Printer Friendly


In connection with a recent work on the Sambandhasamudde[acute{s}]a of the v[bar{a}]kyapad[bar{i}]ya, I consider some major issues concerning Bhartrhari, the tradition he represents, and how a modern scholar might approach the V[bar{a}]kyapad[bar{i}]ya. I discuss theoretical principles which have been set forth as a basis for dealing with Bhartrhari and evidence from the V[bar{a}]kyapad[bar{i}]ya in connection with these principles, chiefly what is referred to as Bhartrhari's perspectivism. I take up in some detail one large issue: the status of the received high language, associated with a group of model speakers called [acute]{s}]ista, who use speech forms characterized as "correct" (s[bar{a}]dhu) linguistic units ([acute{s}]abda) opposed to incorrect linguistic units (as[bar{s}]dhu[acute{s}]abda, apa[acute{s}]abda), that are viewed as corruptions (apabhram[acute{s}]a), with respect to how meanings are understood by users. The principal issue here is: do apa[acute{s}]abdas directly signify meanings for [acute{s}]istas when they communicate with someone using a vernacular, or do these speakers resort to a translation technique such t hat the apa[acute{s}]abda used calls to mind a s[acute{s}]dhu[acute{s}]abda, which then directly signifies a meaning? This topic also involves another important question: how one should consider the Vrtti and other commentaries in relation to what is said in the k[bar{a}]rik[bar{a}] text.

1. Bhartrhari's V[bar{a}]kyapad[bar{i}]ya is, without any doubt the single most important work in the long history of P[bar{a}]ninian grammar, after the works of the three munis: P[bar{a}]nini, K[bar{a}]ty[bar{a}]yana, and Pta[bar{n}]jali. It certainly is the most widely cited early treatise concerning what scholars generally refer to as Indian "philosophy of grammar," not only among subsequent scholars in India in various schools of thought but also among modern scholars. [1] Recent years have witnessed a growing interest in the V[bar{a}]kyapad[bar{i}]ya and Bhartrhari's Mah[bar{a}]bh[bar{a}]syad[bar{i}]pik[bar{a}]. The work under review here, a revised version of Houben's doctoral dissertation, treats principally one section of the third k[bar{a}]nda of the V[bar{a}]kyapad[bar{i}]ya, the Sambandhasamudde[acute{s}]a, [2] where in eighty-eight verses Bhartrhari treats questions concerning relations between linguistic items and meanings.

The subtitle of his work reflects Houben's main concern, to which the central part of the book is devoted: "The Sa[dot{m}]bandha-samudde[acute{s}]a, Translation and Discussion of the K[bar{a}]rik[bar{a}]s" (pp. 137-324), an annotated translation of the Sambandhasamudde[acute{s}]a, followed by an appendix, "Hel[bar{a}]r[bar{a}]ja's Commentary 'Prak[bar{a}]rnaprak[bar{a}][acute{s}]a' on the Sa[dot{m}]bandha-samudde[acute{s}]a" (pp. 325-424). The other two major sections of Houben's book are "Sa[dot{m}]bandha in some early Indian traditions and in Bhartrhari's Philosophy of Language" (pp. 29-79) and "The Third K[bar{a}]nda and the immediate context of the Sambandha-samudde[acute{s}]a" (pp. 81-135). These are preceded by an introduction (pp. 1-27); the book ends with a bibliography (pp. 425-48) and three indices: a general index (pp. 449-53), an index locorum (pp. 453-57), and an index referring to text-critical observations on the Sambandhasamudde[acute{s}]a and Hel[bar{a}]r[bar{a}]ja's commentary (pp. 457-60).

Houben exhibits an impressively wide-ranging erudition together with considerable attention to philological detail and to what other scholars have said. This can be brought home more vividly by noting the contents of this work in greater detail.

In his introduction, Houben takes up Bhartrhari's life and time (pp. 3-10) and the V[bar{a}]kyapad[bar{i}]ya and its interpretation (pp. 11-22), then considers research that has been done on the Sambandhasamudde[acute{s}]a and the notion of sambandha, "relation" (pp. 23-26). The introduction ends with a brief note on the present work (p. 27).

The first major part of this book is divided into four sections. Most of the second section concerning early grammarians is devoted to the Mah[bar{a}]bh[bar{a}]sya discussion on the first part of the v[bar{a}]rttika siddhe [acute{s}]abd[bar{a}]rthasambandhe lokato rthaprayukte [acute{s}]abd[bar{a}]prayoge [acute{s}][bar{a}]strena dharmaniyamah. In the third section, Houben considers views of Mim[bar{a}]ms[bar{a}] (pp. 46-47), vai[acute{s}]esika (pp. 48-53), Buddhism (pp. 53-58), and "other schools: S[bar{a}][dot{m}]khya, Ved[bar{a}]nta" (pp. 58-63) on the notion of "relation." As Houben notes (p. 46), these discussions "[ldots] are mainly based on presently available secondary literature." [3] The last section on sambandha in the V[bar{a}]kyapad[bar{i}]ya consists of four subsections in which the following topics are treated: the assumption that there is a beginningless relation between words and meanings (pp. 64-66), the relation between sounds and signifiers (pp. 66-75), "the intimate relation between [acute{s}]abda 'language', artha 'reality' and j[bar{n}][bar{a}]na 'knowledge'" (pp. 75-77), and "sa[dot{m}]bandha and the primary unit in language" (pp. 77-79). Part two of Houben's work is devoted to a description and discussion of the organization of the V[bar{a}]kyapad[bar{i}]ya's third k[bar{a}]nda.

The third and largest part of Houben's book consists of three sections: "General Remarks" (pp. 139-40), "Analysis of the Sa[dot{m}]bandha-samudde[acute{s}]a" (pp. 141-44), and "Translation and Discussion of the k[bar{a}]rik[bar{a}]s" (pp. 145-324). The translation in turn is divided into two parts, each further subdivided into two parts, covering respectively k[bar{a}]rik[bar{a}]s 1-38, 39-51, 52-59, and 60-88. Houben first gives his translations, then deals with grammatical and semantic issues concerning terms and syntax, and finally discusses the import of k[bar{a}]rik[bar{a}]s--individually and in groups--in the light of what is said elsewhere in the k[bar{a}]rik[bar{a}] text, in the available Vrtti [4] on the first and second k[bar{a}]ndas, and, even more extensively, in light of what Hel[bar{a}]r[bar{a}]ja has to say. This leads to redundancy. In the appendix containing the translation and comments on Hel[bar{a}]r[bar{a}]ja's Prak[bar{i}]rnaprak[bar{a}][acute{s}]a the transliterated text of the eighty-eight k[bar{a}]rik[bar{a}]s appears again, Houben's translations of these k[bar{a}]rik[bar{a}]s is repeated with occasional slight variations to take Hel[bar{a}]r[bar{a}]ja' s interpretation into account, and there is additional discussion of what Hel[bar{a}]r[bar{a}]ja says.

Houben's book affords an opportunity to consider some major issues concerning Bhartrhari, the tradition he represents, and how a modern scholar--from whatever background that scholar be--might approach Bhartrhari's major work. My discussion is organized as follows. First (section 2), I outline the theoretical principles Houben sets forth as the basis for his dealing with Bhartrhari, then (section 3) I consider evidence from the V[bar{a}]kyapad[bar{a}]ya in connection with Houben's principles, chiefly what he refers to as Bhartrhari's perspectivism. I subsequently take up (section 4) one large issue: the status of the received high language, associated with a group of model speakers called [acute{s}]ista, [5] who use speech forms characterized as "correct" (s[bar{a}]dhu) linguistic units ([acute{s}]abda) opposed to incorrect linguistic units (as[bar{a}]dhu[acute{s}]abda, apa[acute{s}]abda), that are viewed as corruptions (apabhra[dot{m}][acute{s}]a), with respect to how meanings are understood by users. The principal issue here is: do apa[acute{s}]abdas directly signify meanings for [acute{s}]istas when t hey communicate with someone using a vernacular, or do these speakers resort to a translation technique such that the apa[acute{s}]abda used calls to mind a s[bar{a}]dhu[acute{s}]abda, which then directly signifies a meaning? This topic also involves another important question: how one should consider the Vrtti and other commentaries in relation to what is said in the k[bar{a}]rik[bar{a}] text. [6] Following this discussion, I take up (section 5) some points concerning Houben's translation. [7]

2. As I said at the outset, I think Houben's book demonstrates considerable learning. It is to be expected, then, that Houben will have his own particular views to propound and devote more attention to the views of some scholars than to those of others. The very breadth of the work and the attention paid to details both in the translations and in the annotations make it difficult to write a general review or to make critical remarks without seeming to be a nitpicker. What is more, in connection with a work such as this, there is the danger of being viewed as a "traditionalist" as opposed to a more open-minded "modern" scholar. Despite these risks, however, Houben's book exhibits qualities--both good and bad--which invite such comments.

2.1 The most important chapter in the introduction is the second, concerning the V[bar{a}]kyapad[bar{a}]ya and its interpretation. Here, Houben expresses his doubts with respect to generally held opinions concerning how one should understand what Bhartrhari intended to set forth and postulates three principles which, he says (p. 16), "[ldots] clarify, in my view, a great deal of Bhartrhari's thought, and particularly of his treatment of the notion of sambandha 'relation'."

Houben accepts (p. 18) that "[ldots] Bhartrhari did have some theoretical preferences [ldots]" At the same time, he expends considerable energy arguing against some scholars who see certain basic ideas maintained and defended in the V[bar{a}]kyapad[bar{i}]ya. Thus, he remarks (p. 15): "According to Peri Sarveswara, the whole of the VP is to be understood on the basis of the first four k[bar{a}]rik[bar{a}]s [ldots] But what would happen if we were to take first other parts of the VP into account and return next to these introductory k[bar{a}]rik[bar{a}]s?" [8] On the same page, Houben later remarks that in the course of his study of the Sambandhasamudde[acute{s}]a he became convinced that "[ldots] attempting to understand Bhartrhari in the framework of "linguistic monism" would always leave some important characteristics of the VP unexplained."

2.2. In the same section of his introduction, Houben postulates the three principles I alluded to earlier. He states them as follows (p. 16): "(a) different perspectives have their own validity; (b) there is a truth or reality behind the overwhelming diversity of words and discourse; (c) there is a permanent relation between linguistic unit and thing-meant."

Principle (a) is adopted as guiding the interpretation of Bhartrhari's work. Houben takes a strong stance in saying that Bhartrhari's approach to issues of language, meaning, and reality may be qualified as "perspectivistic," not simply in that he takes into consideration views proposed by Mim[bar{a}][dot{m}]saka, Vai[acute{s}]esika, and various Buddhistic thinkers but also (p. 16) that "[ldots] the validity of different perspectives is accepted" and that "[ldots] the validity of a single perspective is limited." In Houben's opinion, earlier scholars have failed to appreciate properly Bhartrhari's attitude. Thus, in connection with his first principle, he says (p. 17):

The principle manifests itself first of all in the complex argumentative structure of Bhartrhari's exposition. His awareness of the limits of each single approach to reality explains why the VP has become to a very great extent a sa[dot{m}]graha, an encyclopedia of different existing views on linguistic and philosophical issues. The importance of this encyclopedic approach has not been fully realized by many interpreters and critics of Bhartrhari. In the case of the Sa[dot{m}]bandhasamudde[acute{s}]a we will see that some views which have been widely considered as the view of Bhartrhari on the subject, occupy, in fact, a certain well-defined place in a whole range of different, and within their limits acceptable, angles of approach. Too often, Bhartrhari's statements have been interpreted in a polemic instead of an encyclopedic framework (this point I hope to elaborate at other places, e.g., in Houben, forthcoming, e). [9]

Houben goes on to modify this by noting (p. 18) that Bhartrhari did have some theoretical preferences, although he had a "basically constructive" attitude towards ideas of other thinkers. [10]

Principle (b) is later (p. 19) put in a different manner: "The second principle says that it may be the case that there is a truth or reality behind the overwhelming diversity of words and theoretical discourse." [11] Houben notes that this principle prevents the first one from leading to total skepticism or nihilism. He also remarks, touching more closely on the central topic of his book, that this principle explains "Bhartrhari's arguing for the acceptance of a relation which by definition cannot be expressed as-it-is" in that "both the limits of language and a possible underlying reality are taken into account."

The third principle is one which P[bar{a}]ninian grammarians maintained from the very beginning. Houben goes on (p. 20) to say that this has as a theoretical implication "[ldots]that thought and knowledge of a thing are always intimately, perhaps even inseparably, connected with language."

2.3. In connection with the thesis for which he argues strongly, that Bhartrhari presents other positions in a spirit of equality, Houben also considers, in the last subsection of part 1, an important issue that is the centerpiece of the V[bar{a}]kyapad[bar{i}]ya's second k[bar{a}]nda: whether words (pada) and their meanings (pad[bar{a}]rtha) are to be considered real (sat) constituents of utterances/sentences (v[bar{a}]kya) and utterance meanings (v[bar{a}]ky[bar{a}]rtha) or merely assumed units abstracted formally from indivisible utterances and utterance meanings for the purpose of grammatical description and discussion. The second k[bar{a}]nda of the V[bar{a}]kyapad[bar{a}]ya is concerned with the various ways in which a sentence is defined. Concerning this k[bar{a}]nda, Houben (pp. 78-79) makes three observations that are relevant to his study of the Sambandhasamudde[bar{a}]a. The second of these is as follows (p. 78):

[ldots]in the second K[bar{a}]nda Bhartrhari discusses numerous views in a positive way, in accordance with his 'perspectivism' and his encyclopedic approach to theoretical problems. Much positive attention is therefore also paid to views which accept words as primary units[ldots] [12] One gets the impression that Bhartrhari does not try to establish one view to the exclusion of others, but wants to show different views in contrast with each other. Throughout the second K[bar{a}]nda he is extremely conscious of the value of different viewpoints. That Bhartrhari has still a preference for the sentence as the primary unit is not always very clear in the second K[bar{a}]nda, but it is emphasized in a few strategical places in the third K[bar{a}]nda. There, Bhartrhari refers to the process of apoddh[bar{a}]ra 'abstracting from, analysis' and to the idea of abstracting linguistic or semantic units from more comprehensive ones (3.1.1-2, 3.4.1-2, 3.7.164, 3.13.6, 3.14.144, 3.14.169). Among these places, only 3.1.1 refers to the sentence and the word and smaller linguistic units, in the other [sic] the emphasis is on the sentence meaning and the word meaning and/or word meaning and smaller semantic units.

Houben goes on (pp. 78-79) to make his third point: if the sentence is ultimately considered the primary unit of language, then the Sambandhasamudde[acute{s}]a asks the wrong kind of question. The question asked in this samudde[acute{s}]a is: what sort of relation holds between words and their meanings? "The entire discussion in the Sambandha-samudde[acute{s}]a," Houben argues, "becomes very tentative or hypothetical in this light," and he concludes his third point as follows:

The validity of the discussion is limited by the acceptance of theoretical choices which are not without problems. It may be pointed out, moreover, that in the second K[bar{a}]nda Bhartrhari does not discuss just a single view in which the sentence is primary, but several views. What all views which accept the sentence as primary have in common is that the status of individual words (and corresponding word meanings) is strongly relativized.

Houben ends this section (p. 79) emphasizing that, after all, those who upheld the primacy of constituent words included grammarians, the younger Bhartrhari among them:

If it was so important to Bhartrhari to relativize the status of individual word meanings, one may wonder: who were the thinkers who did attribute a high status to these? Interpreters of the VP have identified these with M[bar{i}]m[bar{a}]msakas. In "Who are Bhartrhari's padadar[acute{s}]ins?" (Houben 1993) it has been argued that the 'upholders of the word' include also grammarians and to some extent the author of the MbhD who may have been the young Bhartrhari.

