12. CHAPTER NINE ABHINAVAGUPTA

CHAPTER NINE ABHINAVAGUPTA

The challenge posed to existing theories of language by

Anandavardhana' s Dhvanyaloka generated great controversy among

the Alamkarikas of Kashmir, and his theory of suggestion

became the major point of contention among his successors;

nearly all later Ālamkārikas clearly position themselves as

either adherents or opponents of this theory. The earliest

extant work which defends Anandavardhana's view of suggested

meaning is the Locana, Abhinavagupta's commentary on the

Dhvanyaloka.1

Abhinavagupta, apart from his contributions to literary

and dramatic theory, wrote many works on Kashmir Saiva

philosophy and tantra and several Saivite stotras;2 as he

lIt is not the earliest such work; : in the Locana Abhinavagupta refers frequently (and usually disparagingly) to the Candrika, an earlier commentary on Anandavardhana's work (for a collection of these references, cf. Kane, History of Sanskrit Poetics, Delhi: Motilal Banarsidass, 1971 [fourth ed.], p. 207). The work is also referred to by Mahimabhatta (Vyaktiviveka 1.5) and Ruyyaka (in his commentary thereon):

2The most comprehensive study of Abhinavagupta's life and works is K.C Pandey's massive Abhinavagupta: An Historical and Philosophical Study (Varanasi: Office, 1963 [second ed.]). Chowkhamba Sanskrit Series This work remains the best available source for general information on Abhinavagupta, but, as Ingalls has noticed (cf. The Dhvanyaloka of Anandavardhana, p. 33), the chronology of his works given by Pandey is seriously flawed. Misled by a false reading in the 1891 Nirnaya Sagar Press edition of the Dhvanyaloka (p. 19), Pandey supposes that Abhinavagupta refers in the Locana to his own Tantraloka (the reference is actually to Anandavardhana's Tattvaloka -- cf. Dhvanyaloka, Kashi ed., p. 67), and therefore believes that the Tantraloka and those of his own works which Abhinavagupta refers to therein must predate the composition of the Locana. In the absence of this supposed reference, the chronology of Abhinavagupta's works remains uncertain.

431

432

himself gives the dates of composition for several of these

works, the period of his literary activity can easily be

fixed His Kramastotra is dated A.D. 990-91, his

Bhairavastotra A.D. 992-93, and his Īsvarapratyabhijña-

vivrtivimarsinI A.D. 1014-15;3 while the dates of the Locana

and the AbhinavabharatI (his commentary on the Natyasastra)

cannot be fixed exactly, they must certainly have been

composed in the latter decades of the tenth century or the

earlier decades of the eleventh.

Abhinavagupta's determined efforts to refute those

objections raised against the theory of dhvani and his

elaborate analysis of the nature of rasa (cf. below, pp.

455ff) proved to be very influential among later followers of

Anandavardhana (Mammata in particular). He has generally been

accorded the highest possible praise by modern scholars of

Alamkārasāstra. Kane describes his commentaries on the

Dhvanyaloka and the Natyasastra as "monuments of learning,

critical insight, literary grace and style".4 De calls them

"remarkable" and observes that "his erudition, reputation and

influence as a great scholar and philosophical writer of his

generation, no doubt, added weight to his championship of

[Anandavardhana's] theory, and contributed a great deal to its

3Kane, History of Sanskrit Poetics, op. cit., p. 242; De, History of Sanskrit Poetics, op. cit., Vol. I, p. 112.

'Kane, History of Sanskrit Poetics, op. cit., pp. 238-9.

433

ultimate exclusive acceptance in later Poetics. "5

Krishnamoorthy describes him as "much more than a mere

commentator explaining the difficulties of text; he is an

original thinker, representing the whole theory in fuller and

more comprehensive form. "6

This last comment is particularly interesting, as it

seems to indicate that Abhinavagupta may be regarded as a

better exponent of Anandavardhana's theory than Anandavardhana

himself. There is at least an element of truth in this

implication; he is certainly more self-assured than

Anandavardhana in his defense of dhvani, and, as we shall see,

is willing to carry the implications of Anandavardhana's

teleological, rasa-centered model of literary aesthetics to

their logical conclusion where Anandavardhana himself

hesitated to do so.

Ānandavardhana, as we have already seen, is sometimes

rather tentative in his confrontation with rival language

theorists. He takes great pains to demonstrate that dhvani

cannot be subsumed under the recognized semantic function of

figurative signification (gunavrtti or laksana) or explained

as a kind of inference. But, in both cases, he seems

unwilling to rely fully on the effectiveness of his own

arguments and claims that, for his purposes, it doesn't really

5De, History of Sanskrit Poetics, op. cit., Vol. I, p. 110 and Vol. II, p. 176.

6Krishnamoorthy, The Dhvanyaloka and its Critics (Mysore: Kavyalaya Publishers, 1968), p. 217.

434

matter whether or not dhvani is reducible to figurative

signification or inference (cf. above, Chapter Five, pp. 230-

232, 242, 247-248). He tries to show why adherents of each of

the major language theories advanced by the philosophers ought

to accept the existence of suggestion as a distinct semantic

function, but carefully avoids taking a definite stand on any

of the issues which divide these philosophers among

themselves; he offers no general language theory of his own,

but hopes to show that his analysis of poetic meaning can co-

exist comfortably with any of the major existing theories. He

does all that he can to highlight the vital role of rasa in

poetry and to show that the most effective literary works are

those which have as their overriding goal the conveying of a

single primary rasa, but nevertheless is ultimately unwilling

to insist on rasa-centrality in poetry or to condemn all non-

rasa-directed poetry as sub-standard. Perhaps the most

striking difference of Abhinavagupta's approach from that of

Anandavardhana is the total absence of this cautious,

concessive, and circumspect attitude.

Abhinavagupta's Divergence from Anandavardhana

Unlike Ānandavardhana, Abhinavagupta does not leave his

reader in the dark as to his views on the major linguistic

controversies of his day. In his exhaustive comments on

Anandavardhana's arguments for the existence of suggested

meaning (1.4)7 and for the distinction of dhvani from

7Dhvanyaloka, Kashi ed., pp. 48-84.

435

figurative expression (1.14-19),8 he sets forth a definite and

specific account of the process of verbal cognition.

For Abhinavagupta this process may be divided into four

phases (kaksyas), corresponding to the four semantic functions

he recognizes. The first phase is that of denotation

(abhidha), which he explains as follows:

The function of denotation operates with respect to word meanings, which consist of universals [ For denotation is the power of conveying meanings [sāmānya] .

dependence on convention [samaya], and convention exists in

only with respect to this [universal], not with respect to the particular [visesa], because particulars are infinite and because a word would deviate from any one particular.9

On hearing a sentence we at first understand, by abhidha, only

the separate meanings of the individual words; the

understanding of a single coherent sentence meaning arises

only subsequently. These individual word meanings are

8Ibid., pp. 141-164.

