from dhvani, since the practice of poets of mature capacity will not be appealing in the absence of an ultimate purport in the form of rasa etc. When there is an ultimate purport in the form of rasa etc., there is nothing which, made subordinate to the desired rasa, will [pragunībhavati] not attain high quality .38

Citra-kävya has formed part of the practice of poets in the

past; such poetry forms a part of the recieved literary

tradition, and the description and analysis of such poetry

therefore is and will remain part of the task of

Alamkārasāstra. But given the properly central role of

suggested meaning, especially rasa, in poetry (as revealed by

Anandavardhana's own literary analysis), dhvani is presented,

prescriptively, as that mode of poetry toward which all

present and future poets should aspire.

A place is retained for citra even in current practice

but, again, it is its marginal character that is stressed.

Although this instruction in the principles of poetry is

38Ibid., p. 497.

287

being made for contemporary poets, beginning poets who need practice may perhaps compose citra; but, for those poets who have attained full development, dhvani alone is poetry.39

The composition of citra-kavya may serve a useful purpose by

aiding a poet in the development of skill in the construction

of figures -- skill which is necessary for any poet -- but the

ultimate goal of all poets should be to compose suggestive,

rasa-directed poetry. Ultimately, then, Anandavardhana is not

willing to entirely banish citra-kävya even from the practice

of contemporary poets.

Thus, for Anandavardhana, all three of the logically

possible modes of poetry -- dhvani, gunibhutavyangya, and

citra -- really exist in the practice of poets both ancient and

contemporary, and he accords to all, even to citra, some

measure of legitimacy. He does all that he can to emphasize

that among these three modes dhvani is overwhelmingly

preponderant in the works of great poets. He strives to

sharply limit the independent scope of the other two modes --

in very many (though not all) cases, gunībhūtavyangya will

overlap with, and be subordinated to, rasa-dhvani; citra-

kāvya, while it may have been freely composed by poets in the

past, should be practiced by contemporary poets only as a

training exercise which will ultimately aid them in composing

dhvani-kāvya. But, however limited their scope may be,

Anandavardhana does admit the existence of significant bodies

39Ibid., pp. 499-500.

288

of poetry in which rasa plays either a secondary role or no

role at all, and sees the analysis of these bodies of poetry,

along with the analysis of dhvani-kavya, as part of the

function of Alamkārasāstra. This admission is problematic

for, as we shall see, Anandavardhana's approach to literary

analysis leaves him with little or nothing useful to say about

works which do not have rasa-dhvani as their ultimate goal.

The Collapse of Aesthetic Criteria in the Absence of Rasa

Anandavardhana's teleological mode of analysis requires

that a literary work have an ultimate and overriding purpose,

namely the suggestion of a single predominant rasa, toward

which all the formal elements of the work, from its individual

phonemes to its overall plot structure, should be subservient.

Nevertheless, given his unprecedentedly empirical orientation

and the close attention he pays to actual literary works, he

cannot help but acknowledge that, in fact, not all works have

rasa as their overriding goal. Anandavardhana is not willing,

as are several later Alamkarikas, to insist on rasa-

directedness as a necessary and constitutive property of

poetry; he does not unqualifiedly dismiss non-rasa-directed

works as non-poetry or defective poetry. In the absence of

rasa-directedness, however, there is no uniform, non-arbitrary

aesthetic criterion by which the employment of particular

formal elements in particular contexts can be evaluated.

Without rasa to serve as the telos of a literary work,

Anandavardhana's aesthetic theory cannot serve as a useful

289

guide either for poets in the composition of new works or for

readers in the analysis and judgement of existing ones.