The tenor of Houben's inquiry which shows through in these passages appears elsewhere, also, as when, in the final chapter of his summary of the third k[bar{a}]nda, he includes the following among nine observations (p. 132):

(2) In the third K[bar{a}]nda, as in the previous ones, there is a tendency to pay positive attention to quite divergent views, and usually there is no absolute commitment to one view to the exclusion of others. The approach to the different problems may be characterized as perspectivistic.

(3) Against the background of this perspectivistic approach, there are still theoretical preferences. It seems possible to locate the preferences evinced in the third K[bar{a}]nda in the context of he discussions in the second K[bar{a}]nda. There, the main problems were: is the primary unit in language the word or the sentence? is the individual word meaning primary or the sentence meaning? The preference evinced there for the primacy of the sentence and sentence meaning implied that individual word meanings, if at all they are accepted, have no independent status. The primacy of the sentence is usually no explicit argument in the discussions in the third K[bar{a}]nda. Only at a few places the reader is reminded of the idea that, in fact, the sentence and the sentence meaning are the primary units. Apart from these places, it may be said that in the third K[bar{a}]nda views in which word meanings are not well-defined individual basic units are generally preferred; there is no strong or absolute commitment to views which would imply word meanings as well-defined individual basic units.

3. The three principles which Houben sets forth are basic to his treatment of the materials in the V[bar{a}]kyapad[bar{i}]ya, and they are put in a manner that would appear to make disagreement difficult. It is nevertheless difficult to see how Houben can maintain some of his claims, at least in the strong versions he formulates, and some of what he says puzzles me.

3.1. Consider Houben's reaction to Peri Sarveswara's appreciation of the introductory k[bar{a}]rik[bar{a}]s (above, 2.1). One may choose to ask what would happen if, instead of starting from the beginning, we studied other parts and came back to the beginning. It is nevertheless reasonable that an author begin his work by presenting his general thesis, and it is also reasonable to let this be a background for one's understanding of what follows.

3.2. The V[bar{a}]kyapad[bar{i}]ya begins with a description of brahman in k[bar{a}]rik[bar{a}]s 1 - 4. Bhartrhari immediately introduces the theme of multiplicity and unity: though brahman is memorialized in the Veda as one, so that there is no separation between it and powers, it appears as though distinct from these powers; and it appears as distinct things by virtue of its several powers. [13] Moreover, the theme of imposition is also brought in at the beginning, where Bhartrhari says [14] that the six modifications--being born and so on [15]--which are the sources of differences in being have their basis in the time-power of brahman, a power on which parts are imposed. [16] Bhartrhari goes on to speak of the means of reaching this brahman, namely the Veda. Once more, he immediately strikes the note of unity versus diversity: although one, the Veda has been memorialized in a form appearing as though it had many "paths." [17]

Subsequently, Bhartrhari outlines in three k[bar{a}]rik[bar{a}]s [18] the topics which are to be covered in his work. Two types of meanings are distinguished: those which have a fixed nature and those which are obtained through extraction of partial meanings from putative complexes. Parallel to these are two kinds of linguistic units: those which are to be explained and those which serve to convey the former. Two kinds of relations are said to link linguistic units with meanings: a relation such that one is the cause and the other an effect and a relation such that a linguistic unit has the property of being naturally apt to make known a meaning. Moreover, such relations can connect linguistic elements and meanings in ways that can have two results: a relation serves as means with respect to merit or merely to the comprehension of meaning; the former holds for correct speech forms, whose use leads to merit, and the latter for incorrect forms in addition to correct ones. There can be no doubt at all that the Vrtti is corre ct when it says that the total content of the work which has been undertaken is summarized in these three verses. [19] There can also be no doubt whatever that the V[bar{a}]kyapad[bar{i}]ya itself distinguishes between linguistic units that are to b explained and those which serve as means to explain them in a grammar--P[bar{a}]nini's Ast[bar{a}]dhy[bar{a}]y[bar{i}]--and between meanings that are fixed and those which are abstracted through analysis. Again, there is the distinction between unity and division, now resulting from abstraction.

No one has denied that Bhartrhari brings in many of points of view throughout the V[bar{a}]kyapad[bar{i}]ya, but one cannot be faulted for considering it appropriate to appreciate the entire work in the light of the introductory k[bar{a}]rik[bar{a}]s. In addition, Bhartrhari makes it clear that he bases himself on a long tradition of usage and grammar. Thus, in VP 1.26, he says not only that the eight topics noted earlier are described in grammar both through indirect indications and direct statements but also that they are now followed according to tradition, to memorialize them. The Paddhati appropriately stresses that Bhartrhari here says he is not doing anything without precedent, only following a tradition in presenting the eight topics at issue, and that he does all this in accordance with tradition, so that there is no invention (utpreks[bar{a}] 'fancy') for the sake of novelty. [20] Bhartrhari goes on to say, in accordance with the Mah[bar{a}]bh[bar{a}]sya, [21] that the correct linguistic units that are a means to merit are established from the tr adition that stems from the [acute{s}]istas. Moreover, no one can render purposeless this settled situation, so that [acute{s}]istas like Panini composed a smrti whose object is the correctness of speech. [22] After stressing again that pure reasoning by inference cannot replace what is established through tradition, Bhartrhari completes the section in question by saying that [acute{s}]istas have undertaken grammatical teaching basing themselves on the teaching that has no author--that is, the Veda--and smrti based on it. [23]

3.3. Houben's third principle as formulated is a restatement of what K[bar{a}]ty[bar{a}]yana and later P[bar{a}]nin[bar{a}]yas have maintained. The theoretical implication Houben draws from this, on the other hand, is not obvious. It is one thing to say that a linguistic item like ghata bears a permanent relation with a referent, a pot; [24] it is another thing to claim that one cannot have knowledge of such an entity without language, that all or most "cognitive episodes" are verbal thoughts at some implicit level. [25] It is well known that for Bhartrhari any cognition of anything in the world is permeated by language. This stance is compatible with accepting a permanent relation between linguistic items and what they signify, but it does not necessarily follow as a consequence of this acceptance.

3.4. I also think some caution is called for in connection with Houben's first principle. As he recognizes, scholars before him have stressed Bhartrhari's dealing with views of different schools of thought in a nonpolemical manner. Bhartrhari does not organize his work as a series of p[bar{u}]rvapaksa to be refuted, thus establishing his siddh[bar{a}]nta. There is no repeated iti cen na, iti siddh[bar{a}]ntah, or iti siddh[bar{a}]ntitam. In this respect, Bhartrhari differs from such scholars as the Paninian grammarians Bhattoj[bar{a}]d[bar{a}]ksita, Kaundabhatta, and N[bar{a}]ge[acute{s}]a; [26] the M[bar{a}]m[bar{a}][dot{m}]sakas Jaimini and V[bar{a}]caspatimi[acute{s}]ra; [27] Naiy[bar{a}]yikas such as Jayantabhatta and Ga[dot{n}]ge[acute{s}]a; [28] and many other defenders of tenets maintained in particular schools. This does not mean, however, that Bhartrhari should be considered not to have held definite views of his own and to have argued--even in the V[bar{a}]kyapad[bar{i}]ya--against other positions. The closest parallel to his way of proceeding is, I think, to be found in the work which serves as Bhartrhari's major source of ide as to be developed, the Mah[bar{a}]bh[bar{a}]sya. Pata[tilde{n}]jali also presents arguments for various points of view--for example, that a verbal base (dh[bar{a}]tu) can be defined as signifying activity (kriy[bar{a}]) and being (bh[bar{a}]va)--and interpretations of given s[bar{a}]tras, without overtly and emphatically presenting a definitive siddh[bar{a}]nta. Yet his very argumentation leads one to see--though not always--his accepted view, since certain alternatives involve such a complex of principles and metarules to be applied that they clearly suffer from what commentators call pratipattigaurava 'prolixity in understanding' and certain others clearly do not harmonize with what is said elsewhere in the Ast[bar{a}]dhy[bar{a}]y[bar{i}], in v[bar{a}]rttikas and in the Bh[bar{a}]sya.

3.5. Although I agree that the V[bar{a}]kyapad[bar{i}]ya text calls for a great deal of study and that for any single student much may indeed remain unexplained, I also think that Houben has applied his first principle in a way that fails to pay appropriate attention to Bhartrhari's defense of certain theoretically important positions. Consider in this context what Houben says about the second k[bar{a}]nda (see 2.3).

In this k[bar{a}]nda, Bhartrhari devotes a large section (VP 2.64-87) to arguments in defense of the word and its meanings as valid. This section ends with a verse saying that all the principles of interpretation dealt with in what has preceded, principles that depend on word-meanings, would not serve to interpret sentences correctly if the individual word did not signify. [29] Thereafter begins a section the first verse of which asserts that there is no contradiction between what has been adduced requiring that one accept as real the individual words of utterances and the position stated earlier, that an utterance is an indivisible unit whose meaning also is indivisible. Even if the utterance meaning is indivisible, it is subject to having partial meanings extracted due to differences in expressive capacity one sees in different utterances. [30] One example will serve to illustrate. If A says to B (1) van[bar{a}]t pika [bar{a}]n[bar{i}]yat[bar{a}]m 'a cuckoo should be brought from the forest' and B does not know what pika means, the latter i nquires about the meaning of this term alone, not about what the utterance means. [31] Therefore, one may claim, the word pikah is a separate independent constituent in (1), with its own distinct meaning. The answer to this is given as follows: a sentence such as (1) is totally distinct from a sentence such as (2) van[bar{a}]d rksa [bar{a}]n[bar{i}]yat[bar{a}]m 'a bear should be brought from the forest.' Nevertheless, people assume a similarity between the two: they know what the extracted words van[bar{a}]t, rksah and [bar{a}]n[bar{i}]yat[bar{a}]m mean and they assume that (1) and (2) share some parts, so that they inquire about the putative part they do not know. But this is merely an analytic procedure applied to a unit that is used as a single indivisible entity in communication. A parallel is drawn between this and the way in which people perceive entities like a gayal (gavaya) and Narasimha--Visnu as "man-lion." Each of these is an entity sui generis. Accordingly, each is the object of a single cognition. Nevertheless, one understands in each a part that is similar to some entity of a different generic class: a cow and a man, respectively. Hence, one also understands (anupa[acute{s}]yati 'sees subsequently') that in each of these there is a part which, though not there, is supposed to be there, a part which is not well known as co[ddot{o}]ccuring with the other and which one has not seen with the other--namely, a horse and a lion. But a person who reasons this way is said to be confused, since what he is perceiving is a single indivisible entity in each case. [32]

Bhartrhari does not just show that the arguments for assuming that words are the true units of communication can be refuted. He also emphatically argues against this position elsewhere, saying that, if a sentence meaning is considered not to arise directly from speech units, then a word meaning too will have to be dissociated from words, with the consequence that the direct relation between words and meanings--which a M[bar{i}]m[bar{a}][dot{m}]saka too assumes to be permanent and not instituted by an individual--also will be given up. [33]

It is generally accepted that Bhartrhari is arguing here against M[bar{a}]m[bar{a}]msakas. This is justified not only by the arguments advanced but also by the fact that Bhartrhari directly alludes to the M[bar{i}]m[bar{a}][dot{m}]s[bar{a}] principle of interpretation according to which the direct expression by an affix that something plays a contributing role in an act takes precedence over what is understood indirectly through inference based on what must obtain in order for a stated provision to be effective or through the co[ddot{o}]ccurrence of terms in an utterance. [34]

It is also noteworthy that Kum[bar{a}]rila reacts to arguments which appear in the V[bar{a}]kyapad[bar{a}]ya. Thus, in connection with sequences like dadhy [bar{a}]naya 'bring some curds' [35] Bhartrhari remarks as follows. Padas often lose through samdhi their presumed primitive forms. This being so, how can one determine the boundaries between padas? And how can one then discriminate a meaning for a word whose boundary has not been grasped? [36] In his [acute{S}]lokav[bar{a}]rttika, Kum[bar{a}]rila sets forth the same argument in more detail while presenting the claims of those who maintain that a sentence alone signifies, as an indivisible unit. One should determine that a meaning M is associated with a word W through reasoning by anvaya and vyatireka: if W is present, M is understood and if W is absent, the meaning M is not understood. Thus, if one has r[bar{a}]j[bar{a}] (nom. s.) meaning 'king', then in the absence of this word, one should not understand this meaning. Yet it is understood when one says r[bar{a}]j[tilde{n}][bar{a}] (inst. s.), r[bar{a}]j[bar{a}] being absent. Similarly, if one associates the meanings 'curds' and 'cow' with dadhi and gauh, respectively, then these meanings should not be understood when one says dadhy atra and g[bar{a}]m. [37] Kum[bar{a}]rila goes on to answer such objections and to uphold the view that padas are the true units. Similarly, Kum[bar{a}]rila reacts to the argument presented in VP 2.16 (see above, with note 33): even though a sentence meaning does not directly derive from words, this does not have as a consequence that the word meanings are not directly related to such speech units. The word meanings can have no other source and are not conveyed merely for their own sake but with the intention of conveying a relational sentence meaning, so that, through these word meanings a sentence meaning is indirectly related to words. [38]

On the basis of the evidence, I think it is difficult to maintain that in the second k[bar{a}]nda of the V[bar{a}]kyapad[bar{a}]ya Bhartrhari is granting equality to the view of those who maintain the principles of interpretation which are based on accepting words as the true units, namely M[bar{i}]m[bar{a}][dot{m}]sakas, or to deny that he is indeed arguing against M[bar{i}]m[bar{a}][dot{m}]sakas.

3.6. As noted earlier ([ss]2.3), Houben says that the "upholders of the word" (padadar[acute{s}]inah) are not just M[bar{i}]m[bar{a}][dot{m}]sakas but also grammarians. In this context, it is worth emphasizing something which Houben is aware of but downplays. [39] That grammarians could and did operate with words as constituents of sentences is beyond doubt. This does not mean, however, that Bhartrhari's arguments against padadar[acute{s}]inah are also aimed at grammarians. The issue actually comes down to something that is brought put clearly in commentaries.

