6. CHAPTER THREE MIMAMSA AND THE TELEOLOGY OF LANGUAGE

CHAPTER THREE MIMAMSA AND THE TELEOLOGY OF LANGUAGE

The most detailed and sophisticated theory of non-poetic

language interpretation in pre-modern South Asia is that

devised by the MImamsakas in their exegesis of Vedic texts.

Many of the rival schools of Sanskrit philosophical and

scientific thought formulated their own theories of language.

But the Mimamsakas, because their raison d'etre was to provide

a thorough exegesis of a substantial body of existing texts,

had to do more than provide an account of the general

principles by which language operates; they were forced to

confront a vast array of specific interpretive problems, and

to devise a system of hermeneutical procedures complex and

precise enough to resolve all such problems. In consequence,

the semantic and interpretive theory they created is, in its

sophistication and subtlety, unequalled in South Asian

intellectual history. While the primary focus of attention in

Mīmamsa is on Vedic texts, the hermeneutical principles

developed in the course of analyzing these texts are

formulated so as to be applicable to language in general, and

are recognized as such by scholars in other fields. For this

reason, Mīmamsa is frequently designated as vākya-sastra ("the

science of sentences").

As we shall see, Anandavardhana's radical reconfiguration

of literary theory and of the process of poetic analysis

draws, in many crucial respects, on the hermeneutical theory

and practice of the Mīmamsakas. We shall therefore, before

66

67

embarking on a detailed analysis of Anandavardhana's work,

examine briefly the Mimamsa approach to textual analysis.

The oldest surviving Mimamsa text is the Mimamsasutra of

Jaimini (c. 200 BC?) . The work is a collection of brief,

often cryptic, aphorisms arranged in twelve adhyayas, each

dealing with a different issue in the interpretation of ritual

texts, and each further subdivided into four or eight pādas.

The Sabarabhasya of Sabara (c. 400 AD) is an extensive

commentary on Jaimini's aphorisms, and serves as the basis for

all later work in the field of MImamsa. The seventh century

produced two important commentators on Sabara's work:

Kumarilabhatta and Prabhakara, founders of the two major

schools of Mīmāmsā (called Bhatta and Prābhakara,

respectively) .1 Kumārilabhatta was author of the

Ślokavarttika (on the first pada of the first adhyaya), the

Tantravarttika (on the remaining pādas of the first adhyāya

along with the second and third adhyayas), the Tuptika (on

adhyayas four through twelve), and the Brhattika.2 Prabhākara

1The two authors are at least roughly contemporary; Prabhäkara is traditionally believed to have been a pupil of Kumārila, and was probably a junior contemporary of his, although some modern scholars (most notably Ganganath Jha) have argued, on doctrinal and stylistic grounds, that Prabhakara was the earlier of the two (cf. Jha, The Prabhakara School of Purva Mimamsa, Delhi: Motilal Banarsidass, 1978 [reprint], pp. 11-17, and Purva-Mimamsa in its Sources, Varanasi, Benaras Hindu University, 1964 [second ed.], pp. 13- 17).

2The last work is not extant, but is frequently cited by later authors both within and outside the field of Mimamsa (for a survey of citations from this work in later Buddhist philosophical literature, cf. Erich Frauwallner, "Kumarila's

68

wrote two complete commentaries on Sabara's Bhasya, called the

Nibandhana and the Vivarana, but more popularly known as the

BrhatI (the "great" commentary) and the Laghvi (the "light"

commentary).3 Nearly all later Mimämsakas identify themselves

as followers of one or the other of these two. While there

are important differences between these two schools of MImamsa

on certain matters of epistemological and cognitive theory,

they are largely in agreement on questions of textual

interpretation.

Mīmamsa, perhaps because it deals so intimately and

exhaustively with the arcane details of Vedic ritual, has

received comparatively little attention from modern scholars

of Indian philosophy, and what little attention it has

received has focussed almost exclusively on matters dealt with

in the first pada of the first adhyaya of the Mimamsasutra

(the so-called tarka- ["reasoning"] päda), in which the

epistemological and ontological views which form the

groundwork of the system are explained and defended. The

standard practice in general surveys of Mimamsa' (and in the

BrhattIka," Wiener Zeischrift für die Kunde des Süd- und Ost- asiens 6 (1962) , pp. 78-90).

3The BrhatI is only partly extant (up to the first pāda of the sixth adhyaya), and the Laghvi is altogether lost

AD) (although the last six adhyayas of Salikanathamisra's (c. 800 commentary on it [called Dīpaśikhāl, containing substantial quotations from the Prabhakara's text, survives).

4E.g. A.B. Keith, The Karma-Mimamsa, Delhi: Munshiram Manoharlal, 1978 Introduction [second ed.] ; Pashupatinath Shastri, to the Purva MImamsa, Chaukhambha Oriental Research Studies 19, Varanasi, 1980 [second ed.]; G.P. Bhatt,

69

summaries of Mīmāmsa views included in most histories of

Indian Philosophy) has been to draw a sharp distinction

between "philosophical topics" (i.e. topics dealt with in the

tarka-pada) and "Mimamsa topics proper" (i.e. the interpretive

questions dealt with in the remaining fifty-nine pädas of the

Mimamsāsūtra) , and to treat the latter topics as of minor

importance or to ignore them altogether. Even those authors

(Jha, Keith) who include a cursory survey of these "MImamsa

topics proper" usually provide only a summary of the

conclusions reached by the Mimamsakas on various points of

Vedic interpretation, rather than a serious examination of the

interpretive principles and methods by which they arrive at

these conclusions.5

This narrow focus on the tarka-pāda is perhaps

understandable; commentaries on this section of the

MImamsasutra tend to engage and counter the views of rival

schools of thought more explicitly than those dealing with the

main body of Jaimini's rules. It is nonetheless unfortunate;

Epistemology of the Bhatta School of PūrvamImamsa, Chowkhamba Sanskrit Studies XVII, Varanasi, 1962; Ganganath Jha, The Prabhakara School of Purva Mimamsa, op. cit. and Pūrva-MImamsa in its Sources, op. cit.

5A rare exception is G.V. Devasthali's Mimamsa: The Ancient Indian Science of Sentence Interpretation (Delhi: Sri Satguru Publications, 1991 [second Devasthali's study rightly treats Mimamsa hermeneutics as a revised ed. ] ) .

general theory of textual interpretation, rather than merely a collection of ad hoc observations on Vedic ritual procedure, and assembles and correlates discussions of various interpretive concepts found throughout the twelve adhyayas of the Mīmāmsa system.

70

the epistemological and ontological postulates defended in the

tarka-pada, while certainly not unimportant, are little more

than a necessary prolegomenon to the real business of Mimamsa.

The most distinctive and ultimately the most influential

contribution of the Mimamsakas to Sanskrit philosophical and

linguistic thought, their theory of textual interpretation,

has been largely overlooked by modern scholars. A thorough

investigation of the principles and methods of text

interpretation devised by the Mimamsakas would be the work not

of one book but of many -- obviously well beyond the scope of

the present study. The most that can be attempted here is a

general examination of some of the interpretive problems posed

by the Mimamsakas and the key concepts and techniques they

devise for dealing with them, focussing particularly on those

elements of Mimamsa interpretive theory which come to play a

significant role in the development of Kashmiri

Alamkāraśāstra.

The Movement from Universal to Particular

One of the key tenets of Mimamsakas is that the Vedas are

eternal and authorless. It is on the basis of their

authorless character that the absolute inerrancy of the Vedas

is asserted. . The accuracy of texts composed by authors is

limited by the knowledge and trustworthiness of those authors.

By contrast, statements made in the Vedas are not made by

anyone (who might potentially be mistaken or deceitful) -- they

simply are. Thus there is nothing which would allow us to

71

call into question the truthfulness of these statements.6 If

the Vedic texts are eternal (and hence authorless), then the

words of which they are composed, and the meanings conveyed by

these words, must likewise be eternal. It is for this reason

that the Mimamsakas must argue that the meaning conveyed by

any word is a universal (akrti, jati), rather than a

particular (vyakti).7

The actual referent of a word such as "cow" must be not

any particular cow or group of cows, but rather the universal

class-property "cow-ness". If, in learning the proper use of

the word "cow", we came to associate it with one or more

individual cows known to us, we could not use or understand

the word to refer to other, previously unknown, cows. If the

word were taken to be associated with all particular cows

(past, present and future), no one could ever understand its

meaning (since no one could ever apprehend all cows for the

purpose of learning this association). The word must, then,

refer to whatever it is that all cows, and only cows, have in

common -- by definition, the universal class-property "cow-

6The epistemological presuppositions of this argument are most clearly articulated and ably defended by Kumārila in his development of the theory of svatahpramanya, the intrinsic validity of all cognitions (cf. Ślokavārttika, Codanāsūtra 33ff. [Prācyabhāratī ed., pp. 41ff.]) . The best English exposition of the theory is found in Bhatt, Epistemology of the Bhatta School of Purva MImamsa, op. cit., pp. 73-111.

7Cf. Ślokavārttika, Ākrtivāda 1 (op. cit., p. 385).

72

ness".8 The meanings conveyed by words can only be universal

class-properties, which are unitary and eternal, rather than

particulars, which are multiple and ephemeral.

But this creates an interpretive problem. Sentences

(Vedic or otherwise) describe or enjoin particular entities or

actions, rather than universal class-properties. When a

person says, "The cow is white," it is a particular cow

(rather than the universal "cow-ness") to which the quality of

whiteness is ascribed; when one person orders another to

"Bring the cow!", it is a particular act of bringing and a

particular cow that are intended; it would be impossible for

anyone to bring the universal "cow-ness". So, while words can

denote only universals, sentences, at least typically, refer

to particulars.9 How can this be so?

The particularity of sentence meaning can be explained as

follows. Sentence meaning is not directly conveyed by the

words which make up a sentence. As Sabara explains, "Words,

having denoted their respective word-meanings, cease

functioning; then the word-meanings, once they have been

understood, cause one to understand the sentence meaning. "10

Words can convey only those universal meanings with which they

ªCf. Śābarabhāsya on Mimamsasutra 1.3.30-35 with Tantravarttika (Anandasrama Sanskrit Series 97, 1970 [second ed.], Vol. II, pp. 229ff.) and BrhatI (Madras University Sanskrit Series 24, 1962, pp. 146ff.) .

9Sābarabhāsya on 1.1.24 (Ānandāśrama Sanskrit Series 97 [second ed.], Vol. I, p. 111-112).

10Tbid., p. 116.

73

are eternally linked; they cannot function collectively to

convey a connected sentence meaning. It is the individual

word-meanings which cause us, by a further mental operation,

to understand the meaning of the sentence as a whole.

Taken separately, individual words convey no useful

information; they can only produce mental impressions of

eternal and ubiquitous class-properties. However, these

universal word-meanings, construed with one another, can, by

a process of mutual specification, cause one to understand

some particular entity in which these universals are

instantiated. For example: having heard the words "cow"

(gauh) and "white" (suklah) juxtaposed with one another, we

understand that the speaker is referring to a particular

entity possessing both "cow-ness" and "whiteness", i.e. a

white cow. Similarly, having heard the words "bring!" (anaya)

and and understood their individual meanings, we

understand that the speaker is issuing a command for a

particular act of bringing, having a particular cow as its

object. Thus, while individual words can have only universals

as their meanings, these word-meanings can, in conjunction

with one another, designate particular entities and actions.

To quote Sabara again, "A word designates a universal; a

sentence designates a particular. The sentence-meaning is the

fixing of word-meanings, which function generally, in some

particular ... Word-meaning is apprehended directly, but

74

sentence-meaning is inferential [anumanika] . "ll It is our

mental reflection on and association of the individual word

meanings, rather than the words themselves, which create an

awareness of the (particular) sentence meaning.

This mental progression from universal to particular

meanings through mutual specification is basic to linguistic

comprehension as the Mimamsakas understand it. It is not only

basic to the process of sentence comprehension, but recurs at

other levels of language as well -- both within the words that

make up sentences and between multiple sentences within a

text. While morphologically simple words such as "cow" can

have only universals as their meanings, it is argued that

morphologically complex words (yaugika-sabdas -- that is, words

whose meanings can be derived etymologically from the meanings

of their parts) can, like sentences, refer to particulars.12

For example: the word goman ("cow-possessor") is composed of

two meaningful parts, the word go ("cow") and the possessive

suffix -man. The first can directly convey only the universal

"cow-ness", and the second only "possessor-ness", but because

these two component meanings delimit one another (just as the

individual word-meanings which make up a sentence do), the

word can refer to a particular possessor of particular cows.

As we shall see (below, pp. 106ff), the individual meanings of

the sentences which make up a particular Vedic text delimit

11Sabarabhāsya on 3.1.12 (op. cit, Vol. IV, p. 53).

12Cf. Tantravarttika, op. cit., Vol. IV, pp. 35ff.

75

one another in much the same manner; one and the same sentence

can convey any of a number of significantly different

meanings, and the precise meaning it conveys in any given

instance (what the Mimamsakas call vacana-vyakti, the

"sentence-particular") can be ascertained only with reference

to its context (prakarana) -- that is, to the surrounding

sentences. So, the progression from morpheme- to word-

meaning, from word- to sentence-meaning, and from sentence- to

text-meaning consists of a succession of stages in which

groups of linguistic elements specify and delimit each others'

meanings so as to produce a single, unified sense.

Unity of Purpose as the Basis for Mutual Specification

Thus it is established that it is only through the

interaction and mutual delimitation of individual semantic

components that language can convey specific sentential and

textual meanings. But what governs this interaction? Just

what is it that leads us to construe a group of word meanings

with one another, and how are we to know which semantic

elements specify which? For the Mimamsakas, that factor which

connects the individual meanings of words and other linguistic

elements is unity of purpose. Unless the individual meanings

of the words in an utterance are linked to some common goal,

there will be nothing to induce the hearers of the utterance

to construe these separate meanings into a connected whole;

only the unrelated word-meanings would be understood. As

Kumarila says, "As long as unity of purpose [prayojana] is not

76

relied upon, the independent word-meanings will not be able to

attain unity. #13

It is an important tenet of the Mimamsakas that every

sentence must be interpreted so as to convey a functionally

unified and coherent meaning. Any interpretation which fails

to maintain the unity of the sentence (ekavākyata) will be

vitiated by the flaw of väkyabheda, "the splitting of the

sentence".14 In fact, this unity is for them constitutive of

the sentence. Jaimini characterizes the sentence as follows:

"A single sentence [exists] because of unity of purpose; it

would stand in need [akänksa] if there were a division [of its

parts] . #15 So it is functional unity that determines the

unity of the sentence. The individual parts of the sentence

connect with the whole because they subserve the overall

purpose of the sentence and are thus "needed" by it. It is

only in response to this "need" (akanksa) that any expressive

element of language can associate itself with any other;

according to Kumarila, "Connection is, in every case,

dependent upon need. "16

Because this need is always driven by some purpose (since

13Tantravarttika on 2.1.46 (op. cit., Vol. II, p. 467) .

14For a summary of these two key Mimamsa interpretive concepts, see Devasthali, op. cit., pp. 163-180; also, below, pp. 79ff.

15Mīmamsasutra 2.1.46 [Ānandasrama Sanskrit Series 97, Vol. II, p. 431).

16Tantravarttika on 2.2.3 (op. cit., Vol. III, p. 24) .

77

anything that is needed is always needed for something),

connections between words or other linguistic elements are, by

necessity, hierarchical. The point is made concisely by

Śālikanathamisra (in his commentary on Prabhakara's BrhatI) ,

"No connection is possible without a relation of primacy and

subordination. #17 Whenever two or more linguistic elements

combine with one another to produce a connected meaning, one

of these elements will be predominant and the others

subordinate to it; the former will be that which stands in

need and the latter the element or elements capable of filling

this need. Elements of equal predominance cannot be connected

with one another, because that need which is the driving force

behind the connection of all linguistic elements does not hold

between them. 18

The hierarchical character of all linguistic relations,

coupled with the required functional unity of sentences

(ekavakyata), makes it necessary that in every sentence there

should be one predominant element, to which all others are

subordinate. As Kumarila puts it, "This being the case [i.e.

the functional unity of the sentence being established], only

one word-meaning will be predominant, and the others will be

17Rjuvimalapañcika on 3.1.1 (in BrhatI, Madras University Sanskrit Series 25, 1964, p. 628).

18Cf. Sābarabhasya and Tantravarttika (op. cit., Vol. IV, pp. 11-113) and BrhatI (Madras University Sanskrit Series 25, op. cit., pp. 698-701) on Mimamsasutra 3.1.22.

78

mentioned as qualifiers [visesanas] [of that] . #19 This

conclusion, of course, raises the further question: in a given

sentence, which element will be predominant?

The Mimamsakas argue that in any sentence it is the

element of "becoming" or "bringing into being" (bhava or

bhavana) expressed by the verb (or, more precisely, the verbal

suffix) which is predominant; all other word- or morpheme-

meanings serve, directly or indirectly, to qualify this

element, and are hence subordinate to it.20 It is from this

element of bhavana that that "need" which determines the

functional unity of the sentence and causes a hearer to

connect the individual meanings of its component words arises.

The "event" expressed by the verb can transpire only through

the working of those already existing entities connected with

it. For example, in the sentence, "Devadatta cooks rice in a

pot by means of fire," the act of cooking expressed by the

verb is made possible only through the interaction of

Devadatta, the rice, the pot, and the fire;21 by contrast,

these four elements, expressed by nouns, do not depend on the

1sTantravarttika on 2.1.46 (op. cit., vol. II, p. 437) .

20Cf. Śābarabhāsya on 2.1.1-4 [Bhāvārthādhikarana] (op. cit., Vol. II, pp. 333-358, especially pp. 340-341 and 355- 358); also Ślokavarttika, Vakyādhikarana 330ff. (op. cit., pp. 661f£ . )

21Because it is only through the interaction of such existing elements that the action expressed by the verb can come into being, these elements are referred to in Sanskrit grammatical and linguistic theory as productive or active factors (kārakas) .

79

act of cooking, or on each other, for their very existence.

They are already established (siddha), while the action is "to

be established" (sadhya). Thus it is the verbal element that

stands in need of the others, and they aid in, and hence are

subordinate to, that "coming into being" which is expressed by

the verb. This reasoning is summed up in the often-cited

bhuta-bhavya-nyaya ("principle of that which exists and that

which must be brought into being"): "When something which

exists and something which must be brought into being are

mentioned together, that which exists is pointed out for the

sake of that which is to be brought into being."22

The "need" generated by the bhavana, the "bringing into

being" expressed by the verbal suffix, is threefold. When we

hear such a suffix it raises the following questions: What is

to be brought into being, by what means is it to be brought

into being, and how (that is, by what specific procedures) is

it to be brought into being?23 The answers to these

questions are not provided by the verbal suffix itself and

must be sought in the surrounding words of the same sentence

and, if necessary, in surrounding sentences; it is in this way

that the need arising from the bhavana unites disparate words

22bhūtabhavyasamuccarane bhutam bhavyayopadisyate; cf. Śabarabhāsya on 3.4.40 (op. cit., Vol IV, p. 423). For a collection of citations and references in Mīmamsa works, see Kevalānandasarasvatī's Mīmamsakosa (Delhi: * Sri Satguru Publications, 1992 [reprint]), Vol. 6, pp. 3063-4.

23Śabarabhāsya, cit., p. and Tantravarttika, op. cit., Vol. II, pp. 350ff. op. Vol. II, 341

80

and sentences into functionally unified wholes.

Consider, for example, a typical Vedic injunction: "One

who desires heaven should sacrifice with the Jyotistoma"

(jyotistomena svargakāmo yajeta) .24 What is it that is to be

brought into being by the action enjoined in this sentence?