Anandavardhana does not dwell at length on the silence of

his brand of literary analysis in the face of non-rasa-

directed literature, but does deal with it in passing. It can

be seen most clearly in his long and detailed discussion of

samghatana (compound-density -- literally "joining together")

and of the factors governing its use in particular poetic

contexts.40 We have seen (above, Chapter Four, pp. 183-186)

that Anandavardhana rejects the effort of earlier Alamkārikas

to account for the gunas ("qualities" -- sweetness [madhurya],

strength [ojas], and clarity [prasāda] ) as properties

belonging to words and their expressed meanings; for him,

gunas must be regarded as properties of particular rasas, not

of the words and meanings which suggest these rasas. In order

to justify this view of the nature of guna, Anandavardhana

takes great care to distinguish it from the closely related

notion of samghatana. Anandavardhana informs us4l that this

samghatana has been defined by others as being of three types:

asamāsa ("without compounds") , madhyamasamasa ("having medium-

length compounds"), and dīrghasamāsā ("having long

compounds"). No mention of this samghatana can be found in

the surviving works of pre-Ānandavardhana Ālamkārikas, but

4°Dhvanyaloka 3.5-9 (op. cit., pp. 309-328).

41Ibid., p. 309.

290

based on Abhinavagupta's commentary on this passage42 it

would seem to have been propounded by Udbhata (most probably

in some lost portion of the Bhamahavivarana).

Udbhata rejected the categorical distinction between

gunas and alamkaras, 43 and the concept of samghatana appears

to have arisen from an attempt to give a more strictly formal

characterization of the features associated with particular

gunas, and so to re-classify them as alamkāras. Phonetic

elements shown to be associated with gunas can with relative

ease be subsumed under the several varieties of alliteration

(anuprasa), but presence or absence of compounds, which plays

a major role in the definition and differentiation of

gunas, 44 has no place among the alamkäras accepted by earlier

theorists or those defined by Udbhata himself in the

42Tbid., p. 310.

43In his Kāvyaprakāśa (Oriental Research Institute Series 122, Mysore, 1977, Vol. 2, p. 213), Mammata paraphrases Udbhata's view as follows: "[Qualities] such as heroism operate through inherence [in a person], [ornaments] such as necklaces operate through contact: therefore [in the case of a person] let there be a difference between ['qualities'] and alamkaras ['ornaments']. gunas [Qualities] such as 'strength' and [ornaments] such as alliteration [i.e. the gunas and alamkaras found in poetry] both operate through inherence [in the words and meanings of a poem]: a distinction between these [can be asserted] only by blindly following [previous authors] like sheep [gaddurikapravahena] ." In his commentary on the passage, Manikyacandra identifies this as Udbhata's view, and Bhamahavivarana. quotes a relevant passage from his

44Long compounds, as we have seen (above, Chapter Four, pp. 184-185), are said to endow a composition with ojas ("strength"); the near or total absence of compounds is thought to endow it with madhurya ("sweetness").

291

Kāvyālamkārasarasamgraha. The analysis of samghatanā

presumably grew out of an effort to rectify this deficiency

and eliminate any need for gunas as a separate category.

Samghatana is a property of the words of a poem; gunas

are, for Anandavardhana, properties of rasas themselves, and

may only metaphorically be said to belong to the words and

meanings which convey these rasas. In order to uphold his

explanation of the distinction between guna and alamkāra,

Anandavardhana must show that gunas are not reducible to

samghatana

Given the acknowledged role of samghatana in imparting a

particular guna to a composition, 45 the relationship between

guna and samghatana may be explained in one of three possible

ways: guna and samghatana may be identical, guna may depend on

samghatana, or samghatana may depend on guna.46

Anandavardhana rejects the first two possibilities on the

grounds that gunas are inseparably associated with particular

rasas, while the several varieties of samghatana are not.

This dissociation between guna and samghatana can be clearly

seen in the verse Yo yah sastram ... ("Anyone among the army of

the Pandavas ... " -- cf. above, Chapter Four, p. 185). Here the

quality ojas ("strength") is clearly present, but does not

45As in the verse cancadbhujabhramita ... ("O queen, Bhima will bind up your hair ... " -- cf. above, Chapter Four, p. 184) , in which the quality ojas ("strength") is dependent, at least in part, on the predominance of long compounds.

46Dhvanyaloka, p. 310ff.