9Ibid., p. 56. Here (and elsewhere) Abhinavagupta's theory of linguistic cognition combines elements of both Mimamsa and Nyaya models. Like the Naiyayikas, he believes the association of words with their denoted meanings to be conventional (rather than natural, as the Mimamsakas assert) . However, he follows the Mimamsakas in insisting that words can denote only universals (whereas the Naiyayikas believe that a word denotes a particular as qualified by a universal). His arguments against accepting particulars as word meanings are the standard ones offered by the Mimamsakas: we could never learn to associate the word "cow" with all particular cows (since their number is limitless); on the other hand, if we learnt to associate it only with one or more particular cows known to us, we would not understand (as we do) when it was used to refer to other, unfamiliar, cows. The facts of usage can be accounted for only if we assume the meaning of the word "cow" to be that which all cows (and only cows) have in common, namely the universal property "cow-ness" -- cf. above, Chapter Three, pp. 71ff.

436

invariably abstract, repeatable universals rather than

particulars.

The semantic function operative in the second phase of

verbal cognition which allows separate words to convey a

unified sentence meaning Abhinavagupta calls tātparya-sakti

("the power [to convey] overall purport") .10 This unified

sentence meaning, unlike the individual word meanings which

give rise to it, is particular rather than universal:

After this [denotation], tätparya-sakti operates with respect to the sentence meaning, in which [the individual word meanings] are construed with one another and which has the form of a particular, in accordance with the principle that "universals, because they could not otherwise be established, cause one to understand the particular. #11

Individual words can denote only universals, but language is,

at least typically, used to refer to and describe particulars.

If one is told to "Bring the cow," one cannot bring the

universal "cow-ness", but only some particular cow; the need

to combine the separate word meanings into a single

intelligible sentence meaning forces us to understand that a

10The first to argue for the existence of tātparya-śakti seems to have been Jayantabhatta (cf. Nyayamañjari [Varanasi: Śivakumārasāstri-Granthamalā 5, 1983], Vol. II, pp. 156, 158- 161). Jayantabhatta's tatparya-sakti serves essentially the same role as does laksana in Kumarilabhatta's theory of sentence meaning (cf. above, Chapter Three, pp. 115-117); through this innovation Jayantabhatta is able to justify the general intuition that laksanā is an anomalous mode of expression (rather than one common to all sentences), albeit at the expense of a certain degree of theoretical economy (he is forced to postulate three semantic f Kumarila can get by with only two). functions, while

1lIbid., pp. 56-7. The quotation is Ślokavārttika (Arthapatti-pariccheda 70cd) . from Kumarila's

437

particular cow and a particular act of bringing are intended.

The same principle applies to all sentences; the universals

denoted by the individual words, by mutually qualifying and

specifying one another, are able to refer to a particular

action or situation. It this capacity of mutual specification

that Abhinavagupta (and Jayantabhatta) call tātparya-sakti.

The third phase of linguistic cognition is that at which

figurative signification (laksana) takes place. It is only

after the operation of tatparya-sakti, Abhinavagupta argues,

that laksana becomes a possibility. We resort to the device

of figurative interpretation only when the attempt at literal

interpretation fails; if the words of a sentence, taken

literally, are unable to convey a coherent sentence meaning

via tätparya-sakti, we try to find coherence by taking one or

more of these words figuratively. For example:

"The boy is a lion" [simho manavakah] -- here, after the appearance of an obstruction to that construal introduced by tatparya-sakti which entered during the second phase, a third function, called laksana, appears, distinct from abhidha and tatparya-sakti and capable of removing this obstruction. 12

Tatparya-sakti and the grammar of Sanskrit demand that we take

the terms "boy" and "lion" as co-referential, yet we know that

there can be no referent which meets both specifications.

Because we know that no one thing can be both a boy and a

lion, we must take one or the other of these two terms

figuratively -- context will ultimately determine which. I£

12Tbid., pp. 58-9.

438

the expression is used to refer to a boy (as it is typically

understood to do), we will understand from it not that the boy

is a lion, but that he possesses certain lion-like qualities

(bravery, ferocity, etc.) .

It is only in the fourth phase of verbal cognition, after

these three semantic functions have done their work, that

suggestion (vyañjana or dhvanana) comes into play.

That power to reveal meanings which is rooted in the understanding of the meanings produced by these three functions and is assisted by the imaginative capacity of the interpreter (which has been prepared1 by awareness of these meanings) is the suggestive function.14

This fourth semantic power, Abhinavagupta insists, cannot be

reduced to or identified with the third, laksana. Those who

would include suggestion within laksana (here Abhinavagupta

surely has Mukulabhatta in mind) admit that there are three

conditions which must hold if laksana is to take place: there

must be some obstacle precluding a literal interpretation of

a word or phrase, there must be some connection between the

literal and the intended figurative meaning of this word or

phrase, and there must be some motive (prayojana) on the

speaker's part for employing a figurative rather than a

literal mode of expression.15

We need not depend on the expression itself to make us

"3pavitrita -- literally "purified".

14Ibid., pp. 62-3.

15Tbid., pp. 59ff. Also cf. above, Chapter Three, pp. 111-112 and Chapter Seven, pp. 336ff.

439

aware of the first two conditions; when we hear the expression

"A village on the Ganges" (gangayam ghosah), we know from our

own experience that a village cannot actually be built on a

river, and also that every river has adjacent to it a bank (on

which a village could be built). The third condition required

for laksana, the prayojana (e.g. the speaker's wish to

indicate the purity of the village through its close

association with the holy Ganges), cannot, Abhinavagupta

argues, be known in a similarly extra-linguistic manner; the

expression itself must somehow convey this intention to us.

The prayojana which motivates a particular case of figurative

expression cannot itself be conveyed by a further operation

figurative expression: if this were so, the second figurative

expression would require its own prayojana, demanding yet

another operation of figurative expression and yet another

prayojana -- an infinite regress would result, and the

expression would thereby fail to convey its intended meaning.

Our understanding of the prayojana of a figurative expression

can be explained only if we allow for an additional semantic

function capable of conveying it (one which does not depend on

the same conditions) -- namely, suggestion. It cannot be

tenably maintained that dhvani is a kind of laksana, since

laksana itself cannot be adequately explained unless one

allows for the existence of suggestion as a distinct semantic

function.

Thus there are, for Abhinavagupta, four semantic

440

functions and, correspondingly, four successive phases of

verbal cognition. By presenting the reader with this detailed

account of the process by which language conveys meaning,

Abhinavagupta not only goes far beyond the apparent intentions

of the author he is commenting on -- Anandavardhana, as we saw

(above, Chapter Five, pp. 248ff), declines to offer a general

language theory of his own, striving rather to show why the

existence of suggestion should be accepted by adherents of all

the major language theories then current, without himself

taking a stand on the major issues which divide these

theorists from one another -- but flatly contradicts what

Anandavardhana does have to say about the workings of

language. Anandavardhana knows and says nothing of the second

of Abhinavagupta's four functions, tätparya-sakti: it is first

mentioned, as far as we know, by Jayantabhatta, who post-dates

him. Furthermore, Anandavardhana explicitly states that there

are three semantic modes : direct expression, figurative

expression, and suggestion. 16 Abhinavagupta seems untroubled

by the contradiction between Anandavardhana's view and his

own, and remains silent on this point.