It might at first seem as if it is only the sacrificial

performance that is to be brought into being; it is, after

all, the verbal root denoting sacrifice (vyaj) to which the

optative (injunctive) suffix -eta, the element expressive of

bhavana, is appended. But this will not do. If the

injunction is to function properly (i.e. if it is to motivate

its hearers to action), then the ultimate object to be

achieved by its performance must be something desirable in

itself. The only element in the sentence suitable for this

role is heaven (svarga-); thus we must understand attainment

heaven to be the ultimate result of this sacrificial

performance. 25 People will undertake sacrificial action (or

24Cited in Sabarabhasya on 6.1.1 (op. cit., Vol. V, p. 173) .

25Śabarabhāsya, op. cit .. Vol. II, p. 340 and Tantravarttika, op. cit., Vol. II, p. 350. Prabhakara's view of the sacrificer's motivation is more complex; he agrees that the expectation of some desirable result (phala) is necessary to prompt one to undertake a sacrificial action enjoined by the Veda, but nevertheless argues that it is primarily the sense of obligatoriness created by the injunctive verbal suffix, rather than the ultimate result, which motivates the sacrificer. Some Western scholars (beginning with Stcherbatsky) have seen this notion of obligatoriness as analogous to Kant's categorical imperative. This discussion of motivation is one of the few points of Prabhakara's system to have received serious attention from modern scholars; cf., for example, Hajime Nakamura, "Problem of the Categorical

81

any action at all, for that matter), only if they can expect

to accomplish some desirable result by doing so. If the thing

to be brought into being by the performance of this Vedic

injunction were merely the sacrificial act itself, no one

would be prompted to action by it; the injunction would then

serve no purpose whatsoever. The injunction will be

purposeful only if we understand the sacrifice to be capable

of satisfying the stated desire of the sacrificer; thus in the

sentence, "One who desires heaven should sacrifice with the

Jyotistoma," the object of the bhavana, the thing to be

brought into being, must be heaven, understood as the ultimate

result of the sacrifice here enjoined.

The sacrificial performance itself, expressed by both the

verbal root Vyaj ("sacrifice") and the word jyotistoma (the

proper name of a particular sacrifice), must be understood as

fulfilling the second "need" generated by the bhavana, that of

a means or instrument by which the result can be brought into

being.26 Thus the injunction, "One who desires heaven should

Imperative in the Philosophy of Prabhakara School: A Brief Note, " in R.C. Dwivedi ed., Studies in Mīmāmsa (Delhi: Motilal Banarsidass, 1994), pp. 169-183 (itself a summary and critique of Th. Stcherbatsky, "Über die Nyayakanika des Vacaspatimisra und die Indische Lehre vom kategorischen Imperativ," in Beitrage zur Literaturwissenschaft und Geistesgeschichte Indiens, Bonn: Kommissionsverlag Fritz Klopp, 1926, pp. 369- 80), and Irene Wicher, "Svargakäma," Wiener Zeitschrift für die Kunde Südasiens 38 (1994), pp. 509-522. Prabhakara's main discussion of the question is found in BrhatI on sūtras 6.1.1- 3 (Madras University Sanskrit Series 26), pp. 1-54.

26Tantravarttika, op. cit., Vol. II, pp. 351-352. Because the intended result does not appear immediately at the conclusion of the sacrificial performance, the sacrifice

82

sacrifice with the Jyotistoma," properly analyzed, can be

restated as follows, "One should bring heaven into being by

means of the Jyotistoma sacrifice" (jyotistomena yāgena

svargam bhavayet) .

There is nothing in the sentence itself to fulfill the

third need generated by the bhavana, that of a specific set of

procedures to be followed in order to bring about the intended

result. We know that we are to perform the Jyotistoma

sacrifice in order to attain heaven, but we do not know yet

precisely what is entailed by the performance of this

sacrifice. For this information, we must turn to the

sentences surrounding this primary injunction, sentences which

will specify the preparation and offering of various

sacrificial materials, the deities to which these materials

are to be offered, the ritual formulas (mantras) to be recited

in the course of these offerings, and so on. It is only when

all three "needs" of the bhavana have been filled in this way

that the Vedic injunction becomes fully comprehensible and,

thus, performable.

Since the complete understanding of the bhavana generated

by a single Vedic injunction requires the hearer to draw on

information from multiple surrounding sentences, it should be

cannot be regarded as the immediate cause of the result. Thus the Mimamsakas postulate, as the immediate result of the sacrifice, an imperceptible force which they call apūrva (the "new thing" created by the sacrifice); the sacrifice produces the apurva, and the apurva, at some time in the future, produces the desired result. Cf. Sabarabhāsya, op. cit., Vol. II, p. 340 and pp. 358ff.

83

obvious that all bhavanas do not operate equally and

independently. All of the sentences which surround a primary

injunction, and from which the specific procedures of the rite

it enjoins are arrived at, must have some bhavana of their

own; however, these differ from the bhavana conveyed by the

primary injunction in that they are ultimately subordinated to

it. On this basis the Mimamsakas divide the verbs contained

in the Veda, which express bhavana, into two main classes,

primary (pradhana) and secondary (guna) .27 Verbs expressive

of actions which entail the modification or preparation of

materials28 to be used in other actions are regarded as

secondary. Such actions do not in and of themselves produce

any beneficial sacrificial result (phala) for their

performers; they are to be performed because they make

possible (by providing suitable materials) other, primary,

actions which do produce such results.

As with the connection of elements within a single

sentence, connections between multiple sentences are

ultimately determined by unity of purpose within a given text.

The actions enjoined by secondary injunctions have the

production or modification of substances as their immediate

object but, as these substances themselves play an essential

27Sabarabhāsya on Mīmamsasutra 2.1.6-8, op. cit., Vol. II, pp. 379ff.

28Including preparation by the recitation of appropriate mantras or by the mental dedication of substances to particular deities.

84

role in the action enjoined by some primary injunction, the

secondary injunctions ultimately subserve the purpose of the

primary injunction with which they are connected (i.e., the

production of a particular result for the sacrificer). And,

once again, we see that the connections between sacrificial

elements (and, thereby, between the texts which enjoin

them) 29 are necessarily hierarchical. A secondary action is

subordinate to the substance it modifies, the substance is

subordinate to the primary action in which it is used, and the

primary action is subordinate to the sacrificial result it is

meant to produce. 30 Mult e actions may be seen as

constituting a single, complex, ritual performance only

insofar as they all, directly or indirectly, subserve a single

ultimate purpose; likewise, multiple sentences may be seen as

constituting a single, coherent text only insofar as they all

ultimately subserve a single ultimate purpose (i.e., the

injunction of a complex ritual performance which will produce

a particular sacrificial result for its performer).

So, the process by which individual sentence meanings

combine to form a functionally unified text parallels that by

29In Mimamsa text analysis, there is always a parallelism between the hierarchical organization of the elements of a ritual action and that of the textual elements by which they are described or enjoined: if action A is subordinate to action B, then the sentence or sentences enjoining action A are subordinate to those enjoining action B.

30The chain of subordination may be extended indefinitely; the performance of any secondary action may require materials which may themselves require ritual preparation by further secondary actions.

85

which individual word meanings combine to form functionally

unified sentence meanings. This parallelism is highlighted by

the Mimamsakas' use of the term maha-vakya ("great-sentence")

to designate a collection of sentences which form a

functionally unified text. At all levels of language, it is

only subservience to a single purpose which allows disparate

semantic units to combine into coherent wholes.

Sentential Coherence and the Splitting of Sentences

The need to read sentences as functionally coherent

wholes plays a central role in Mimamsa text analysis. Any

interpretation which requires a sentence to serve multiple

injunctive functions will be vitiated by the flaw of vākya-

bheda ("splitting of the sentence"), and hence must be

rejected as untenable. As we shall see, determining the

precise nature of the unity required of sentences is a complex

matter which requires careful analysis; nevertheless, the

requirement of functional unity within sentences provides one

of the most basic tools the Mimamsakas bring to bear in the

resolution of particular interpretive problems.

One of the seminal treatments of vakya-bheda in the

Mīmamsa system is found in the Vajapeyādhikarana (MImāmsāsutra

1.4.6-8).31 The example under discussion is the following

sentence: vājapeyena svārājyakamo yajeta ("One who desires

sovereignty should sacrifice with vajapeya"). The word

31Sābarabhāsya and Tantravarttika, op. cit., Vol. II, pp. 290-294, and Brhati, Madras University Sanskrit Series 24, pp. 222-228.

86

vajapeya ordinarily signifies a kind of gruel made from rice

or other grains; can we understand it here as enjoining the

use of this substance in some sacrifice, or must we take the

word vajapeya to be the proper name of a sacrifice? At first

glance, the answer seems obvious. Sacrifice, by definition,

entails the offering of some substance to some deity. The

word vajapeya denotes a substance suitable for such an

offering. The sentence, then, may be interpreted as enjoining

a sacrifice, having sovereignty as its result, and in which

vajapeya is the substance to be offered.

But this will not do. If the sentence were interpreted

in this way, it would suffer from the flaw of vakya-bheda.

This reading requires the sentence to enjoin two quite

separate things: the performance of the sacrifice for the sake

of sovereignty, and the use of vajapeya as the offering-

material in this sacrifice. In effect, it treats the

utterance as if it were made up of two separate sentences:

"One who desires sovereignty should sacrifice" (svarājykāmo

yajeta), and "One should sacrifice by means of vājapeya"

(vājapeyena yajeta) . 32

It would not be possible to understand the word yajeta

("One should sacrifice"), uttered only once, as construing

with both svārājyakama and vajapeya, as the word would be

called upon to play different and mutually exclusive

functional roles in connection with these two; it would

32Śābarabhāsya, ibid., p. 292.

87

function as a predicate (vidhi) in the first sentence, and as

a subject of predication (anuvada -- i.e. that in connection

with which something is predicated) in the second.

Kumārila explains the problem as follows: in the

expression, "One who desires sovereignty should sacrifice, "

the act of sacrifice is predicated (vidheya) and presented as

something to be taken up (upādeya) as the means to a

particular result (i.e. sovereignty), and is therefore

subordinate (gunabhuta) with respect to that result; the

result (sovereignty), by contrast, is made the subject of

predication (anūdya or anuvadya) and presented as the point of

reference (uddesya) with a view toward which the sacrifice is

taken up, and is therefore predominant (pradhanabhūta) with

respect to the sacrifice.33 On the other hand, in the

expression, "One should sacrifice by means of vajapeya, " it is

the vajapeya that is predicated, that is to be taken up (as a

means to accomplishing the sacrifice), and is therefore

subordinate; the sacrifice expressed by the word yajeta is the

subject of this predication, the point of reference with a

view toward which the väjapeya is taken up, and is therefore

predominant with respect to it.

33Tantravärttika, ibid., p. 292. These two incompatible sets of semantic roles come to be known in the later Mimamsa literature as the viruddha-trikas, the "contradictory triads"; cf., for example, (on 1.4.8) Somesvarabhatta's Nyayasudha (in Mīmamsadarsanam, PracyabharatI Granthamala 22, Vol. II [1986] ) , p. Madhava' s Jaiminīyanyayamalavistara (Vrajajīvan Prācyabhāratī Granthamāla 35, 1989 [reprint] ), pp. 103,

55-56, and Bhavanatha's Mimamsanayaviveka (New Delhi: Rastriya Samskrta Samsthanam, 1977), p. 201.

88

Thus, if, in the sentence vajapeyena svārājykāmo yajeta

("One who desires sovereignty should sacrifice with

vajapeya"), the word vajapeya were taken to denote the

substance by means of which the sacrifice is to be performed,

then the word yajeta ("One should sacrifice") would be called

upon to play two contradictory semantic roles within one and

the same sentence. The requirement of sentential unity would

thereby be violated, and the sentence would suffer from the

flaw of vakya-bheda. This interpretation is therefore

dismissed as unacceptable; we must find a way to make sense of

the terms of the sentence without forcing a sentential split.

If the word väjapeya is understood to be the proper name

of a sacrifice, rather than to designate a particular

sacrificial material, the difficulty will vanish. One will

recall that for the Mimamsakas a sentence such as "One who

desires heaven should sacrifice" (svargakamo yajeta) , properly

analyzed, can be restated as "One should bring heaven into

being by means of sacrifice" (yagena svargam bhavayet).

Analogously, "One who desires sovereignty should sacrifice

with the Vajapeya" (vājapeyena svārājyakāmo yajeta) can be

reformulated as "One should bring sovereignty into being by

means of the sacrifice [named] Vajapeya" (vājapeyena yāgena

svārājyam bhavayet). On this reading, the term vajapeya and

the verbal root Vyaj ("sacrifice"), despite their disparate

syntactical treatment in the Vedic injunction, are co-

89

referential (samanadhikarana) .34 Thus, the sentence enjoins

a single bhavana with a single result (sovereignty) and a

single means (the sacrifice called Vajapeya) -- hence, no

vākya-bheda.35

Thus any given sentence can contain only a single

injunction directed toward a single result. But what exactly

constitutes a "single injunction"? Need the "single thing"

enjoined by a sentence be confined to a single word (such as

yajeta, "One should sacrifice") or a set of co-referential

elements (such as the name vajapeya and the verbal root Vyaj) ?

The answer is no. Only a single action can be enjoined by a

given Vedic sentence, but such an action may sometimes be

enjoined along with one or more of those productive factors

(kārakas -- cf. above, p. 78) which are directly linked with

it, and subordinate to it. This is known in Mīmāmsa parlance

as visista-vidhana ("qualified injunction").

As an example, consider the following sentence: "One

34On the co-referentiality of sacrificial names with the verbal root vyaj, see especially the Udbhidadhikarana (1.4.1- 2) -- Sabarabhasya and Tantravarttika, op. cit., Vol. II, pp. 268-277, and Brhati (Madras University Sanskrit Series 24), pp. 180-196. Sacrificial names are not, properly speaking, injunctive. They (typically, at least) convey no new information about the actions they designate; they perform a referential function within Vedic texts, serving to link sentences connecting a sacrifice with a particular result (as here) with others which specify materials or procedures for the same sacrifice.

35By contrast, if the word väjapeya were taken to denote a substance, the sentence would express two separate bhavanās, one with sovereignty as its result and sacrifice as its means, and another with sacrifice as its result and vājapeya as its means.

90

purchases the soma with a red, pink-eyed, one-year-old [cow] "

(arunayā pingāksyā ekahāyanyā somam krīnāti) .36 Here, it is

argued, "One purchases with a red, pink-eyed, one-year-old

[cow] " forms a single, complex predicate, enjoined in relation

to the soma, a substance to be used as a sacrificial offering

as part of the Jyotistoma (the sacrifice being described in

the passage in which the foregoing sentence is found). It

might be objected that accepting such a complex predicate

would expose one's interpretation to the same flaw of vākya-

bheda discussed above. The act of purchasing would be

predicated (vidheya) with respect to the soma, but would serve

as the subject of predication (anuvada) in relation to which

its qualifiers (the pink-eyed, one-year-old cow and the

redness of the cow) are enjoined;"7 it would thus be forced

to play contradictory semantic roles, just like the word

"sacrifice" in "One who desires sovereignty should sacrifice

with vajapeya." How can this or any sentence be interpreted

as a "qualified injunction" without destroying the unity of

the sentence meaning?

36The sentence is analyzed in one of the most important and most frequently cited sections of the Mimamsa system, the Arunādhikarana (3.1.12) -- cf. Sabarabhasya and Tantravarttika, op. cit., Vol. IV, pp. 33-65, and Brhati, Madras University Sanskrit Series 25, pp. 658-668; cf. also Mīmamsakośa, op. cit., pp. 529-622. Other important discussions of visista- vidhana are found in the Paurnamasyadhikarana (2.2.3-8 -- 'cf. Tantravarttika, op. cit., Vol. III, pp. 32-27) and the Paśusomādhikarana (2.2.17-20 -- cf. Tantravārttika, ibid., pp. 74ff) .

37Cf. Tantravarttika, op. cit., Vol. IV, p. 55 (on 3.1.12) .

91

The avoidance of vakya-bheda in cases of qualified

injunction is explained as follows: while, properly speaking,

only a single element can be enjoined by a single sentence,

that element which is enjoined (in this sentence, the act of

purchase), is qualified (visista) by several subordinate

elements. The qualified action can be enjoined only if its

qualifiers (visesanas -- e.g. the one-year-old COW, the

redness, etc.) are likewise understood to be enjoined.

Thereby, a sentence containing a qualified injunction, even if

it can (in accordance with the principle of ekavākyatā)

explicitly enjoin only a single element, can cause one to

understand, by implication, the injunction of those qualifiers

connected with it. As Kumārila explains, "This [i.e. the

injunctive word in a qualified injunction], having enjoined

once and ceased functioning, because its own meaning would

otherwise be incompatible, gives rise [by implication] 38 to

other injunctions bearing on each of the qualifiers; thereby,

without abandoning the unified form understood from the Vedic

38This kind of implication is what the Mimamsakas call arthapatti -- the necessary presumption of something otherwise unknown. When one knows of the existence or occurrence of something which would not be possible without the existence or occurrence of a second, unknown, thing, one may legitimately presume that this thing exists or has occurred. For example, if we are told that "Fat Devadatta doesn't eat by day, " we may legetimately presume that he does eat at night -- otherwise his fatness would be inexplicable (cf. Ślokavarttika, Arthapatti 51ff). The Mimamsakas argue that arthapatti is a means of valid knowledge in its own right, and must be recognized as something Ślokavärttika, distinct from inference (anumāna) -- Arthapatti Śalikanathamisra's cf.

pp. 273-278. Prakaranapañcika (Benares Hindu University Darsana Series 4), 10-35 and

92

recitation, it is able to achieve the same result as if it

were repeated [in connection with each of these

qualifiers] . #39 Thus, the sentence, "One purchases the soma

with a red, pink-eyed, one-year-old [cow]" is able to enjoin

the act of purchase, the pink-eyed, one-year-old cow, and the

redness of the cow, without violating the requirement of

sentential unity.

This theory of visista-vidhana obviously provides a

powerful interpretive tool for dealing with complex

injunctions without violating the requirement of ekavākyatā;

indeed, it might at first seem, too powerful. What is to

prevent one from interpreting a sentence such as "One who

desires sovereignty should sacrifice with vajapeya" as a case

of "qualified injunction"? On this interpretation, a single

element (sacrifice), qualified by a particular offering

material (vajapeya), would be enjoined with a view toward a

single result (sovereignty) . Hence there would be no

splitting of the sentence, and no reason to interpret the word

vājapeya as the name of a sacrifice, rather than as

designating a substance to be used in the sacrifice -- the

accepted interpretation of this and similar sentences, and the

argument for the very existence of sacrificial names, would

collapse.

Obviously, such an interpretation cannot be accepted; but

39Tantravarttika, op. cit., Vol. II, p. 33; cf. also Tantravārttika, Vol. IV, pp. 95-96 (on 3.1.15).

93

how can one justify the disparate treatment of sentences such

as "One purchases the soma with a red, pink-eyed, one-year-old

[cow]" (where visista-vidhana is allowed) from those such as

"One who desires sovereignty should sacrifice with vajapeya"

(where it is not)? The question is not dealt with as clearly

as one might wish in the Mimamsa literature, but the crucial

difference would seem to be this: In the sentence "One

purchases the soma with a red, pink-eyed, one-year-old [cow] ",

despite the complex, multi-partite predicate, there is only a

single thing which must be brought into being (i.e., the act

of purchasing) and, consequently, a single bhavana. This

purchasing is performed for the sake of the soma (i.e., in

order to make it available for later sacrificial use), and is

hence subordinate to it, but the soma itself is not produced

by this action; the only new thing brought into being by the

performance here enjoined is the act of purchase itself. By

contrast, in the sentence "One who desires sovereignty should

sacrifice with vājapeya," (if vājapeya is taken to denote a

substance) there are two distinct things to be brought into

being, sacrifice (by means of vajapeya) and sovereignty (by

means of sacrifice) : there is no way these two products can be

connected with a single bhavana, and hence no way to arrive at

a unitary sentence meaning. It is the impossibility of

arriving at a single bhävana with a single result which

prevents one from interpreting this and similar sentences as

94

cases of qualified injunction.40

To return to the example in question: in the sentence

"One purchases the soma with a red, pink-eyed, one-year-old

[cow]," the ability of "One purchases with a red, pink-eyed,

one-year-old [cow] " to operate as a single, complex predicate

depends on its having a functionally unified, hierarchical

structure of its own (just as complete sentences and higher

order textual units must be hierarchically ordered towards a

single purpose if they are to convey a coherent sense). The

complex predicate must contain a single predominant element,

to which all its other components are subordinated. As

before, it is the active, verbal element which assumes the

predominant role. It is primarily the act of purchase which

is predicated here; the other elements of the complex

predicate serve only as qualifiers of this action, and are

hence subordinate to it.