Let us begin with the k[bar{a}]rik[bar{a}] in which Bhartrhari contrasts the adherents of the two views in question, VP 2.57: abhedap[bar{u}]rvak[bar{a}] bhed[bar{a}]h kalpit[bar{a}] v[bar{a}]kyav[bar{a}]dibhih \ bhedap[bar{a}]rv[bar{a}]n abhed[bar{a}][dot{m}]s tu manyante padadar[acute{s}]sinah "Differences preceded by nondifference are posited by those who maintain that the utterance is the unit of communication; those who maintain that the word is the unit of communication, on the other hand, consider nondifference to be preceded by differences." The previous k[bar{a}]rik[bar{a}] sums up two alternatives under the position that there is an indivisible utterance: [40] whether one assumes that there is a permanence of putative composites or of a single generic unit, those who uphold this position say that a single entity has a single meaning, [41] which never deviates from it. [42] The first half of VP 2.57 thus links with this preceding verse, stating that those who maintain the primacy of a sentence as a single meaningful unit that is indivisible and has a single indivisible meaning nevertheless countenance div ision of such units into constituents, only these different constituents are fictitiously posited (kalpit[bar{a}]h) and necessarily based on the true, whole, units. Those who maintain the opposed view say that the "wholes" are composite and are based on their constituent units. Under this view, it is appropriate that only words are based on the status of being real and sentences that are said to be single units are "fictitiously" posited. Punyar[bar{a}]ja emphasizes this contrast. [43]

Bhartrhari goes on in subsequent verses to treat the related issue of whether one should consider the continuously recited Vedic texts (sa[dot{m}]hit[bar{a}]p[bar{a}]tha) as the source of the analyzed texts (padap[bar{a}]tha) or vice-versa. [44] The Vrtti on VP 2.58 notes some views which merit mention here. First, an absolute opposition is made such that either the samhit[bar{a}] or the pada is permanent. According to some, the padas are products of padak[bar{a}]ras (lit., 'makers of words') or identical with words known from the oral tradition handed down; for others, it is the samhita which is a product of a human tradition and the padas are eternal. Still others hold that both are equally eternal, but in different ways: the pada text is handed down as an eternal entity which serves to convey something else and the samhita text is handed down as an eternal entity that is thereby conveyed. Finally, some maintain that there is only one eternal tradition handed down, and that these are simply two aspects ([acute{s}akt[bar{i}] 'powers'), of being divided and undivi ded, which play the roles of being what conveys and what is to be conveyed. [45]

3.7. With this, it is appropriate to come back to what is said in the Vrtti on VP 1.2-26 (see [ss]3.2), since, as Hel[bar{a}]r[bar{a}]ja rightly points out, the second k[bar{a}]nda is where Bhartrhari reaches conclusions concerning the sentence as an object of explanation and its meaning as a fixed meaning. [46]

The Vrtti explains in detail what is meant by apoddh[bar{a}]rapad[bar{a}]rtha ('meaning gotten through extraction'), sthitalaksano 'rthah ('meaning whose character is fixed'), and anv[bar{a}]khyeyah [acute{s}]abdah ('linguistic unit that is to be explained'). An apoddh[bar{a}]rapad[bar{a}]rtha--the meaning, e.g., of a case marker in a noun or a tense marker in a verb--has the following properties. It is absolutely fused in a single whole meaning, so that it is extracted from a putative complex in a guise that is gotten by inference and assumed, and only in this guise does its distinction from other abstracted part meanings become relevant. Such a partial meaning discriminated from other partial meanings has a form that is beyond normal communication. [47] Not merely is such a form outside the realm of actual communication, it is established generally in a way that the Vrtti speaks of as a fancy, thus emphasizing that it is an invented entity. In accordance with what they have understood, grammarians establish such partial meanings, which they reach th rough repeated exposure to them from their traditions. [48] Thus, under the assumption that different utterances contain the same component because of similarity in form, grammarians abstract component elements assumed to occur in what is actually an impartite linguistic unit and they do this for the purpose of carrying out grammatical operations which account for such whole units. The partial meaning that goes beyond normal communication is then adopted as what is signified by these component linguistic items abstracted through reasoning by anvaya and vyatireka. [49] Clearly, this abstracted meaning (so 'yam apoddh[bar{a}]rapad[bar{a}]rthah) enters into grammatical talk, as when a grammarian like P[bar{a}]nini derives complex units from posited components. The Vrtti remarks that it also takes part in everyday talk involving parts, which is similar to that of a grammar. [50] That is, in everyday speech also speakers act as though sentences like those cited in [ss]3.5 spoke of a distinct object, agent, and so on, separable from actions.

Vrsabha notes appropriately that the Vrttik[bar{a}]ra says atyantasa[dot{m}]srstah, with ayanta- 'absolutely', in order to exclude a whole preceded by parts which are combined. [51] In addition, since the whole meaning from which parts are extracted is a single whole without actual parts, the form in which such a part meaning is abstracted is said to be assumed, something to be inferred. That is, to begin with there are no real parts associated with distinct meanings, so that reasoning through anvaya and vyatireka that such and such a partial meaning is associated with a particular part of a larger unit is an assumption, not a given fact. [52] Further, in normal interaction, as when some one acts upon being told to do something or refrains from doing something, such communication takes place through whole utterance meanings, not discrete part meanings. [53]

The Vrtti also says, with respect to the linguistic units characterized as those to be explained, that for some the explanation has the word as its boundary, for others the sentence. [54] That is, some grammarians derive words like purusa-s ('the man') and pac-a-ti ('cooks') individually, as though they could stand alone, and others consider the derivational procedure immediately to involve words as related to each other in sentences. [55] Now, saying a unit is to be explained amounts to saying its meaning has a fixed character. In fact, the Vrtti later says that in grammar the meaning of either a pada or a sentence is considered to be fixed. [56] On the other hand, the Vrtti also emphasizes that a meaning is fixed only relative to a sentence. Meaning divisions within padas vary. They are not necessarily linked to units with fixed boundaries, since they are abstracted in various ways by different grammarians, who divide padas differently. [57]

That is, so far as actual communication is concerned, the utterance or sentence (v[bar{a}]kya) is viewed as the unit of communication and the unit of meaning is a single, indivisible utterance meaning. Utterances are broken up into words and words into smaller units for the sake of grammatical description, and such sub-sentential units are associated with meanings abstracted from utterances, but only in a grammar ([acute{s}][bar{a}]stre) can one legitimately consider a word an upper boundary of description or a word meaning a fixed meaning, just as only grammarians deal with items like bases and affixes.

3.8. The V[bar{a}]kyapad[bar{i}]ya k[bar{a}]rik[bar{a}]s make the very same points. Thus, after listing eight different views concerning what constitutes a sentence and a sentence meaning, as well as considering definitions of sentences proposed by K[bar{a}]ty[bar{a}]yana and in M[bar{a}]m[bar{a}][dot{m}]s[bar{a}], the second k[bar{a}]nda starts by emphasizing the unity of the sentence and its meaning. Bhartrhari invokes as parallels the cognition of a variegated picture and the picture itself. As there is a single cognition which takes the entire picture into its purview, and this is then divided in imitation of the different things seen, there is the understanding of a sentence meaning in the same manner: there is one understanding, which is then artificially divided. There is a picture, which has a single overall form, but one describes it by means of colors blue and so on, which are distinct in character and represented as such. In the very same manner, a single sentence, which semantically is totally independent, is accompanied by an explanation by means of other words, which ar e semantically dependent. [58] Further, the extraction of words in a sentence is possible in the same manner that bases, affixes, and so on are divided from each other in a word. [59] Similarly, at the beginning of the third k[bar{a}]nda Bhartrhari says that words are divided into two, four, or five classes by different thinkers and that such a division is made only on the basis of abstracting from sentences, just as one extracts bases, affixes, and so on from words. He also notes different positions concerning individual and generic property when one abstracts word meanings. [60]

Of course, verses 24-26 are not the absolute beginning of the first k[bar{a}]nda. However, they constitute a summary of the general topics to be covered. Consequently, it is appropriate to say that Bhartrhari introduces the first k[bar{a}]nda with a statement of topics he will take up and each of the next k[bar{a}]ndas with a statement of his general thesis: that the unit of actual communication is the sentence, associated with a sentence meaning, that words and word meanings are abstracted from such sentences through the same reasoning grammarians use to abstract bases, affixes, and so on from words.

3.9. In sum, I consider that the evidence from the V[bar{a}]kyapad[bar{i}]ya [61] supports the position that Bhartrhari does indeed have a well-conceived theoretical stance, which he upholds, and that he does indeed argue against scholars, such as M[bar{i}]m[bar{a}][dot{m}]sakas, who do not accept the primacy of the sentence and of sentence meaning. That Bhartrhari also discusses many different views and variations on them should not obscure this point. Nor should one expect him constantly to remind his audience of this central thesis. It is sufficient that this is established firmly and that a diametrically opposed view is refuted in the second k[bar{a}]nda, the subject of which is the sentence and its meaning. It is also sufficient to begin the third k[bar{a}]nda with a statement which again emphasizes that classes of meanings and the particular sets of words with which they are associated are abstracted from sentences and their meanings. The remainder of the third ka[bar{a}]da is devoted to discussing such individual categories as well as the ways in which com plex terms--compounds and derivates with taddhita affixes--are viewed.

Finally on this topic, we have to confront the third point Houben makes in his discussion of sambandha and the primary unit of language (see [ss]2.3), namely that if the sentence is ultimately considered the primary unit of language, then the Sambandhasamudde[acute{s}]a asks the wrong kind of question. I consider that the material considered above is sufficient to show that it is Houben who has missed the point. Bhartrhari can maintain that the sentence is the real unit of actual communication and still accept that through analysis one can and should abstract words and word meanings. This is necessary in order to carry out a grammatical description. Further, in everyday life people also accept words and word meanings, though here too these are to be considered abstracted from sentences and sentence meanings. In addition, there are scholars who accept the word and its meanings as the true units of communication. Accordingly, it is appropriate that, once he has established the primacy of the indivisible sentence, Bhart rhari proceeds to discuss classes of words and their meanings. Moreover, it is appropriate to begin with a discussion of generic property and individual as word meanings, since, as Hel[bar{a}]r[bar{a}]ja points out, different scholars maintain that one or the other is the meaning of all abstracted words. [62] This includes finite verb forms, since, under the points of view Bhartrhari takes up, these too can signify a generic property that characterizes all instances of a given action and an individual instance of action (kriy[bar{a}]) can also be viewed as a being (satt[bar{a}]). [63] Given all this, it is proper to take up in the third samuddesa the possible views concerning a relation between words and word meanings.

4. In VP 1.25 (see [ss]3.2, with note 18), Bhartrhari speaks of correct and incorrect linguistic units in connection with merit and the comprehension of meaning. This point is linked with a series of issues, which were objects of discussion starting at least with Pata[tilde{n}]jali, concerning terms like (3) go (nom. sg. gauh) considered correct linguistic units (s[bar{a}]dhu[acute{s}]abda) and related terms like (4) g[bar{a}]v[bar{i}], gan[bar{i}], got[bar{a}], gopotalik[bar{a}], considered incorrect linguistic units (as[bar{a}]dhu[acute{s}]abda, apa[acute{s}]abda), both used in the same meaning ('cow'). The following issues are treated:

A. Are the types (3) and (4) equally old or is one to be considered derived from the other?

B. If one is derived from the other, does type (3) derive from type (4) or (4) from (3)?

C. At the time that both types (3) and (4) are in use, does everyone understand the meaning in question when either (3) or (4) is used?

D. Is there a direct word-meaning relation between (3) and the meaning and also between (4) and the same meaning or is only one directly related, the other indirectly related, and for what reasons?

It is agreed that both types of terms convey given meanings. On other points, there are disagreements. The discussions concerning terms of types (3) and (4) can be summarized as follows. [64]

1. If both (3) and (4) are inheritances from time immemorial, they both simply signify the meanings in question.

2. On the assumption that terms of type (4) are corruptions (apabhram[acute{s}]a) of (3), a conclusion compatible with this is that for those persons who know both types and are members of an [acute{e}]lite stratum, the [acute{s}]istas:

2a. Terms of type (3) directly signify meanings.

2b. Terms of type (4) indirectly signify meanings, through the intermediary of type (3). [65]

3. The fact remains that there are persons who normally communicate with terms of type (4). Accordingly, under 2, one has to assume further that the corruption which gave rise to (4) led to institutionalized terms that directly signify meanings at least for such speakers. Nevertheless, there are still two possibilities open:

3a. Type (4) now simply is part of usage and directly signifies, even for [acute{s}]istas.

3b. Though institutionalized, type (4) is, by virtue of its origin, still considered to signify only erroneously.

4. These options are of import for considering whether or not both types of items bear [acute{s}]akti which in turn has to do with what one defines as s[bar{a}]dhu.

4a. If both types (3) and (4) bear [acute{s}]akti and being s[bar{a}]dhu consists by definition in bearing [acute{s}]akti, then the distinction between s[bar{a}]dhusabda and apa[acute{s}]abda is eliminated. This is something neither grammarians nor others find desirable.

4b. Under 3b, it is possible not only to maintain that being s[bar{a}]dhu consists by definition in bearing [acute{s}]akti but also still to maintain the distinction in question, since items of type (4) do not truly bear [acute{s}]akti because they still are considered to signify through error.

5. While maintianing 3a, the distinction between types (3) and (4) can still be maintained, if being s[bar{a}]dhu (s[bar{a}]dhutva) is defined in a different manner:

5a. s[bar{a}]dhutva is a property of that which can be explained according to the procedures of an authoritative grammar like P[bar{a}]nini's.

5b. S[bar{a}]dhutva is a property of that which is appropriate to the production of merit; that is, the use of a s[bar{a}]dhu term entails merit which the use of an equivalent apa[acute{s}]abda does not.

4.1. Pata[tilde{n}]jali brings in (3) and (4) several times during discussions in the Paspa[acute{s}][bar{a}]. Thus, he notes that teaching correct linguistic units is briefer than teaching incorrect ones, since for each correct linguistic unit there are many incorrect ones. (3) and (4) are cited as examples of this situation. [66] Now, in Paspa[acute{s}][bar{a}] vt. 6, K[bar{a}]ty[bar{a}]yana remarks that if there is merit in the knowledge of correct linguistic units there is also demerit (j[tilde{n}][bar{a}]ne dharma iti cet tath[bar{a}]dharmah). Commenting on this, Pata[tilde{n}]jali makes two points. First, demerit obtains for one who knows correct linguistic units because such a person also knows incorrect ones. That is, even [acute{s}]istas, who use s[bar{a}]dhu terms, are presumed at least to know apa[acute{s}]abda terms for everyday interaction. Further, because there are many incorrect terms for each correct one, greater demerit obtains. [67] Pata[tilde{n}]jali also makes the well-known and important observation concerning such terms: given that there is the same comprehension of a meaning through a correct linguistic unit and an incorrect one, the grammar serves to establish a restriction intended for merit: the meaning in question should be expressed by means of a correct term, not an incorrect one. [68]

It is noteworthy that Pata[tilde{n}]jali not only contrasts correct and incorrect linguistic units using the respective terms [acute{s}]abda and apa[acute{s}]abda but that when speaking of one single item like go as opposed to the group of items (4), he characterizes the latter as apabhra[dot{m}][acute{s}]a with respect to the former. That is, these are not merely treated as incorrect speech elements contrasted with correct ones, they are also considered somehow to be corruptions with respect to the correct speech items. Pata[tilde{n}]jali thus takes a stand on the questions A and B. Since items of the type g[bar{a}]v{bar{i}] [69] are considered to be corruptions, it follows that only the correct linguistic units are to be considered eternal, so that only these bear an eternal relation with meanings. [70]

In addition, both Pata[tilde{n}]jali and K[bar{a}]ty[bar{a}]yana indicate--though not directly in connection with the issue of [aute{s}]abda versus apa[acute{s}]bda--how one might consider at least some apa[acute{s}]abdas to have arisen. In [acute{S}]ivas[bar{u}]tra vt. 1, K[bar{a}]ty[bar{a}]yana gives several reasons why l must be taught in P[bar{a}]nini's aksarasam[bar{a}]mn[bar{a}]ya despite its restricted occurrence. One of these is to account for the citation of terms that result from incapacity. To illustrate, Pata[tilde{n}]jali gives the example ltaka. Someone is named rtaka, so that this term is to be used when referring to him. Say some br[bar{a}]hmana woman has used ltaka instead and has done this out of incapacity, because she is unable to pronounce r. This is an error, so that ltaka as pronounced by the woman is a corruption of rtaka. On the other hand, if someone else says, "the br[bar{a}]hmana woman says ltaka," that person is quoting, so that he is not himself using an incorrect form. [71]

Further, since Pata[tilde{n}]jali states explicitly and K[bar{a}]ty[bar{a}]yana implicitly assumes that one who knows correct speech terms also knows their incorrect counterparts, question C is answered at least in part: at the time that both types of terms were in use, the [acute{s}]istas who used s[bar{a}]dhu[acute{s}]abdas also could use and understand apa[acute{s}]abdas.

The last question noted (D) is not dealt with in the Mah[bar{a}]bh[bar{a}]sya, but all the issues are considered elsewhere.