This is far from the only area in which Abhinavagupta

diverges from Anandavardhana's own statements and apparent

intentions in commenting on his work. Anandavardhana's work,

as we have seen, drew a great deal of criticism from defenders

of existing poetic and linguistic theories; the task

16Tbid., p. 429.

441

Abhinavagupta sets for himself in his commentary on the

Dhvanyaloka is not only to explain the text as fully as

possible but to defend it in the face of the many objections

raised against it. In the course of this defense he sometimes

seems to be trying to save Anandavardhana from himself:

resorting to strained and far-fetched interpretations to

justify apparently confusing or contradictory statements or

arguments made by Anandavardhana, and downplaying or ignoring

Anandavardhana's arguments where they seem inadequate and

substituting his own.

Sometimes it is merely a seemingly careless formulation

on Anandavardhana's part which Abhinavagupta feels a need to

justify. For example, in Anandavardhana's definition of

dhvani there is an apparent grammatical error:

That type of poetry in which either a meaning or a word, subordinating itself or its own meaning [respectively], suggest [sic.] that [non-expressed] "dhvani" by wise people.17 meaning is called

The subject of the relative clause "either a meaning or a

word" ought to require a verb in the singular, yet

Anandavardhana has used the dual form vyanktah (" [these two]

suggest") ; the error was apparently criticized by

Bhattanāyaka18, and Mahimabhatta provides an extensive

criticism of both Anandavardhana's formulation and

17Ibid., p. 103.

18Cf. Locana, op. cit., p. 104.

442

Abhinavagupta's attempted defense of it.19

Rather than simply admit that Anandavardhana is guilty of

a small error in phrasing, Abhinavagupta provides an

explanation which he sees as justifying the apparently

aberrant usage. Word and (expressed) meaning, he argues,

always co-operate in the communication of suggested meaning.

In some types of dhvani suggestion will depend primarily on

meaning while in others word will be the predominant

suggestive factor; nevertheless word (in cases of the first

type) and meaning (in cases of the second) will always play

some role in the process of suggestion, albeit in a

subordinate capacity. The primacy of either meaning or word

in any particular case of dhvani justifies the disjunctive

phrasing of the subject ("either a meaning or a word"), while

the fact that both must always operate in any case justifies

the use of a dual verb.20 The interpretation is extremely

strained; whether or not the contention that word and meaning

always co-operate in suggestion is a valid one, there is no

indication here or elsewhere in the Dhvanyaloka that

Anandavardhana holds this view.

A similar case may be ound in Abhinavagupta's

explanation of karika 1.2. Anandavardhana, introducing the Ãnar distinction between directly expressed and understood (i.e.

suggested) meaning, makes the following statement:

19Vyaktiviveka, Kashi Sanskrit Series 121, pp. 94-6.

20Ibid., p. 104.

443

That meaning, honored by connoisseurs, which is established as the soul of poetry, is regarded as having two types [bhedau], called "understood" [pratiyamāna] .21 "expressed" [vācya] and

Anandavardhana's description of meaning, in both its expressed

and suggested varieties, as "the soul of poetry" is troubling,

given his repeated statements elsewhere that dhvani alone is

the soul of poetry. 22

Again, rather than allow that Anandavardhana has spoken

somewhat carelessly, Abhinavagupta tries to justify the text

as written through a rather convoluted interpretation of the

kārikā:

The term "poetry" is not applied merely by virtue of

utterances; therefore he says "honored by connoisseurs". meaning, since it is not applied to everyday or Vedic

This single meaning [i.e. that which is honored by connoisseurs] is analyzed by discerning people into two components. For, if the nature of meaning is the same [in everyday and Vedic language and in poetry], why should connoisseurs honor only one particular type? There must be some difference [in poetic meaning]. That difference is the understood portion which, since it is the basis of its differentiation, is established by discerning people as the soul [of poetry]. But some, their minds confused by its intertwining with directly expressed meaning, dispute the separate existence of this [understood meaning], just as materialists dispute the separate existence of the soul. Therefore, having begun by using the word "meaning" [artha] in the singular and having referred, by the adjective "honored by connoisseurs" [sahrdaya-slaghya] to the reason [for its differentiation from everyday and Vedic meaningl, he

21Ibid., pp. 42-43.

22It is not clear if this seeming misstatement, like the one just discussed, was criticized by other writers before Abhinavagupta's time. Mahimabhatta was later to attack Anandavardhana's statement as both false and inconsistent with his own remarks elsewhere in the Dhvanyaloka (cf. Vyaktiviveka, op. cit., p. 94). Dhvanyaloka of Anandavardhana, op. cit., p. 75. Cf. also Ingalls, The

444

says, by way of analysis, that this [meaning] has two types, i.e. parts [amsau] -- not that these two are both "souls" of poetry.23

Considered as a unity, without a view toward dividing it into

parts, meaning in general may be regarded (by the

undiscriminating) as the soul of poetry. However, by

describing this meaning as "honored by connoisseurs",

Anandavardhana subtly indicates that it is more specifically

the suggested component of meaning that is to be so regarded;

it is the role of suggested meaning in poetry which makes

connoisseurs partial to it, and thus it is the suggested

component which motivates the use of the term "soul". Thus,

there is no real conflict with Anandavardhana's statements

elsewhere -- here too he is really saying, albeit less

straitforwardly than he usually does, that suggested meaning

alone is the soul of poetry. Here, as in the case discussed

previously, Abhinavagupta, in his effort to defend

Anandavardhana's phrasing, ends up by reading a great deal

more into Anandavardhana's meaning than his own words would

seem to warrant.

Such cases of commentarial excess, arising as they do

from mere difficulties in the wording of Anandavardhana's

text, might be dismissed as of relatively minor importance.24

There are, however, other, more significant points on which

23Ibid., pp. 43-45.

24Although they do sometimes have significant doctrinal consequences -- e.g Abhinavagupta's view that word and meaning co-operate in all cases of suggestion.

445

Abhinavagupta deviates from Anandavardhana's apparent

intentions in explaining and defending his dhvani theory.

Anandavardhana, for example, states that there is no need to

defend his theory against the potential objections of the

grammarians: since the concept of dhvani (he claims) is itself

drawn from the grammarians' own analysis of language, how

could there possibly be any conflict between his theory and

theirs?25 Abhinavagupta realizes that this will not do; the

grammarians, given their belief in the indivisible,

metaphysical unity of sentences and sentence meanings, cannot

accept the multiplicity of semantic functions called for by

Anandavardhana's theory. Therefore, the potential objections

of the grammarians must be considered and refuted.

As Abhinavagupta sees it, all the arguments

Anandavardhana raises against the Prābhākara Mīmamsakas'

objections to the theory of suggestion (cf. above, Chapter

Five, pp. 207ff.) are equally applicable to those the

grammarians might raise. While the Prabhākaras do not, like

the grammarians, regard the sentence as an indivisible unity,

they do argue that words, acting collectively, directly convey

whole sentence meanings, 26 and that there is therefore no

need to postulate multiple semantic operations to explain

25Dhvanyaloka, pp. 443-4; cf. also pp. 132-3 and above, Chapter Five, p. 207.

26Rather than conveying a set of individual, unconnected word-meanings which are combined into a unified sentence meaning by a subsequent mental process, as the Bhatta's would argue -- cf. above, Chapter Three, p. 116.