292

depend on any unusual density of long compounds.

Ojas is, as we have seen, a property of particular rasas

(in the present case, raudra ["fury"] ). Any verse which

effectively conveys such a rasa will, by virtue of that fact

alone, possess the quality ojas. A high density of long

compounds may in some cases serve to convey such a rasa, and

consequently to endow a verse with ojas, but is not a

necessary condition for the production of this quality. As

particular gunas can be produced even in the absence of the

varieties of samghatana conventionally associated with them,

they cannot be said to depend on samghatana. Where a

particular variety of samghatana aids in the production of a

particular quality, its role will be subordinate and hence

dependent.

By demonstrating the disjunction in practice between

particular qualities (e.g. ojas) and the particular varieties

of samghatana typically associated with these qualities (e.g.

a high density of long compounds), Anandavardhana is able to

bolster his argument that the production of gunas in a

literary work ultimately depends not on the use of specifiable

sets of formal language elements but rather on the suggestion

of particular rasas. But the disjunction between guna and

samghatana raises an additional question: if the employment of

a given variety of samghatana is not (like the presence a

particular guna) invariably associated with a specific rasa or

set of rasas, what is it that governs, or ought to govern, the

293

use of particular varieties of samghatana in particular

contexts?

Anandavardhana offers several factors which ought to

govern a poet's use of samghatana, factors which are partly

dependent on rasa and partly not. The type of samghatana to

be employed in any given verse or passage should be restricted

so as to be appropriate to its speaker and the expressed

meaning it conveys,47 and appropriate also to the genre of

the work in which it occurs.48 In the case of speaker and

that of expressed meaning, "appropriateness" is determined

largely on the basis of rasa; the appropriateness of

particular types of samghatana to particular genres has, as we

shall see below, no apparent basis in rasa.

As a general rule, when the speaker of a verse (either

the poet, or some character created or depicted by the poet)

is endowed with emotion in the form of rasa or bhäva, and when

this rasa or bhäva is the overriding object to which the

expressed meaning of the verse is subordinate, samghatanā

should be restricted to asamasa ("without compounds") or

madhyamasamasa ("having medium-length compounds"). When the

rasa in question is pathos (karuna) or love-in-separation

(vipralambha-srīgara) it should be limited to asamasa alone.

These restrictions are justified as follows:

When rasa is the main thing to be communicated, factors

47Ibid., p. 318.

48Ibid., p. 323.

294

which obstruct or conflict with the cognition of this rasa are to be totally avoided. And samghatana which contains long compounds [dīrghasamāsa], because these compounds can be interpreted in various ways, sometimes obstructs the cognition of rasa; thus, the excessive employment of this [dirghasamasa] is not appealing. This is especially so in the case of dramatic poetry and, apart from this, in the case of karuna or vipralambha- srīgara. For, because of the delicateness of these two rasas, if there is even a slight lack of clarity in word or meaning, the disturbed. cognition of these rasas will be But when some other rasa, such as raudra ["fury"] is to be communicated, samghatana with medium- length compounds or sometimes, when concerned with the actions of steadfast and arrogant hero, even with long compounds will not be flawed, if employed in dependence on an expressed meaning which is suitable to the rasa in question and necessarily connected with the introduction of this samghatana. Thus even this [dīrghasamāsa] is not entirely to be avoided. 49

A high density of long compounds creates a potential for

ambiguity from which the obstruction of rasa may result. It

is consequently to be avoided generally, and particularly when

one intends to suggest a rasa such as karuna which is

especially susceptible to such obstruction. Even in the case

of rasas such as raudra, which are normally thought to be

fostered, rather than impaired, by a high compound-density,

dīrghasamāsā samghatana should be employed only where

appropriateness to both speaker and expressed sense demands

it.

The sort of restrictions Anandavardhana imposes on the

use of samghatana in rasa-directed poetry should be quite

familiar; they are decidedly similar to those limitations

Anandavardhana elsewhere imposes on the use of sound-based

49Ibid., pp. 320-321.