4.2. Let us begin with M[bar{i}]m[bar{a}][dot{m}]sakas and Naiy[bar{a}]yikas.

4.2.1. As a p[bar{u}]rvapaksa, Jaimini states that there cannot be any settled division with respect to linguistic items such that only items of the type (3) go or of the type (4) g[bar{a}]v[bar{i}], gon[bar{i}], got[bar{a}], gopotalik[bar{a}], etc., should be used, because there is no teaching concerning the situation where such usage would arise. [72] That is, there is no teaching--such as P[bar{a}]nini's grammar--whose authority is accepted in this sphere, whereby one could decide the issue. [acute{S}]abara's arguments are as follows. The question at issue is: are (3) and (4) equally means of knowledge with respect to the object possessed of a dewlap and so on? In other words, should one consider go to signify the object in question as the single term with unbroken tradition of usage and then consider the other terms (4) corruptions, or are all the terms to be considered equally without beginning? [73]

The immediate answer is that all should be considered to signify the object in question without beginning. The reason given is that the meaning in question is understood from (4) and that there was similarly a relation between them and this meaning a hundred years ago, earlier than that, and earlier still, so that (4) and their relation with this meaning has no beginning. Moreover, it has already been established that there is no creator of a relation between words and their meanings, this being a fixed eternal relation. [74] Consequently, both (3) and (4) are to be considered s[bar{a}]dhu in the sense that they bring about (s[bar{a}]dhayanti) the comprehension of the same meaning and both may be used in speaking. (3) and (4) thus have the status of synonyms, much like hasta, kara, p[bar{a}]ni 'hand' [75] Such terms are uttered for a direct purpose--to convey a particular meaning--not for some as yet unseen result that is to be brought about, and there is no teaching that instructs one to pronounce them for such a purpose. There fore, it cannot be established that one term is s[bar{a}]dhu and the others are as[bar{a}]dhu. [76]

4.2.2. Jaimini and [acute{S}]abara refute the p[bar{u}]rvapaksa and establish a siddh[bar{a}]nta on the basis of several arguments. A linguistic item can be such that an error has a part in it, since it is produced through articulatory effort. [77] It is thus possible to discriminate between (3) and (4) by considering the latter to result from errors in attempting to produce (3). A person may intend to jump on to dry land yet fall in the mud, to touch water once or once to spit out water with which he has rinsed his mouth, yet accidentally do this twice. In the same way, (4) can have come into use due to error and need not be part of an unbroken tradition of usage. [78]

It is also contrary to principle that, unless otherwise definitively established, one assume more than one term for a single meaning. [79] Moreover, the choice is not arbitrary, since there is an authority concerning such usage that allows one to make a decision: the special authority of the learned. [80] A particular term that such authoritative persons teach alone as correct should be understood to be correct. [81] Further, people understand the meaning in question from (4) due to the similarity of these terms to (3), S0 that (4) do not have the capacity that (3) has to signify the same object in question. [82] Apabhram[acute{s}]a terms like (4) play a role in the understanding of a meaning by bringing to light the capacity which properly belongs to their origins alone. [83] [acute{S}]abara describes what is at play as follows. Due to incapacity, someone pronounces g[bar{a}]v[bar{i}] when he wishes to pronounce a form of the term go, such as gauh. Someone else understands that this person means to refer to an animal with a dewlap and so on, and that to this end he wants to pronounce gauh but instead pronounces g[bar{a}]v[bar{i}]. Thus learning from this situation, others also say g[bar{a}]v[bar{i}] when what they wish to refer to is the same animal. Thereby, this animal is understood from g[bar{a}]v[bar{i}], etc. For g[bar{a}]v[bar{i}] and so on are similar to go. [84] Finally, Jaimini likens this to the situation where someone uses a form with a certain ending that is not appropriate, yet one understands what that person intends to say, thus recalling the form with the appropriate ending. [85] For example, someone might say a[acute{s}]makair [bar{a}]gacch[bar{a}]mi "I am coming from A[acute{s}]maka," using an instrumental instead of an ablative form. The instrumental form a[acute{s}]makaih is heard. However, in accordance with what is intended, the appropriate form a[acute{s}]makebhyah is recalled, and from this the meaning "from A[acute{s}]maka" is understood. Similarly, when one hears g[bar{a}]v[bar{i}] and so on, one has a recollection of go, and from this one understands the animal with a dewlap, and so on. [86]

4.2.3. The situation where a hearer interprets an incorrect form in the way noted can thus be treated as the first step in establishing for later generations that items of type (4) directly signify, without an intermediate step of recalling an equivalent of type (3). Such usage is then traced back to accidents in conversations where one speaker makes a mistake, another knows what that person intends to say, concludes that he meant to use a particular term and himself understands the intended meaning from that term. Still other persons assume that the form which was accidentally used actually directly signifies the meaning in question. This scenario is envisaged by others also.

Commenting on Prabh[bar{a}]kara's Brhat[bar{i}] to JS, [87] [acute{S}][bar{a}]lik[bar{a}]n[bar{a}]tha says that what Prabh[bar{a}]kara intends to convey is the following. A person A, wishing to utter the word go, utters g[bar{a}]v[bar{i}] instead, due to a fault in his articulating speech organs or to not being mindful or a similar reason. Through context, the adult B with whom A is talking understands what A intends to say, so that he understands the object signified by go and continues his conversation with A. The conversation is witnessed by two other people, who have not acquired the understanding that go is related as signifier to the object in question. These two mistakenly determine that B had directly understood this meaning from g[bar{a}]v[bar{i}] and not through the intermediary of go. They therefore mistakenly conclude that g[bar{a}]v[bar{i}] is itself a signifier, and in this conviction they continue to use this term in their conversations with others. Following their usage, other children then converse using g[bar{a}]v[bar{i}], since they had not acquired the knowledge associating any other word with the object in question. In this way, the use of g[bar{a}]v[bar{i}] is established as having a particular beginning. [88]

4.2.4. Similar considerations are found later also, as in the Tattvacint[bar{a}]mani, where Gange[acute{s}]a argues against assuming that apabhra[dot{m}][acute{s}]a terms as much as sa[dot{m}]skrta terms directly signify. Having concluded that only a samskrta term bears the signifying relation ([bar{s}]akti), which is secondarily transferred to apabhra[dot{m}][acute{s}]a terms, Gange[acute{s}]a has an opponent object: how can mlecchas and others of their kind, who do not know sa[dot{m}]skrta terms, impose the [acute{s}]akti of such terms on other terms? This is answered by recounting how one considers the erroneous attribution of sakti to apabhra[dot{m}][acute{s}]a terms in times long past. [89] Due to neglect, some person A, instead of using go as he should, uses g[bar{a}]v[bar{i}] A does this while conversing with B, who already has learned (vyutpannah) the signifying relation between go and a cow. B understands what A intends, so that he concludes that the latter meant to use go, and he understands the meaning 'cow' from this correct term, thus carrying out his conversation with A. There is also a bystander, a child who wants to learn (vyutpitsuh) the relation between the word and the meaning in question. He considers that B has understood 'cow' from g[bar{a}]v[bar{i}] alone, so that he acquires the understanding of g[bar{a}]v[bar{i}] as bearing the signifying relation with a cow (go[acute{s}]aktatvena). Moreover, this individual then serves as a model for others to learn that g[bar{a}]v[bar{i}] is so related to a cow. In this way, it is established erroneously that apabhramsas have direct signifying relations with meanings. [90]

4.3. It is thus admitted that at some stage apabhra[dot{m}][acute{s}]a terms also directly signify meanings. Now, even if one insists that the signifying relation that holds between terms of type (4) g[bar{a}]v[bar{i}] etc., is ultimately due to an error, the fact remains that this relation does hold. At the stage where g[bar{a}]v[bar{i}] and such do signify, persons who use such terms and do not know their sa[dot{m}]skrta equivalents-- indeed do not know Sanskrit at all--cannot be said to understand a meaning through recollecting a sa[dot{m}]skrta term. Moreover, at this stage there can no longer be any question of error. Later P[bar{a}]nin[bar{i}]yas set out such arguments in opposition to Naiy[bar{a}]yikas and M[bar{i}]m[bar{a}][dot{m}]sakas.

4.3.1. Kaundabhatta presents the following arguments. [91] Under the assumption that [acute{s}]akti consists in a term's bringing a meaning to mind (bodhakatvam), [92] it is possible to say that vernacular terms like g[bar{a}]v[bar{i}] also have [acute{s}]akti, because these too convey meanings. Since there is thus no difference between items of types (3) and (4) in that both have this property, the latter also can be considered to be s[bar{a}]dhu, contrary to the accepted view. [93]

The answer Naiy[bar{a}]yik[bar{a}]s and others have to this objection is given in the k[bar{a}]rik[bar{a}] as[bar{a}]dhur anum[bar{a}]nena v[bar{a}]cakah kai[acute{s}]cid isyate.[ldots] "Some consider it appropriate that an as[bar{a}]dhu term convey [94] a meaning by calling to mind [95] a s[bar{a}]dhu term." [96] In addition, four k[bar{a}]rik[bar{a}]s from the v[bar{a}]kyapad[bar{i}]ya are cited in support of this position. According to these verses, an apabhra[dot{m}][acute{s}]a item that is used where a s[bar{a}]dhu term is to be used serves as a means whereby a given meaning is expressed, but not directly: it is separated from the apabhra[dot{m}][acute{s}]a term by the s[bar{a}]dhu term. Apabhra[dot{m}][acute{s}]a terms thus are causes for the understanding of meanings to arise through the intermediary of recalling s[bar{a}]dhu terms; they bring the meaning of a correct term to one's understanding by apparently identifying with it. [97] An analogy is drawn with how adults understand a baby. A child learning to say amba amba 'mommy, mommy' [98] makes a mistake in speaking and might say something indistinct, like bambamba. [99] Those who know the proper linguistic units, however, determine the meaning meant once the distinct form has been recalled. Apabhra[dot{m}][acute{s}]a terms, moreover, are not accepted, in teaching that continues tradition, [100] by authoritative persons who serve as models, in the way that synonyms are accepted. [101]

Of course, actual usage is the main source for concluding that a given term has the capacity to signify a certain meaning and this is the same for both types of terms. Kaundabhatta therefore goes on to summarize reasons for considering that apabhra[dot{m}][acute{s}]a terms do not directly signify meanings, as follows. First, if terms of both types (3) and (4) were considered directly to signify, one would have to conclude that all of them possess the capacity to signify, which involves prolixity. Moreover, they cannot be considered synonyms in the same way that s[bar{a}]dhu terms with the same meaning are said to be synonymous. The latter occur as synonyms in all speech areas, but terms like g[bar{a}]v[bar{i}] gon[bar{i}], and so on are restricted in their distribution, each used in a different area. Accordingly, Naiy[bar{a}]yikas and M[bar{i}]m[bar{a}]msakas conclude that being a s[bar{a}]dhu item consists in having the capacity directly to signify. [102]

Kaundabhatta then notes that in the second half of the verse cited above--that is, v[bar{a}]cakatv[bar{a}]vi[acute{s}]ese v[bar{a}] niyamah punyap[bar{a]]payoh--Bhattoji states his own view. The following is said to be the intention of this passage. If apabhra[dot{m}][acute{s}]a terms did not have the capacity to signify, there would not be any verbal cognition from them at all. Nor does such a cognition arise from recollecting the associated s[bar{a}]dhu term, since speakers of a lower stratum, p[bar{a}]maras, who do not know s[bar{a}]dhu terms, also understand meanings from apabhra[dot{m}][acute{s}]a terms and not from s[bar{a}]dhu terms. Moreover, such understanding does not arise due to erroneous assignment of signifying capacity, since no factor enters into play that would cancel what could be the false assumption that apabhrap[dot{m}][acute{s}]a terms convey meaning, so that the knowledge that such terms directly convey meanings is not falsely acquired and the knowledge that arises from hearing the terms also is not an error. [103] To buttress this position with what is said by earlier authority, two more k[bar{a}]rik[bar{a}]s from the V[bar{a}]kyapad[bar{i}]ya are invoked. As interpreted by Kaundabhatta and his commentators, the first verse says the following. Apabhra[dot{m}][acute{s}]a terms acquired the status of being established as everyday usage due to a continuous tradition of speaking among faulty speakers. [104] Among such speakers, a s[bar{a}]dhu term does not signify. [105]

Of course, this means that being s[bar{a}]dhu cannot be considered equivalent to being a meaning signifier. Hence, being s[bar{a}]dhu is said to consist in lending itself to the production of merit, and being as[bar{a}]dhu is said to consist in lending itself to the production of demerit. The last part of k[bar{a}]rik[bar{a}] 38 of the Vaiy[bar{a}]karanasiddh[bar{a}]ntak[bar{a}]rik[bar{a}] thus states that a restriction is provided by the grammar with respect to merit and demerit. [106]

4.3.2. N[bar{a}]ge[bar{s}]a too argues strongly that [acute{s}]akti resides not only in s[bar{a}]dhu terms but also in apabhra[dot{m}][acute{s}]a terms, because verbal exchange, which is the main source for learning the relation between speech units and meanings, is the same for both. [107] That is, whether a form like gauh or g[bar{a}]v[bar{i}] is used, one learns from usage that each is related as signifier to a cow.

In addition, N[bar{a}]ge[acute{s}]a also argues against those who maintain that a meaning is understood from an apabhra[dot{m}][acute{s}]a term through recollecting a related s[bar{a}]dhu term, [108] and he too has his opponents cite VP 1.177ab and 179 (see note 101) in support of this position. [109] Anum[bar{a}]nena in VP 1.177ab signifies a type of knowledge, namely remembering, so that the verse speaks of a recollection whose objects are s[bar{a}]dhu terms. Those who know s[bar{a}]dhu terms have such a knowledge of these from an apabhra[dot{m}][acute{s}]a because the latter can recall the former due to its similarity with it, since it has such a s[bar{a}]dhu term as its source.

The first arguments advanced against this view concern people who know both s[bar{a}]dhu and apabhra[dot{m}][acute{s}]a usage. To begin with, such people are known to understand what apabhra[dot{m}][acute{s}]a terms signify even without recalling corresponding s[bar{a}]dhu terms. There are also some who may not know particular s[bar{a}]dhu terms for certain meanings. Yet they still understand what is meant by the apabhra[dot{m}][acute{s}]a terms. The position being maintained would entail the unjustifiable consequence that those who do not know s[bar{a}]dhu terms signifying given meanings could have no understanding of those meanings from the apabhra[dot{m}][acute{s}]a terms. [110] One could, to be sure, say that upon hearing an apabhra[dot{m}][acute{s}]a term for which they do not know a s[bar{a}]dhu equivalent, such persons would assume that there must be some such term, so that they too understand a meaning only indirectly through such recollection. [111] This is not acceptable. The recollection of a s[bar{a}]dhu term from an apabhra[dot{m}][acute{s}]a cannot be considered truly to bring a meaning to mind. For a verbal cognition to take pla ce appropriately, one must perceive directly a signifying element characterized by a particular sequence of sounds that defines the item in question as signifier of a given meaning. This is not true of anything remembered, which is not perceived as actually uttered at the moment. [112] Nor does one have a recollection so vague that the entity signified by the apabhra[dot{m}][acute{s}]a item in question could be considered as referred to by a pronoun. [113] For example, if someone says gagar[bar{i}]m [bar{a}]naya "bring a pot," one does not understand the meaning of gagar[bar{i}]m [114] by recollecting t[bar{a}]m ('it' [fem. acc.]) Finally, and most generally, it is known that only an actually uttered term has the property of producing a verbal knowledge, so that it is not possible for such a knowledge to arise from a recalled s[bar{a}]dhu term that is not actually uttered. [115]

N[bar{a}]ge[acute{s}]a next takes up arguments which invoke error in connection with persons who know only apabhra[dot{m}][acute{s}]a usage. The claim is advanced that p[bar{a}]maras use a word like gagar[bar{i}] instead of the s[bar{a}]dhu word ghata 'pot' and understand this meaning from it. However, this results from a continuous error. Thus, one accounts for an erroneous attribution of [acute{s}]akti to gagar[bar{i}], although one does not learn a true [acute{s}]akti for this term. At some time, someone mistakenly used gagar[bar{i}] instead of ghata and this error continued up to the present. This is unacceptable. Mistaking one term for another requires some similarity. The [acute{s}]akti that is said to reside in a term like ghata is learned not only with respect to an object qualified by the generic property of being a pot but also associated with a distinct sequence of sounds. One can see no property common to gagar[bar{i}] and ghata so that claiming an error that continues to the present cannot be justified. [116]

This problem is circumvented in the manner outlined earlier ([ss][ss]4.2.1-4). As before, moreover, here too the V[bar{a}]kyapad[bar{i}]ya (1.177ab) is invoked. [117] This too is refuted on the grounds that it is not possible to decide on the basis of given terms that certain ones are apabhra[dot{m}][acute{s}]as relative to particular sa[dot{m}]skrta terms simply on the basis of what each signifies. Nor is it appropriate to say that those are sa[dot{m}]skrta terms which [acute{s}]istas acknowledge to occur universally, as opposed to apabhra[dot{m}][acute{s}]as. which have restricted dialect distribution, since it is accepted that [acute{s}]ista usage also has dialect distribution: [acute{s}]avati is used as a verb meaning 'go' in the Kamboja country but in the [bar{A}]rya country only the derivate [acute{s}]ava- 'corpse' is used. Accordingly, apabhra[dot{m}][acute{s}]as have [acute{s}]akti. [118]

What is more, N[bar{a}]ge[acute{s}]a finally notes, it is because of this that one sees that when women, [acute{s}][bar{u}]dras, and children have a doubt concerning the meaning of a s[bar{a}]dhu term used, they determine what is meant through the apabhra[dot{m}][acute{s}]a. [119]

Further, N[bar{a}]ge[acute{s}]a conceives of the property of being s[bar{a]]dhu (s[bar{a}]dhutva) in the same way as Kaundabhatta ([ss]4.3.1): this is a particular generic property, revealed by grammar, that resides in a speech unit qualified by a particular meaning and defines such a unit's having the capacity to produce merit. [120] The Paramalaghuma[tilde{n}]j[bar{u}]s[bar{a}] also considers that s[bar{a}]dhutva can consist simply in being what is to be explained by P[bar{a}]nini's grammar. [121]

4.4. As has been shown, the V[bar{a}]kyapad[bar{i}]ya is invoked in the course of arguments supporting two positions: that apa[acute{s}]abdas/apabhra[dot{m}][acute{s}]as signify only indirectly, through the intermediary of recollected s[bar{a}]dhu[acute{s}]abdas, and that both equally signify. Let us now consider more closely VP 1.175-83 and 27 together with the Vrtti.