446

linguistic cognition. In several instances Abhinavagupta

points out that the objections of the grammarians against the

existence of suggestion are essentially the same as those

raised by the Prabhakaras, and can be undermined by the very

same arguments.27 Here Abhinavagupta's defense of suggestion

conflicts directly with Anandavardhana's own assertion that

conflict between his own language theory and that of the

grammarians cannot arise and need not be considered as a

potential threat to the dhvani theory.

In his discussion of the differentiation of suggestion

from laksana (figurative expression), Abhinavagupta again

departs in important ways from Anandavardhana's own arguments

in defense of his theory. As we have seen, Abhinavagupta, in

presenting his own theory of linguistic cognition, relies

primarily on the threat of infinite regress in fending off the

objections of those who would identify suggestion with

laksana -- it is generally agreed that laksana can take place

only when the hearer recognizes some motive (prayojana) on the

speaker's part for resorting to a non-literal mode of

expression: if this prayojana were itself conveyed by a second

operation of laksana, then this second laksana would require

its own prayojana, requiring yet another laksana, and on and

on. Therefore, laksana itself is explicable only if one

accepts an additional semantic function (i.e. suggestion)

capable of conveying the prayojana required for laksana. This

27Locana, pp. 67, 415-16.

447

argument is nowhere made by Anandavardhana, but Abhinavagupta

relies on it almost exclusively in challenging those who would

subsume suggestion within laksana and, in the process,

substantially undermines some of Anandavardhana's own

arguments for the differentiation of these two semantic

functions.

Anandavardhana argues that any attempt to identify

laksana with dhvani will be undermined by the twin flaws of

over-extension (ativyäpti) and under-extension (avyapti) ; that

is to say, that there are cases of laksana which cannot be

regarded as examples of dhvani and cases of dhvani in which

laksana plays no role. 28 There are many common, everyday

kinds of figurative expression which, even when employed in

poetry, contribute no special beauty and cannot be regarded as

examples of dhvani. For example:

Wilted on both ends by contact with her swelling breasts and hips; fresh in between, not having made contact with the middle of her body; disarranged by the flailing movements of her creeper-like arms -- this bed of lotus petals tells [vadatil of the affliction of this emaciated woman.29

Here the word vadati ("tells") is used figuratively -- the bed

does not actually speak, but reveals the desparate state of

the woman who has lain there -- but this figurative usage is

merely an everyday manner of speaking. Anandavardhana argues

that this and other such everyday figurative expressions

28Cf. Dhvanyaloka, pp. 142-52.

29Tbid., p. 143 -- cf. above, Chapter Five, p. 224.

448

cannot be regarded as instances of dhvani, and ultimately

denies that there is any suggested content (even of the

subordinate variety) in such expressions.30

As Abhinavagupta recognizes, the claim that no suggestion

takes place, even in such poetically uninteresting cases of

figurative expression, creates a problem. Anandavardhana

accepts the view that laksanā always requires some

prayojana,31 and argues that this prayojana can only be

conveyed by suggestion.32 If this is so then suggestion will

exist wherever there is laksana, even in such everyday,

aesthetically insignificant cases as Anandavardhana is here

discussing. Abhinavagupta resolves the apparent contradiction

by explaining that Anandavardhana does not really mean that

there is no suggestion at all in laksana of this sort:

"Because [in such cases of laksanā] the prayojana is not

regarded as important, suggestion has nothing useful to do --

this is [Ānandavardhana's] idea. "33 On this explanation,

Anandavardhana does not really deny that suggestion exists in

such cases, but merely dismisses it as insignificant and

30Ibid., pp. 142-5, 426.

31Except for purely conventionalized, "dead" metaphors, such as the use of the word lavanya (literally "saltiness") in the sense of "beauty" or "charm" (cf. above, Chapter Five, p. 208) -- in such cases a particular figurative sense becomes so closely and invariably associated with a word that it becomes, for all practical purposes, the primary meaning of the word.

32Cf. Dhvanyaloka, pp. 148-50 and above, Chapter Five, pp. 226-230.

33Locana, p. 142.

449

unworthy of consideration.

Explaining the example discussed above, Abhinavagupta

says :

In the metaphorical expression " [The bed of lotus petals] tells", the prayojana is the understanding that [the bed] clearly reveals [the woman's condition]. If this were stated explicitly and without concealment, what loss of beauty would there be? Or, when it is communicated in a concealed manner [as it is here], what additional beauty is produced? It is with just this in mind that he [Anandavardhana] will later say "A word, functioning suggestively, will fall within the scope of the term dhvani when it reveals some beauty which could not be conveyed by any other mode of expression" 146].34 " [1.15 -- p.

The figurative usage of the word "tells" does suggest a

prayojana, the clarity with which the bed evinces the lovelorn

state of its former occupant, but this suggestion plays no

aesthetically significant role -- one might equally well have

stated explicitly that the bed clearly reveals this. As

Ānandavardhana's own remark indicates, it is the unimportance

of suggestion, not its total absence, which excludes such

figurative usage from the realm of dhvani.

Anandavardhana's argument for the under-extensiveness

(avyapti) of laksana with respect to dhvani is similarly

curtailed and adjusted by Abhinavagupta. Ānandavardhana

asserts unambiguously that there can be no laksana involved in

any of those cases of dhvani classified as vivaksita-vācya

("in which the directly expressed meaning is intended

34Ibid., pp. 143-4.

450

literally") ;35 in such cases of dhvani, as their name would

imply, the directly expressed meaning of the word or phrase

which gives rise to the suggested meaning is not set aside or

displaced, as it would be in a case of laksanā.

Abhinavagupta, however, is not prepared to accept this claim

without significant qualification.

As his first example of vivaksita-vācya dhvani

Anandavardhana gives the following verse:

What was the name of the austerity he performed, and on what mountain, and for how long, that he, this little parrot, bites a bimba-fruit as red as your lip, young woman?36

Anandavardhana does not himself explain his understanding of

the verse, but the idea seems to be this -- here the great

beauty and desirability of the lip of the woman addressed is

suggested; if even the privilege of tasting a fruit which

resembles this lip can be won only by the practice of

unimaginably long and arduous austerity, the kissing of the

lip itself must be considered a great privilege indeed. As it

is not impossible to understand the speaker's question

literally, the literal meaning of the verse is not displaced,

but merely supplemented by this additional suggested sense,

and the verse may thus be considered a case of vivaksita-vācya

35Dhvanyaloka, pp. 151-2.

36Dhvanyaloka, p. 138. śikharini kva nu nāma kiyac ciram kimabhidhānam asāv akarot tapah/ taruni yena tavādharapātalam daśati bimbaphalam sukasāvakah// (Dharmakīrti -- Subhasitaratnakosa 439)

451

dhvani.