This understanding of the structure of complex predicates

forces the MImamsakas to challenge and ultimately reject some

of our more basic intuitions regarding the relationship

between words in a sentence. Consider the three subordinate

terms in "One purchases with a red, pink-eyed, one-year-old

[cow]" (arunayā pingāksyā ekahāyanyā krīnāti). Two of these

40At is the explanation given by Bhavanätha, in the only explicit discussion of the question I any rate, this

have been able to find. Bhavanatha explains the distinction between the two sentences in his commentary on the Arunadhikarana (3.1.12) -- cf. Bhavanatha, Mimamsanayaviveka,

506-7. unpublished manuscript, Adyar Library 1.A.27, Vol. II, pp.

95

terms, pink-eyed (pingaksya) and one-year-old (ekahayanya),

are morphologically complex and are therefore to be

interpreted, for reasons already discussed, 41 as referring to

a particular substance (here understood, by context or

implication, to be a cow), qualified by the attributes denoted

by the components of these words (pink eyes and one year of

age) . Both words are marked with the instrumental case-

ending, indicating that they denote the means by which the

action of purchasing is to be performed. The substance they

denote can be exchanged for the soma to be purchased, and is

thereby able to perform the role assigned it by the

instrumental case-marking; thus there is no difficulty in

connecting these terms with the verb "One purchases" (krināti)

as subordinate elements.

But consider the remaining term in the predicate, "red"

(arunaya). This word denotes not a substance but a quality,

a color. At first, it seems natural to assume that "red" is

a qualifier of the pink-eyed, one-year-old cow; color is,

after all, a property of substances. But the Mīmamsakas'

analysis of the structure of complex predicates does not allow

for such an interpretation. If "red" is to form part of a

single, functionally unified predicate, it must be interpreted

as a qualifier not of the cow, but of the primary element of

the predicate, the act of purchase. If one were to interpret

the word "red" as a qualifier of the cow, rather than of the

41Cf. above, p. 74.

96

act of purchase, a splitting of the sentence would result; the

cow would be enjoined in connection with the act of purchase,

yet would serve as the subject of predication (anuvada) in

connection with which the quality of redness is enjoined.42

The cow cannot serve as qualifier and qualified in the same

sentence without creating a vakya-bheda. Furthermore, the

word "red", like "pink-eyed" and "one-year-old", is marked

with the instrumental case-ending; thus its referent, like

theirs, must be directly connected with the act of purchase

and must serve as a means toward the accomplishment of this

action.43

But against this interpretation the following objection

can be raised. Redness'is a property which can belong only to

substances; it is, in and of itself, formless (amūrta). It

would not be possible for such a formless, insubstantial

property to serve as a means of purchasing, or of any other

action. 44 Given this impossibility, the fact that the word

"red" is marked with the instrumental suffix cannot be

regarded as significant -- as Sabara's hypothetical opponent

puts it, "An unperformable thing cannot be enjoined even by a

hundred sentences. #45 "Redness" can in no way be used as a

means of purchase; if it is to convey any useful meaning

42Cf. Śābarabhāsya, op. cit., pp. 51, 55, and 62.

43Cf. Śabarabhāsya (on 3.1.12), op. cit., Vol. IV, p. 50.

44Ibid., pp. 40-41.

45Ibid., p. 43; cf. also p. 50.

97

(which, as part of the uniformly authoritative Veda, it must)

it must be connected with some substance -- either with the cow

mentioned here, or with other substances mentioned in the same

context (prakarana) .

Sabara answers the objection as follows: It is true that

qualities such as color must always qualify some substance or

other. But this connection between quality and substance is

not what the Vedic injunction expresses. sses. The The word "red"

(arunaya), like the words "pink-eyed" (pingaksya) and "one-

year-old" (ekahāyanya) (which refer directly to the cow), is

marked with the instrumental case-ending, indicating that it

is a means for the act of purchasing. This is a direct,

explicit statement of the Veda, and cannot simply be set

aside; we must find some way to make sense of this overt

declaration.

accomplish The meaning conveyed by the Veda is such: "One should purchasing by means Of the quality of redness. " One must try to find a way SO that [purchasing] can be accomplished by this [redness]. And a quality cannot accomplish any action except by qualifying some [material] means [sadhana] - - implication, thus, by [redness] becomes qualifier of the [material] means [i.e. the cow] 46

All the Vedic injunction tells us is that the act of

purchasing the soma is to be performed by means of a pink-

eyed, one-year-old cow, and by means of the quality of

redness. Our own understanding of the nature of qualities

leads us to understand that the redness mentioned must belong

46Ibid., p. 62.

98

to some substance and thence to connect the redness with the

pink-eyed, one-year-old cow mentioned in proximity to it; but

this connection is not directly conveyed by the Vedic

injunction itself. What is here enjoined is a single action

(purchase), qualified by a set of subordinate factors which

serve as a means to this action; thus "One purchases with a

red, pink-eyed, one-year-old [cow]" can be understood as a

single, complex, but functionally unified, predicate, enjoined

in relation to the soma. Again, it is the need to understand

all units of language as hierarchically ordered toward a

single purpose which governs interpretation.

Any meaning-elements which cannot be integrated into such

a functionally unified structure must be considered incidental

to the actual purport of the Vedic text which conveys them,

and must consequently be disregarded. Such elements, although

they are actually expressed (vacya) by the Vedic text in

question, are not part of what is "intended to be expressed"

(vivaksita) by it.

Consider the following example. In a passage describing

the Jyotistoma sacrifice, this sentence is found: "One wipes

the spoon with a woolen strainer" (dasapavitrena graham

sammärsti) .47 Here the following question is raised: Should

47The example is in Grahaikatvādhi-karana discussed (MImāmsāsūtra at length the 3.1.13-15) -- Sabarabhsya and Tantravarttika, op. cit., Vol. IV, pp. 65-96, cf.

and Brhati, Madras University Sanskrit Series 25, pp. 668-686. The discussion immediately follows that of "One purchases the soma with a red, pink-eyed, one-year-old [cow]," to which it provides a counter-example.

99

one wipe one and only one spoon, or all of those spoons to be

used in the Jyotistoma sacrifice? The answer may seem obvious

at first. The word spoon (graha-) is marked with the

accusative singular case-ending (-m) ; part of what this suffix

conveys is that the noun to which it is attached refers to a

single entity; hence only one spoon should be wiped.48 But

the Mimamsakas argue that, in spite of the use of the singular

suffix, all spoons connected with the Jyotistoma should be

wiped; the singleness expressed by the suffix is not "intended

to be expressed" (vivaksita) by the Vedic injunction, and

hence must be dismissed as insignificant.49

In defense of this conclusion the MImamsakas turn, once

again, to the notion of vakya-bheda. The sentence in question

enjoins the act of wiping in connection with the spoon or

spoons designated by the word graha -; if the singleness

48Śabarabhāsya, ibid., pp. 70-71.

49But what does it even mean to speak of expressive intent (vivaksa) in connection with the eternal, authorless Vedic texts? Sabara does not directly address the question, but Kumārila considers the problem in some detail, and offers several possible explanations (cf. Tantravarttika on 3.1.13 -- op. cit., Vol. IV, 65-70) . The intentionality to the Vedas may be metaphorical; just as we pp. ascription of

may say of a river bank which is on the verge of collapse, "It wants to fall" (pipatisati), likewise we may say of a particular Vedic text that it "wants" or "does not want" to express some particular component of its expressed meaning, depending on whether or not this component can or cannot form part of a coherent injunction, without literally ascribing consciousness of desire to the text. Alternatively, the intentionality referred to may be that of competent Mīmāmsakas and ritualists, who have transmitted and interpreted the Vedas from time immemmorial. Or, one might even suppose that the Vedic texts are themselves conscious beings capable of literally holding intentions.

100

expressed by the nominal suffix were enjoined in connection

either with the act of wiping or with the spoon/s, a splitting

of the sentence would result. As Śabara explains, if this

singleness were enjoined in connection with the act of wiping,

it would have to be either predominant or subordinate with

respect to that action.50 The former would be impossible.

There is no way in which the act of wiping could assist or

modify "singleness" at all. And even if there were, this

"singleness" itself could serve no useful purpose here or

elsewhere.51 The spoons in connection with which the act of

wiping is enjoined are used as sacrificial implements in the

performance of the Jyotistoma, a rite which has heaven as its

result; thus, via its role in preparing these spoons for

ritual use, the act of wiping ultimately assists in the

achievement of a desirable result. The "singleness" expressed.

here, unlike the spoons, does not aid in the performance of

any fruitful ritual act, so, if the act of wiping were

enjoined with respect to it, this act, too, would be

ultimately useless.

Nor, Sabara argues, can this "singleness" be enjoined as

a subordinate element in connection with the act of wiping.

"Singleness" is a formless (amūrta) quality, and thus cannot

assist in accomplishing the act of wiping. Here, an obvious

50In accordance with the principle that relations between elements of a text are invariably hierarchical -- cf. above, pp. 76-77.

51Śabarabhāsya, ibid., pp. 84-85.

101

objection arises: it has just been established, in the

discussion of "One purchases the soma with a red, pink-eyed,

one-year-old [cow]," that even a formless quality such as

redness can (indirectly) assist in the accomplishing of an

action, by qualifying some substance to be used in that

action. Śabara responds as follows:

Even if [in other sentences] some other formless [quality] accomplishes an action by qualifying its means [sadhana], nevertheless, this is not possible [here]. For here the spoon is not [enjoined] for the sake of wiping; here wiping is enjoined for the sake of the spoon. For the spoon has a purpose [in the larger ritual context]. One would have to postulate a purpose for the wiping [if it were not performed for the sake of the spoon]. If the spoon assisted the act of wiping, then [the singleness] would assist [the wiping] by means of assisting this [spoon]; thus the singleness would be an assistor of the act of wiping. But this is not the case. Therefore there is no connection between singleness and wiping.52

The crucial difference between this case and that of the red

cow lies in the inversion of the hierarchical relationship

between substance and action.

In the case of the red cow, the insubstantial quality

redness can be understood to assist in the act of purchase

because it qualifies the particular material to be used as the

means of purchase (i.e. the cow). In this sentence, the spoon

is not enjoined as a means for accomplishing the act of

wiping -- rather, the wiping is to be performed for the sake of

the spoon (or spoons). Other sentences in the passage

describing the Jyotistoma enjoin the use of the spoons to pour

52Ibid., p. 85.

102

liquid sacrificial offerings -- thus they serve a known purpose

in the larger ritual context. The same cannot be said for the

act of wiping. No result is mentioned in connection with this

action; if it were not presumed to be performed in order to

prepare the spoons for use elsewhere in the Jyotistoma, one

would have to postulate for it some unspecified transcendent

result. In accordance with the principle of economy of

hypothesis (kalpana-laghava), one ought to avoid postulating

such unstated results wherever possible. Thus one must

conclude that the wiping is to be performed for the sake of

the spoon/s. Furthermore, the word spoon (graha-) is marked

with an accusative case ending, indicating that it is not a

means for act of wiping, but rather its direct object -- that

toward which the act of wiping is directed.53 Since the

53The Sanskrit grammarians define the direct object as "that which is most desired by the actor" (kartur Ipsitatamam -- cf. Panini's AstadhyayI 1.4.49) . Nevertheless, it is the usefulness of the spoons in the larger ritual context, rather than the accusative case ending, which ultimately determines its predominance with respect to the act of wiping. When there is no such usefulness in the larger context, even a substance expressed by a word in the accusative case will, perforce, be understood as a means for the action expressed in the sentence, and hence as subordinate to it; in such cases, the accusative case-ending must be interpreted figuratively as expressing not predominance with respect to the action, but relation to the action in general (karaka-samanya -- cf. below, pp. 114-115) . qualities associated with this substance (number, gender, Hence any

color, etc.) will likewise be subordinate to the action, and will consequently be "intended to be expressed" (vivaksita). E.g. in the sentence "One should sacrifice a domestic animal (pasum alabheta) the word for "domestic animal" (pasu) is marked with the masculine, singular, accustive case-ending -m. The animal in question is not used elsewhere in the ritual context (such that the act of sacrifice could be understood as a preparation of the substance for subsequent ritual use);

103

spoon or spoons are not subordinate to the act of wiping, the

singleness expressed by the nominal suffix cannot, by

qualifying this spoon/s, indirectly subserve the act of wiping

(as redness can subserve the act of purchase by qualifying the

cow). Hence singleness cannot be enjoined in connection with

the act of wiping.

Nor can it be enjoined in connection with the spoon/s.

If the act of wiping and the quality of singleness,

unconnected with one another, were both enjoined in connection

with the spoon/s, another splitting of the sentence would

result. One cannot argue that the spoon/s is/are enjoined in

connection with two subordinate elements, forming a case of

qualified injunction (visista-vidhana) ; here, the spoons are

not enjoined at all. The spoons are enjoined in other

sentences describing the procedures of the Jyotistoma, as a

means for pouring liquid offerings. In this sentence the

spoons, a known element, are not part of what is predicated;

rather they serve as the subject of predication (anuvada),

that in relation to which the act of wiping is predicated

here, the animal is the material offering which makes the act of sacrifice possible, and therefore must be understood, despite the use of the accusative case, as a means for the act of sacrifice. Hence those qualities mentioned in connection with it (i.e. the singleness and masculine gender expressed by the suffix) are intended to be expressed, and one must sacrifice a single, male animal. The sentence is discussed as a counter-example to "One wipes the spoon" (under Mimamsasūtra 3.1.15 -- cf. Sabarabhasya and Tantravarttika, Vol. IV, op. cit., pp. 95-96) and at more length in the Pasvekatvadhikarana (MImamsasutra 4.1.11-16 -- cf. Sabarabhasya and Tuptika, op. cit., Vol. V, pp. 12-16 and BrhatI, Madras Sanskrit Series 25, op. cit., pp. 971-974).

104

(vidheya) and enjoined as something to be performed. If, in

this single sentence, the act of wiping and the quality of Ehe a singleness were each predicated in relation to a single

subject of predication, the requirement of sentential unity

would be violated.

Thus there is no way to interpret this sentence such that

the quality of singleness can be integrated with the other

elements of the sentence to form a functionally unified whole.

As Sabara says, " [Here], taking the spoon as its point of

reference, the act of wiping is enjoined in connection with

it. If, in addition, connection with singleness were enjoined

with respect to either the substance [i.e. the spoon/s] or the

act of wiping, then, because two connections would be

enjoined, the sentence would be split ... Thus this singleness

is unconnected with either. #54 Since it cannot be integrated

into the sentence meaning without producing a vakya-bheda, we

must understand that singleness, although it is expressed by

the nominal suffix, is "not meant to be expressed"

(avivaksita) ;55 it is incidental to the overall purport of

54 Śabarabhāsya, op. cit., Vol. IV, p. 86.

55The concept of vivaksa ("expressive intent") plays an important role in the Sanskrit grammatical tradition as well as in MImamsa. It is important, for our purposes, to note the very different uses to which the notion of vivaksā is typically put in these two traditions. The grammarians' primary task is to explain the derivation of proper morphological and syntactic structures, and they are consequently interested in the determinant of grammatical form -- role of vivaksā as a in certain cases, the presence of a specific expressive intent determines the selection of one possible form, which expresses the intended

105

the sentence, and so must be disregarded. Hence all the

spoons employed in connection with the Jyotistoma are to be

wiped.

The Ultimate Meaning of a Text

As we can see, it is only in terms of the larger

purposive structure of an entire text that the exact

significance of, and the precise relations between, the

various elements expressed by the individual words and

morphemes of a given sentence can be determined. Considered

in isolation from its context, even a comparatively simple

sentence such as "One wipes the spoon, " could bear any of a

wide range of possible significations -- the act of wiping,

qualified by the spoon/s and/or the singleness expressed by

the nominal suffix, could be enjoined as a single complex

predicate; either the wiping, or the singleness, or both,

could be predicated and enjoined in connection with the

spoon/s; the wiping, qualified by singleness, could be

enjoined in connection with the spoon/s, or, qualified by the

sense, over another, which does not (cf., for example, Patañjali's Mahabhasya on Panini 1.2.69, where the role of vivaksa in determining the formation of ekasesa compounds is discussed). By contrast the Mimamsakas, being primarily concerned with the interpretation of existing texts, use the notion of vivaksa to differentiate between the relevant and irrelevant meanings of a particular expression in a specific context -- of the many possible senses an expression actually does convey, some are "meant to be expressed" [vivaksita] and some are not (and may hence be disregarded). It is in the latter, Mimamsa-derived, sense that Anandavardhana applies the notion of vivaksa to poetic analysis, using it to classify the meanings conveyed by particular verses as either intended or unintended -- cf. below, Chapter Five, pp. 198-203 and Chapter Six, pp. 285-286.

106

spoon/s, could be enjoined in connection with singleness.

Kumārila lists dozens of possible variations in the precise

injunctive meaning which might be understood from this single

sentence.56

The job of the Mimamsaka is, by a careful analysis of the

functional connections between such a sentence and other

surrounding ones, to settle upon one of these many possible

meanings as contextually appropriate; one must determine which

elements of the sentence are predicated (vidhi) and which form

the subject of predication (anuvada), which are primary and

which are subordinate, which are qualified and which are

qualifiers, which are "intended to be expressed" (vivaksita)

and which, if any, are "not intended to be expressed"

(avivaksita). The specific meaning arrived at when all of

these variables have been fixed for a particular sentence in

a given context is called by the Mimamsakas the "sentence

particular" (vacana-vyakti) . As the term itself would

suggest, the process by which sentence-meanings mutually

specify one another to produce a specific textual meaning is

analogous to that by which individual word meanings within a

sentence mutually specify one another so as to refer to

particular entities.

As we have seen, according to the Mīmāmsakas'

understanding of verbal signification, an individual word can

refer only to a universal class property (jati, akrti), never

56Tantravarttika, op. cit., Vol. IV, pp. 75-76.

107

to a particular (vyakti) ; however, the several class

properties denoted by the individual words of a sentence can,

by specifying and delimiting one another, designate one or

more particulars in which these class properties are

instantiated. Similarly, a text consists of a collection of

sentences, each of which, taken individually, could convey any

of a wide range of meanings; functioning collectively,

specifying and delimiting one another, the general meaning

conveyed by each can be reduced to a particular sentence

meaning (vacana-vyakti), in which the relations of subject and

predicate, primary and subordinate, etc. are definitely fixed.

The determination of the precise vacana-vyaktis of

particular sentences in particular contexts is fundamental to

the Mimamsa project of Vedic interpretation. As Kumārila

says

In a single sentence, out of the many meanings understood from the hearing of stems, suffixes, and words, which is comprised in the injunction or useful [in connection with what is enjoined], and which is not ?-- in considering this, because all [of these meanings] are seen [in the sentencel, it is only by the differentiation of vacana- vyaktis that one can determine which is enjoined [vihita] or used [viniyukta], and which is not. 57

Thus it is only by determining the vacana-vyaktis of the

sentences that make up a Vedic text that one can ascertain

definitely what is to be done and what is not to be done in

57Tantravarttika, ibid., p. 74. For more on vacana-vyakti cf., for example, Sabarabhasya on 1.1.1 (op. cit., pp. 5-6), 3.1.12 (Vol. IV, p. 55), and 6.1.1 (Vol. V, p. 174); Tantravarttika on 1.3.7 (op. cit., Vol. II, p. 135), 1.4.10 (p. 299), 2.2.6 (Vol. III, p. 36), 2.2.10 (p. 46), 2.2.16 (p. 54-55), and 3.1.8 (Vol. IV, p. 26).