4.4.1. The first two of these k[bar{a}]rik[bar{a}]s state what authorities consider appropriate (icchanti 'desire, wish') to call an apabhra[dot{m}][acute{s}]a: a linguistic unit, such as gon[bar{i}], that is devoid of the purification endowed by proper grammatical formation, uttered when one wishes to use, for example, gauh. Authorities consider such an item an apabhra[dot{m}][acute{s}]a under particular circumstances: when it is linked to a particular meaning. Terms like asva and gon[bar{i}] are treated as apabhra[dot{m}][acute{s}]as if they are used in particular meaning: asva 'horse' instead of a[acute{s}]va, and gon[bar{i}] 'cow' instead of go. Such items are s[bar{a}]dhu terms, however, when used in another sphere, that is, in other meanings. In all such cases, being s[bar{a}]dhu is determined by a difference in the meaning that conditions usage. [122] For example, gon[bar{i}] used with reference to a receptacle that holds a certain amount of grain, etc., and asva referring to someone who does not have money (a-sva) are s[bar{a}]dhu. [123] In addition, gon[bar{i}] and asva can be s[bar{a}]dhu terms used also with reference to a cow a nd a horse, for a reason other than what usually determines the use of these terms for such animals: a cow is likened to a gon[bar{i}] because it has a lot of milk, so that it shares a property--holding a large amount--with the receptacle called gon[bar{i}], and a horse is spoken of as lacking money. [124]

VP 1.175 speaks of terms considered to be apabhra[dot{m}][acute{s}]a. Moreover, a term like g[bar{a}]v[bar{i}] is an apabhra[dot{m}][acute{s}]a in that it is used when a speaker intends to say gauh. That is, this is viewed as a corruption, and the Vrtti remarks that apabhra[dot{m}][acute{s}]as like g[bar{a}]v[bar{i}] are used due to a speaker's incapacity, inattention, or similar cause. Further, these corruptions have sources, namely the s[bar{a}]dhu[acute{s}]abdas which a speaker intends to use in the first place. The Vrtti on VP 1.175 not only notes this but also cites the author of the Sa[dot{n}]graha, who says that any apabhra[dot{m}][acute{s}]a has a correct linguistic unit as a source. Moreover, ultimately, there is no independent apabhra[dot{m]][acute{s}]a lacking such a source (aprakrtih): every apabhra[dot{m}][acute{s}]a has a s[bar{a}]dhu term for its source. It is, of course, undeniable that not every use of apabhra[dot{m}][acute{s}]a terms is due to error, since some speakers simply use g[bar{a}]v[bar{i}] and so on as normal everyday terms. This is explained as a generalization. Due to their becoming well known, some apabhra[dot{m}][acute{s}]as achieve the status of being normal everyday term s and thereby gain independence. [25]

4.4.2. Once apa[acute{s}]abdas like g[bar{a}]v[bar{i}] are considered ultimately to be corruptions of s[bar{a}]dhu terms, with which they coexist in a setting where speakers of the accepted norm must interact with others who use apa[acute{s}]abdas normally, two views immediately are possible. First, one may refuse to grant status to apabhra[dot{m}][acute{s]]as, so that for [acute{i}]stas a translation situation obtains: they interact with those who use apabhra[dot{m}][acute{s}]as normally but they understand in terms of their own usage--something akin to a "pidgin." Alternatively, they may accept a true diglossic status, using the accepted norm among themselves and interacting with others in their own vernacular. Of course, this is a question of degree, and if the speakers of the vernacular do not also control the model speech at least to some extent the [acute{s}]istas themselves actually must interact in the vernacular.

VP 1.177 operates with the model in which apabhra[dot{m}][acute{s]]as signify indirectly. They serve to make meanings understood, but only by bringing to mind s[bar{a}]dhu terms, with which they are seemingly identified; only in this manner do they serve to bring to light the meanings of such terms. [126] The Vrtti brings out how this indirect signification takes place by invoking the parallel of gestures like constricting one's eyes. [127] Apabhra[dot{m}][acute{s}]as used in the sphere of s[bar{a}]dhu terms make a meaning understood through the intermediary channel of s[bar{a}]dhu terms, just as gestures such as constricting the eyes convey meanings through people's being acquainted with conventions, so that the gestures themselves seem to take on the form of these conventions and are well established as such. [128] Although these gestures appear to convey meanings directly, this is only because conventions have been set such that they are understood to convey what certain utterances signify; the utterances which describe what these gestures will convey by co nvention are directly connected with the meaning, and it is only because one identifies the gestures and the conventional utterances that they seem to convey meaning directly. [129] Similarly, apabhra[dot{m}][acute{s}]as do not directly convey meanings. They only appear to do so because they are identified with the true direct signifiers. The Vrtti notes explicitly that VP 1.177 is stated in order to say that apabhra[dot{m}][acute{s}]as do not have direct signification. [130]

4.4.3. Although apa[acute{s}]abdas might be considered corruptions of s[bar{a}]dhu[acute{s}]abdas, they nevertheless do indeed signify meanings, even should one insist that they do so indirectly. Therefore, the question comes up: why are apabhra[dot{m}]{acute{s}]as like g[bar{a}]v[bar{i}] and so on not recognized as synonyms of s[bar{a}]dhu[acute{s}]abdas like go? In this context, the behavior is invoked of authoritative persons referred to as [acute{s}]ista, who serve as models for speech and other modes of behavior. In connection with such issues, which depend on lore traditionally handed down, no governing cause is stated other than the established custom of how these [acute{s}]istas behave. If g[bar{a}]v[bar{i}] and so on were indeed merely alternative expressions to go and so on, [acute{s}]istas like P[bar{a}]nini would allow them also to be made known by rules of their grammars--that is, would allow for their explanation by such rules--and they would use them. [131] They do neither. Further, a s[bar{a}]dhu term is one which is used for a meaning, which is what prompts the use of words, on the assumption that it directly signifies such a meaning (pratyaksapaksena), and a meaning which directly prompts the use of words is conveyed by s[bar{a}]dhu terms. [132] Hence, Bhartrhari says what he does in VP l.l78. [133]

4.4.4. The next three k[bar{a}]rik[bar{a}]s concern how apabhra[dot{m}][acute{s}]as originate and are propagated to the point where they attain full status among certain speakers. The first situation is the familiar one. A child learning to talk makes mistakes because it does not yet have the full capacity of all its articulatory organs; even though it exerts itself to speak clearly in the desire to produce the correct word it has been told, it produces an indistinct sound. Adults listening to what the child is saying, on the other hand, determine the distinct word that is at the source of the indistinct sound made by the child and consider only that to be connected with a meaning, not its corruption produced by the child. Similarly, some meaning is expressed by an apabhra[dot{m}][acute{s}]a which is used when a s[bar{a}]dhu term should be used, but this is not directly expressed. It is separated from the apabhra[dot{m}][acute{s}]a by the s[bar{a}]dhu term. [134]

The Vrtti here speaks of speech or language which has become mixed. In this language, apa[acute{s}]abdas are used in the sphere of s[bar{a}]dhu terms, but [acute{s}]istas, who know grammar, understand s[bar{a}]dhu terms through them, and it is only through these s[bar{a}]dhu terms that they consider the meaning as being expressed. An as[bar{a}]dhu term is thus nothing more than a means of knowing other terms, just as smoke is a means of knowing fire. [135]

There are others, however, for whom the reverse holds, as is stated in VP 1.181, the Vrtti on which says the following. Apabhra[dot{m}][acute{s}]as, being used repeatedly by women, [acute{s}][bar{u}]dras, c[bar{a}]nd[bar{a}]las and such, reached the status of being fixed among negligent speakers, so that conversation using them became more commonly established among such speakers. Moreover, now when a doubt comes up consequent on someone's use of a s[bar{a}]dhu term, one determines what is meant by means of the apabhra[dot{m}][acute{s}]a of that term. Further, people thus consider only the as[bar{a}]dhu term to be the direct signifier and they set the s[bar{a}]dhu term on the side of what serves to recall another term. [136]

4.4.5. The final k[bar{a}]rik[bar{a}]s of the first k[bar{a}]nda deal with three positions. The first of these verses begins with a transition from what was said in the previous k[bar{a}]rik[bar{a}]. It speaks of this divine speech which has been defiled [137] by incapable speakers and goes on to note the contrary thinking on this issue of those who view speech as not being eternal (anityadar[acute{s}]in[bar{a}]m). [138] There are thus far two positions. A third position is brought up in the second k[bar{a}]rik[bar{a}]. Under this view, there is a continuous unbroken stream of both sorts of speech units--s[bar{a}]dhu[acute{s}]abda and apa[acute{s}]abda--so that there is no question of one deriving from the other. Nevertheless, these come down distinguished from each other, so that if a certain term is used when one means to express a meaning through another that term does not signify that meaning. [139]

The Vrtti on VP 1.182 begins by recounting what is known from accepted tradition. In the beginning, the speech of self-luminous men was as free from any apabhra[dot{m}][acute{s}]as as it was from falsehood and such. Over time, however, due to its being associated with the memory of continuous repetition of earlier faults, speech went on getting mixed with apabhramsas to the point where this reached the state of being the norm, so that for speakers of this time such speech was treated as original, not a deviation from a norm. [140] This is one extreme view. Others go to another extreme. They maintain that there is no continuum of eternal s[bar{a}]dhu speech forms (anityav[bar{a}]dinah) [141] and accordingly do not accept that s[bar{a}]dhu terms are sources of merit. These people say that s[bar{a}]dhu speech is established purely by convention, just as rules are established in contests between wrestlers, and that the set of s[bar{a}]dhu terms is derived from the vernacular original in that it has its source in that. [142] Moreover, this set of s[bar{a}]dhu forms is c onsidered a modification that is established later, [143] and is determined by confused men [144] according to accent, grammatical formation, and such. [145]

Finally, the Vrtti considers another position, that both types of speech forms equally come down in an unbroken tradition. Even those for whom there is no primal age or a divine speech which was unmixed with apa[acute{s}]abdas accept an unbroken tradition, handed down by the [acute{s}]istas, establishing a difference between s[bar{a}]dhu and as[bar{a}]dhu terms, just as they accept such a tradition, similarly handed down, establishing that some women may be approached and others not. Since this distinction is thus set, whether a well-established asadhu item like g[bar{a}]v[bar{i}] or one which, like asva, is a mistake and not well established, is used when one wishes to signify something using another particular term--namely go or a[acute{s}]va--both fail to signify in and of themselves. Whether a meaning is understood through the intermediary of a s[bar{a}]dhu term or, as with conventional gestures such as constricting one's eyes, there is an immediate understanding of a meaning for people who have become confused by repeated use of as[bar{a}]dhu terms, what comes about is an understanding but nothing more. [146]

This is understandable from the point of view of a P[bar{a}]nin[bar{i}]ya, for whom the grammar serves to establish a restriction such that in particular circumstances one who can should express himself using s[bar{a}]dhu terms only. To be sure, there are and may well always have been as[bar{a}]dhu terms also, but for this person they do not signify if a speaker with whom he is engaged in conversation intends to communicate in the high speech. Note, moreover, that, in consonance with the entire tradition that the V[bar{a}]kyapad[bar{i}]ya represents, the Vrtti does not countenance still another view, which possibly could also be entertained under the thesis that both s[bar{a}]dhu and as[bar{a}]dhu terms come down in an unbroken continuum. This is that someone wishing to speak in the vernacular and accordingly to use a term like g[bar{a}]v[bar{i}] or gon[bar{i}], might instead mistakenly use a term like gauh. [147] For, even if apa[acute{s}]abdas are not considered to be deviants of s[bar{a}]dhu[acute{s}]abdas and are granted equal antiquity, they are not granted equal status. There is always an assumption th at, if a mistake in usage takes place, it consists in using an apa[acute{s}]abda instead of an intended [acute{s}]abda. This is maintained not just by grammarians but also by M[bar{i}]m[bar{a}][dot{m}]sakas; see [ss]4.2.2. This is also understandable in terms of how these usages coexisted. At least from the eras of K[bar{a}]ty[bar{a}]yana and Pata[tilde{n}]jali, correct Sanskrit usage coexisted with vernacular usage, viewed as relatively incorrect, and Bhartrhari considers this distinction to be carried on by [acute{s}]istas. When speaking vernaculars to contemporaries, then, such speakers would hardly be viewed as using Sanskritic forms by mistake. Nor was a [acute{s}]ista's use of apa[acute{s}]abdas in informal contexts viewed as damning. On the contrary, it is in the context where chaste usage is absolutely required--especially in ritual--that lapses into vernacular are censured.

By ending the first k[bar{a}]nda in the way he does, Bhartrhari nicely rounds things out. For he thus harkens back to a k[bar{a}]rik[bar{a}] (VP 1.27) which immediately follows his summary of what will be dealt with (see [ss]3.2, with notes 18, 19). The Vrtti on VP 1.27 draws a parallel between the established distinction of s[bar{a}]dhu and as[bar{a}]dhu usage, which comes down in an unbroken tradition, and other such established traditions. These concern: other means of achieving merit, which are positively enjoined; practices like killing living beings, telling lies, and stealing, which are forbidden; and acts like hiccuping, laughing, and scratching, which are neither enjoined nor forbidden. These all come down in unbroken traditions and are not subject to doubt. [148] In a comparable manner, VP 1.158 reiterates that P[bar{a}]nini's grammar is a smrti, a work which hands down a memorialized tradition, that is determined by the uninterrupted tradition of usage by [acute{s}]istas. [149] The Vrtti on this k[bar{a}]rik[bar{a}] again draws a parallel with the distinctions carried on in other smrtis--between what may be eaten or not eaten and so forth--and grammatical tradition. [acute{S}]istas do not transgress the way of proper behavior that is set forth in these other traditions, and the grammar is a similar smrti, whose domain is what speech should and should not be used. [150]

4.4.6. As was noted earlier (see [ss]4.3.1, with note 96), VP 3.30 also plays an important role in discussions concerning s[bar{a}]dhu and as[bar{a}]dhu items. What the first half of this verse says accords with what was said in VP 1.180cd ([ss]4.4.4, with note 134). The second half, on the other hand, assumes that s[bar{a}]dhu[acute{s}]abdas and apa[acute{s}]abdas do not differ in that both signify, so that a restriction is stated with respect to merit and demerit: only the use of s[bar{a}]dhu[acute{s}]abdas gains one merit. This is reconcilable with the position that apabhra[dot{m}][acute{s}]a terms have come to gain established status over time, as portrayed in VP 1.181, except that here the understanding of meaning directly from an apa[acute{s}]abda is not restricted to those speakers that are considered incapable. This is also reconcilable with the third view treated at the end of the first k[bar{a}]nda, such that [acute{s}]abdas and apa[acute{s}]abdas are both considered to have come down in a continuous stream.