Commenting on this verse Abhinavagupta at first supports

what appears to be Anandavardhana's understanding of the

verse, but then relinquishes this position and concedes that

even here there is laksanā:

Here there are only three semantic operations -- denotation, tätparya, and suggestion -- since laksana, the third,37 cannot take place at an intermediate stage, due to the absence of [its necessary preconditions, i.e.] the blocking of the primary meaning, etc. Or, let us grant that there is laksana based on similarity38 at an intermediate stage -- since there is a blocking of the primary meaning, due to the implausibility of this particular question being asked without any reason. Nevertheless, the prayojana of this [laksana] can only be suggested, and will enter in the fourth phase. In the previous case,39 laksana is the main factor assisting in the suggestive operation, but here denotation and tātparya-śakti [are the main factors]; because the understanding of the suggested meaning arises from the beauty of the sentence meaning [rather than specifically from the metaphorical identification of the speaker and the parrot], the function of laksana assists only slightly. But in the case of that suggested meaning which is understood without [asamlaksyakrama -- i.e. the srngara-rasa suggested by the apparent sequence

versel there is not even a glimmer of laksana, precisely because of the lack of apparent sequence, as we shall explain later.40 So, in this second type [of dhvani, i.e. vivaksita-vācya] as well, there are four semantic functions. 41

37That is, the third in sequence of the four semantic functions admitted by Abhinavagupta -- cf. above, pp. 434ff.

38I.e. the similarity of the speaker, who hopes to kiss the woman's lip, with the parrot eating the bimba-fruit.

39I.e. the example of avivaksita-väcya dhvani which Anandavardhana has just given and with which this verse is being contrasted; the example is the verse "suvarnapuspam prthivim ... " (cf. above, Chapter Seven, pp. 374-375) .

4°Cf. Locana, pp. 154ff.

41Locana, p. 140.

452

As we have seen (above, Chapter Seven, pp. 336ff),

Mukulabhatta, as part of his effort to show that all the

varieties of dhvani discussed by Anandavardhana may be

explained as cases of laksana, construes as broadly as

possible the "impossibility" of literal meaning required for

laksanā to take place; no logical contradiction is required --

if an expression, understood literally, would serve no useful

communicative function, or would be incommensurable with the

character or presumed intention of the speaker who uses it,

this is enough to cause us to reject this literal meaning and

look for a suitable figurative sense. Abhinavagupta,

obviously sensitive to this expansive understanding of the

kind of obstacles to literal meaning which may prompt

figurative interpretation, grants that there is laksana even

in this verse: as the speaker's questioning his lover about

the parrot, understood literally, would be pointless and

absurd, one must assume that he means to communicate something

further, namely, by implicitly identifying his own situation

with that of the parrot, to compliment his beloved on the

beauty of her lip and indicate his own desire. The presence

of laksana even here does not threaten the case for the

separate existence of suggestion: laksanā requires a

prayojana, and this prayojana, for the reasons already given,

can only be suggested.

Thus, even in a case of vivaksita-vacya dhvani, where

Ānandavardhana clearly wishes to exclude it, Abhinavagupta

453

acknowleges the presence of laksana. His argument is similar

to that offered in his discussion of the role of suggestion in

non-poetic figurative usage: when Anandavardhana says that

there is no laksana involved in vivaksita-vācya dhvani, he

does not really mean that laksana is entirely absent in such

cases, only that it plays a relatively unimportant role. It

is only in the suggestion of rasa and rasa-like states (bhava,

rasabhasa, etc.) that Abhinavagupta totally disallows any role

for laksanā.

Thus we see that Abhinavagupta substantially undermines

Anandavardhana's arguments for the over-extension and under-

extension of laksana with respect to dhvani, instead relying

primarily on an argument of his own to prove the distinction

between suggestion and laksana -- that the prayojana required

by laksana could not itself be conveyed by laksana without

creating an infinite regress, and must therefore be conveyed

by a distinct semantic function, namely suggestion.

Bhattanayaka and Abhinavagupta's Theory of Rasa

Abhinavagupta's most significant departure from

Anandavardhana is his unqualified insistence on the pre-

eminence of rasa, and rasa alone, in all poetry. As we have

seen (above, Chapter Six, pp. 271ff), Anandavardhana, while he

tries to show that rasa is generally the predominant object of

great poetry, and while his teleological mode of poetic

analysis is fully applicable only to rasa-directed poetry, is

ultimately unwilling to insist that all poetry must have rasa

454

as its overriding purpose. He allows for the existence of

poetry in which suggested meaning, including rasa, is of

secondary importance (gunībhūtavyangya-kavya) and in which

such meaning plays no role at all (citra-kāvya)

Abhinavagupta, by contrast, has no qualms about insisting on

the effective communication of rasa as the essential and pre-

eminent goal of all poetry.42

Anandavardhana proclaims that dhvani is the "soul" of

poetry; this includes not only rasa-dhvani, but vastu- and

alamkara-dhvani as well. Abhinavagupta is frequently at pains

to show that, in spite of Anandavardhana's application of this

label to dhvani in general, it is really rasa-dhvani alone

that is the soul of poetry. "This rasa-dhvani, " he says, "is

truly dhvani -- this alone is, strictly speaking, the soul [of

poetry] ".43 He explains Anandavardhana's inclusion of vastu-

and alamkara-dhvani as follows:

Rasa alone is really the soul. But, vastu- and alamkāra- dhvani always culminate in rasa and are thereby elevated above the directly expressed meaning -- it is with this idea that he makes the general statement, "Dhvani is the

42Abhinavagupta's divergence from Anandavardhana in this area has been frequently noted by modern scholars; cf. De, History of Sanskrit Poetics, op. cit, Vol. 2, p. 177 -- "Having realized the importance of Rasa in poetry, Abhinava, however, went a step further than the Dhvanikara [i.e. the author of the karika portion of the Dhvanyaloka, supposed by De to be different from Anandavardhanal and Anandavardhana in boldly setting it up as the only essence or aesthetic foundation of poetry"; cf. also Krishnamoorthy, The Dhvanyaloka and its Critics, op. cit., p. 221.

43Locana, op. cit., p. 52. Similar pronouncements abound in the Locana -- cf. pp. 39, 175, 461, 473, 484-5, 538.

455

soul of poetry. "44

Even in cases where a particular rasa plays a subordinate role

(e.g. in the alamkara rasavat -- cf. above, Chapter Four, pp.

172ff), it must always ultimately subserve some other rasa:

When rasas etc. are ornaments [alamkāras], the situation will be just the same as when similes and so on are ornaments; there must necessarily be something else which is ornamented. And even if this [thing ornamented] is a mere narrative element [vastu-matra], this itself, because it will end by becoming a vibhava etc., 45 will have rasa etc. as its ultimate purport. Thus in every case rasa-dhvani alone is the soul [of poetry] .46

For Abhinavagupta rasa-directedness is not merely a prevalent

feature of much or most great poetry (as it is for

Ānandavardhana), but is an essential and constitutive feature

of all poetry.

Abhinavagupta not only insists on the centrality of rasa

in poetry far more confidently and unambiguously than

Anandavardhana was prepared to do; he also provides a far more

detailed analysis of the nature of rasa than can be found

anywhere in Anandavardhana's work. Anandavardhana concerns

himself largely with the epistemology of rasa: What are the

semantic processes by which the words of a poem cause us to

understand a particular rasa? How do the specific formal

features of poetic language enhance or detract from this

44Ibid., p. 85.

45I.e. one of the aesthetic factors (vibhavas, anubhavas, and vyabhicari-bhavas) suggestive of rasa -- cf. above, Chapter Four, p. 130.