108

performing the ritual enjoined by that text.

So, in the comprehension of any complete, coherent text

(Vedic or otherwise) -- that is, a text whose parts are

hierarchically ordered toward a single ultimate goal -- the

hearer or reader of the text must pass through a series of

interpretive stages. One first comprehends the universal

meanings of individual words and morphemes. Certain of these

meanings (i.e. the bhavanas expressed by verbal suffixes -- cf.

above, pp. 78ff) generate a psychological expectancy which one

seeks to fill by connecting these meanings with those of

surrounding words or morphemes. Groups of words, united by

their capacity to fulfill this expectancy generated by

bhavana, specify one another and form particular, coherent

sentence-meanings. These sentence meanings are likewise

joined with one. another by their capacity to fulfill the

expectancy generated by some bhavana (since, if the three

"needs" generated by a bhavana cannot be filled by the words

within a sentence, one must look to surrounding sentences for

the requisite information -- cf. above, p. 82), and further

specify one another so as to convey particular vacana-vyaktis.

At this point all semantic components of the text are

integrated into a single, functionally unified structure,

organized toward a single end (in the case of a Vedic text,

the injunction of a complex ritual act which will produce a

particular sacrificial result for its performer)

This final stage of textual comprehension, in which the

109

precise, hierarchical relations of all semantic components of

a text, both within and between sentences, have been

definitely ascertained, the Mīmamsakas call paryavasāna58

("completion" -- as opposed to avāntara-vākyārtha, 59

"intermediate sentence meaning", a meaning understood from a

part of a text which must ultimately be revised or restricted

in the light of context). Paryavasäna is the end point of the

hermeneutical process, the point at which all interpretive

questions have been settled and the ultimate, functionally

integrated meaning of the entire text becomes clear.

Figurative Expression60

It is only in connection with the need to arrive at a

single, functionally unified, textual meaning that the

question of figurative or non-literal expression (laksanā or

gunavrtti) 61 arises. Each of the individual words and

58Cf., for example, Sabarabhsya on 3.1.12 (op. cit., Vol. IV, p. 56); Slokavārttika, Vakyadhikarana 342 (op. cit., p. 664); Tantravärttika on 1.4.24 (op. cit., Vol. II, p. 328), 2.1.1 (pp. 334, 338, 342, 349), 2.2.6 (Vol. III, p. 33), 2.2.18 (p. 72), 2.2.19 (p. 79), 2.2.21 (p. 83), and 3.1.14 (Vol. IV, p. 84); BrhatI on 1.3.33 (pp. 168, 171) .

59Cf., for example, Tantravarttika on 1.2.7 (op. cit., Vol. II, p. 16) and 1.3.7 (p. 135) .

60The most important discussions of figurative expression can be found under MImamsasutra 1.4.23 (cf. Sabarabhasya and Tantravarttika, op. cit., Vol. II, pp. 113-127 and BrhatI, Madras University Sanskrit Series 24, pp. 255-267) and Mimamsāsutra 3.2.1 (cf. Sabarabhasya and Tantravarttika, op. cit., Vol. IV, pp. 123-130, and BrhatI, Madras University Sanskrit Series 25, pp. 707-716) .

61The MImamsakas (especially Kumārilabhatta and his followers) draw a distinction between gunavrtti (metaphorical expression) and laksanā (non-metaphorical figurative

110

morphemes of which any given text is composed will express a

single literal meaning -- that meaning with which it is

eternally and invariably connected. As we have seen, these

meanings, mutually specifying one another, combine to form

coherent (that is, uni-functional) sentential and textual

meanings. Any meaningful component of a sentence or text

which, interpreted literally, could not be satisfactorily

integrated within the overall meaning of that sentence or

text, must be interpreted as conveying some secondary or

figurative meaning which can be integrated within that overall

meaning. Thus, figurative interpretation is always a last

resort -- whenever it is possible to do so, one must interpret

expression -- metonymy, synecdoche, etc.); nevertheless, the processes of gunavrtti and laksana are basically the same and are governed by the same contextual conditions. Kumārila explains the distinction as follows: "The use [of a word] in a meaning necessarily connected with its literal meaning is called figurative expression [laksanā]; but usage arising from contact with indicated qualities [laksyamāna] [gunas] which are figuratively

(Tantravarttika, op. cit., Vol. II, p. 313). is metaphorical [gauna] " Metaphorical expression involves a two-stage figurative process. In the sentence "Devadatta is a lion" (simho devadattah) the word "lion" first figuratively indicates the qualities associated with a lion (fierceness, bravery, etc.) ; the word "lion", indicating these qualities, refers, by a further figurative indication, to Devadatta (understood as the possessor of these qualities). "Devadatta is fierce, brave, etc." Thus we understand the sentence to mean So, while metaphorical usage is a multi-stage process and is thus somewhat more complex than metonymical or other figurative expression, it involves the same basic operation -- the replacement of a contextually inapplicable literal sense with a closely connected meaning which is suitable to the context.

111

literally. 62

There are three conditions which must be met if one is to

interpret any word or expression figuratively. First, as just

discussed, there must be no possibility of understanding the

word or expression in its literal sense

(mukhyarthasambhava) .63 Second, there must be some

connection (sambandha -- e.g. physical proximity, similarity,

cause/effect relation) 64 between the literal meaning of the

word or expression and the figurative meaning to be understood

from it -- otherwise any word could figuratively convey any

meaning at all, and there would be no way of settling on a

particular meaning as the correct one. Third, the meaning to

be understood figuratively must have some purpose

(prayojana) -- that is to say, it must, unlike the primary

meaning of the expression in question, connect usefully with

the overall meaning of the text in question and thus form part

62Thus, the process by which one moves from literal to

91). figurative interpretation is one of arthapatti (cf. above, p. Where an expression, interpreted literally, would make no sense, one must assume that some figurative meaning is intended -- otherwise, the use of words or expressions which can convey no contextually appropriate literal meaning would not be explicable. Cf. K. Kunjunni Raja, Indian Theories of Meaning (Madras: Adyar Library and Research Centre, 1977 [second ed.]), pp. 258-9.

63Cf. Devasthali, Mimamsa: The Ancient Indian Science of Sentence Interpretation, op. cit., pp. 58-59 and K. Kunjunni Raja, Indian Theories of Meaning, op. cit., pp. 258ff.

64Cf. Śābarabhāsya (on 1.4.23), op. cit., Vol. II, p. 316. Mimamsasutra 1.4.23 provides a list of possible connections which may hold between literal and figurative meanings. Similar lists may be found in Nyayasutra 2.2.62 and in Patañjali's Mahabhasya on Panini 4.1.48.

112

of a functionally integrated, coherent text.65 So the

literal meaning, which cannot connect with the overall

purposive structure of the text, is replaced by a second,

related, meaning which can connect with it.

For example, consider the sentence, "The grass-bundle is

the sacrificer" (yajamanah prastarah) .66 If both words in

this sentence were interpreted literally, we would have to

understand the sentence either as assigning the role of the

grass-bundle to the sacrificer, or as assigning the role of

the sacrificer to the grass-bundle. The grass-bundle is used

to form a bed on which the sacrificial spoons are placed when

not in use; if the sacrificer were used as a bed for the

spoons, he would be unable to perform the actions required of

him in the course of the ritual performance. The grass-bundle

is insentient, and incapable of performing the actions

required of the sacrificer. Thus, considered in the light of

the larger ritual context, there is no way to derive an

acceptable meaning from the sentence if both its component

words are understood literally.

However, if the word "sacrificer" is interpreted

figuratively, the interpretive problem can be resolved. The

65Cf. Śābarabhāsya on 6.4.43 (op. cit., Vol. V, p. 297) and Devasthali, op. cit., p. 66. The same three conditions for figurative expression are recognized by Anandavardhana and the later Ālamkārikas but, as Devasthali notes, their understanding of prayojana differs markedly from that of the Mīmāmsakas (cf. below, Chapter Five, p. 228) .

65 Sabarabhāsya, op. cit., Vol. II, pp. 113ff.

113

grass-bundle assists in the accomplishment of the sacrificer's

purpose (the performance of the sacrifice and the enjoyment of

the appropriate sacrificial result), and may on this ground be

metaphorically described as "the sacrificer". The

identification of the grass bundle with the sacrificer

emphasizes the importance of its role in the sacrifice and may

thus serve as a form of praise; this praise helps to motivate

the sacrificer to use the grass-bundle as part of the

sacrificial performance. Thus, if the word "sacrificer" is

interpreted figuratively, the sentence can be seen as serving

a useful function and subserving the overall purpose of the

text. 67

This replacement of a figurative meaning for an

unconstruable literal one can occur at any level of language.

Words, whole sentences, or even individual morphemes may, if

their literal meaning cannot be integrated with those of

surrounding elements, be understood as conveying a secondary

or figurative sense. Figurative expression at the word level

67The interpretation of this and similar sentences as praising elements of the sacrifice is a crucial element of the Mīmāmsā to Vedic exegesis The Mīmāmsakas' understanding of the nature of the Veda demands that any approach

element of any Vedic text subserve, directly or indirectly, the injunction of one sacrificial action or another. Yet many Vedic sentences do not directly enjoin anything -- e.g. statements of fact, narratives of the acts of the gods, etc. Such statements, it is argued, figuratively and understood as praising particular sacrificial must be interpreted

acts or substances (or as dispraising acts or substances to be avoided), and thus indirectly subserving one sacrificial injunction or another. Such passages are known as arthavādas. Cf. Śabarabhasya on 1.2.1-18 (op. cit., Vol. II, pp. 1-32) ; also Jha, Purva-Mimamsa in its Sources, op. cit, pp. 177-182.

114

is seen in the sentence just discussed. The figurative

interpretation of whole sentences must often be resorted to in

the interpretation of arthavāda passages -- passages which

consist of merely declarative statements, and thus,

interpreted literally, could not be seen as subserving the

overall purpose of the text in which they occur (namely, the

injunction of some particular sacrificial action); such

statements are to be understood as figuratively conveying

praise for those elements of the sacrifice with which they are

connected and hence indirectly subserving the overall purpose

of the texts to which they belong.68

For figurative expression at the morpheme level, consider

the sentence "One offers the grains as an oblation" (saktuñ

juhoti) .69 Here the word "grains" is marked with the

accusative case-ending. The accusative case ordinarily

indicates that the referent of the word marked by it is

predominant with respect to the action expressed by the verb

(cf. above, pp. 102-103) . But here the accusative ending

cannot be interpreted literally. The grains mentioned in this

sentence serve no purpose in the larger ritual context; if the

offering as an oblation were performed for their sake, it too

would fail to connect usefully with the larger ritual context,

and the sentence as a whole would be purposeless (anarthaka).

68Cf. Kunjunni Raja, op. cit., p. 269.

69Cf. Śābarabhāsya on 2.1.11-12 (op. cit., Vol. II, pp. 384-389) .

115

Thus, the grains cannot be predominant with respect to the act

of offering, and the literal interpretation of the accusative

case-ending must be rejected as untenable. We must interpret

the case-ending as figuratively conveying general connection

with the action (karaka-samanya), rather than the specific

connection with the action literally expressed by the

accusative. 70

So, figurative signification may operate at any level of

textual organization, but the resort to figurative

interpretation is always driven by the same basic need -- the

need to integrate all the components of a given text into a

hierarchically ordered unit directed toward a single ultimate

purpose. It is only when the literal meaning of any component

cannot be fitted into such a framework that the device of

figurative interpretation is acceptable.

However while, in theory, figurative interpretation may

be resorted to only when absolutely necessary, in practice,

the need to resort to it proves to be pervasive. As we have

seen (above, pp. 70ff), the Mimamsakas argue that individual

words can directly refer only to universal class-properties;

it is only through the interaction and mutual specification of

these universal meanings that words, construed in functionally

unified sentences, can designate particulars. Śabara takes

this process to be one of figurative signification. Words can

literally express only universals; when they are used to refer

70Cf. Tantravarttika, op. cit., Vol. II, pp. 386-387.

116

to particulars (as they generally are) they are being used

figuratively. As he says, "People figuratively employ

[upacaranti] the word domestic-animal [pasu], which expresses

a class-property [akrti], with reference to a particular

substance having horns, a tail, hair, and four feet."71

Kumarila and the Bhatta-Mimamsakas, following this line of

reasoning, argue that all sentence-meanings are conveyed

figuratively. 72 The construing of word-meanings intoa

unified sentence-meaning always involves the replacement of

unconnected, universal meanings with particularized and

contextually delimited meanings connected with these

universals -- hence sentence meanings are invariably conveyed

71Śabarabhāsya (on 2.2.17), op. cit., Vol. III, p. 68.

72However, Prabhakara and his followers do not agree that all sentence meaning is conveyed figuratively. The Prābhakaras' understanding of the nature differs markedly from that of the Bhattas. sentence-meaning Prabhākara held the view that the words of a sentence express their meanings already construed with one another (anvitabhidhana-vada) . This is contrasted with the Bhatta view that the words of a sentence express their individual meanings, which are only subsequently construed (abhihitānvaya-vāda) into a coherent sentence-meaning Thus for Prabhakara the understanding of the sentence-meaning does not arise, by a process of figurative expression, after the individual word-meanings have been understood: the complete, construed sentence-meaning is conveyed directly and immediately by the words themselves. It is hard to see how this theory can be squared with statements of Sabara such as that quoted above, and with the Mīmamsa view

as their that individual words can have only universal class-properties meanings; nevertheless, the theory has been vigorously defended by the later Prabhakaras, most notably by Śālikanāthamiśra in his Vakyarthamatrkavrtti (in Prakaranapañcikā, ed. A. Subrahmanya Sastri,Benares Hindu University Darsana Series 4, 1961, pp. 376-450). On abhihitānvaya-vāda and anvitābhidhāna-vāda, cf. also Kunjunni Raja, Indian Theories of Meaninq, op. cit., pp. 193-213.

117

by a process of laksana.73 One may resort to figurative

interpretation only when forced to do so, but, in practice,

one is always forced to do so. Thus the device of figurative

interpretation is basic to all language exegesis. And, like

all Mimamsa exegetical techniques, it is guided in every case

by the need to integrate all elements of a text into a single,

uni-functional whole.

Prospectus

As we shall see, the MImamakas' teleological, goal-

oriented method of textual interpretation serves as a model

for Anandavardhana in his radical approach to the analysis and

criticism of literary texts. Anandavardhana adopts the

Mīmamsa view that any coherent text must be understood as

directed toward a single, ultimate goal; for him, all the

components of a work poetry (ranging from the individual

phonemes to the overall plot-structure) ought to be selected

and designed so as aid in the conveying of some particular

rasa.

In connection with his incorporation of the central,

teleological principle of Mīmāmsā hermeneutics into

Alamkarasastra, Anandavardhana applies many of the fundamental

interpretive concepts and devices of Mimamsa in the realm of

literary analysis. Beginning with his work, the paired terms

73Cf. Vacaspatimiśra, Tattvabindu (ed. V. Ramaswami Sastri, Annamalai University Sanskrit Series 3, 1936), pp. 131, 153 (where he quotes a statement of Kumarila to this effect), 156-7, 161; cf. also Kunjunni Raja, Indian Theories of Meaning, op. cit., pp. 218-219.

118

primary and subordinate (guna/pradhana, anga/angin),

predicate and subject of predication (vidhi/anuvada), intended

and unintended meanings (vivaksita-/avivaksita- artha) become

part of the basic conceptual vocabulary of Alamkārasastra.

One of the most striking aspects of Anandavardhana's

reformulation of Alamkāraśāstra (albeit one which has drawn

little notice from modern scholars) is the sudden

transformation in the idiom of literary analysis ushered in by

his work. Anandavardhana not only introduces a new aesthetic

theory and a new approach to literary analysis, he does so in

what is, for the Alamkarikas, a new language -- that is, by and

large, the language of Mimamsa hermeneutics.

ANANDAVARDHANA' S DHVANYĀLOKA AND THE TELEOLOGY OF POETIC LANGUAGE

We have seen that growing recognition of the importance

of rasa in poetry becomes problematic for the Alamkärikas of

early ninth century Kashmir. The anomalous status of rasa in

poetic theory reaches its apex in the work of Rudrata (c.

825). Rudrata's Kāvyalamkāra presents two separate schemata

of literary aesthetics, one based on the formal analysis of

poetic language in the tradition of such earlier authors as

Bhämaha, Dandin, and Udbhata, and the other based on the

analysis of rasa (and those aspects of plot and

characterization connected with it) deriving primarily from

the tradition of Natyasastra (dramaturgy). These two modes of

analysis are presented independently, and no attempt

whatsoever is made to relate them to one another. It is in

the Dhvanyaloka of Anandavardhana that these two modes are for

the first time successfully integrated.

Anandavardhana is mentioned in Kalhana's Rajatarangini as

one of the court poets of Avantivarman of Kashmir (r. 855-

883),1 and his literary activity can thus be dated from mid

to late ninth century. His known poetic works are

Arjunacarita (a Sanskrit mahākāvya), Visamabānalila (a Prakrit

poem), and Devisataka (one hundred Sanskrit verses dedicated

1Rajataranqinī 5.34.

119

120

to the goddess Durga), of which only the last is extant.2 He

was also the author Of several philosophical works: the

Tattvāloka, 3 and a commentary on Dharmottara's

Pramanaviniścayatīka. The Dhvanyaloka is his only known work

in the field of Alamkarasastra. The Dhvanyaloka consists of

a series of mnemonic verses (kārikās) with a running prose

commentary (vrtti) .4

Ānandavardhana integrates formalist and rasa-based modes

of literary analysis by viewing each literary work as a

functionally unified whole directed toward a single goal, the

conveying of a particular rasa to the readers or hearers of a

poem. Rather than attempt to treat rasa within the framework

2Published in Kavyamalagucchaka 9, Nirnayasagar Press, 1893. Interestingly, the work is a citra-kāvya ("picture poem", one which depends on complex verbal tricks for its effect), a genre which Anandavardhana himself dismisses as mere "imitation of poetry" (kāvyānukāra).

3The work is mentioned by Abhinavagupta in his commentary on the Dhvanyaloka, but little is known of its contents.

4There is a long running controversy regarding the authorship of these kārikās. Buhler, Jacobi, De and Kane, among others, contend that the verse portion of the Dhvanyaloka is a self-sufficient work (of unknown authorship), and that Anandavardhana is the author of only the prose commentary. distinction The claim is based on the terminological drawn by Abhinavagupta between karika-kara ("author of the kārikā") and vrtti-kāra ("author of the commentary") in his own commentary on the Dhvanyaloka (despite the fact that Abhinavagupta himself cites both kārika and vrtti as Anandavardhana's work in his commentary on the Natyasāstra). A detailed discussion of the arguments surrounding this issue is beyond the scope of the present work. Krishnamoorthy (Dhvanyaloka and its Critics, Kavyalaya Publishers, 1968, pp. 46-95) provides the best available survey of the controversy, and capably for Ānandavardhana's authorship of both kārika and vrtti. argues

121

of alamkära-analysis, as earlier theorists had attempted to

do, he sees it as the outcome of a previously unrecognized

process of suggestion (dhvani). Although it plays a pivotal

role in all great literature, suggested meaning has, according

to Anandavardhana, been entirely overlooked by previous

theorists. He proceeds to demonstrate not only that this

suggested meaning is pervasively present in the classics of

Sanskrit and Prakrit literature, but that the entire array of

formal elements meticulously defined and described by his

predecessors becomes aesthetically effective only insofar as

it aids in the conveying of this suggested meaning.

By establishing suggested meaning generally, and rasa in

particular, as the ultimate goal of all aesthetically

effective literature, Anandavardhana departs radically from

pre-existing modes of literary analysis. Prior Ālamkārikas

approached the analysis of literary language by attempting to

isolate those discrete elements which give beauty to

particular verses or expressions, and to describe and

categorize these elements as exhaustively as possible.