295

figures such as alliteration (anuprāsa) and rhyme (yamaka).50

Here, however, he explicitly considers, as he does not do in

his examinations elsewhere of context-based restrictions on

the use of the alamkäras and other formal poetic elements, the

status of literature which does not have the evocation of rasa

as its dominant object.

When the poet, being the speaker, is devoid of rasa or bhava, the samghatana is unrestricted; likewise, when the speaker is a character depicted by the poet and is devoid of rasa or bhava, the samghatana is similarly unrestricted.51

In stark contrast to the detailed set of regulations governing

the use of samghatana in rasa-directed literature, these brief

remarks reveal the total silence of Anandavardhana's mode of

literary analysis when faced with literature in which rasa is

not predominant. The restrictions he sets forth for the use

not merely of particular varieties of samghatana, but of

alamkāras and all other poetic elements, depend entirely on

the presence of a dominant rasa, which serves as the telos,

the overriding purpose, of a literary work. In the absence of

such a telos, there is no criterion in terms of which the use

of particular elements may be evaluated. If a literary work

is not ultimately rasa-directed, there is nothing to be said,

positive or negative, about the use of a particular variety of

samghatana (or any other poetic element) in a given context.

Anandavardhana's close examination of the factors

50Cf. above, Chapter Four, pp. 174-177.

51Ibid., pp. 319-320.

296

determining the use of samghatana has still more disturbing

implications for his literary theory: even where rasa is

predominant, there are factors unrelated to it which partly

govern the use of samghatana. In addition to the limitations

on the use of samghatana imposed by appropriateness to the

speaker and to the expressed meaning of a given verse or

passage (which, as we have seen, have their ultimate basis in

rasa) Anandavardhana outlines a set of genre-based

restrictions on samghatana which are largely independent of

the requirements of rasa.

Anandavardhana lists a total of eleven verse- and two

prose-genres, and notes briefly the restrictions on samghatana

appropriate to each.

In muktakas,52 if the poet is intent on introducing rasa into his composition, will depend on that appropriateness [of samghatana] [rasa] . This has been shown already.53 Otherwise, In sandānitakas etc., samghatana is unrestricted .... 54 because of appropriateness tc their forceful composition, samghatana is only that with medium or long compounds; if they occur in larger compositions, one should conform to appropriateness as stated for that composition. But, in a paryayabandha, 55 samghatana is only that with no compounds or medium

52Single verses which do not form part of any larger composition.

53I.e. in the above discussion of appropriateness to speaker and expressed meaning.

54Sandānitakas, visesakas, kalāpakas, and kulakas -- the next four genres listed. They are syntactically connected clusters of, respectively, two, three, four, and five or more verses; they may stand alone, like muktakas, or form part of some larger composition.

5According to Abhinavagupta, independent verses dealing with a single topic. a set of syntactically

297

compounds; even if, sometimes, depending on appropriateness to the meaning, samghatana with long compounds is also employed, parusa ["harsh"] and gramya ["rustic"] vrttis56 should be avoided. In parikatha, 57 samghatana is unrestricted, because there is no great intent to introduce rasa into the composition, as it is mere plot that is presented. In khandakatha and sakalakathā, 58 which are found chiefly in because they consist mostly of kulakas and so on, 59 Prakrit,

there is no difficulty even if long-compounded samghatana is employed, but the appropriateness of vrtti according to rasa should be adhered to. In a mahākāvya, if rasa is the ultimate purpose, appropriateness is according to rasa; otherwise, samghatana is unrestricted .... In drama [abhineyartha], however, introduce rasa should be made .... a thoroughgoing effort to In an ākhyayika, 60 for the most part, samghatana is only medium-compounded or long-compounded, because the beauty of prose depends on its forceful composition, akhyayika, this forceful composition and because, in an is at its most intense. But, even though prose possesses a high degree of forceful composition, in a katha, appropriateness of samghatana such has been spoken of for the introduction of rasa should be followed.61

This list of genre-based restrictions, however accurately

it may reflect the actual practice of poets in composing works

56Cf. above, Chapter Four, pp. 189-190.