46Locana, p. 197.

456

awareness in particular contexts? About the ontology of rasa

he has virtually nothing to say; despite the crucial role of

rasa in his model of literary aesthetics, Anandavardhana never

specifically tells us what rasa is.

Abhinavagupta, by contrast, considers at great length the

question of the specific nature and psychology of rasa. His

most elaborate exploration of this question occurs in his

commentary on Bharata's Natyasāstra, 47 and a briefer

treatment of the same issues is found in the Locana;48

Mammata includes a summary of Abhinavagupta's discussion in

his Kāvyaprakāsa,49 and this in turn becomes the basis for

most later Alamkarikas' remarks on the nature of rasa.

Drawing primarily on the tradition of dramaturgy (Natyasastra)

rather than Alamkarasastra, he examines in succession the

views of rasa offered by four earlier theorists: Lollata,

Śankuka, Bhattanāyaka, and Bhattatauta (his own teacher, whose

view of rasa he accepts as his own) .50

47AbhinavabharatI on Natyasastra 6.31 (Gaekwad's Oriental Series XXXVI [second ed.], pp. 2 272-86); this passage of Abhinavagupta's commentary has been separately edited and translated by Raniero Gnoli under the title, The Aesthetic Experience According to Abhinavagupta (Instituto Italiano per il Medio ed Estremo Oriente, Rome, 1956).

48On Dhvanyaloka 2.4; op. cit., pp. 180-90.

49Mammata, Kāvyaprakāsa 4.28ff. (ed. Rāmagovinda Sukla, SarasvatIbhavana Granthamala 46 [Varanasi, 1990], pp. 59-71.

5°No works by any of these authors survive. Lollata was the author of a commentary on the Spandakarika -- composed either by Kallata (Mukulabhatta's father) or his teacher Vasugupta -- and therefore can be dated no earlier than the mid- to late ninth century. Śankuka is mentioned by Kalhana

457

Lollata, we are told, held that rasa is merely an inten-

sification of a persistent emotional state (sthāyibhāva) --

e.g. srngara-rasa is a highly intensified form of love (rati) .

This rasa exists in the character portrayed in a drama, and

also in the actor, by virtue of his power of concentration

(anusamdhāna) .51 According to Abhinavagupta this was also

the view of the "ancients" (ciramtanas); he cites several

remarks from Dandin's Kavyadarsa (2.181, 183) which reflect a

similar understanding of rasa.52

Śankuka believed rasa to be inferred by the audience to

exist in the actor.53 The vibhāvas, anubhāvas, and

vyabhicarabhavas which gives rise to rasa (cf. above, Chapter

(Rājataranginī 4.705) as the author of a Bhuvanabhyudaya, describing the civil war in which Ajitapīda poem,

was deposed from the throne of Kashmir (AD 850-51), and was thus probably more or less contemporary with Lollata, whose view of rasa he criticizes. Both wrote commentaries on Bharata's Natyasastra, frequently alluded to by Abhinavagupta in his own commentary on the same work. Bhattanayaka may possibly be the same as the "Nayaka" mentioned by Kalhana (4.159) as a contemporary of King Samkaravarman (r. AD 883- 902). In any case, he must be later than Sankuka and earlier than Bhattatauta -- late ninth to mid-tenth century. He was the author of the Hrdayadarpana (or Sahrdayadarpana), a work attacking Anandavardhana's dhvani theory. Bhattatauta was the author of the Kavyakautuka, on which Abhinavagupta wrote a commentary (both works are lost).

51AbhinavabharatI, op. cit., p. 272; cf. also Locana, p. 184.

52Ingalls suggests, rightly I think, that Anandavardhana's view of rasa was basically the same (cf. The Dhvanyaloka of Anandavardhana, Introduction, pp. 16-19. As Abhinavagupta's own remarks indicate, this appears to have been the general view of rasa among the early Alamkarikas, and Anandavardhana's silence on the question would seem to indicate acquiesence.

53AbhinavabharatI, pp. 172ff.

458

Four, p. 130) are imitations of the causes, effects, and

associated mental states (respectively) which are connected

with particular emotions in real life; witnessing the

depiction of these factors, one infers the existence in the

actor of the appropriate emotional state. One does not

imagine these imitated factors or the mental states inferred

from them to be real, but neither does one see them as

delusory. Drama demands a distinct mode of awareness, one

which Śankuka compares to our awareness of a picture of a

horse: we recognize what we see as a horse, without falsely

imagining it to be a real horse.54

Bhattanayaka rejects both of these views of rasa as

inadequate in that they fail to explain the emotional response

of the audience to plays and poems which convey rasa.55 We

do not respond to the poetic or dramatic depiction of

particular emotional states as we would upon encountering them

in the real world. If srngara-rasa were an emotional state

actually produced or inferred by us to exist in the character

portrayed or in an actor, we would, on witnessing a private

love scene, experience not disinterested enjoyment but

embarrassment, disgust, envy, etc. (just as we would on

witnessing such a scene in real life). If the rasas generated

by a play or poem were perceived merely as emotional states

belonging to others, they would not involve us emotionally as

54AbhinavabharatI, p. 276; also Locana, p. 186.

55AbhinavabharatI, pp. 276ff .; Locana, pp. 180ff.

459

they do, but would leave us indifferent. On the other hand,

if we, through a process of identification, experienced these

emotions as our own, our responses would still be

inexplicable. Audiences are delighted even by works conveying

rasas such as karuna ("pathos"); if we experienced the

emotions of the characters as our own, such works would

produce only pain and audiences would avoid them.

Ānandavardhana's claim that rasa is suggested, according

to Bhattanayaka, will suffer from the same difficulties as

Lollata's and Sankuka's theories. Is this "suggested" rasa

experienced as one's own, or as belonging to some other? In

either case, our affective responses to drama and poetry will

be inexplicable, for the reasons stated above.56

Rasa, then, must be an experience distinct both from

witnessing or inferring an emotion to exist in another and

from experiencing it oneself. For Bhattanayaka, the essential

and distinctive feature of rasa is precisely that it

transcends the dichotomy between self and other. The

emotional states of the characters presented in a play or

poem, along with those factors (vibhavas, anubhavas, and

vyabhicaribhavas) associated with them, undergo a process of

generalization (sadhāranīkarana) ; we apprehend them neither as

pertaining to ourselves nor as pertaining to others, but

stripped of all personal associations. It is this

generalized, de-personalized emotional experience which is

56 AbhinavabharatI, p. 276.

460

called rasa. Thus we respond to these rasas neither as

indifferent bystanders nor as if we were personally affected

(e.g. actually suffering when suffering is depicted), but with

a kind of depersonalized enjoyment (bhoga) which Bhattanayaka

likens to consciousness of the supreme brahman.57

For Bhattanayaka production of this rasa-experience is

the essential and defining element of poetry. According to

Bhattanayaka there are three separate operations involved in

poetic expression. The first is direct expression (abhidha),

which it shares in common with other forms of discourse.58

The second he calls bhavakatva ("activation", or "bringing

into being"), his term for the generalization of vibhavas,

etc. which produces rasa. The third is bhogakrttva, the

production of pleasure in an audience by means of rasa.59

The latter two operations are specific to poetic language, and

serve to differentiate it from other varieties of language.