Anandavardhana's teleological model of literary aesthetics

necessitates a complete re-configuration of the descriptive

categories and analytic methods of his predecessors. If

poetry is geared solely toward the communication of rasa, then

the discrete, isolable elements of poetic language described

by earlier Alamkarikas are to be employed not for any

intrinsic beauty they possess, but only for their ability to

122

contribute toward this overriding goal.

In this chapter I will examine Anandavardhana's departure

from earlier modes of analysis and his attempt to revalue

existing poetic categories by integrating them into a

unifocal, rasa-centered model of literary aesthetics. His

arguments for the existence of a realm of suggested meaning

unrecognized by earlier critics, and for the treatment of rasa

as a type of suggested meaning, will be briefly surveyed. I

will then discuss his application of the Mimamsakas'

teleological approach to text-analysis in the realm of poetry,

his establishment of the communication of a single rasa as the

goal which determines the unity and aesthetic coherence of a

particular literary work, and the mechanisms by which the many

rasas existing in such a work are subordinated to a single

overriding rasa. We will consider Anandavardhana's use of

hierarchical methods of text-analysis to differentiate the

various types of suggested meaning from the alamkāras (figures

of speech) described by previous theorists. The subordination

of alamkaras and other recognized features of literary

language to the communication of rasa will then be considered

in detail.

Proof of the Existence of Dhvani ("Suggestion") in Poetry

Ānandavardhana's first order of business, as he commences

the Dhvanyaloka, is to establish that the poetic phenomena he

analyzes and shows to be central in both the production and

appreciation of great poetry are totally distinct from the

123

features of poetic language treated of by his predecessors in

the field of Alamkārasāstra. A radical departure is made from

the field as it has so far been known, and it is essential

that Anandavardhana demonstrate that the categories and

methods of previous Alamkärikas are inadequate to the kind of

analysis pursued here. He argues that the most crucial

feature of great literature is dhvani, the suggestion or

revelation of some meaning which is not directly stated, and

that this mode of expression, despite its centrality, has been

entirely overlooked by earlier poetic theorists.

In the vrtti on the first verse of the Dhvanyaloka,

Anandavardhana examines three possible arguments against the

existence of dhvani, all of them designed to uphold the

adequacy of earlier aesthetic categories. First, it may be

argued that poetics has already provided an exhaustive

characterization of the possible features of poetic language.

Poetry consists of words and their meanings. Factors which

lend beauty to these words and meanings (sabdālamkāras and

arthalamkaras, respectively) , have been defined and

illustrated. The arrangement of phonemes and words for

aesthetic effect (vrttis and gunas) and the differences of

style that depend upon them (rītis) have likewise been

thoroughly analyzed. This characterization of the linguistic

features operative in poetry is total, and there is simply no

scope for the introduction of any additional concepts such as

124

"dhvani " 5

The second argument, while closely related to the first,

attempts to preclude the possibility of dhvani on definitional

and sociological grounds. The categories outlined by the

Ālamkārikas are constitutive of poetry; anything which falls

outside their scope would be excluded by definition from the

realm of poetry. "The defining property of poetry is that it

consists of words and meanings which are pleasing to the

hearts of connoisseurs (sahrdayas). This is not possible for

any mode of composition apart from the procedures already

described (by the Ālamkarikas) ."6 The dhvani-theorist may

stipulate that only those who accept his view can truly be

called connoisseurs, but this will not render his view

acceptable to the general community of cultured persons.

The third argument would seem to follow from the first

two. "It is not possible that this dhvani could be anything

new, since anything which does not fall outside the realm of

beauty will be included in the causes of poetic beauty already

described. If, on the other hand, you are merely giving a new

name to one of these recognized causes of beauty, there is no

significance in this."7 The dhvani-theorist is placed on the

horns of a dilemma. Existing poetic categories (alamkāras,

5Dhvanyaloka [ed. Pattabhairama Sastrī, Kashi Sanskrit Series 135, Benares, 1940], pp. 16ff.

6Ibid., p. 22.

7Ibid, pp. 24-25.

125

gunas, etc.), construed broadly, will encompass any feature of

sound or meaning which gives beauty to a poem. To place

dhvani outside these categories is to deprive it altogether of

aesthetic significance. To accept its inclusion within these

categories is to render it superfluous -- no more than a re-

designation of what has already be described. At best, dhvani

can be seen as a heretofore undescribed sub-type within one or

another of these categories. "Even if, given the infinite

variety of modes of expression, there is some minor type that

has not been described by the well-known definers of poetry,

we see no reason for people to dance about, saying 'Dhvani,

dhvani', closing their eyes in the false belief that they are

connoisseurs. Varieties of alamkara by the thousands. have

been and continue to be shown by great people, and we do not

hear of them behaving in this way. "8

These arguments all serve to render the formalist

paradigm of previous Alamkarikas immune to any fundamental

revision. Theoretical innovation and progress are possible,

but only at the level of detail: existing categories may be

more precisely defined, recognized figures of speech may be

analyzed into subdivisions, and so on. Any challenge to the

basic principles and methods of aesthetic analysis, however,

is summarily ruled out. This form of analysis and its object,

poetry itself, are well-known and established forms of

discourse, and anything falling outside their recognized

8Ibid, pp. 25-26.

126

confines will be of no significance for the production or

appreciation of aesthetically effective literature.

Anandavardhana's primary defense against this line of

criticism hinges on a re-examination of poetic practice.

Dhvani, it is claimed, does not emerge from the re-

interpretation of the concepts and categories of earlier

Ālamkārikas, but is concerned with a realm of poetic meaning

which, although deeply ingrained in the works of the greatest

poets, has heretofore remained totally unknown to the

"definers of poetry" (kavyalakşanavidhayinah -- the term by

which Anandavardhana generally refers to his formalist

predecessors) .

[Karikā:] That meaning, honored by connoisseurs, which is established as the soul of poetry , having is regarded as two "understood" types , "expressed" [vacya] and [pratIyamana] ... Of these two, expressed called

meaning, which is well-known, has been analyzed in many ways, by means of sub-divisions -- simile, etc. -- by others. [Vrtti:] By the definers of poetry. [Kārikā: ] Therefore it will not be dealt with extensively here. [Vrtti:] It will only be reviewed insofar as this is useful. [Kārika:] But the "understood" is a distinct element in the language of great poets, which appears separately from the well-known parts [of a poem], like beauty [lāvanya] in women.10

Prior analysts of meaning in poetry have focussed on that

'Anandavardhana's description pratIyamana artha as the "soul" of poetry has been severely of both vācya and

criticized, and created difficulties for his defenders. It is in apparent contradiction with his frequent description of dhvani alone (i.e. pratīyamana artha) as the soul of poetry. In his commentary on the passage Abhinavagupta attempts, through a convoluted interpretation, to show that here also only the pratIyamana artha is taken as the soul -- cf. below, Chapter Nine, pp. 443-444.

10Dhvanyaloka, op. cit., pp. 42ff.

127

which is directly stated -- the realm of dhvani, of suggested

meaning, has not been misunderstood by them, but has rather

remained beyond their notice.1l This point is underlined by

the analogy of female beauty. It is impossible to give a

satisfactory account of personal beauty by even the most

detailed description of physical features. There is always a

certain quality which will elude this sort of analysis.

Similarly, in poetry, the whole is greater than the sum of its

parts. The analysis and description of the formal features of

a poem, however precise, will not provide a full account of

its aesthetic effect.

Anandavardhana sets himself the task of showing that this

suggested meaning is pervasively present in the recognized

classics of Sanskrit and Prakrit literature and, furthermore,

that it is the predominant feature of these works, playing a

central role in their aesthetic effectiveness. In the light

of this task, exemplification takes on a prominence it has not

held in previous Alamkāra works. Earlier Ālamkārikas such as

Bhāmaha (occasionally), Vamana and Udbhata (in the

Bhamahavivarana) quote example verses from existing literature

rather than compose examples of their own, but these citations

are employed as illustration, rather than evidence. Given

1lThis, as we shall see, is something of an overstatement, and will be substantially qualified in Anandavardhana's subsequent analysis. The three varieties of suggested meaning recognized by Anandavardhana (vastu, alamkāra, and rasa) are all dealt with here and there by earlier alamkära-theorists. They are not, however, grouped together and recognized as constituting a separate realm of meaning, as they are here.

128

Anandavardhana's radical departure from existing categories,

the example verses he cites become vitally important for his

argument. The proof of the existence of dhvani, and thus the

overall legitimacy of this inquiry, depends upon them.

Thus, even before providing a definition of dhvani,

Ānandavardhana gives a variety of examples to establish the

existence of the realm of suggested meaning with which his

analysis will be concerned. Ānandavardhana identifies and

exemplifies three possible varieties of suggested meaning:

vastu (narrative element), alamkära (figure of speech), and

rasa (aestheticized emotion). The following Prakrit verse is

given as an example of the first variety:

Wander confidently, Monk; that dog [which frightened you] has been killed today by the lion who dwells in the creeper-bower on the banks of the GodavarI. 12

Here, a woman wishing to meet her lover in secret, fearing

that they will be interrupted, addresses a wandering

mendicant. Although the literal meaning of the sentence is a

command that the monk should wander, its actual intent is to

magnify his fear and prevent him from wandering near her

trysting place. The actual intent of her statement is thus

the opposite of its explicit intent. Therefore, the expressed

and suggested senses of the verse are clearly distinct.

Figures of speech can likewise be understood from a verse

12Dhvanyaloka, op. cit., p.52. bhama dhammia visattho so sunao ajja mario dena/ golānaikacchakudangavāsinā dariasīhena// (Hala, SattasaI 2.75)

129

without being directly stated:

"Having already obtained Śri, why would he again produce in me the pain of churning? I cannot imagine for him, his mind free of lassitude, his former sleep. Why should he build a bridge again, being attended by the lords of all the islands?" The ocean, when you approach, appears to tremble, as if entertaining such doubts.13

The churning of the ocean from which Śri (the goddess who

personifies royal prosperity) emerges; sleeping afloat on the

ocean (during the time of pralaya, when the worlds are

dissolved); the construction of a bridge over the ocean to the Lanka -- are all well known feats of the god Visnu. From the

ocean's imagined doubts we understand that he mistakes the

king eulogized in the verse for Visnu, and from this we

understand a metaphorical identification (rupaka) of the king

with the god. As Visnu is not even mentioned explicitly, the

rūpaka cannot be said to be directly expressed.

The third and most important variety of suggested

meaning, rasa, differs from the first two types in that it can

never be directly expressed. In vastu- and alamkara-dhvani,

a narrative element or verbal figure which could have been

13Ibid., p. 261. Prāptaśrīr esa kasmāt punarapi mayi tam manthakhedam vidadhyā- nnidrām apy asya purvām analasamanaso naiva sambhāvayāmi/ setum badhnāti bhūyah kimiti ca sakaladvīpanāthānuyāta- stvayy ayāte vitarkān iti dadhata ivābhāti kampah payodheh// Anandavardhana does not cite an example of suggested alamkāra when proving the existence of dhvani in the first chapter, but refers the reader to a later portion of the work ("agre darsayisyate" p. referring to the extended discussion of alamkära-dhvani in chapter two (2.20-30). The 78),

verse quoted above is the first example given of alamkära- dhvani based on the power of meaning (arthasaktimula), and becomes a stock example of suggested alamkāra for later authors.

130

stated directly is expressed only by implication. Rasa,

however, constitutes a realm of "meaning" which is necessarily

suggested. For example:

Siva, his composure somewhat disturbed, like the ocean when the moon begins to rise, cast his eyes on the face of Uma [ParvatI], whose lower lip was [red and round] like a bimba-fruit.14

Here Siva's growing love for ParvatI and the srngara-rasa

which depends upon it are not explicitly mentioned. According

to the principles of dramaturgy, each rasa is conveyed by a

particular set of aesthetic factors (vibhavas, anubhavas,

vyabhicaribhavas and sthayibhava15) . In the above example,

the presence of ParvatI and the beauty of her lip are

vibhavas, Siva's gazing at her face is an anubhava, and his

loss of composure a vyabhicāribhāva, all of which, in

combination, serve to convey śrngara-rasa.

In claiming that rasa is necessarily suggested,

Anandavardhana challenges the view of Udbhata (the most

prominent of Anandavardhana's predecessors in Kashmiri

14 Tbid., p. 249. haras tu kimcitparivrttadhairyas candrodayārambha ivāmburāsih/ umāmukhe bimbaphaladharosthe vyapārayām asa vilocanāni// (Kālidāsa, Kumārasambhava 3.67)

15Vibhävas are those factors which cause an emotion in a character, either serving as the object of this emotion (in the case of love, one's beloved), or heightening it (a moonlit night, a secluded garden, etc.). Anubhävas are the visible symptoms of such an emotion. Vyabhicaribhavas are those transient emotional states which accompany a more abiding emotion (in the case of love: embarrassment, etc.). The sthayibhava is the basic emotional anxiety, expectancy,

state of the character on which the rasa is based (e.g. desire in the case of srīgara ["love"] or grief in the case of karuna ["pity"] ) .

131

Alamkārasāstra), according to whom a rasa may be conveyed in

poetry by the mention of its own name (svasabda), as well as

by the aesthetic factors appropriate to that rasa.16 It is

not invariably the case, argues Anandavardhana, that a verse

in which the name of a rasa or of the emotion it is based upon

will convey that rasa.17 Where such words are used in a

verse conveying rasa, they are invariably found to be

accompanied by the factors appropriate to that rasa. Even

without such explicit mention, the rasa will be communicated

by the same factors. It follows therefore that it is these

factors, rather than the designation of the rasa by name, that

convey this rasa, even where such designation is present.

The concept of rasa originated in the realm of dramatic

analysis, where it is seen to be conveyed not merely by the

text of a play, but by the full range of performance elements

(gesture, costume, etc.). Because Anandavardhana's concern

is with poetry, which depends on words alone for its aesthetic

effect, his analysis problematizes for the first time the

nature of rasa as a purely linguistic product. "Meaning" is

that understanding which follows from a particular set of

words. If, upon hearing or reading a passage of poetry, we

become aware of a particular rasa, this must be considered a

part of the meaning of the text, albeit one which can never be

stated explicitly.

16Kāvyālamkarasārasamgraha 4.3, ed. N.D. Banhatti, p. 52.

17Dhvanyaloka, op. cit., pp. 81ff.

132

Anandavardhana sets out to show that this rasa is

pervasively present in, and is in fact the essential

characteristic of, great poetry. He traces the centrality of

rasa to the very origin of poetry itself.

This very [suggested] meaning is the essence of poetry, although it [poetry] may be beautified through the various arrangement of words and meanings of many sorts. And thus the grief of the first poet, Valmiki, arising from the cry of a heron [krauñca] distraught through separation from its dead mate, was transformed into verse. For grief [soka] is the emotion which serves as a basis for karuna rasa [pathos]. Although other varieties of understood meaning [besides rasa] are shown [elsewhere], they are indicated here by means of rasa and bhava alone, because these are most important.18

Poetry is thus born from emotional experience. Both of the

Sanskrit epics are seen by Anandavardhana as primarily

intended to convey a single rasa.

In the Rāmayana karuna rasa has been mentioned by the Ādikavi [the '"first poet" Valmīki] himself, saying "grief became verse" [sokah ślokatvam ägatah]. And this [karuna rasa] has been carried to its fulfillment by him, constructing his work so as to end with [Rama's] final separation from Sita. Likewise in the Mahabharata, which bears the appearance of both a sastra and a kāvya, the purusartha ["human aim"] liberation and santa rasa [quiescence] have been shown as the primary things intended to be communicated by the great sage [Vyāsa], by composing a conclusion which produces despair through the distasteful end of the Pandavas and the Vrsnis [Krsna's clan], and [thereby] showing the production of vairagya [dispassion, the emotion on which santa rasa is based] to be the primary intent of his composition.19

In order to establish rasa as a variety of meaning, it is

necessary to demonstrate that it exists as something distinct

from the expressed meaning which gives rise to it. Although

18Ibid., pp. 87-90.

19Ibid., pp. 529-30.

133

the precise mechanism by which rasa is conveyed to an audience

is a major focus of dispute in the tradition of Natyasastra,

it is universally agreed that the rasa experience depends on

the presentation (visually or verbally) Of vibhavas,

anubhāvas, and vyabhicāribhāvas. 20

Some would argue that rasa is not itself a variety of

meaning, but is a property of expressed meanings which convey

the proper vibhavas, etc. "The expressed meaning appears as

consisting of rasa, etc.21, not as something separate along

with rasa, etc. #22 If this were the case, Ānandavardhana

retorts, then the rasa conveyed by a verse would be apparent

to all who understand its expressed meaning. He resorts to a

visual analogy: a white body appears as consisting of

whiteness. One cannot see the body without seeing its

whiteness. However, in poetry, this is not the case. The

appreciation of rasa requires an aesthetic sensitivity

(sahrdayatva) not common to all people, and may not be

possible even for one who understands the literal meaning of

a verse. The opponent counters with a further analogy:

genuineness is a property of gemstones, but one that is

20In accord with Bharata's dictum: vibhāvānubhāva- vyabhicarisamyogad rasanispattih (Gaekwad Oriental Series 36, Vol. I, p. 272) .

21The "etc." is meant to include bhava, rasabhasa and bhavabhasa, rasaprasama and bhavaprasama, those rasa-like or rasa-based features employed by Udbhata to explain the alamkāras preyasvat (Kavyalamkarasarasamgraha 4.2), ūrjasvin (4.5), and samahita (4.7) -- ct. above, Chapter Two, pp. 52-58.

22Ibid., p. 402.

134

apparent only to specially trained perceivers. Likewise, rasa

may be a property of the expressed meaning of a poem which is

apparent only to the cultivated listener.

In the opponent's example, says Anandavardhana, the

genuineness of the gem does not appear as anything apart from

its nature. This analogy does not hold for rasa, which is

apprehended as something separate from the vibhāvas,

anubhavas, and vyabhicaribhavas which give rise to it.

For no one believes that the vibhavas, anubhavas, and vyabhicaribhavas themselves are rasas Therefore awareness of rasa, etc. is invariably connected with [but not identical with] awareness of the vibhavas, etc. Because these two awarenesses are established as effect and cause, sequence must necessarily exist between them, but this is not apparent, because of its brevity. Thus it has been said [in 2.3] that rasas, etc. are suggested without apparent sequence.23

It is this rapid succession between the expressed meaning and

the rasa derived from it which leads to the misapprehension.

that there is no distinction between them. The separate

existence of rasa is confirmed, however, by the fact that it

may be conveyed even by wordless songs or gestures, where

there is no expressed meaning whatsoever.24

In the case of other types of dhvani, such as the example

of vastu-dhvani quoted above ("Wander confidently ... " ), the

23Ibid., p. 404. There is no reason to suppose, as Ingalls (The Dhvanyaloka of Anandavardhana, p. 540) and the editors of the Kashi text do, that the words "ity alaksyakrama eva santo vyangyā rasādayah" are a citation from some earlier work; they appear to be nothing more than a parapharase of Anandavardhana's own earlier remarks.

24Ibid., p. 417.

135

suggested meaning contradicts the literal sense of the verse,

and the succession between the two is easily apparent. As

Abhinavagupta rightly points out, it was for precisely this

reason that this type of dhvani was considered first, as the

distinctness of the two meanings cannot be questioned in such

an instance.25 In the case of rasa-dhvani, however, the

second meaning does not displace the first, and hence the

distinction between the two may not be immediately noticeable.

The case for this unnoticed succession is bolstered by analogy

with more commonplace modes of verbal comprehension: "Just as

word meaning, conveying a sentence meaning by its own

capacity, is not separately perceived when it has fulfilled

its function ... Likewise, this suggested meaning appears

immediately in the truly-seeing [tattvärthadarsini] minds of

the aesthetically sensitive, who turn their attention from the

directly expressed meaning. #26 Communication takes place

through sentence meaning, and the meanings of individual words

are merely a means through which such meaning is conveyed.