4.4.7. To summarize what can be said on the basis of the V[bar{a}]kyapad[bar{i}]ya passages considered here: Bhartrhari recognizes, as did his predecessors, that both s[bar{a}]dhu[acute{s}]abdas and apa[acute{s}]abdas are used and that the latter predominate. He also takes three possibilities into consideration concerning their relative status:

Apa[acute{s}]abdas are accounted for as corruptions (apabhra[dot{m}][acute{s}]a) of s[bar{a}]dhu[acute{s}]abdas, and at one primeval time there was a divine speech unsullied by such impurities.

Both apa[acute{s}]abdas and s[bar{a}]dhu[acute{s}]abdas have existed as far back as one can go, and there is no use insisting that the former derive from the latter through some sort of error.

What people call apa[acute{s}]abdas are actually part of the natural speech of people, without the adornment of grammar, and s[bar{a}]dhu[acute{s}]abdas have arisen from them via grammatical analysis.

Bhartrhari also has to admit, as did Pata[tilde{n}]jali, that even [acute{s}]istas can understand meanings from apa[acute{s}]abdas as well as s[bar{a}]dhu[acute{s}]abdas. That is, although there was among [acute{s}]istas an accepted high speech, they also used vernaculars in their everyday encounters. Here too, however, there are different approaches:

There is, to begin with, what we may call the translation thesis. A [acute{s}]ista considers the apa[acute{s}]abda a corruption and understands a meaning only indirectly, by recalling a s[bar{a}]dhu[acute{s}]abda that is linked directly as signifier with the intended meaning. For those who are not part of the [acute{s}]ista tradition, it is not only true that meaning is understood directly from what [acute{s}]istas call apa[acute{s}]abdas, it is also true that even if they can communicate with [acute{s}]istas, in case of doubt they understand through their apa[acute{s}]abda, just as the [acute{s}]ista may understand through the intermediary of s[bar{a}]dhu[acute{s}]abda.

A [acute{s}]ista might have to admit that he can indeed understand a meaning directly from an apa[acute{s}]abda, even if this is a corruption and all the more so if it has as unending a tradition of usage as does any s[bar{a}]dhu[acute{s}]abda, but that is all he will admit. He will not go so far as to grant that the apa[acute{s}]abda actually signifies in the same manner as does a s[bar{a}]dhu[acute{s}]abda. [151]

4.5. Hel[bar{a}]r[bar{a}]ja's comments on VP 3.30 reflect closely what is said in the k[bar{a}]rik[bar{a}]s and Vrtti discussed in [ss]4.4. He begins with what is, in effect, a paraphrase of VP 1.1 82 [152] and notes that for learned persons [153] speech that has become defiled signifies not directly, as a corrupt speech, but only once it has had its original undefiled form made known. He specifies that the learned do not determine a meaning directly from an apa[acute{s}]abda, so that for them there is no relation between apa[acute{s}]abdas and meaning. [154] Hel[bar{a}]r[bar{a}]ja then alludes to the tradition mentioned in the Vrtti on VP 1.182 concerning the primeval time when speech was devoid of apabhra[dot{m}][acute{s}]as and to the Sa[dot{n}]graha's statement that any apabhra[dot{m}][acute{s}]a has a correct linguistic unit as its source, while emphasizing that apabhra[dot{m}][acute{s}]as do not signify, that they only bring to mind a s[bar{a}]dhu term due to similarity, and that it is from s[bar{a}]dhu terms that meaning comprehension results. [155] He immediately goes on, however, to speak of what prevails nowadays: impurities of speech have gained general currency in the same manner as falsehood and such, due to the prevailing lack of merit, so that meaning is indeed directly understood from apa[acute{s}]abda without the intervention of s[bar{a}]dhu terms. Although this is so, one concludes that only s[bar{a}]dhu terms are means of attaining merit, on the authority of teachings that say one should not use mleccha speech--apa[acute{s}]abdas--that one should speak only using s[bar{a}]dhu terms. Accordingly, the grammar follows suit and teaches these alone. [156] Further, since it is accepted that a diversity of entities is preceded by a unity and it is also true that there is a multiplicity of human views because they are not restricted, one concludes that a group of apa[acute{s}]abdas like g[bar{a}]v[bar{i}], gon[bar{i}], and so on, has a single s[bar{a}]dhu[acute{s}]abda (go) for a source and not vice versa. [157] Thus, an apa[acute{s}]abda has as its source a s[bar{a}]dhu[acute{s}]abda, which is the object of knowledge. [158] In addition, as the ultimate stage of knowledge is identical with the undifferentiated Brahman, [159] so the stage of the s[bar{a}]dhu[acute{s}]abda is the stage of knowledge, and as differentiation in the ultimate knowledge is false (vitathah 'contrary to fact'), so is the apabhra[dot{m}][acute{s}]a stage of speech that consists in impurities false in contrast to the true form of speech that lacks these impurities. Accordingly, there is a difference in conception (vikalpah) that depends on whether one is considering the ultimate or something else. [160]

Hel[bar{a}]r[bar{a}]ja then goes on to consider the Mah[bar{a}]bh[bar{a}]sya's evam ih[bar{a}]pi sam[bar{a}]n[bar{a}]y[bar{a}]m arthagatau [acute{s}]abdena c[bar{a}]pa[acute{s}]abdena ca (see 4.1), concerning which he says the following. This is stated only with respect to the stage of speech characterized as ignorance. Since Pata[tilde{n}]jali states arthagatau ("there being a comprehension of a meaning"), [161] the possibility that he might be saying that apa[acute{s}]abdas signify is not granted respect. [162] What is meant, says Hel[bar{a}]r[bar{a}]ja, is the following. Granted, to be sure, in the stage of speech characterized as ignorance, people communicate for the most part with apa[acute{s}]abda so that there may well be just a comprehension of meaning through both [acute{s}]abda and apa[acute{s}]abda. According to Hel[bar{a}]r[bar{a}]ja, then, it is considered that Pata[tilde{n}]jali's statement, of course, shows he concedes that one understands a meaning through both, but not that an apa[acute{s}]abda is thereby on a par with a s[bar{a}]dhu[acute{s}]abda as a signifier. However, Hel[bar{a}]r[bar{a}]ja continues, because such usage is generally current, Bhartrhari says it does not make a difference between [acute{s}]abda and apa[acute{s}]abda so far as concerns comprehension of meaning, and the grammar establishes a restriction concerning merit and sin. [163]

4.6. Houben emphasizes VP 3.3.30 and the associated k[bar{a}]rik[bar{a}]s from the first k[bar{a}]nda in arguing that Hel[bar{a}]r[bar{a}]ja has not understood Bhartrhari's position, that the author of the Vrtti need not represent what Bhartrhari thought, and that those who have followed these commentators are mistaken. Thus, arguing against Virendra Sharma, Houben says (p. 25):

Sharma notes that Kaunda Bhatta and N[bar{a}]ge[acute{s}]a hold corrupt forms to be directly expressive, and say that it is the view of the grammarians. Sharma argues that this would be an alteration of the traditional view of the grammarians (1977:239-249). In my view, however, Kaunda Bhatta and N[bar{a}]ge[acute{s}]a agree on this point with both Bhartrhari and Pata[tilde{n}]jali (cf. discussion of k[bar{a}]rik[bar{a}] 30), while it is Hel[bar{a}]r[bar{a}]ja who (following the ancient Vrtti) deviates from the traditional grammarian's view.

The discussion of VP 3.3.30 to which Houben refers actually is fairly short, approximately five pages (pp. 237-42).

4.6.1. Concerning VP 3.3.3Ocd, in particular, Houben says (p. 238): "The second line of k[bar{a}]rik[bar{a}] 30 of our chapter, devoted to the other view, tells us two things: (1) there is no difference between correct and incorrect words in being significative or expressive of a meaning; (2) there is a restriction with regard to merit and demerit." He goes on to note that the same points are made in VP 1.27, then remarks (p. 239): "The point that there is no difference between correct and incorrect word in being significative, is moreover elaborated in k[bar{a}]rik[bar{a}]s 181-183." Houben's position is that one should not accept the interpretation of the k[bar{a}]rik[bar{a}]s found in the Vrtti (p. 239):

Usually, these k[bar{a}]rik[bar{a}]s are interpreted on the basis of the ancient Vrtti. However, on the basis of the k[bar{a}]rik[bar{a}]s in their own context one has to arrive at different conclusions, conclusions which are moreover in perfect accord with some relevant remarks by the author of the MBhD. This passage is therefore one of the reasons to keep what is said in the Vrtti strictly separate from what is said in the k[bar{a}]rik[bar{a}]s, in whichever way one decides the authorship of these two works. [164]

Houben nevertheless admits that to him parts of VP 1.182-83 (see notes 138-39) are not absolutely clear and remarks (p. 239): "What is not directly clear from the Sanskrit, is which opposite opinion is being referred to in 182cd. Nor is the exact meaning of ubhayes[bar{a}]m and avicched[bar{a}]d clear." [165]

There is much in Houben's discussion that I find less than acceptable. Let me begin with a general point. I consider it objectionable that in a book of 460 pages, with verses cited twice and translations repeated, the author could not take a few pages to present in full his arguments concerning VP 1.18 1-83; that instead he merely tells us he intends to discuss problems on another occasion, although he wishes readers to accept his conclusion that the Vrtti does not represent what Bhartrhari intended. The arguments Houben does set down, moreover, are not cogently formulated. Thus, he says (p. 240):

In the light of the preceding k[bar{a}]rik[bar{a}]s 181-182, however, it is very likely that 183 was intended to refer also to the tradition of 'incorrect' Prakrit words (which were the original and correct words according to the other group). In view of the fact that in some circles in Bhartrhari's time (fourth or fifth century CE) the 'incorrect' Prakrit forms were cultivated, and in the light of 181, which seems to refer to this situation, k[bar{a}]rik[bar{a}] 183 would then also refer to someone who wanted to pronounce an 'incorrect' Prakrit word, but knew only the corresponding 'correct' Sanskrit word.

The first of these verses clearly says that apabhra[dot{m}][acute{s}]as gained established status over time, with the result that for some speakers s[bar{a}]dhu terms are not signifiers. Houben does not demonstrate, however, how the two k[bar{a}]rik[bar{a}]s make it "very likely" that the final verse intended to convey what he claims. Moreover, although it is beyond dispute that vernaculars were indeed generally used, it does not follow that VP 1.183 thereby concerns speakers who wished to use a vernacular form but mistakenly used a Sanskritic one instead. Houben should have taken into consideration the repeated statement in the first k[bar{a}]nda k[bar{a}]rik[bar{a}]s and Vrtti that the distinction between s[bar{a}]dhu and as[bar{a}]dhu terms is an accepted tradition and the view among P[bar{a}]nin[bar{i}]yas and others that misuse involves the use of apa[acute{s}]abdas where s[bar{a}]dhu[acute{s}]abdas should be used.

4.6.2. Houben's arguments concerning the particular issue whether as[bar{a}]dhu terms signify or not could also do with more precision. He says (pp. 240--41):

What is important is that k[bar{a}]rik[bar{a}] 183 contains no indication whatsoever that Bhartrhari would not [emphasis in original] accept that in some circumstances "incorrect" words express their meaning directly. It was precisely the point of 181 that 'incorrect' words may be expressive among some speakers. And at two other places, 27 in the first K[bar{a}]nda and 30 in the Sa[dot{m}]bandhasamudde[acute{s}]a, Bhartrhari allows that "incorrect" words express their meaning directly.

The Vrtti, however, suggesting that in 182 a different view is being discussed (different from the two views referred to in 181), [166] denies that "incorrect" words can be expressive. On this view, according to the Vrtti, the "incorrect" word, whether it has become well-established (as in the situation described in 181) or not (as in 175--180, where the speaker does intend to pronounce the correct word), is by no means expressive of the meaning (nav[bar{a}]cakau bhavatah).

What VP 1.27 says is that as[bar{a}]dhu terms are those which are contrary to s[bar{a}]dhu ones--either in that they are not established from a tradition handed down by [acute{s}]istas or in that they are not means of achieving merit--although there is not a distinction between both types in that they both convey meaning. [167] This does not say explicitly that both signify meaning directly. Even under the thesis that as[bar{a}]dhu terms signify indirectly, it is still true that they signify, so that they do not differ in this respect from s[bar{a}]dhu terms. [168] Similarly, VP 3.3.30 also says that there is no distinction between the two types of terms in that they both have the property of being signifiers, but again this does not necessarily mean that both directly signify. Nor can one say that v[bar{a}]caka refers only to a direct signifier. This would be incompatible with VP 1.178, which says of as[bar{a}]dhu terms that they are not directly signifiers and uses s[bar{a}]ks[bar{a}]d av[bar{a}]cak[bar{a}]h. If v[bar{a}]caka meant only 'which signifies directly', then av[bar{a}]caka would refer to an item that does not signify directly, so that s[bar{a}]ks[bar{a}]t 'directly' would be otiose. Similarly, abhi dh[bar{a}] can mean 'signify, express', without specifying whether this is done directly or indirectly, so that VP 1.180 can say that a certain meaning is signified by an as[bar{a}]dhu term indirectly, as separated from its signifier by the correct term (s[bar{a}]dhuvyavahitah), which directly signifies it. Of course, v[bar{a}]caka and abhidh[bar{a}]yaka can also refer specifically to direct signifiers. Thus, VP 1.183, in which abhidh[bar{a}]yakah occurs, says with respect to a term used when one wishes to use another one that it does not signify the meaning. The verse specifies the particular condition under which this holds. The Vrtti (see note 146) uses v[bar{a}]caka, but also goes on to stress that in both cases there is a mere understanding (sampratyayam[bar{a}]tram) of the meaning in question. That is, an apa[acute{s}]abda like gon[bar{i}], which has become established, seemingly signifies, but this is in the way that conventional gestures convey meanings, and an apa[acute{s}]abda like asva for a[acute{s}]va signifies only through the intermediary.

In connection with the same issue, Houben (pp. 241--42) appeals to the Mah[bar{a}]bh[bar{a}]syad[bar{i}]pik[bar{a}]:

As for the MBhD, its author evinces an awareness of both the view that incorrect words are not expressive (MBhD 1:10.14--15) and the view that they are (MBhD 1:26.13--14, 27.34--35). The author of the MBhD, as the k[bar{a}]rik[bar{a}]-author and unlike the author of the Vrtti and Hel[bar{a}]r[bar{a}]ja, does not advocate an absolute denial of the possibility of incorrect words being in some circumstances expressive.

In footnote 380 (p. 241), appended to this passage, Houben criticizes Virendra Sharma, saying: "Sharma (1977: 147--148) mentions only MBhD 1:10.14--15 as 'proof' of Bhartrhari's denial of direct expressive powers to incorrect words, and neglects the other two places which would have compelled him to modify his position." In footnote 377 (p. 241), appended to the text where Houben says "Bhartrhari allows that 'incorrect' words express their meaning directly," he remarks: "Exactly the same attitude is evinced in the MBhD 1: 26.13-14 and 27.4-5." The D[bar{i}]pik[bar{a}] passages are as follows:

I.10.14-15: apabhra[dot{m}][acute{s}][bar{a}] ye [te] tv apy apraty[bar{a}]yak[bar{a}]h \ esa eva pakso naite praty[bar{a}]yayant[bar{i}]ti "But terms that are apabhra[dot{m}][acute{s}]as [ldots] are not meaning conveyors. The position is just this: these do not convey meaning."