Thus rasa is not only an important and desirable feature to be

employed in poetry, it is itself constitutive -- it is what

makes poetry poetry.

Bhattanāyaka's bhavakatva, as has been noted

57AbhinavabharatI, p. 277; Locana, p. 183.

58Abhidha must here be construed broadly enough to include figurative usage and any other modes of non-literal expression which are not specifically poetic.

59Locana, pp. 182.

461

previously,60 is formulated on the analogy of the Bhätta

Mīmāmsakas' sabda-bhāvana;61 both terms refer to what

language does, over and above what it says. But there is a

crucial difference between the two, one which again emphasizes

the distinction between poetic and non-poetic language. For

the Mīmamsakas, sabda-bhavana is the power of language to

induce its hearer to undertake a particular action (e.g.

sacrifice) ; Bhattanayaka's bhavakatva does not lead those who

hear a poem to perform a particular action, but rather creates

in them a particular kind of pleasurable experience. The

understanding one derives from ordinary language is a means to

an end; it provides information on the basis of which

particular courses of action will be deemed favorable or

unfavorable. Thus all such language serves, directly or

indirectly, to urge one to undertake or refrain from

particular acts.62 Poetic language is special in that the

awareness we derive from it is directly pleasurable and is

therefore an end in and of itself. The enjoyment stemming

60Cf. Ingalls, The Dhvanyaloka of Anandavardhana, op. cit., Introduction, pp. 35-6; also Sivaprasad Bhattacharya, "Three Lost Masterpieces of the Alamkara Sastra" [PracyavanI, Calcutta, 1944], reprinted in Studies in Indian Poetics, [Calcutta: Firma KLM Private Ltd., 1981], p. 43.

61"Verbal" bhavana -- as opposed to artha-bhavana, the "bringing into being" of a particular result by means of sacrificial action.

62This is generally acknowledged even by those who are not, like the Mimamsakas, primarily concerned with the explicitly injunctive language of the Vedas; cf., for example, the Buddhist logician DharmakIrti's Pramanavarttika 3.93ff. (op. cit., pp. 217ff) .

462

from this rasa-experience is the primary obect of poetry; any

practical or moral instruction derived from such works is

secondary.63

Bhattanayaka criticizes Anandavardhana's dhvani theory,

and Abhinavagupta therefore treats him as an opponent, but

most of the central and innovative features of Bhattanayaka's

rasa theory are incorporated in Abhinavagupta's own analysis

of rasa. 64 As we have already seen, he agrees with

Bhattanāyaka in taking rasa (and rasa alone) to be the

necessary and defining feature of poetic language. He takes

Bhattanayaka's critique of Lollata's and Sankuka's analyses of

rasa as valid and follows him in taking the generalization

(sadharanīkarana) of rasa and its determinants as the crucial

feature which distinguishes rasa from real-world emotional

experience and accounts for the pleasurable character of the

rasa-experience.65 He accepts Bhattanayaka's likening of the

experience of rasa to the experience of the supreme

brahman 66 He argues that poetry does impart moral and

practical instruction, but agrees with Bhattanāyaka in making

63Locana, p. 183.

64As Ingalls observes: "One might say ... that Abhinava has taken over most of the new ideas of Bhattanāyaka but trimmed them here and there so that they may fit into the terminology and the general view of Ananda. " -- The Dhvanyaloka of Anandavardhana, Introduction, p. 37.

65AbhinavabharatI, pp. 279, 285-6.

66Locana, p. 190.

463

pleasure the primary and distinctive goal of poetry.67

By contrast, Abhinavagupta's areas of disagreement with

Bhattanayaka are quite limited. His major objection to

Bhattnayaka's rasa-theory concerns Bhattanayaka's insistence

that the process by which poetry generates the rasa-experience

cannot properly be described as vyañjana ("suggestion" or

"manifestation"68), as Anandavardhana would have it. As

Abhinavagupta sees it, rasas already exist in potential form

(sakti-rupa) in aesthetically sensitive persons and are only

manifested (vyakta or abhivyakta) by poetry or drama. These

latent emotional states are the product of traces (vāsana)

left by experiences from previous lives.69 The function of

poetry or drama is to remove those obstacles (vighnas) which

prevent us from experiencing these latent states.70 One of

these obstacles is the awareness of particular limitations of

place and time pertaining to oneself or another; thus

Bhattanayaka's "generalization" is part of the very process of

"manifestation" or "suggestion" which he rejects. What

67Locana, pp. 40-41, 190.

68Cf. above, Chapter Five, p. 204.

69Bhattanāyaka rejects this view, arguing that if it were so there would be no basis, even in the experience of our past lives, for identification with the deeds and emotional experiences of superhuman or divine heroes such as Rāma (Abhinavabhārati, p. 276, Locana, p. 181). Abhinavagupta denies this: our past-life experiences are without beginning and hence infinitely variable -- he quotes a passage from the Yogasutra [4.9-10] in support of this view (Locana, pp. 187- 8).

70 AbhinavabharatI, pp. 280ff.

464

Bhattanayaka calls bhavakatva and bhogikarana are in fact

nothing apart from this process of suggestion (vyañjana,

dhvanana) .71

Despite Abhinavagupta's positioning himself as a critic

of Bhattanayaka, the two are quite close in their views on the

nature of poetry and on the role of rasa in poetry; indeed, in

certain key respects Abhinavagupta's views are closer to

Bhattanayaka's than to those of Anandavardhana, whom he

professes to follow. Anandavardhana declares his threefold

(vastu-, alamkāra-, and rasa-) dhvani to be "the soul of

poetry"; Bhattanayaka and Abhinavagupta both part from this

characterization and make rasa alone the pre-eminent and

essential feature of poetic language -- Bhattanayaka by

rejecting Anandavardhana's dhvani and substituting his own

bhavakatva (like dhvani, bhāvakatva is a semantic funtion

specific to poetic language, but one whose sole object is to

convey rasa) , Abhinavagupta by re-interpreting

Anandavardhana's remarks in an effort to show that he really

meant (despite his own statements to the contrary) to make

rasa-dhvani alone "the soul of poetry". Although his

teleological analysis of poetry would seem to demand it,

71Locana, p. 188. If it is the latent pre-existence of rasas in the hearers of a poem which justifies the description the process by which they are brought to experience as we "suggestion" of vastus and alamkaras, which cannot be said to "suggestion" (vyañjanā) , how are to explain the

pre-exist in this way? Abhinavagupta (who, as we have already seen, marginalizes vastu- and alamkära-dhvani and raises rasa- dhvani to a position of exclusive predominance) does not deal with the question.

465

Ãnandavardhana was ultimately unwilling to make the

predominance (or even the presence) of rasa a necessary

feature of poetry. Bhattanayaka and Abhinavagupta, however,

make rasa an essential feature, indeed the constitutive

feature, of poetic language -- rasa is not merely a highly

desirable element of poetry or one which generally plays a

predominant role in the best poetry (as it is for

Anandavardhana); it is what makes poetry poetry, what

differentiates it from other forms of discourse. Both argue

that the primary object of poetry is to produce pleasure

directly in its hearers and that it can do so only by

conveying rasa; thus for them rasa is necessarily the

overriding goal of all poetic language.