When the sentence is understood, they have no further role to

perform, and are no longer perceived as distinct from their

product. Likewise, the expressed meaning which conveys rasa

is only a means to this end. The minds of aesthetically

competent listeners (or readers) will proceed rapidly to the

appreciation of the suggested rasa, and the expressed meanings

25Ibid., p. 53.

26Ibid., pp. 101-2.

136

which convey this rasa will no longer be apparent as something

distinct.

As we have already seen, this notion of aesthetic

competence plays an important role in Anandavardhana's

understanding of poetry. Previous theorists devoted

considerable attention to the qualifications required for the

composition of poetry, but Anandavardhana is the first to

seriously examine the requirements for the audience of poetry.

His term for the aesthetically competent audience member is

sahrdaya (literally, "having the same heart" -- i.e. capable of

sharing the aesthetic experience crafted by the poet). The

word is employed by Vamana27 and Udbhata28 in the general

sense of a cultured or sensitive listener, and was therefore

in current use well before Anandavardhana's time. However,

the existence of a select group of qualified listeners and the

nature of these qualifications assume an unprecedented

importance in the Dhvanyaloka.

One of the proofs given for a separate realm of suggested

meaning in poetry is the existence of a select group capable

of appreciating it fully. If poetry depended only on direct

expression for its effect, then anyone sufficiently acquainted

with the rules of language and the meanings of words would be

capable of understanding poetry. But, mere study of grammar

27cf. Kavyalamkārasūtravrtti, Kashi Sanskrit Series 209, fourth ed., 1989, p. 25

28cf. Gnoli ed., Udbhata's Commentary on Bhamaha, Istituto Italian per il Media ed Estremo Oriente, 1962, pp. 20, 39.

137

and dictionaries is not adequate to provide a student with the

capacity to appreciate poetry.29 If it were, the entire

enterprise of Alamkarasastra would clearly be unnecessary.

The fact that such training is inadequate makes it clear that

there must be something to be known from poetry other than

words and the meanings they directly express, and this

something is nothing other than suggested meaning.

There is a certain circularity involved in this notion:

only those capable of understanding suggested meaning are

admitted as an aesthetically competent audience for poetry,

while the existence of such a community is cited as evidence

for the existence of suggested meaning. Ānandavardhana is

alive to the possibility that this attempt to delineate a

distinct class of sahrdayas will be seen as arbitrary. The

second of the three arguments against the existence of dhvani

discussed above makes just such a charge. In attempting to

show that this group of sahrdayas is more than an arbitrarily

stipulated subset of the community of cultured persons,

Anandavardhana describes and attacks a rival conception of

both dhvani and sahrdaya.

"The term dhvani is applied to that poem in which an indefinable beauty of certain words and meanings appears, understood in dependence on some special quality of the perceiver, like the genuineness of certain gems. " -- This definition of dhvani which someone has given is incorrect and should not be so stated.30

29Dhvanyaloka, op. cit., p. 93.

30Tbid., pp. 517-18.

153

are applied to the expressed meanings of sentences, then they

must apply equally to any suggested meanings that arise from

these expressed meanings, rasas included. Even if one does

not accept that rasa can be the direct meaning of a verse

(sāksātkāvyartha), one must at least admit that it is

occasioned by this meaning. On this view, rasas occasioned by

the anuvada portions of a verse will be anuvada, those

occasioned by the vidhi portion, vidhi. In the example given

above ("Thrown off, when clinging to their hands ... "), the

benediction ("May the arrow-fire of Siva ... burn away your

sins") forms the vidhi of the verse. The punning comparison

of the fire with a rejected lover and the description of its

burning the women of the Triple City supports this benediction

by emphasizing the power of this fire, so this portion of the

verse, along with the vipralambha-srīgara- and karuna-rasa

suggested by it, are regarded as anuvada.

Contradictory rasas may sometimes subserve one another

through their occurrence in separate characters. The

depiction of fear in a hero's opponents will intensify the

vIra-rasa (heroism) which his deeds convey to the audience.55

Likewise, depicting the pitiable sufferings of a king's

defeated enemies, and of their wives and other dependents,

will heighten the audience's impression of the king's power

and the vira-rasa which depends upon it. For example:

"Why do you laugh? Seeing me again after so long will

55Ibid., pp. 386-7.

154

you not come to me? Pitiless man! Why this addiction to travel? Who has drawn you away?" Speaking thus in their dreams, firmly embracing the necks of their lovers, the women of your enemies, wake, weep loudly, [finding that] the circle of their arms is empty.56 as they

Such verses, elaborately and often sadistically portraying the

sufferings of enemy women after their husbands have been

defeated or slain, constitute a rather substantial genre of

Sanskrit muktakas (single verses which do not form a part of

any larger work). In such cases the karuna-rasa evoked by the

situation of the women will be subordinate to the overall

object of the verse, the praise of the king's skill and power

in the destruction of his enemies.

Even where karuna-rasa is the main meaning of a sentence

(vākyarthībhūta), factors evocative of śrīgāra may be

introduced in such a way as to heighten this primary rasa,

rather than detract from it. For example:

This is that hand that pulled at my belt, that pressed my swelling breast, that touched my navel and thighs and buttocks, that loosened the knot of my garment.57

The verse is from the Striparva of the Mahabharata, in which

the Pandava and Kaurava women mourn the slain heroes and the

funeral rites are performed for them. It is spoken by the

56Ibid., p. 193. kim hāsyena na me prayāsyasi punah prāptaś cirād darśanam keyam niskaruna pravāsarucitā kenāsi durīkrtah/ svapnāntesv iti te vadan priyatamavyāsaktakanthagraho buddhvā roditi riktabāhuvalayas tāram ripustrījanah// (Subhasitaval 2570)

57Ibid., p. 377. ayam sa rasanotkarsī pīnastanavimardanah/ nābhyūrujaghanasparsī nīvīvisramsanah karah// (Mahabharata 11.24.17)

155

wife of Bhurisravas, one of the heroes slain in the Bharata

war, upon finding his severed arm on the battlefield. Her

erotic recollections serve only to intensify her grief at the

death and mutilation of her husband, and thus heighten the

karuna-rasa of the verse and the passage that contains it.

Thus, through careful analysis of a wide variety of

examples, Ānandavardhana endeavors to show that the

subservience of various, sometimes contradictory, rasas to a

single primary rasa is already well established in poetic

practice. He prescribes the centrality of a single rasa as a

crucial feature of a poetic work, but makes every effort to

ground this prescription in a detailed descriptive account of

existing literary works.

Having concluded this intensive and sophisticated

analysis of the interrelation and unifocality of multiple

rasas however, Ānandavardhana makes a significant

qualification of this teleological principle of analysis:58

the factors (vibhavas, etc.) which convey srngara-rasa are,

generally speaking, pleasing, and may be introduced anywhere,

even when they conflict with the main rasa, either to interest

those to be instructed by a poem (vineyas)59, or simply to

58Ibid., pp. 397ff.

59Although it is generally acknowledged that poetry may provide moral instruction to its listeners, this function of poetry receives little Alamkarikas, who are more concerned with poetry's unique and explicit attention from most

definitive function, that of producing pleasure in an audience. Anandavardhana deals with the question nowhere apart from the offhand reference we are now considering.

156

make the poem more beautiful. For example:

True, beautiful women are pleasing; true, wealth is enjoyable; but life is unstable, like the wandering sidelong glance of a drunken woman.60

The main rasa of the verse is santa (quiescence), but the

sidelong glance is suggestive of śrīgara. There is no

apparent functional relationship between the two rasas here,

and by the standards Anandavardhana articulated earlier, the

introduction of elements of śrīgāra would be pronounced

defective, being not only irrelevant but in fact contradictory

to the santa-rasa which is primarily intended by the verse

(given that santa-rasa has as its emotional basis the

cessation of all desires). Anandavardhana accepts, however,

that such comparisons add to, rather than detract from, the

beauty of this and similar verses, and makes a special

dispensation ..

We shall see that this is not the only case in which

Anandavardhana resists the implications of his own model of

aesthetic effectiveness. There are elements of literary and

genre convention which are not explicable in terms of, and

even conflict with, Anandavardhana's general aesthetic

criterion, but which he is nevertheless unwilling to dismiss

Later Kashmiri authors such as Abhinavagupta and Kuntaka discuss the matter in somewhat more detail, but it remains essentially marginal to the central concerns of Kashmiri Alamkāraśāstra.

60 Tbid., p. 399. satyam manorama ramah satyam ramya vibhutayah/ kimtu mattanganapangabhangalolam hi jīvitam/7 (Subhāsitaratnakośa 1608)

157

as irrelevant to the composition of good poetry (cf. below,

Chapter Six, pp. 292-300) .

The Differentiation of Dhvani from Alamkāra

The introduction of the concepts of sub- and

superordination enables Anandavardhana to differentiate the

phenomena he designates by the term dhvani from those dealt

with by earlier Alamkärikas in their efforts to explain the

role of non-literal meanings (including rasa, although this

was not discussed as type of "meaning" before

Anandavardhana). Although these Alamkarikas, in describing

figures of speech which cause us to understand a fact or

another figure not explicitly stated, or which communicate a

particular rasa or rasa-like experience to an audience, were,

without realizing it, dealing with all three varieties of

suggested meaning, their analysis, according to

Anandavardhana, remains steadfastly focussed on the expressed

meaning even when treating such figures, and in the examples

they provide for such figures the suggested meaning is almost

invariably subordinate to the expressed.61

Ānandavardhana argues that alamkaras must, by definition,

be dependent and of only secondary importance. The root-

metaphor of Alamkārasāstra, as its name would suggest, is that

of physical ornamentation. The words and meanings of a poem

are regarded as its "body" (sarira), and the alamkāras are

ornaments which beautify it, as bracelets and other ornaments

61Ibid., pp. 108ff.

158

beautify a body. An ornament must invariably ornament

something, and is thus inherently other-directed. This fact

alone excludes the alamkaras from the realm of dhvani.

Anandavardhana defines dhvani as follows:

That type of poetry in which either a meaning or a word, subordinating itself or its own meaning [respectively], suggests that [non-expressed] meaning is called "dhvani" by wise people. 62

The expression "subordinating itself or its own meaning" is,

as Anandavardhana himself explicitly points out,63 intended

precisely to exclude alamkāras which incorporate some

suggested sense from the realm of dhvani.

Thus the root-metaphor of Alamkarasastra, that of

ornamentation, plays a major role in Anandavardhana's

demonstration of the inadequacy of his predecessors' theories

and defense of his radical departure from them. The

differentiation, analysis, and illustration of the various

alamkaras had been a primary focus of attention for all

previous Alamkārikas, even those (i.e. Vamana and Rudrata) who

had given prominence to other concepts (guna and rasa,

respectively) in their analyses. By claiming that alamkāras

are, necessarily and by definition, of only secondary

importance, Anandavardhana breaks decisively with literary

theory as it has so far existed in India.

In Ānandavardhana's teleological, MImamsa-derived mode of

62Ibid., p. 103.

63Tbid., p. 108.

159

text interpretation, the relative predominance among the

verbal and semantic elements of a verse and the hierarchical

relations between these elements are the primary focus of

analysis. By contrast, earlier Ālamkārikas devote themselves

primarily to the definition and categorization of these

isolable elements without regard to the functional

relationships between them in particular contexts.

Consequently they show no interest in the primacy or

subordination of the non-literal meanings found in the

alamkāras they describe. Ānandavardhana's claim that their

concern is solely with subordinate suggestion (rather than

with genuine cases of dhvani, in which the suggested meaning

predominates) rests on two pillars: the term "alamkāra" itself

(which is taken to imply subordination), and the examples they

provide for their alamkaras, in which, he claims, the

expressed meaning is generally, if not invariably,

predominant.

In order to make such an argument, of course, it is

necessary to articulate some criteria on the basis of which we

can distinguish a predominant meaning from a subordinate one.

Anandavardhana employs two such criteria, neither entirely

unambiguous or unproblematic. The first is what he calls

"being the [overall] meaning of a sentence" (vakyarthibhava or

vakyarthatva). The term is nowhere defined or explained, and

we must examine its application to get some idea of the sense

in which Anandavardhana employs it.

160

The following verse is given as an example of the

alamkāra samāsokti ("condensed expression" -- in which the

explicit subject of a verse is described in terms equally

applicable to a second, unstated, subject) .

The moon, becoming red/passionate, touched the face of night, with its twinkling stars/trembling eyes, so that her [i.e. the night's] cloak of darkness was condensed/bunched up [samasta] and fell away in the east/in front, unnoticed by her, because of her own redness/passion.4

The literal sense of the verse is simply a natural description

of redness of the rising full moon and the reddening of the

surrounding night sky. But, through punning use of the word

raga (redness/passion), the adjectives applied to the moon and

the night are constructed so that they would apply equally

well to a pair of lovers.

Anandavardhana's comment on the verse is as follows: "In

cases such as this the expressed meaning is understood to be

predominant and is only accompanied by a suggested meaning,

since the night and the moon alone -- on which the behavior of

female and male protagonists have been superimposed -- are the

meaning of the sentence. "65 Here the relevant distinction

64Ibid., p. 109. upodharāgena vilolatārakam tathā grhītam sasinā nisāmukham/ yathā samastam timiramsukam taya puro 'pi rāgād galitam na laksitam// The verse, traditionally ascribed to Panini (probably not the famous grammarian), is cited as an example in Udbhata's Bhamahavivarana, in his commentary on Bhamaha's explanation of samāsokti (2.79-80) (cf. Gnoli, op. cit., pp. 34-5).

65Ibid., pp. 109-11.

161

seems to be one of reference. The verse is really about the

night and the moon; the suggestion that they are like a pair

of lovers may add a certain beauty to an otherwise mundane

description, but this does not affect their centrality as the

referents of the verse.

Elsewhere the term vakyarthibhava is employed more

generally in the sense of the being the intent or purport of

a sentence (what the Mimamsakas would call tatparya) . In

commenting on the verse "Ksipto hastavalagnah" ("Thrown off,

when clinging to their hands ... " -- cf. above, p. 151), for

example, Anandavardhana explains that it is the overwhelming

power of Siva that is the väkyartha, rather than the karuna-

rasa arising from the plight of the women of the Triple-city

or the vipralambha-srngära introduced by the comparison of

Śiva's arrow-fire with a rejected lover. Although the verse

is primarily devoted to description of the pitiable plight of

the women, its overall purport (as an invocation of Siva's

blessings) establishes praise of Siva as the vakyartha.

A second criterion applied in determining the relative

predominance of meanings derives from their comparative

beauty. "The intent to express either a literal or a

suggested meaning as primary is dependent on its having a

greater degree of beauty [cārutvotkarsa] ."66 In alamkāras

that convey some suggested meaning, it will, generally

speaking, be the expressed meaning on which the beauty of the

66Tbid., p. 114.

162

verse primarily depends. For example:

Although he is called by his companions, although he has woken and said "All right", although he wants to get up, the traveller doesn't relax his contracted posture.67

This verse is given as an example of the alamkāra visesokti

(in which a complete set of causes fails to produce an

expected effect) .68 There are two sub-varieties of this

alamkara as explained by Udbhata (Kavyalamkarasarasamgraha,

5.5-6). In one variety the reason for the anomaly is

explicitly stated, in the other it must be understood. In

cases of the latter type, such as the verse quoted, there is

thus (in Anandavardhana's terminology) a suggested vastu (fact

or narrative element). In this case, the reason for the

traveller's failure to get up is not given, and must be

surmised by the reader on contextual grounds. However, the

beauty of the verse, according to Anandavardhana, does not

derive from this suggestion, but depends upon the expressed

meaning itself.69 Whatever we may imagine to be the cause

67Ibid., p. 117. āhūto 'pi sahāyair om ity uktvā vimuktanidro 'pi/ gantumana api pathikah sankocam naiva śithilayati// The verse is attributed in several anthologies to one Bharścu (sometimes Bharvu or Bhascu), of whom only a few stray verses are known. Cf. Raghavan, Bhoja's Śrngaraprakasa, Madras, Punarvasu, 1963 [second ed.], pp. 817-18.

68A literal translation of the visesokti is difficult. It could perhaps be rendered as "the statement of term

special (or anomalous) case."

69It is not even clear what this suggested meaning is. Anandavardhana does not explain his own understanding of the Abhinavagupta, commenting on this passage (p. 117) explains that, according to Udbhata, we should understand from verse.

the verse that it is the morning cold that prevents the

163

for the traveller's reluctance to rise, it is the (directly

expressed) description of this reluctance which gives beauty

to the verse.

Anandavardhana is prepared to concede that among the

multitude of alamkäras defined and described by Bhamaha and

others there may possibly be a few genuine examples of dhvani.

In the alamkāra paryāyokta (oblique statement), a fact or

event is hinted at without being explicitly mentioned.

Bhämaha gives the following example:

Either in houses or on the road, we do not eat food which learned brahmans do not eat.70

The words are spoken by Krsna to his cousin and arch-rival

Śisupäla in declining his offer of food, and we are to

understand from this refusal that he knows the food to be

poisoned. The definition of paryayokta is general enough that

traveller from getting up. Abhinavagupta finds this explanation unsatisfying and understands the suggestion of the verse to be that the traveller has been dreaming of his beloved and is thus reluctant to wake. He claims that this is Anandavardhana's view of the verse also, but given his silence, it seems more likely that he accepted Udbhata's interpretation as valid. Although there is no trace of this verse or any apparent comment upon it in Gnoli's reconstructed text, Udbhata's explanation of the verse is presumably drawn from the Bhamahavivarana. The portion of the text where we would expect to find it (on Bhamaha 3.23-24) is lost, and would occur somewhere between fragments 39 and 40 of Gnoli's edition.

70 Bhamaha, Kavyalamkara, ed. and tr. P.V. Naganatha Shastry, Delhi, Motilal Banarsidass, 1970 [reprint], p. 57. grhesv adhvasu va nannam bhunjmahe yad adhītinah/ na bhuñjate dvijah ... (3.9) The verse fragment is, Bhamaha tells us, from a work called Ratnaharana, otherwise unknown.

164

it might include instances in which the suggested meaning is

primary (although Anandavardhana argues that in examples such

as Bhamaha's this is not so). If this is the case,

Anandavardhana contends, paryayokta should be included in

dhvani rather than dhvani, via paryayokta, being subsumed

under the alamkāras.71 Dhvani has a much wider scope than

paryayokta which is, at best, equivalent to a particular

variety of vastu-dhvani. If examples of this or any other

known alamkara are found in which the suggested meaning is

predominant, these are not, properly speaking, alamkāras at

all, and should be reinterpreted as instances of dhvani. Such

cases are rare in any case, and cannot pose any serious threat

to the case for dhvani as a distinct mode of poetry.

The cases we have been considering involve the suggestion

of a fact (vastu). Suggested alamkaras and rasas have also

been dealt with by previous Alamkärikas, and these too must be

carefully differentiated from true instances of dhvani.

Virtually all arthalamkaras which can be directly expressed

may be suggested as well.72 Udbhata and others have shown

many cases in which one alamkara, expressed directly, causes

us to understand another. A great many of the recognized

71Dhvanyaloka, op. cit., pp. 118-19. Pratiharenduraja (c. 950), in his commentary on Udbhata's Kavyalamkarasarasamgraha, will argue that all instances of dhvani cited by Anandavardhana can, in fact, be subsumed under one or another of the alamkaras described by Udbhata, and that all instances of vastu-dhvani can be treated as cases of paryayokta or aprastutaprasamsa (cf. below, Chapter 7, pp. 369ff) .

72Ibid., p. 257ff.

165

alamkāras depend upon an understood simile (upamā) -- rūpaka

("metaphorical identification"), in which the upameya (the

thing compared, e.g. a woman's face) is identified with the

upamana (the standard of comparison, e.g. the full moon) ;

sasamdeha ("doubt"), in which the speaker implies a simile by

pretending to confuse the upameya with the upamāna; dīpaka

("illuminator"), in which syntactic parallelism implies

similitude between the upamāna and upameya. Vamana even

attempted to demonstrate that all arthalamkāras are ultimately

permutations of the simile (upamaprapañca), and hence depend

upon some implied comparison.73 Bhamaha indicates that the

figure atisayokti ("hyperbole") lies at the root of all

alamkaras.74

As before, the crucial issue for Anandavardhana is the

relative primacy. of the expressed and suggested meanings.