I.26.13-14: evam arth[bar{a}]vabodha[dot{m}] prati sarvo 'rthah [acute{s}]abdam apa[acute{s}]abda[dot{m}] ca prayu[dot{n}]kte \ tatra [acute{s}]abden[bar{a}]sau praty[bar{a}]yyo netereneti niyamah kriyate "In the same way, any meaning provokes both a [acute{s}]abda and an apa[acute{s}]abda for the understanding of a meaning. This being so, a restriction is formulated: that is to be made understood with a [acute{s}]abda, not with the other."

I.27.4-5: yady apy esah paksah sy[bar{a}]t s[bar{a}]dhuvad apa[acute{s}]abd[bar{a}] api v[bar{a}]cak[bar{a}]h ity evam api drsta[dot{m}] phalam tulyam arth[bar{a}]vabodhah \ iha tv adrsta[dot{m}] phalam abhyudaya iti "Though this view too is possible, that apa[acute{s}]abdas signify as s[bar{a}]dhu terms do, nevertheless, they have the same direct result: the comprehension of a meaning. In this case, however, there is an unseen result: prosperity."

Clearly, the second passage speaks of a restriction that a meaning is to be conveyed (praty[bar{a}]yyah) by one term and not by another, but it does not expressly say anything about an apa[acute{s}]abda signifying directly. The last passage does indeed use v[bar{a}]cakah. However, as noted earlier, this too does not mean that the term in question necessarily signifies directly. Accordingly, I think Houben's use of "directly" is exaggerated. It is also an exaggeration to insist that the Vrtti does not allow that apa[acute{s}]abdas do signify directly under particular circumstances. The Vrtti on VP 1.181 (note 136) certainly does admit, as does the k[bar{a}]rik[bar{a}], that for speakers among whom apabhra[dot{m}][acute{s}]as have become established, they signify and s[bar{a}]dhu terms do not.

In sum, I find that Houben's discussion of this issue lacks cogency. In addition, I consider less than straightforward the strategy of argumentation adopted in approaching the texts in question. Towards the beginning of his book, Houben makes it clear (p. 7) that he considers the author of the Vrtti to be someone different form the author of the k[bar{a}]rik[bar{a}]s, whom he considers identical with the author of the Mah[bar{a}]bh[bar{a}]syad[bar{i}]pik[bar{a}]. Subsequently (p. 13), he recommends caution and remarks: "Even for someone who would like to establish continuity and unitary authorship of the two works, it is necessary to make a sharp distinction between the two in order to prove this point." In his commentary on VP 3.3.30, he maintains the sharp distinction between the works but, as can be seen from what I have said, he also abandons a great deal of his caution. Here Houben argues on the basis of his own interpretations, which are supported by saying "it is very likely that" VP 1.183 meant to say what he thinks it did and, negatively, that this k[bar{a}]rik[bar{a}] does not contain any indication that Bhartrhari would not accept that apa[acute{s}]abdas can signify directly. Houben does not demonstrate that what he considers likely is supported by evidence and is not merely a feeling on his part. Yet, starting from his own interpretation, he goes on to argue that Hel[bar{a}]r[bar{a}]ja has somehow misrepresented what the k[bar{a}]rik[bar{a}] text says.

4.6.3. The materials thus presented constitute for Houben (p. 241) "[ldots] the background against which Hel[bar{a}]r[bar{a}]ja's commentary on 30 should be evaluated." As noted in [ss]4.6.2, Houben characterizes Hel[bar{a}]r[bar{a}]ja as advocating "an absolute denial of the possibility of incorrect words being in some circumstances significative." Here again, he has exaggerated and his presentation lacks subtlety. Hel[bar{a}]r[bar{a}]ja's comments on VP 3.3.30 (see [ss]4.5) immediately set the tone by speaking of the learned (vidv[bar{a}][dot{m}]sah). For him, it is they who do not determine a meaning directly from an apa[acute{s}]abda, so that there is no relation between such a term and a meaning. Now, when Hel[bar{a}]r[bar{a}]ja says that apabhra[dot{m}][acute{s}]as are not signifiers (av[bar{a}]cak[bar{a}]h), he clearly means they are not direct signifiers, since he immediately goes on to note that such terms bring s[bar{a}]dhu terms to mind, from which one has a comprehension of meaning. He then speaks of the stage at which apabhra[dot{m}][acute{s}]as have become established, and relates this to what Pata[tilde{n}]jali says. In addition, Hel[bar{a}]r[bar{a}]ja makes explicit the circumstances under which apa[acute{s}]abdas signify meanings directly, just as do s[bar{a}]dhu[acute{s}]abdas: at the stage alluded to in VP 1.181 and the Vrtti thereto. It is also with this very stage in mind that Hel[bar{a}]raja first brings in Pata[tilde{n}]jali's comments (see [ss]4.1) that a restriction is established such that one should use only s[bar{a}]dhu[acute{s}]abdas to signify meanings, which are equally understood from the use of such terms and apa[acute{s}]abdas.

Houben remarks in particular on what Hel[bar{a}]r[bar{a}]ja says concerning the Mah[bar{a}]bh[bar{a}]sya's sam[bar{a}]n[bar{a}]y[bar{a}]m arthagatau [acute{s}]abdena c[bar{a}]pa[acute{s}]abena ca. In his comments on VP 3.3.30, Houben notes (p. 241): "From the use of arthagatau 'in the understanding of meaning' Hel[bar{a}]r[bar{a}]ja wants to infer Pata[tilde{n}]jali's reluctance to accept that incorrect words have a capacity to express the meaning." Later (p. 366), Houben translates the pertinent phrase from Hel[bar{a}]r[bar{a}]ja's commentary as follows: "From the expression arthagati, "understanding of meaning" [it is clear that] [Pata[tilde{n}]jali] is unwilling to accept that incorrect words have expressive power towards their meaning." He also remarks (p. 366, note 697): "Note the plural in reference to Pata[tilde{n}]jali, whereas Bhartrhari is referred to by Hel[bar{a}]r[bar{a}]ja in singular." Of course, having Hel[bar{a}]r[bar{a}]ja say that Pata[tilde{n}]jali does not wish to accept that apa[acute{s}]abdas signify directly suits Houben's thesis, under which Hel[bar{a}]r[bar{a}]ja in effect denies what Pata[tilde{n}]jali actually intended and interprets the issue in accordance with the V[bar{a}]kyapad[bar{i}]yavrtti, as opposed to the V[bar{a}]kyapad[bar{i}]ya itself. On the other hand, Houben simply asserts that Hel[bar{a}]r[bar{a}]ja's avakarnayanti refers to something Pata[tilde{n}]janli did. He does not show that Hel[bar{a}]r[bar{a}]ja elsewhere comparably refers to the Bh[bar{a}]syak[bar{a}]ra in the plural. In fact, Hel[bar{a}]r[bar{a}]ja frequently refers to Pata[tilde{n}]jali, using bh[bar{a}]syak[bar{a}]ra and bh[bar{a}]syakrt and, as far as I can ascertain, always in the singular. [169] In view of the evidence, it is obviously not appropriate to consider that avakarnayanti in the Prak[bar{i}]rnaprak[bar{a}][acute{s}]a on VP 3.3.30 has reference to Pata[tilde{n}]jali. Instead, it is proper to consider this an instance of an impersonal third plural form (see note 162). That is, Hel[bar{a}]r[bar{a}]ja here is reporting an opinion he knows of, one which agrees with what is said in the Vrtti on VP 1.183 (see note 146). Hel[bar{a}]r[bar{a}]ja also notes immediately thereafter that the author of the text speaks of a non-difference in usage due to the fact that apabhra[dot{m}][acute{s}]as have become established and that under these circumstances the grammar provides a restriction such that one should use only s[bar{a}]dhu terms to signify meanings in order to gain merit.

4.6.4. All this is in accord with what is said throughout the P[bar{a}]ninian tradition. K[bar{a}]ty[bar{a}]yana begins by saying that the grammar serves to establish a restriction intended for merit. Pata[tilde{n}]jali explains this and says that the restriction is stated showing that, although there is the same understanding of meaning through an apa[acute{s}]abda as well as a [acute{s}]abda, only the use of the latter results in merit. It is possible that in saying this Pata[tilde{n}]jali considered that both types of terms signified meanings directly. On the other hand, it is also important to see that Pata[tilde{n}]jali does not discuss this issue, so that a definitive conclusion is not possible. One can only surmise that this point may not even have been a source of contention for Pata[tilde{n}]jali.

As I have noted ([ss]4.4), Bhartrhari elaborates on the issue of s[bar{a}]dhu and as[bar{a}]dhu usage in a manner which can be understood as reflecting thinking that occurred over the years not only among grammarians but also among M[bar{i}]m[bar{a}][dot{m}]sakas and others (see [ss]4.2), including differences of opinion on whether apa[acute{s}]abdas signify directly or indirectly and for whom, as well as on whether apa[acute{s}]abdas are to be considered truly corruptions (apabhra[dot{m}][acute{s}]a) of s[bar{a}]dhu[acute{s}]abda and how they came about if they are so treated. On the other hand, Bhartrhari does not enter into discussions concerning relative brevity and prolixity in assuming that several apa[acute{s}]abdas signify a meaning signified by a single [acute{s}]abda, although we know this dispute goes back much earlier, since one finds it in Jaimini's M[bar{i}]m[bar{a}][dot{m}]s[bar{a}]s[bar{u}]tras (see [ss]4.2.2). Further, by the time of Hel[bar{a}]r[bar{a}]ja, the question of relative brevity and prolixity, centering around whether a [acute{s}]akti should be assumed for an apa[acute{s}]abda distinct from a s[bar{a}]dhu[acute{s}]abda had probably already come to have considerable importance in arguments, although this does not play an important role in the V[bar{a}]kyapad[bar{i}]ya or in Hel[bar{a}]r[bar{a}]ja's commentary. [170] As shown above ([ss][ss]4.2.4-4.3), this issue came to play a large role in discussions concerning s[bar{a}]dhu and as[bar{a}]dhu terms. In this context, let us consider N[bar{a}]ge[acute{s}]a's comments on the Prad[bar{i}]pa to Bh[bar{a}]sya I.8.21-22. As I have pointed out (note 168), Kaiyata contrasts two positions: that apabhra[dot{m}][acute{s}]as have come to be established through usage, so that they signify directly without calling to mind their s[bar{a}]dhu counterparts, and that they simply signify in the same way as s[bar{a}]dhu terms. N[bar{a}]ge[acute{s}]a associates the first position with the view already set forth in VP 1.181, but he puts this in terms familiar from Ny[bar{a}]ya discussions: apabhra[dot{m}][acute{s}]as convey meaning through erroneous attribution of [acute{s}]akti [acute{s}]aktibhramena). He also describes, in a manner that is familiar, how this comes about (see [ss][ss]4.23-4.2.4): A mistakenly says g[bar{a}]v[bar{i}] where gauh should be used and B understands the intended meaning by remembering the s[bar{a}]dhu term gauh, but a bystander C takes it that B has understood the meaning directly from g[bar{a}]v[bar{i}], thus mistakenly attributing to this the capacity to signify a cow. The error that has this source is subsequently continued. [171] N[bar{a}]ge[acute{s}]a then notes that Kaiyata brings in the second position because there is no decisive factor to show that apabhra[dot{m}][acute{s}]as gained their ability to signify in this manner, so that [acute{s}]akti resides also in vernacular terms. [172]

It is patent, I think, that although N[bar{a}]ge[acute{s}]a does indeed subscribe to the position that apabhra[dot{m}][acute{s}]a terms signify as directly as do s[bar{a}]dhu terms, he does not say exactly what Bhartrhari says. He is concerned with the situation, noted by Hel[bar{a}]r[bar{a}]ja (see [ss]4.5, with note 156), such that apa[acute{s}]abdas have already achieved currency and some speakers communicate with these only. The arguments in the Ma[bar{n}]j[bar{u}]s[bar{a}] are also specifically aimed at M[bar{i}]m[bar{a}][dot{m}]saka and Naiy[bar{a}]yika opponents, and the issue of prolixity in assuming separate [acute{s}]akti relations between individual apa[acute{s}]abdas and a single meaning occupies an important position. In addition, recall that, although Kaundabhatta cites VP 3.3.30 from Bhattoji's Siddh[bar{a}]ntak[bar{a}]rik[bar{a}]s (see [ss]4.3.1), he does not attribute the K[bar{a}]rik[bar{a}] to Bhartrhari, despite the fact that he subsequently cites verses which he does explicitly attribute to the V[bar{a}]kyapad[bar{i}]ya. Moreover, the first half of VP 3.3.30 now is said to represent the position of Naiy[bar{a}]yikas and others, the second half Bhattoji's siddh[bar{a}]nta. Given that Bhartrhari quite unpolemically entertains the view that apa[acute{s}]abdas signify through calling to mind their s[bar{a}]dhu equivalents, Kaundabhatta's presentation too cannot be said to agree in full with what Bhartrhari says. I therefore consider Houben's bald assertion (p. 25) that "[ldots]Kaunda Bhatta and N[bar{a}]ge[acute{s}]a agree on this point with both Bhartrhari and Pata[tilde{n}]jali[ldots]" inappropriate and lacking in perspective.

4.6.5. In sum, Houben's discussion of VP 3.3.30 and related k[bar{a}]rik[bar{a}]s which concern different views on the status of s[bar{a}]dhu and as[bar{a}]dhu terms is so intent on demonstrating that Bhartrhari definitely allowed for as[bar{a}]dhu terms directly to signify that he overlooks something he himself makes the very basis of his own study, Bhartrhari's "perspectivism." With respect to this issue, as with regard to other major disputes, Bhartrhari does indeed present various points of view. He does not polemically attack other positions, but he shows a clear preference for one: whether as[bar{a}]dhu terms are considered truly corruptions of s[bar{a}]dhu terms, even if both come down in uninterrupted transmission, the latter are definitely given higher status. This is the tradition of [acute{s}]istas, to which he adheres.

5. To my knowledge, Houben's is the first English translation of the Prak[bar{i}]rnaprak[bar{a}][acute{s}]a on the Sambandhasamudde[acute{s}]a. As he notes (p. 23), the k[bar{a}]rik[bar{a}]s were translated into English earlier by K. A. Subramania Iyer (1971), who also summarizes Hel[bar{a}]r[bar{a}]ja's comments. Concerning this work, Houben says (p. 23): "His translations suffice to give a general impression of the subject matter, but are not always precise and are sometimes more based on Hel[bar{a}]r[bar{a}]ja's commentary than on the k[bar{a}]rik[bar{a}]s of the VP."

In some cases, Houben's claim is justified, and he has based his translation on a rigorously established text, so that scholars must be grateful for his effort in producing a disciplined translation of two difficult texts. Houben deserves thanks also for the detailed discussions which follow the translations of k[bar{a}]rik[bar{a}]s. Nevertheless, as I think I have demonstrated, these discussions show that Houben is at times unjustifiably intent on attributing misrepresentations to the V[bar{a}]kyapad[bar{i}]yavrtti and Hel[bar{a}]r[bar{a}]ja's commentary. Despite their rigor and usefulness, some of the translations also do not do full justice to the original and in fact do not compare all that well with those of K. A. Subramania Iyer. I am aware that translations are happy hunting grounds for nitpickers and what many would consider among the most excellent could be subject to criticism at the hands of a determined critic. With all due deference, however, I think it appropriate to consider three examples in Houben's translation to make my point.

5.1. Houben's translation of the VP 3.3.1 (pp. 145, 331) is:

The cognition of the speaker, the external thing meant and the own form [of the word] are understood through words which are uttered. The relation of these (namely, the cognition, external thing meant and own form) [with the words which are uttered] is well-established [p. 331: "thing-meant"].

K. A. Subramania Iyer's translation of the same verse is (1971: 76)

From words which are uttered, the intention of the speaker, an external object and the form of the word itself are understood. Their relation is fixed.