The Special Status of Aesthetic Language

For Abhinavagupta, poetic language differs from ordinary

language not only in its ultimate function, but in its nature.

He makes much of the extraordinary (alaukika) character of the

rasa-experience and of the processes by which poetic language

produces this experience. Anandavardhana and Bhattnayaka both

allow for the existence Of semantic modes which are

specifically poetic;72 Abhinavagupta lays even greater stress

on the unique character of poetic expression and, as we shall

see, concludes, on the basis of this unique character, that

72Or, at any rate, semantic modes which assume a unique significance in poetic language. Ānandavardhana argues for the existence of suggestion even in ordinary language, but claims that it is only in poetry that it assumes an important communicative role (cf. above, Chapter Five, pp. 218-221).

466

poetic language need not be subject to the same requirements

for sentential coherence that are applied to ordinary

language.

In discussing the means by which the various aesthetic

factors (vibhavas, etc.) give rise to rasa, Abhinavagupta

makes it clear that this process has no parallels outside the

realm of poetry, and that the absence of such parallels should

be regarded as an asset rather than a flaw in his theory of

rasa.

Objector: Do the vibhävas and so on act with respect to this [rasa] as informative [jñapaka] or productive [karaka] causes?73 Reply: They are neither informative nor productive causes; they are serviceable to the enjoyment [of rasa] [carvanopayogin] . 74 Objector: But where else do we see this [type of causality]? Reply: It is precisely because it is not seen [elsewhere] that we call it "extraordinary".75

Elsewhere he describes the absence of any non-poetic parallel

for the process which leads to the rasa-experience as an

"ornament" [bhūsana] of his theory. 76 The sui generis

character of this process should not be raised as an objection

against this theory of rasa; it is an asset precisely because

it serves to make clear the special and extraordinary status

of rasa.

73I.e. do the vibhavas etc. make one aware of (already existing) rasa, or do they produce rasa where it did not exist before

"4Through the process of "generalization" discussed above (pp. 458-460) .

75Locana, p. 158.

"6AbhinavabharatI, p. 285.

467

Because of the unique nature and function of poetic (i.e.

rasa-directed) language, those semantic principles and

restrictions arrived at through the analysis of everyday or

Vedic language do not necessarily apply to it. One of the

central semantic principles developed by the Mīmamsakas is the

requirement that any sentence express a single, unified

meaning; any sentence which fails to do so (or any

interpretation of a Vedic sentence which would make this

sentence fail to do so) will suffer from the flaw of vākya-

bheda ("splitting the sentence" -- cf. above, Chapter Three,

pp. 85ff) . Responding to the contention that a sentence

communicating rasa along with its directly expressed

meaning77 would suffer from vakya-bheda, Abhinavagupta argues

that vakya-bheda, while it must be avoided in sastras, does

not, in the context of poetry, constitute a flaw at all.78

77Rather than as a second meaning which subsequently displaces the directly expressed meaning (as happens in the case of figurative expression)

78Locana, pp. 159-60. Abhinavagupta's treatment of this objection differs markedly from Anandavardhana's response to the very same charge (cf. Dhvanyaloka, pp. 421-2 and above, Chapter Five, pp. 214-216) . Anandavardhana nevers questions the applicability of vakya-bheda or other Mimamsa principles of text analysis to poetic language. He argues that sentences conveying rasa do communicate a unitary meaning. Rasa will always be either superordinate (in dhvani) or subordinate (in gunibhutavyangya) with respect to the expressed meaning which gives rise to it; either the directly expressed meaning (in the latter case) or the rasa (in the former) will be the predominant meaning toward which the sentence is ultimately directed. This integration of the various elements of meaning in a single purposive framework is all that is required for "unity of sentence-meaning" as the Mimamsakas understand it; thus there is no vakya-bheda in such sentences.

468

If a sastric sentence did not convey a single, definite

meaning, we could have no faith in any information or

instruction derived from it, and it is for this reason that

vakya-bheda is considered a defect. Because poetry is not

meant to convey particular information about the world or to

motivate its hearers to particular actions, it does not

require this sort of unified meaning and hence is not subject

to this flaw.

[The understanding derived from poetry] is not like that derived from sastra, e.g. "I am enjoined in this action, I should do this", or "I have already achieved this goal",79 because in these cases [one's understanding] is ordinary [laukika] in that it is directed toward things to be done in the future; but here [i.e. in poetry] the savoring of the vibhavas and so on arises, like a magical flower, having its essence at that very moment, and not connected with earlier or later times.80

The value of the understanding derived from poetic

language is not contingent on anything outside of itself. All

that is required of poetic language is that it give pleasure

(as it will do if it conveys rasa); the meaning derived from

it need not correspond to any extra-poetic reality for its

effectiveness, and hence poetic language is not subject to the

same requirement for sentential coherence or the same need for

verification as sastric language. "These [poetic] sentences

do not require validity so as to motivate [hearers] by

communicating a true meaning (as do sentences enjoining

sacrifices such as the agnistoma), because they culminate only

79And hence need not act further in this matter.

80Locana, p. 160.

469

in pleasure. "81 Thus poetic language, because its value does

not reside in practical utility, need not be subject to the

same rules applied to other forms of discourse.

While much Of Abhinavagupta' s rasa-theory and the

arguments he offers against the anti-dhvani theorists prove to

be tremendously influential, this attempt to exempt poetic

language from Mimamsa-derived strictures on textual coherence

is not followed up by later Ālamkärikas. As we shall see, the

application of Mimamsa-derived interpretive principles to the

analysis and appraisal of poetic language, initiated by

Anandavardhana, is continued and expanded by the later

Ālamkārikas.

Conclusion

In the course of his commentary on the Dhvanyaloka,

Abhinavagupta departs in many important respects from the

stated views of Anandavardhana; but in doing so he often seems

to carry the implications of Anandavardhana's theory through

to their logical conclusion in a way Anandavardhana himself

was reluctant to do. Anandavardhana, while his teleological

model of textual coherence and the critical principles he

develops from it would seem to demand it, was not prepared to

insist on rasa-directedness as an essential feature of all

poetry; Abhinavagupta has no qualms about doing so.

As we have seen, Anandavardhana's dhvani-theory provoked

a great deal of criticism, and Abhinavagupta expends a great

81Ibid., p. 455.

470

deal of effort in attempting to fend off these attacks. Yet,

in doing so, he often substantially diverges from

Ānandavardhana's own views, curtailing or neglecting

Anandavardhana's arguments where he finds them to be

problematic and substituting arguments of his own. In

formulating his rasa-theory Abhinavagupta draws at least as

much on the work of the anti-dhvani theorist Bhattanāyaka as

on Anandavardhana's own views on rasa. In this rapprochement

between Bhattanayaka and Abhinavagupta we see the beginning of

a general consensus -- one which cuts across the lines of the

dhvani controversy -- regarding the essentiality of rasa as the

paramount goal of all poetic language. This consensus is also

reflected in the views of Ānandavardhana's next major critic,

Mahimabhatta, whose work will be examined in the next chapter.