Alamkara-dhvani can be said to exist only where the expressed

meaning is merely a vehicle to convey the understood alamkāra,

on which the beauty of the verse primarily depends (as in the

example "Praptasrir esa ... " ["Having already obtained Śri ... "]

quoted above, p. 129). The suggested simile in alamkāras such

as rūpaka and sasamdeha is invariably subordinate to the

expressed meaning. The identification of distinct things with

one another, or the confusion between them, is only

73Vamana, Kavyalamkarasutra, 4.3.1-33 Series 209, pp. 157-187 (Kashi Sanskrit

74Bhamaha, op. cit., p. 45. Quoted by Anandavardhana (p. 467) -- cf. below, Chapter Six, pp. 276-277.

166

comprehensible if based on some sort of similarity between the

them. But this understood simile has no special beauty of its

own, and is merely a necessary prerequisite for the

understanding of the expressed alamkara.

That latent element of hyperbole or exaggeration which

Bhämaha finds in all alamkāras is similarly subordinate.

"That alamkara in which hyperbole [atisayokti] finds a place

through the imaginative capacity [pratibha] of the poet will

attain great beauty; any other will be mere ornament

[alamkāramātra] . "75 The understood hyperbole is thus

necessary for the beauty of the expressed alamkāra, and is

therefore subordinate to it.

Thus, just as in the case of suggested vastus,

Anandavardhana employs the concepts of primacy and

subordination to differentiate cases of suggested alamkära

found here and there in the works of earlier Alamkārikas from

alamkära-dhvani, which he claims to define and describe for

the first time.

The case of alamkära-dhvani is, however, in one important

respect, considerably more problematic than that of vastu-

dhvani. The suggestion of a vastu in a verse invariably

depends on its expressed meaning. Alamkaras differ in that

their suggestion may, accoring to Anandavardhana, be either

dependent on the power of the (expressed) meaning

(arthasaktimūla) or dependent on the power of word

75Dhvanyaloka, op. cit., pp. 467-68.

167

(śabdaśaktimūla). The category of śabdaśaktimula-dhvani

presents special problems and remains an area of serious

controversy (particularly in Mahimabhatta's Vyaktiviveka -- cf.

below, Chapter Ten, pp. 512ff).

The term sabdasaktimula ("rooted in the power of word")

is explained as follows: "Although our understanding in such

a case is concerned with meaning [arthi], it is produced by

the capacity [samarthya] of a word [or words] capable of

connection with two meanings -- thus it is considered to be

rooted in the power of word. "76 The function of polysemy and

double entendre in poetry had already been extensively

explored by previous Alamkärikas, largely under the rubric of

the alamkāra slesa ("pun"), but also in such figures as

yamaka, lātānuprāsa, punaruktavadābhāsa, etc. It is not at

all clear at first glance whether, in the case of a double

entendre, one should regard both meanings as directly

expressed (albeit by the same word or words), or one meaning

(the one appropriate to the context) as expressed and the

other as suggested by the same words. The differentiation

between the figure slesa and sabdasaktimula-dhvani thus

becomes problematic and requires a careful analysis of the

role of double meaning in both contexts.

Anandavardhana begins the discussion of this question by

considering a hypothetical objection which dismisses the

category of sabdasaktimula-dhvani as essentially redundant.

76Ibid., p. 412.

168

"If, where a second meaning appears by the power of a word or

words, this is described as a type of dhvani, then the figure

slesa would lose all scope [visaya] ."77 The new category

overlaps entirely with one already well established, and is

hence superfluous.

Anandavardhana, in response, argues that the two cannot

be considered identical. In the first place, sabdaśaktimūla-

dhvani is restricted to cases in which it is an alamkära which

is suggested. Where the same words produce an understanding

of two vastus -- two matters of fact -- this would be accepted

as a case of slesa, and it cannot therefore be said that this

alamkära would be deprived of all scope.

This argument is, however, insufficient. It is an

essential element of the theory of slesa as it has been

developed by Bhamaha and subsequent authors that a verse

containing a slesa should make us understand not only two

separate meanings, but some relationship between them, which

will take the form of some other alamkara. Bhämaha (3.17)

explains that slesa is of three types, because it indicates

either sahokti ("statement together" -- a figure in which

actions are simultaneously ascribed to two separate things by

a single expression; cf. Bhamaha 3.39ff), upama ("simile"), or

hetu ("cause"). Udbhata (4.9-10), as is often the case,

reduces Bhamaha's observations to a more general principle,

and states that the words employed in ślesa produce an

77Ibid., p. 235.

169

impression (pratibha) of some other (unspecified) alamkära.

Thus it is well established that some further alamkāra

must always be understood in cases of slesa. Anandavardhana,

here as elsewhere, seems reluctant to disturb the recognized

concepts and categories of earlier Alamkarasastra, preferring

to see the phenomena he embraces under the concept of dhvani

as entirely separate from the realm of surface features dealt

with by his predecessors. It is thus essential for him to

show that this well-known aspect of slesa is in some way

insufficient to account for all instances where alamkāras are

understood through double entendre.

It is Anandavardhana's contention that in all instances

of slesa both the double entendre and the alamkära which

depends upon it must, in spite of this dependence, be regarded

as directly expressed. For example:

Her breasts, even without a necklace [hāra], are naturally attractive/necklaced [harin]: for whom would they not produce astonishment?78

Here we understand the figure virodha (" [apparent]

contradiction"), and the understanding of this figure is

clearly dependent on the polysemic word harin. The literal

sense of the word in this case is "attractive", but because

the word also means "having a necklace [hära]" there is an

apparent verbal contradiction: the woman's breasts are both

78Ibid., 236. tasya vina 'pi hārena nisargad eva hārinau/ janayām āsatuh kasya vismayam na payodharau// (Subhasitāvali 1534)

170

with and without a necklace. Our awareness of this virodha

alamkara, however, depends on the word api ("even"). As

Abhinavagupta points out in his comment on the verse, if this

word were removed, there would be no appearance of

contradiction -- the breasts are beautiful, and they have no

necklace. It is only the verbal cue provided by the

concessive particle api that leads us to ascribe a second

meaning of the word harin. In all instances of expressed

ślesa there will an explicit verbal cue of some sort -- what

Anandavardhana calls a nibandhana (a "cause" or "basis") --

which leads us to understand the double meaning.

In true cases of sabdasaktimula-dhvani, however, there

will be no such cue. The presence of a polysemic word or

words will, by itself, produce an awareness of a second

(suggested) meaning. For example:

At that time the great season/Great Time [i.e. Śiva in his aspect of world-destroyer] called summer, whose loud laughter appeared in the form of the white of its blooming jasmine/whose high laughter79 is white like blooming jasmine, came forth, ending/destroying the two months/cosmic age of the season of flowers [spring] .80

The literal sense of the verse can be understood without any

difficulty, and there is no particle or other word which would

prompt us to look for a double meaning. Nevertheless, the

presence of so many attributes which would be applicable to

79An attribute commonly ascribed to Śiva.

80Tbid., p. 241. atrāntare kusumasamayayugam upasamharann ajrmbhata grīsmābhidhānah phullamallikādhavalāttahāso mahākalah. (Bana, Harsacarita [Kane's ed., p. 21])

171

Siva as well as to the summer suggests such a double meaning

to us, on the basis of which we further understand a simile:

the hot season, bringing an end to the flowers of spring, is

like Śiva when he destroys the world.

Anandavardhana explains the process by which we

understand these meanings thus:

In these examples, when a second, non-contextual meaning appears through the power of word, [one thinks] "A sentence should not express unconnected meanings": thus, on the basis of implication [samarthya], one concludes that the contextual and non-contextual meanings are related as upameya [thing compared] and upamana [standard of comparison]. Thus this pun is hinted at by the meaning, not touched on by any word.81

Because a sentence must by definition express a single,

unified meaning, the double entendre, once noticed,

necessitates that we find some connection between the second

meaning and the overall sense of the sentence.82 This

connection will take the form of some alamkāra (typically

metaphor or simile, as above, but other alamkäras may also be

suggested in this way -- cf. Dhvanyaloka, op. cit., pp. 245-7) .

Because our understanding of this alamkära rests entirely on

B1Ibid., 244.

82Although Ānandavardhana here makes use of the Mīmāmsā notion of ekavākyata ("sentential unity" -- cf. above, Chapter Three, 85ff) sabdasaktimula-dhvani, Mahimabhatta (c. 1050 AD) will later pp. 76, to defend his explanation of

use the very same notion to show the complete impossibility of this variety of dhvani; precisely because there is no verbal cue which prompts the search for a double meaning in such expressions, there is nothing to connect their two possible meanings with one another, and hence no way to derive a propositionally unified sense from them -- cf. below, Chapter Ten, pp. 512-515.

172

our need to find coherence in the overall verse-meaning,

rather than upon some specific verbal cue, instances of

sabdasaktimula-dhvani can be clearly differentiated from cases

of expressed ślesa.

Just as the suggested vastus and alamkaras dealt with by

previous Ālamkarikas have been shown (because of their

subordination to the expressed meanings which give rise to

them) to lie outside the scope of dhvani, likewise the rasas

and rasa-based phenomena these Alamkarikas have shown to

operate in figures such as rasavat ("possessing rasa")83 can

be differentiated from true cases of rasa-dhvani on the basis

of their subordinate status. Rasa-dhvani is defined as

follows:

Where the various beautifying factors of word and meaning [sabdālamkāras and arthalamkāras, as well as other poetic features such as gunas] have rasa, etc. as their goal this will be a case of dhvani [dhvaner visayah] 84

Alamkaras such as rasavat are defined contrastively:

But in a poem where, given the predominance of some other sentence meaning, rasa, etc. is subordinate, this rasa, etc. will be an alamkāra -- this is my view.85

This subordinated rasa may occur either by itself, as in

the verse "kim hasyena ... " ("Why do you laugh ?... ", cf. above,

pp. 153-154) or in combination with some other alamkāra, as in

the verse "ksipto hastavalagnah ... " ("Thrown off, when

83Cf. above, Chapter Two, pp. 50ff.

84Dhvanyaloka, op. cit., p. 180.

85Ibid., p. 191.

173

clinging to their hands ... ", cf. above, p. 151), where the

subordinate rasa, vipralambha-srīgāra (love in separation)

appears by means of, and as intimately connected with, the

double entendre through which the same attributes are ascribed

to Śiva's arrow-fire and to the spurned lover with whom it is

compared. In both these cases, the rasas conveyed are

subordinated to the overall purport of the verse in question,

namely the praise of some unnamed king and of Siva,

respectively. Where the rasa itself is the overall meaning or

object of the sentence (vakyarthibhuta), however, it cannot be

described as an alamkära, an ornament. An ornament is, by

definition, that which imparts beauty to something else;

nothing can be said to ornament itself.86 Where the rasa

serves as the main meaning of a verse, any co-occurrent

alamkāras will ornament it. Therefore the scope of rasa-

dhvani is clearly demarcated from that of alamkāras such as

rasavat.

The analysis of the subordinate/superordinate

relationship (angangibhäva) between the parts of a verse or a

composition thus becomes the basis for differentiating all

three varieties of dhvani from related phenomena explored in

the work of earlier Ālamkārikas. This reliance on the binary

opposition of primacy and subordination to distinguish dhvani

often produces a doubling of prior categories and subdivisions

of alamkāra within the analysis of dhvani. For instance, the

86Ibid., p. 197.

174

various modes of blending (samkara) and juxtaposition

(samsrsti) by which multiple alamkaras combine in a single

verse (as defined by Udbhata, Kavyalamkarasarasamgraha 5.11-13

and 6.5)87 are applied without alteration in Anandavardhana's

analysis of the combination of multiple varieties of dhvani

(Dhvanyaloka, 3.43) 88 Ānandavardhana's subdivision of rasa

and rasa-like phenomena into rasa, bhāva, rasābhāsa, and rasa-

praśānti recapitulates the categories developed by Udbhata

(Kavyalamkarasarasamqraha 4.1-5 and 4.7) to explain the

alamkaras rasavat, preyas, ūrjasvin, and samāhita

(respectively) and their rootedness in rasa.89

The Subordination of Alamkära to Rasa-dhvani

This careful demarcation of the realm of alamkāra from

that of dhvani depends on the necessary subordination of any

alamkara to some alamkārya, some thing-to-be-ornamented,

ideally rasa. Thus, the primacy of a single rasa in any

composition can serve as a criterion for the evaluation of

particular alamkāras in particular settings, just as it can

serve for the evaluation of the interrelationship of rasas

with one another.

The foremost requirement for any alamkara, according to

Ānandavardhana, is that it should not obstruct our

87Kāvyālamkārasārasamgraha, op. cit., pp. 67-71, 78-9.

88Dhvanyaloka, op. cit., pp. 501-16.

89Kavyālamkārasārasamqraha, op. cit., pp. 50-56 -- cf. above, Chapter Two, pp. 52-58.

175

appreciation of the rasa of a given verse or composition.

Alamkäras, even if they are well constructed and beautiful in

and of themselves, may easily serve as a distraction and thus

impair the audience's appreciation of the primary rasa. This

danger is especially acute in the case of śrīgära-rasa, which

is said to be more "tender" (sukumara) than other rasas.

For this [srīgära], because it consists of the full development of desire, and because desire can be interrupted even by some minor cause, is the tenderest of all rasas, and will not bear even for a moment the introduction of any contradictory factor.90

In accordance with this view Anandavardhana severely

restricts the application of the sabdalamkaras which have

received such a great deal of attention from earlier authors

(especially Dandin and Rudrata). The persistent use of a

single type of alliteration (anuprasa), for instance, should

be avoided in cases where srīgara is the predominant rasa

(although it is acceptable where srīgara is subordinate) .91

Furthermore, alliteration, like other features of word and

meaning, will be approved or rejected in a particular context

on the basis of its appropriateness (aucitya) to the dominant

rasa of the passage in question. Repetition of those phonemes

or groups of phonemes described as "harsh" (parusa -- cf.

Udbhata, Kavyalamkarasarasamgraha, 1.492) will be an obstacle

in passages meant to convey srīgara, but the same sort of

9Dhvanyaloka, op. cit., pp. 396-7.

91Ibid., p. 218-19.

2Kāvyālamkārasarasamgraha, op. cit., p. 5.

176

alliteration will be desirable when rasas such as bībhatsā

(disgust) are meant to be conveyed.93

Yamaka (the repetition of a group of phonemes so as to

give two separate meanings), difficult verbal puns (duskara-

sabda-slesa), and other such complex verbal figures should be

entirely avoided even by a skilled poet when śrngara is

predominant. The reasoning for such specific prohibitions is

explained by resort to a more general principle:

Only an alamkara achieved without a separate effort [aprthag-yatna-nirvartya], whose composition can be effected through [the poet's] fixation on rasa, is acceptable in cases of dhvani.94

Difficult sabdālamkāras such as yamaka demand great

effort from both poet and audience, and consequently do not

allow one to appreciate the verse through a single cognitive

effort. We must first "decode" the double meanings and other

verbal puzzles involved in such figures, and can only

subsequently savor the rasa conveyed by the verses containing

them. In contrast, figures based on meaning (arthalamkāras),

even complex ones, can emerge naturally in the work of a poet

intent on rasa.

This [distinction] is appropriate, since rasas are introduced by particular expressed meanings and the words that convey them. And alamkāras such as rūpaka [i.e.

these arthalamkäras] are the expressed meanings which reveal

[bahiranga] to the manifestation of rasa. [rasas]. Therefore these are not extrinsic But in the case of yamakas and other difficult sabdālamkāras this

"Dhvanyaloka, op. cit., p. 303.

94Ibid., p. 219.

177

extrinsicality is established.95

Even where verses containing yamaka are found to evoke rasa,

this rasa will invariably be subordinate, since the figure

itself will demand the greater share of the listener' s

attention

While all arthālamkāras may be employed in verses or

passages primarily intended to evoke rasa, they too must

conform to certain restrictions if they are to contribute

effectively to this rasa, rather than constitute an obstacle

(rasa-vighna). Firstly, the alamkāra must be intended as a

subordinate element, rather than as the primary object of the

verse. For example:

Again and again you touch her trembling glance as it shifts to the corner of her eye; you sound sweetly, hovering near her ear, as if telling something in secret; although she strikes at you with her hands, you taste her lip, the essence of pleasure: we," because we seek the truth, have failed, but you, O bee, are successful.97

The verse is spoken by King Dusyanta, the hero of Kalidasa's

95Ibid, p. 222.

96The speaker, Dusyanta, although referring to himself alone, uses the plural; this may be simply an instance of "the royal we" -- not uncommon in Sanskrit, and not restricted to royalty -- but Kuntaka feels that the use of the plural here emphasizes Dusyanta's exclusion from imtimacy with Śakuntala (in contrast to the intimate proximity of the bee). Cf. VakroktijIvita, ed. Krishnamoorthy, p. 117, and below, Chapter Eight, p. 406.

97Ibid., p. 224. calāpāngam drstim sprśasi bahuso vepathumatīm rahasyākhyayīva svanasi mrdu karnāntikacarah/ karau vyādhunvatyah pibasi ratisarvasvam adharam vayam tattvānvesān madhukara hatās tvam khalu krtī// (Kālidāsa, Abhijñānasākuntala 1.21)

178

Abhijñanasakuntala, as he covertly observes Sakuntala. He has

fallen in love but hesitates to speak to her, being unsure

whether she is a ksatriya or a brahman (and hence not

legitimately accessible as a lover). He watches her fighting

off an approaching bee, which he imagines as a rival for her

affections. Hence the natural description (svabhavokti) 98 of

the bee is subordinate to the vipralambha-srīgara conveyed by

the verse.

The following is cited as a counter-example:

. [Visnu], who, by a forceful command in the form of a blow from his discus, made the love-play of Rahu's wives devoid of the unfettered play of embracing, so that only kissing remained [possible] .99

98Abhinavagupta (pp. 224-5) notes that earlier commentators have denied that the alamkära in this verse is svabhavokti, and have interpreted Anandavardhana's comment on this so as to reflect this view. The compound bhramarasvabhavokti ("natural description of the bee") is re- analyzed as a bahuvrihi: bhramarasvabhave uktir yasya (" [an alamkäral to which there belongs a statement regarding the nature of the bee"). This alamkara is, in their view, rūpakavyatireka ("metaphor combined with contrast"); the bee is metaphorically identified with a rival lover, and the fourth pada contrasts him (as successful in love) with Dusyanta (as a failure). This interpretation of the compound is, however, extremely strained, and Anandavardhana's comment can without serious difficulty be read in its more natural sense. The bee's actions are described literally. The metaphorical identification of the bee with a lover is not explicitly stated, and thus the rūpaka (and the vyatireka which depends upon it) must be regarded as suggested rather than expressed. The expressed alamkara of the verse would thus be svabhāvokti. .

99Ibid., p. 225. cakrābhighātaprasabhājñayaiva cakāra yo rāhuvadhūjanasya/ ālinganoddāmavilāsavandhyam ratotsavam cumbanamātraśesam//

Ruyyaka, in his commentary on the Vyaktiviveka (Kashi Sanskrit Series 121, p. 315), identifies the source of the verse as the (lost) Hayagrivavadha of Bhartrmentha. He gives the correct

179

Here the alamkāra is paryāyokta ("oblique statement"). When

the asura Rahu attempted to swallow the elixir of immortality

(after the gods and asuras had churned it from the ocean),

Visnu decapitated him with his discus; because he had begun to

swallow the elixir, he survived as a disembodied head. By

referring to an effect (that Rahu's wives can only kiss and

never embrace him) the verse obliquely causes us to understand

the cause (that Visnu decapitated Rahu). Although, via this

alamkāra, the verse does convey rasa (vira-rasa, based on

Visnu's heroic deed), the figure itself draws our attention,

and the beauty of the verse is based primarily upon it, rather

than on this rasa. Thus, the alamkara in this case cannot be

said to be subordinate.