57According to Abhinavagupta, the relating of a number of stories concerned with a single purusartha ("human aim" -- they are traditionally four: dharma [loosely, righteousness or good conduct], artha [wealth], kama [physical pleasure], and moksa [spiritual liberation] ).

58" Partial narrative" and "complete narrative", respectively.

59I.e., clusters of syntactically connected verses, as described above.

60The two main prose genres recognized in Alamkārasastra are akhyayika and katha. The criteria for differentiting these genres are not firmly settled. Bhamaha (1.25-29) offers several formal and content-based restrictions on the two genres, but these are challenged by Dandin (1.23-30), who rejects altogether the distinction between the two genres.

61Dhvanyaloka, pp. 324-327.

298

of particular genres, is, in terms of Anandavardhana's

aesthetic theory, totally inexplicable. Even where he does

give a reason for the restriction of samghatana in a

particular genre, the reason itself cries out for explanation,

and receives none. Why should the beauty of verse-clusters,

such as sandānitakas, and prose compositions (akhyāyikas and

kathas) depend on "forceful composition" to a greater degree

than that of other genres? Why, despite this need for

forceful composition, should rasa-based constraints on

samghatana remain predominant in kathā, but not in ākhyāyikā?

Why should rasa and rasa-based constraints on samghatana be

rejected in favor of mere plot-development in parikatha and

not in other genres? No attempt is made to provide any

rationale for the aesthetic preferability of particular

varieties of samghatana in particular genres. What is

presented seems to be nothing more than a set of purely

arbitrary stipulations.

It would appear that, in this investigation of

samghatana, theory has given way to observation.

Anandavardhana finds that there are, in practice, certain

genre conventions governing the use of samghatana, and accepts

these conventions as binding despite the fact that they cannot

be justified or explained within the framework of his poetic

teleology. This is not the only instance in which this

occurs. We have already seen (cf. above, Chapter Four, pp.

155-157) that, after concluding a detailed investigation of

299

conflict between primary and subordinate rasas and the means

by which it may be avoided or dissolved, Anandavardhana makes

a special dispensation: those factors (vibhavas, etc.)

associated with srngara-rasa may be employed in any poetic

context, even when, by the standards he has just articulated

they conflict with the primary rasa of the text or passage in

question. This provision seems to accord well with the actual

practice of Sanskrit poetry -- srīgara-rasa and the factors

associated with it tend to be employed pervasively, even in

seemingly incongruous settings -- but is theoretically

unjustifiable and stands in direct contradiction to

Anandavardhana's own analysis of conflict between rasas.

Anandavardhana's close attention to the actual practice

of poetry thus raises two major difficulties for his

teleological model of literary aesthetics, both revealed most

clearly in his analysis of the factors governing samghatana.

First, there exists a body of poetic work, however small, in

which rasa is not of primary importance and may even be

totally lacking. When confronted with such poetry

Anandavardhana's mode of literary analysis loses all

explanatory and critical power; in the absence of a

predominant rasa to serve as an ultimate goal, there is no

standard of aesthetic value to guide poets in composing or

critics in evaluating a literary work. Second, even where

rasa is predominant, there remain certain conventions of

poetic usage which have no bearing on the capacity of a work

300

to effectively convey its primary rasa. Such conventions can

have no justification within the teleological framework of his

literary theory, but Anandavardhana is nonetheless unwilling

to dismiss them as illegitimate or irrelevant to the

composition of good poetry.

Conclusion

Thus one can discern in the Dhvanyaloka a certain tension

between theory and observation. Ānandavardhana, by

introducing into Alamkārasāstra the teleological mode of

discourse analysis developed by the Mimämsakas in their

analysis of Vedic texts, is able to create a coherent theory

of literary aesthetics which has tremendous explanatory power.