A further condition for the subordination of alamkāra to

rasa is that it must be taken up or abandoned in accordance

with the requirements of this rasa, rather than simply for the

sake of any intrinsic beauty. The taking up of an alamkāra on

the proper occasion is illustrated as follows:

reading of the verse as ekabhighata- rather than cakrabhighata-, and explains that the referent of the pronoun yo ("who, what" [masc.]) is Visnu's discus Sudarsana rather than Visnu himself; on this reading we would translate, " ... [The discus Sudarsana] which, by a forceful command in the form of a single blow ... " Given that Ruyyaka apparently had access to the full text of the work, we have no reason to doubt his correction. It is interesting, however, that the apparent misreading recurs consistently in the many citations of the verse in the works of later Alamkarikas (Pratīharenduraja, Kuntaka, Mahimabhatta), suggesting that they are citing Anandavardhana's citation rather than the original work (Ruyyaka himself quotes the Anandavardhana's verse with

Alamkārasarvasva reading, without comment, in his [Kāvyamala 35, p. 106]) .

180

Having blossoms/longing which are/is unrestrained, having a pale color/complexion, flower/yawn100, showing its own motion by reason of the suddenly beginning to

ceaseless rising of breezes/revealing her own exhaustion by the ceaseless rising of her sighs; by looking at this garden-creeper, twined round a madana-tree/full of love, which is like another woman, I will surely make the queen's face red with anger.101

The verse is spoken by Udayana, the hero of Harsa's Ratnavall.

Although the queen's jealousy is here only imagined, based on

the punning comparison of a creeper with another woman, the

verse foreshadows the queen's real jealousy when Udayana falls

in love with the princess Ratnavali. The slesa and the simile

that depends upon it are thus introduced not simply because

they are beautiful in and of themselves but because their use

at this juncture serves to magnify the rasa developed in the

composition as a whole.

The fullest possible development of rasa may, in some

cases, necessitate that one alamkāra, after it has been

introduced, be abandoned in favor of another. For example:

You are red/passionate because of your new shoots, I because of the valued qualities of bees/arrows come to you, and, shot from the bow of the my beloved;

god of love, to me as well; a blow from the foot of a beautiful woman would delight you102 and likewise me;

100Yawning (jrmbha) longing. is regarded as a symptom sexual

101Ibid., p. 226. uddāmotkalikām vipāndurarucam prārabdhajrmbhām ksanād āyāsam svasanodgamair aviralair ātanvatīm ātmanah/ adyodyanalatām imam samadanam narīm ivanyam dhruvam pasyan kopavipātaladyuti mukham devyāh karisyamy aham// (Harsa, Ratnavali 2.4)

102The asoka tree is believed to burst into bloom when kicked by a young woman.

181

everything is the same for us, O asoka, except that the creator has made me full of grief [saśoka] .103

Aśoka is the name of a particular type of tree, but

etymologically means "without grief"; the speaker, separated

from his lover, contrasts his grieving state with that of the

tree. The first three quarters of the verse contain a series

of puns (slesa) through which we understand a simile (upama)

between the speaker and the tree, but in the fourth this

upama-slesa is abandoned in favor of the alamkāra vyatireka

("contrast") (this vyatireka is likewise based on slesa,

arising from the punning use of the word asoka, but

Anandavardhana does not remark on the point) . The simile-

based slesa conveys vipralambha-srngära by highlighting the

speaker's feelings for his lover, but the abandonment of this

alamkāra and the introduction of a new alamkāra, vyatireka,

intensifies this rasa by emphasizing the wretched state of the

speaker in her absence.

Another important condition for the subordination of an

alamkära to rasa is that the poet should not attempt to

develop it too fully (atinirvahanaisita). In the verse "kopāt

komalalola -... " ("His beloved, having bound him firmly ... ",

103Tbid., p. 227. raktas tvam navapallavair aham api slaghyaih priyāya gunais tvām āyānti śilīmukhāh smaradhanurmuktas tatha mām api/ kāntāpādatalāhatis tava mude tadvan mamāpy avayoh sarvam tulyam asoka kevalam aham dhātrā saśokah krtah//

The verse is taken from Yasovarman's (lost) Ramabhyudaya -- cf. V. Raghavan, Some Old Lost Rama Plays (Annamalainagar: Annamalai University, 1961), pp. 21-22.

182

cf. above, p. 150), the metaphorical identification of the

hero with a prisoner is not carried through to its fullest

possible extent. As Abhinavagupta points out, 104 in addition

to identifying the lover's arm with a fetter, the poet could

have identified the lover with a huntress (vyadhavadhu), the

bed-chamber with a jail or a cage, and so on. To fully

develop the metaphor in this way would spoil the vipralambha-

srīgāra which is the primary object of the verse.

If an alamkära is developed fully, the poet must

carefully ensure that it continues to serve the purpose of

nourishing the main rasa throughout. For example:

I see your limbs in the syama-creepers, your gaze in the tremulous glance of the deer, the brightness of your cheek in the moon, your hair in the peacocks' masses of tail-feathers, the play of your eyebrows in the slender wave of the river; alas, shy one, there is no likeness of you in any one place. 105

Here the figure106 is carried through to its full extent,

through the speaker's comparison/identification of a series of

features of the natural environment with those of his absent

lover. The likeness of these individual features is, however,

used to heighten the vipralambha-srngara of the verse, through

104 Ibid., p. 232.

105Tbid., pp. 232-3. śyāmāsv angam cakitaharinīpreksane drstipātam gandacchayām sasini sikhinām barhabhāresu kesān/ utpaśyāmi pratanusu nadīvīcisu bhrūvilāsān hantaikastham kvacid api na te bhīru sādrsyam asti// (Kālidāsa, Meghaduta 101)

106According to Abhinavagupta, (in which attributes of one thing are ascribed to another), but one utpreksa

could also describe it as a suggested upamā or rūpaka.

183

its culmination in the speaker's lamenting the absence of the

whole to which they belong.

Generally, then, the introduction, development, and

termination of alamkaras should be contingent upon the

requirements of the primary rasa of a verse or a larger work.

Like the ornaments used to decorate the human body, the

ornaments of a poem are to be valued not for their own beauty

but for the potential they have to beautify something else.

Rasa, as the ultimate goal of poetry, is the thing to be

beautified by these ornaments. If poets employ alamkāras for

their own intrinsic beauty, without regard for their ability

to aid in the conveying of rasa, the aesthetic impact of their

works will suffer as a consequence.

Deviation from the specific prescriptions made here for

the use of alamkāras, or from the more general principle of

subordination which underlies them, will produce the flaw of

rasa-bhanga ("breaking of the rasa"). Ānandavardhana claims

that examples of such misapplication of alamkāras can

frequently be found even in the works of great poets, but

declines, out of respect, to cite specific examples. 107

The Subordination of Other Poetic Elements to Rasa-dhvani

Although they are not treated as extensively as

alamkāras, other features of poetry examined by previous

Ālamkārikas are similarly shown to be subordinate to or

dependent upon rasa. Gunas ("qualities") are, for

107Tbid., p. 233.

184

Anandavardhana, first and foremost qualities of rasa. These

qualities can be said to belong to words and meanings only

insofar as these words and meanings communicate particular

rasas.

Like Bhamaha before him, Anandavardhana rejects the list

of ten gunas set forth in the Natyasastra (and accepted by

Dandin and Vamana) and limits the number of gunas to three:

madhurya ("sweetness"), ojas ("strength"), and prasāda

("clarity"). Mādhurya is a quality of srngara-rasa,

especially of vipralambha-srngara, and of karuna-rasa, because

these rasas are especially delightful to the hearts of

connoisseurs. 108 Rasas such as raudra ("fury" ) are

characterized by ojas. Prasada is a quality common to all

rasas, and poetry is said to possess this quality insofar as

it is able to clearly communicate rasa.

It is because rasa can be conveyed by both sound and

sense that a division between qualities of word and meaning

(such as that made by Vamana) can be drawn. For example:

O queen, Bhima will bind up your hair, his hands red with the clumped and densely matted blood of Duryodhana, whose two thighs will have been crushed by a fierce club-blow propelled by his [i.e. Bhima's] whirling arm. 109

The phrases "his hands red with the clumped and densely matted

108Tbid., pp. 205-208.

109Tbid., p. 210. cancadbhujabhramitacandagadabhighata- samcūrnitoruyugalasya suyodhanasya/ styanavabaddhaghanasonitasonapāni- ruttamsisyati kačāms tava devi bhīmah// (Bhattanārāyana, Venīsamhāra 1.23)

185

blood" and "his two thighs crushed by a fierce club blow

propelled by his whirling arm" are each expressed by a single

compound. The use of such lengthy compounds, quite

independently of their meaning, heightens the raudra-rasa

conveyed by the verse; hence the verse would be said to

possess the sabda-guna ojas.

A verse may, on the basis of its meaning, convey raudra,

even in the absence of the syntactic or phonological features

which would normally convey such a rasa:

Anyone among the armies of the Pandavas who, taking pride in the strength of his own arm, bears a weapon; anyone in the Pañcala clan, be he a child, an old man, or one still in the womb; anyone who witnessed this act; anyone who stands in my way while I am moving in battle; blind with rage, I will be the death of such a person, even if he himself is the death of all the world. 110

This verse lacks the long compounds which characterized the

previous verse, but its meaning still clearly conveys raudra-

rasa. So such a verse possesses ojas as an artha-guna.

Hence, gunas can be shown to depend not merely on specific

features of word and meaning but, ultimately, on the rasa

evoked by a verse or composition.

It is only by virtue of this direct dependence on rasa

that gunas can be differentiated from alamkäras. Here again

Anandavardhana employs the basic metaphor of Alamkārasāstra,

110Ibid., p. 211. yo yah śastram bibharti svabhujagurumadah pāndavīnām camūnām yo yah pāñcalagotre sisur adhikavayā garbhasayyam gato vā/ yo yas tatkarmasāksī carati mayi rane yaś ca yaś ca pratīpah krodhāndhas tasya tasya svayam api jagatām antakasyāntako ham// (Bhattanārāyana, Venīsamhāra 3.32)

186

that of body (sarira) and ornamentation (alamkara), as part of

his argument for the necessity and centrality of dhvani in

poetry.

Those [features] which depend on this primary [angin] meaning, i.e. rasa etc., are qualities [gunas], just as heroism and so on [are qualities in a person]; those which depend on subordinate "limbs"], i.e. words elements [anga -- lit.

alamkäras, like bracelets and so on [which ornament the and expressed meanings, are

limbs of a person] .111

As we have seen, alamkaras, although they may ultimately

assist in conveying rasa, function primarily at the level of

direct expression. Any suggested meaning they convey will

necessarily be subordinate (otherwise they would not be

alamkāras at all, but instances of one or another type of

dhvani). They are thus analyzable as features of words and

their expressed meanings. Gunas are properties of rasas

themselves; one may describe the words and meanings which

convey a sweet rasa as "sweet", but this is to speak

metaphorically. If gunas were actually, like alamkāras,

properties of words and their expressed meanings, there would

be no reason to designate or treat them separately from

alamkāras. The distinction between these two categories of

aesthetic phenomena is explicable only on the basis of their

relation to rasa-dhvani.

Anandavardhana shows that other distinctions employed by

previous Alamkärikas are similarly explicable only in terms of

his teleological, rasa-oriented model of poetic effect.

illIbid., p. 204.

187

Dosas, flaws which impair the aesthetic effectiveness of a

poem, are extensively analyzed by Ānandavardhana's

predecessors in the field. For the most part these flaws are

not context-dependent and are to be unconditionally avoided in

composing poetry. Grammatical errors, metrical deviation,

obscure or technical vocabulary, and so on, will always

detract from the effectiveness of a poem. However Bhamaha

describes a sub-group of defects, involving harshness of sound

or sense, which may be acceptable or even desirable in certain

instances. 112 Bhamaha gives examples of the unobjectionable

use of words which would ordinarily be rejected as

phonetically harsh (srutikasta) or semantically objectionable

(śrutidusta).

Because they possess such exceptions, Anandavardhana

designates this group of flaws as anitya-dosas ("variable

defects") . 113 Bhamaha gives no clear explanation for these

exceptions, and no indication is given as to how one is to

discriminate between acceptable and unacceptable uses of such

112Cf. Kāvyālamkāra, 1.47-59. There are four defects dealt with here: śrutidusta ("objectionable to the ear" -- use of obscene, scatological, or otherwise objectionable words), arthadusta ("objectionable in meaning" -- conveying an obscene sense without the use of such words) , kalpanādusta ("objectionable in construction" -- in which the juxtaposition of words reproduces the form of an obscene srutikasta ("harsh to the ear" -- use of words regarded as word), and

phonetically harsh or ugly).

113Dhvanyaloka op. cit., p. 214. While Bhamaha gives no examples of acceptable instances of arthadusta kalpanadusta, Anandavardhana seems to refer to the entire or

group of defects as anitya.

188

words or constructions. For Anandavardhana the explanation is

simple: such phonetically or semantically harsh elements must

be avoided when srngära (or some similarly tender rasa) is to

be conveyed. In verses conveying other rasas, such as

bībhatsa ("disgust") or raudra ("fury"), such usages will not

pose any aesthetic problem, and may even serve to intensify

these rasas. The centrality of rasa provides the necessary

criterion allowing one to determine the acceptability of such

usages in particular contexts. Thus, like the distinction

between guna and alamkara, the distinction between variable

and invariable defects (anitya- and nitya-dosas) can be

satisfactorily accounted for only if the centrality of rasa-

dhvani is accepted.

Like alamkāras and gunas, other aesthetically relevant

features of poetic language dealt with by earlier theorists

are shown to be either dependent on or reducible to rasa-

dhvani. Ritis ("styles"), those particular constellations of

gunas which Vamana regards as the essential factors in

producing poetic beauty, represent for Anandavardhana a

confused and inadequate effort to theorize dhvani

(specifically rasa-dhvani) .

The rītis -- Vaidarbhi, Gaudiya, and Pañcali -- have been introduced by those unable to explain the true nature of poetry -- explained here by the introduction of dhvani -- which appeared to them indistinctly [asphuta]. The true nature of poetry was only indistinctly, slightly apparent to the formulators of the definitions of ritis; here it is defined clearly: no purpose would be served by giving

189

a definition of riti as well. 114

We have already seen that gunas, the constituent elements of

Vamana's ritis, are properties of rasa alone. For a poem to

be endowed with a particular set of gunas (and thus possess a

particular riti) is simply for it to convey a particular rasa

endowed with these qualities. Thus rīti is merely a

roundabout way of describing rasa-dhvani, and is entirely

reducible to it. The concept of riti, once the real nature of

rasa and its central role in poetic enjoyment is understood,

becomes superfluous.

Those aspects of poetry designated by the term vrtti

("method, mode") in the works of earlier theorists are shown

to function in subordination to the requirements of rasa. The

term is applied by Udbhata to the three varieties of

alliteration (parusa, "harsh"; upanāgarika, "sophisticated,

urbane"; and gramya, "rustic") ; 115 Bharata describes four

vrttis (kaisikī, sāttvatī, ārabhatī, and bhāratī), modes of

presentation governing plot and language as well as gesture,

costume, and a variety of other elements of dramatic

performance.116 There is no apparent historical connection

between these two sets of categories, and the designation of

both by the term vrtti seems to be mere coincidence, but

114Dhvanyaloka on 3.46 (op. cit., p. 517) .

115Kavyalamkarasarasamgraha 1.4-7 (op. cit., pp. 4-6) .

116Nātyasastra, 20.1ff, Vol. III, Parimal Sanskrit Series #4, Parimal Publications, 1983, pp. 80ff.

190

Anandavardhana treats them collectively.

Vrtti is said to be a function. The vrttis kaisiki, etc. [described by Bharatal are a function residing in the expressed meaning [of a poem] which is appropriate and in conformity with rasa. Upanagarika, etc. [i.e. Udbhata's vrttis] reside in the words.117

Vrtti is, generally speaking, the arrangement of elements in

a poem in such as a way as to convey a particular rasa;

Udbhata's vrttis do SO phonetically and Bharata' s

semantically, but they are alike in their subordination to and

dependence on the requirements of rasa. The employment of

either sort of vrtti in an improper context (that is, where it

is not appropriate to the dominant rasa) is mentioned by

Anandavardhana as one of the factors which damages the

capacity of a work to convey rasa.118 Like alamkāras, then,

vrttis are to be employed not for any independent, intrinsic

beauty but because of their capacity, in the proper context,

to heighten the particular rasa conveyed by a passage.

The entire range of concepts and categories applied to

literature by earlier practitioners of Alamkārasāstra is thus

integrated in a single, all encompassing model of aesthetic

effect. Rasa is established as the overriding goal of

literature in general, and the predominance of a single rasa

as the standard for the aesthetic coherence of a particular

literary work. The features of composition defined and

described by previous Ālamkärikas are either shown to function

117Dhvanyaloka on 3.33 (op. cit., p. 401).

118Dhvanyaloka on 2.18-19 (op. cit., p. 364) .

191

as subordinate elements in the communication of rasa

(alamkāras and vrttis) or entirely absorbed into the

description of rasa-dhvani and hence rendered superfluous

(guna and rīti). As we have seen, this teleological model of

literary aesthetics and the accompanying hierarchical analysis

of the elements of poetic texts is without precedent in the

area of Alamkarasastra, and has its roots in the Mīmāmsa

technique of Vedic text analysis. Ānandavardhana's

importation of the key elements of Mimamsa interpretive

technique into Alamkāraśāstra leads to an increasingly

detailed application of the terminology and methods of Mimamsa

to poetic analysis in the works of the Alamkärikas of tenth

and eleventh century Kashmir (both followers and opponents of

Anandavardhana's dhvani theory), and these developments will

be examined in subsequent chapters.

Eager to deflect the charge that he is merely re-naming

and re-describing the elements and categories outlined by

earlier Ālamkārikas, Ānandavardhana tries to establish that

dhvani depends on suggestion (vyanjana) a mode of verbal

meaning entirely distinct from the literal and metaphorical

modes recognized and treated by previous language theorists.

In doing so he challenges not only established poetic theory

but the philosophy of language in general. Grammarians,

Mīmāmsakas, and logicians (Naiyāyikas as well as Buddhists

such as Dharmakirti), while they disagree violently on many

aspects of language theory, uniformly accept only literal and

192

metaphorical modes of meaning. Hence Anandavardhana must

defend his theory of poetic language not only against the

attacks of fellow aestheticians, but also against the views of

the major schools of linguistic philosophy. Anandavardhana's

encounter with general linguistic theory, unprecedented in its

depth and intensity, has important consequences for the future

development of Alamkarasastra in Kashmir, and in Chapter Five

we will examine this encounter in detail.

We have also seen that, in the light of Anandavardhana's

claim to have discovered a new variety of poetic meaning, the

citation of example verses in the Dhvanyaloka takes on an

evidential function it has not borne in earlier works on

poetic theory. Anandavardhana must not only illustrate the

varieties of dhvani he defines, he must prove that dhvani is

pervasively present in the recognized classics of Sanskrit

literature, despite the failure of earlier poetic theorists to

notice or describe it. This gives his work an empirical

orientation, geared more toward the description of existing

literature and analysis of its aesthetic effect than that of

previous works of Alamkārasāstra. This empirical approach

opens up new possibilities in the area of literary criticism

(possibilities which will be more fully exploited by Kuntaka

in his Vakroktijīvita), but at the same time creates

theoretical problems for Anandavardhana, as he is forced to

acknowledge that there exists a significant body of respected

poetry in which rasa is not the overriding aesthetic goal.

193

Ānandavardhana is unwilling to deny all literary value to such

works, yet has difficulties explaining them in terms of his

teleological model of poetic language. The tension between

theory and criticism in the Dhvanyaloka and the problems

presented by poetry in which suggested meaning is subordinate

(gunībhuta-vyangya-kavya), or that in which it is altogether

absent (citra-kävya) will be examined in Chapter Six.