The vast array of unexplained and seemingly arbitrary rules

for poetic composition advanced by previous Alamkārikas is

replaced by a single, overriding principle: any poem should

have a single, predominant rasa, and the selection and

arrangement of all the formal elements of this work should be

done in such a way as to intensify, rather than hinder, the

capacity of the work to convey this rasa.

Observation of actual poetic practice, however, reveals

that not all poetry conforms to the expectations generated by

this theory. Not all poems have a primary rasa, or, indeed,

any rasa at all. Even when there is a single, primary rasa,

the employment of particular phonetic and semantic features in

particular contexts is not always unequivocally governed by

the requirements of this rasa.

301

This tension could easily be resolved. One could simply

endorse without reservation the critical implications which

follow from this aesthetic theory. Any coherent literary work

must have the evocation of a single, dominant rasa as its

ultimate goal. Any poetic work which does not have such a

goal is defective poetry or is not poetry at all, properly

speaking. Poets who allow their use of particular elements in

particular contexts to be guided by arbitrary genre

conventions or other factors unrelated to the requirements of

their primary rasa are committing an error, and their works

will suffer aesthetically as a result.

Alternatively, one could allow oneself to be guided by

one's observation of literary practice, and restrict the

applicability of this teleological model of poetry

accordingly. A poem may, and frequently does, have as its

goal the evocation of a single predominant rasa, and when this

is the case certain restrictions on the use of alliteration,

metaphor and various other features of sound and sense ought

to apply. A poet may legitimately choose, however, to direct

his efforts toward something other than the evocation of rasa,

and in such cases these restrictions will not be applicable.

Rasa certainly plays an important role in poetry generally,

but need not be all-important in all cases.

Anandavardhana is unwilling to resolve the apparent

tension between his theory and his observation of poetic

practice in either of these ways. Instead, he does all that

302

he can to minimize the scope of this dilemma, by attempting to

show that most poetry is rasa-directed and that most of the

accepted principles of poetic composition can be explained and

justified in terms of the need to effectively suggest rasa.

Nevertheless, from Anandavardhana's work, which exercises a

decisive influence on virtually all subsequent Alamkārasāstra

in Kashmir, there emerge two possible courses for further

investigation which are somewhat at odds with one another.

The vast majority of Kashmiri Alamkarikas in the ensuing

centuries follow the first of the courses outlined above,

supporting or at least tacitly accepting Anandavardhana's

rasa-centered poetic teleology and either ignoring the

existence of poetry which does not conform to the expectations

generated by this model or rejecting any poetry which does not

so conform as necessarily defective. The attention of these

Alamkārikas comes to be directed almost entirely toward the

theoretical issues raised by Anandavardhana's approach to

literary analysis (chiefly the question of whether this model

requires the postulation of a previously unrecognized and

uniquely poetic mode of signification [vyañjana], as

Anandavardhana contends, or can be accounted for solely in

terms of the generally recognized categories of philosophical

or literary language analysis); the kind of empirically

oriented analysis and critique of existing literary works

which we find in the Dhvanyaloka is largely neglected in their

works.

303

The one notable exception to this tendency in post-

Ãnandavardhana Alamkārasāstra is Kuntaka who, in his

Vakroktijīvita (cf. below, Chapter Eight), concentrates on

this kind of empirically oriented literary analysis and

generally ignores or sidesteps the theoretical controversies

which, for his contemporaries, have become the primary focus

of interest in Alamkāraśāstra. Among the Kashmiri

Ālamkārikas, it is only in his work that we find serious

(though largely tacit) resistance to the teleological approach

to poetic analysis and the concomitant insistence on rasa-

centrality as a necessary condition of all good poetry. revolut Anandavardhana's revolutionary approach to literary

analysis initiated a dramatic transformation in

Alamkarasastra, not merely challenging the conclusions drawn

by earlier authors in the field but establishing a new

understanding of what Alamkārasāstra was, altering

significantly the questions that could be asked regarding the

nature and function of literary works and the requirements

which theories of literary aesthetics were expected to meet.

In the following chapters we will examine the development of

Kashmiri Alamkārasāstra in the wake of this transformation.