Aside from the use of parentheses and brackets in one, both translations convey just about the same information, although I think one would have to accept that the second is clearer and more felicitous. One phrase which immediately strikes a reader in Houben's translation is cognition of the speaker, as opposed to K. A. Subramania Iyer's intention of the speaker. Here Subramania Iyer follows Hel[bar{a}]r[bar{a}]ja, who paraphrases j[tilde{n}][bar{a}]na[dot{m}] prayoktuh with prayoktur abhipr[bar{a}]yah. This is appropriate. For the translation conveys more precisely what the k[bar{a}]rik[bar{a}] speaks of. Bhartrhari is not talking about any cognitive process or result of such a process. He is talking about a knowledge which a speaker has in his mind and wishes to convey to someone in words. Admittedly, this is a relatively minor point. Yet a translation should certainly aim to convey to readers the content of the original text in a way that is both precise and understandable, and currently fashionable jargon that fails to accomplish this aim should be avoided.

5.2. Consider now the same authors' translations of VP 3.36ab (Houben, pp. 176, 341; K. A. Subramania Iyer, p. 81):

Houben: As regards samyoga (connection) and samav[bar{a}]ya (inherence), they (are tacchabd[bar{a}]h:) are called by that word (sc. 'relation') because they have (as it were) that property (sc. dependence).

Subramania Iyer: Conjunction and inherence are called relations because they have the attribute thereof.

Here again, I think it is fair to say that Subramania Iyer's translation not only is more felicitous but also more immediately conveys precisely what the text intends. The point made is that conjunction (sa[dot{m}]yoga) and inherence (samav[bar{a}]ya) are spoken of as relations because they have a property which defines what a relation is, namely the property of being dependent. Subramania Iyer translates the Sanskrit taddharmanoh "because they have the attribute thereof," but Houben translates "because they have (as it were) that property (sc. dependence)." He also devotes much of his commentary on the k[bar{a}]rik[bar{a}] to explaining this term, as follows (p. 177):

In this interpretation, the first word toddharmanos is a Bahuvr[bar{i}]hi going with sa[dot{m}]yogasamav[bar{a}]yayoh. Hel[bar{a}]r[bar{a}]ja takes it as an upam[bar{a}]-bahuvr[bar{i}]hi, and tad in the compound as a reference to sa[dot{m}]bhandha, the topic of the preceding three k[bar{a}]rik[bar{a}]s. This gives the following: samyoga (connection) and samav[bar{a}]ya (inherence) have as it were the property (sc. dependence) of this (relation). That the compound is interpreted as an upam[bar{a}]-bahuvr[bar{i}]hi is not without reason, for if it was a simple Bahuvr[bar{i}]hi, there would be the suggestion that relation still does have an own property. And since upam[bar{a}]-bahuvr[bar{i}]his are not uncommon in Sanskrit, it is not far-fetched to interpret the compound this way, in a context in which identities and near-identities are of crucial significance. A translation without 'as it were' or equivalent would use less words, yet say more than warranted on the basis of the Sanskrit compound. In my translation, tad- in the compound is taken as a direct reference to the property of dependence. In that case, if 'as it were' is omitted there is still no suggestion that relation (pure and simple) is an entity having its own property. Because samyoga (connection) and samav[bar{a}]ya (inherence) are dependent in some respects, but independent in other respects, the words 'as it were' have been added between parentheses.

Houben reiterates his difference with Hel[bar{a}]r[bar{a}]ja in a note appended to his translation of the Prak[bar{i}]rnaprak[bar{a}][acute{s}]a. [173] I confess that I find the discussion cited above confusing. VP 3.3.5 says that there is no term which signifies a relation qua relation (svadharmena 'in its own quality') [174] and that, since a relation is absolutely dependent, [175] its characteristic form is not referred to by any particular nominal term. Accordingly, a relation does indeed have a property, dependence (p[bar{a}]ratantrya).

Moreover, Houben does not justify choosing his interpretation of taddharmanoh. He does not demonstrate why one should choose to say that tad in this compound "refers to dharma in 5 and to atyantaparatantratva in 4." Hel[bar{a}]r[bar{a}]ja's interpretation is at least well grounded in Bhartrhari's own diction. Taddharmanos tu t[bar{a}]cchabdyam in VP 3.3.6 reflects a reasoning which is formulated in several places, both in grammar and in Ny[bar{a}]ya. For example, in the Mah[bar{a}]bh[bar{a}]sya on 4.1.48, Pata[tilde{n}]jali says that a term x is used with reference to some Y that is not X, a proper referent of x, for four reasons: because Y is located in or on X, because Y has a property or properties that X has, because Y is located near X, and because Y is accompanied by X. [176] In consonance with such usage, VP 3.3.6 says that conjunction and inherence are termed (t[bar{a}]cchabdyam 'the property of having that for word', i.e., being so designated) sambandha because they have a property that characterizes a relation. The property in question is being dependent on something else (p[bar{a}]ratantryam), as Hel[bar{a}]r[bar{a}]ja rightly notes: one considers that the defining feature of a relation is being dependent; conjunction and inherence have this feature with respect to substances and qualities, so that the term sambandha is used for them. [177] On the other hand, these are not absolutely dependent in that they can be independent entities served by other, dependent, ones. Hence, Hel[bar{a}]r[bar{a}]ja takes taddharmanoh in the k[bar{a}]rik[bar{a}] as equivalent to tasyeva dharmo yayoh. [178] Taddharman used here is then equivalent to taddharmadharma 'which has a property that is (like) the property of X', just as ustramukha ('camel-face') used of someone who has a face like that of a camel is tantamount to ustramukhamukha and tatk[bar{a}]la used with reference to vowels that have the time duration of a given vowel is equivalent to tatk[bar{a}]lak[bar{a}]la. [179] Of course, this means that tad of taddharmanoh refers to a relation (sambandha): taddharmah = tasya dharmah 'property of that' = sambandhasya dharmah 'property of a relation'; taddharmadharma = taddharma iva dharmo yasya 'something which has the property which is the property of that'. This makes good sense, since in VP 3.3.4-6 Bhartrhari is speaking of a relation and what its basic characteristic is, then speaks of what conjunction and inherence have in common with this.

5.3. Finally, consider Houben's translation of a passage from the beginning of the Prak[bar{i}]rnaprak[bar{a}][acute{s}]a on 3.3.2 (123.11-2): svar[bar{u}]p[bar{a}]vivekenaiva hy arthapar[bar{a}]mar[acute{s}]o 'bhidh[bar{a}]nam ucyate vrddhavyavah[bar{a}]r[bar{a}]t tathaiva sambandhavyutpatteh. Hel[bar{a}]r[bar{a}]ja here says that only referring to a meaning as not distinguished from the term that signifies it is spoken of as signifying, and he gives a reason for this: because from the usage of elders the relation between signifier and meaning is learned ([ldots] -vyutpatteh) in this way alone. Houben translates (p. 333): "For we speak of 'expression' only if the thing-meant is grasped as being not different from the own form, because the relation arises in exactly that way [with the thing-meant being not different from the own form of the word] from the usage of the elders." He thus interprets vyutpatti here to mean 'arising'. This is confusing. If Houben understands that in Hel[bar{a}]r[bar{a}]ja's view a word-meaning 'relation' arises in the sense of being produced, then Hel[bar{a}]r[bar{a}]ja is made to contradict wh at he says in his commentary on VP 3.3.1, where he emphasizes that the relation is not a matter of agreed convention established by men. How, then, should one understand that "the relation arises [ldots] from the usage of the elders"? One does not have to make the effort. For Hel[bar{a}]r[bar{a}]ja is clearly using vyutpatti here in a well-known sense, with reference to learning a relation. Similarly, vyutpanna refers to one who knows the relation between a s[bar{a}]dhu term and a given meaning, and vyutpitsu refers to someone who wishes to learn such a relation; see [ss]3.7, with note 49; [ss][ss]4.2.3-4, and note 110.

6. In his work, Houben has undertaken both to translate rigorously the k[bar{a}]rik[bar{a}]s of the Sambandhasamudde[acute{s}]a and to explain these in a historical and theoretical perspective. He has in addition not only translated Hel[bar{a}]r[bar{a}]ja's commentary on these verses but also attempted to demonstrate that in important ways Hel[bar{a}]r[bar{a}]ja has misrepresented what Bhartrhari meant to say and that he has done this by accepting interpretations found in the Vrtti. While attempting to maintain a neutral stance towards the question whether the author of the Vrtti is Bhartrhari himself, moreover, Houben nevertheless makes it clear that he considers the Vrtti to have misrepresented what is said in important k[bar{a}]rik[bar{a}]s of the V[bar{a}]kyapad[bar{i}]ya's first k[bar{a}]nda.

Houben is to be admired for his ambitious undertaking and for his learning. For reasons given above, I nevertheless consider that his undertaking has not succeeded in some important respects. I think he exaggerates what he calls Bhartrhari's "perspectivism." He also depends too often on vague argumentation and assumption when he attempts to demonstrate that the Vrtti and Hel[bar{a}]r[bar{a}]ja have misrepresented Bhartrhari's views. And, for all its rigor, Houben's translation at times either fails to convey the intent of the V[bar{a}]kyapad[bar{i}]ya clearly or actually misunderstands what Bhartrhari and Hel[bar{a}]r[bar{a}]ja say.

I suggest that, instead of seeking to find "our" interpretations of what Bhartrhari says as opposed to what we consider misrepresentations on the part of commentators like Hel[bar{a}]r[bar{a}]ja, we would do well patiently to consider with more receptive minds what all these authors say as well as full evidence from scholars representing other schools of thought on common topics of discussion.

This is a review-article of: The Sambandha-Samudde[acute{s}]a (Chapter on Relation) and Bhartrhari's Philosophy of Language: A Study of Bhartrhari's Sambandha-samudde[acute{s}]a in the Context of the V[bar{a}]kyapad[bar{i}]ya with a Translation of Hel[bar{a}]r[bar{a}]ja's Commentary, Prak[bar{i}]rna-prak[bar{a}][acute{s}]a. By JAN E. M. HOUBEN. Gonda Indological Studies, vol. II. Groningen: EGBERT FORSTEN, 1995. Pp. 460 + xv. I wish to express here my gratitude to scholars who helped me by reading a draft of this review. I have adopted numerous suggestions Ashok N. Aklujkar made regarding both style and content. Edwin Gerow suggested important stylistic revisions, which I have also adopted. Claus Oetke helped me sharpen thoughts on "perspectivism" and translation. Had it been possible, I would have taken his cue and gone much more deeply into both these issues in general and in connection with Bhartrhari specifically. Jon Yamashita made several suggestions and corrected many typographic errors.

(1.) For bibliographic information see Cardona 1976: 295-305; forthcoming, [ss]4.2; and Ramseier 1993.

(2.) On the cover and the title page sambandha appears--with m for the bindu representing anusv[bar{a}]ra--but elsewhere sa[dot{m}]bandha. I shall uniformly write sambandha.

(3.) I consider this section the weakest part of Houben's book. Thus, in subsection 3.1 (pp. 46-47), concerning M[bar{i}]m[bar{a}]ms[bar{a}], he does not even mention Jaiminis[bar{a}]tra, where the issue is taken up whether terms like g[bar{a}]v[bar{i}], gon[bar{i}] are to be granted authority in the same way that go is; see below, [ss]4.2.

(4.) In what follows, I shall refer to the Vrtti on the V[bar{a}]kyapad[bar{i}]ya and to the Vrttik[bar{a}]ra, although I accept that Bhartrhari is the author of both works, as well as of the Mah[bar{a}]bh[bar{a}]syadipik[bar{a}]. Recent arguments that have been proposed to show that Bhartrhari and the Vrttik[bar{a}]ra are distinct are not acceptable, in my opinion. For literature and arguments against some recent claims, see Cardona forthcoming, [ss]4.2.3.

(5.) These [acute{s}]istas are br[bar{a}]hmanas characterized not only by their speech but also by their moral behavior, and they inhabit a particular area in the subcontinent. See Cardona 1997: 550-54 ([ss]834). In a more general perspective, [acute{s}]istas are the carriers of Vedic traditions governing behavior.

(6.) Another topic that brings up this question is that of how word and meaning are identified with each other. Due to limitation of space, I do not take this up.

(7.) A comment is in order concerning the physical aspects of this book. In general, the production is good. There are typographical errors, as is to be expected in any book of this size, but misprints are relatively few and mostly self correcting. One error that is not appears on page 241, note 380, where a reference is given to "Sharma (1977: 147-148)." The correct reference is to pages 247-48. More surprising is the fact that in both copies which I received--one a review copy, the other a complimentary copy from the publisher--pages 145-60 are missing and pages 161-76 are duplicated. Fortunately, M. M. Deshpande did me the favor of copying and sending the missing pages, for which I thank him.

(8.) I have omitted only a reference to Peri Sarveswara Sharma's article.

(9.) The article alluded to has been published: Houben 1992-93.

(10.) Similarly, Houben 1992-93: 2.

(11.) Italics in the original.

(12.) I have omitted references Houben gives here to two articles by him.

(13.) VP 1.2: ekam eva yad [bar{a}]mn[bar{a}]ta[dot{m}] bhinna[acute{s}]aktivyap[bar{a}][acute{s}]ray[bar{a}]t \ aprthaktve 'pi [acute{s}]aktibhyah prthaktveneva vartate.

(14.) VP 1.3: adhy[bar{a}]hitakal[bar{a}][dot{m}] yasya k[bar{a}]la[acute{s}]aktim up[bar{a}][acute{s}]rit[bar{a}]h janm[bar{a}]dayo vik[bar{a}]r[bar{a}]h sad bh[bar{a}]vabhedasya yonayah.

(15.) j[bar{a}]yate 'is born', asti 'is, exists', viparinamate 'changes', vardhate 'grows', apaks[bar{i}]yate 'diminishes', vina[acute{s}]yati 'perishes, ceases to be'. E.g., Nirukta 1.2.

(16.) adhy[bar{a}]hita- is equivalent to adhy[bar{a}]ropita-, as the Paddhati notes (18.9-10): adhy[bar{a}]hit[bar{a}]h adhy[bar{a}]ropit[bar{a}]h kal[bar{a}]h yasy[bar{a}]h.

(17.) VP 1.5: pr[bar{a}]ptyup[bar{a}]yo 'nuk[bar{a}]ra[acute{s}] ca tasya vedo maharsibhih \ eko 'py anekavartmeva sam[bar{a}]mn[bar{a}]tah prthak prthak.

(18.) VP 1.24-26: apoddh[bar{a}]rapad[bar{a}]rtha ye ye c[bar{a}]rth[bar{a}]h sthitalaksan[bar{a}]h anv[bar{a}]khyey[bar{a}][acute{s}] ca ye [acute{s}]abd[bar{a}] ye c[bar{a}]pi pratip[bar{a}]dak[bar{a}]h \k[bar{a}]ryak[bar{a}]ranabh[bar{a}]vena yogyabh[bar{a}]vena ca sthit[bar{a}]h \ dharme ye pratyaye c[bar{a}][dot{n}]ga[dot{m}] sambandh[bar{a}]h s[bar{a}]dhvas[bar{a}]dhusu \\ te lingai[bar{a}] ca sva[bar{a}]abdai[acute{s}] ca [acute{s}]stre 'sminn upavarnit[bar{a}]h \ smrtyartham anugamyante kecid eva yath[bar{a}]gamam.

(19.) VPVr 1.24-26 (65.1): trisv apy esu [acute{s}]lokesu prastutasya prastutasya parisam[bar{a}]ptih.

(20.) Paddhati 64.21-23: smrtyartham iti : na may[bar{a}] ki[tilde{n}]cid ap[bar{u}]rvam kriyate kintu smaran[bar{a}]rtham astapad[bar{a}]rthisamanugama iti prakaranaprayojanam [bar{a}]ha yath[bar{a}]gamam iti: notpreksay[bar{a}] api tv [bar{a}]gam[bar{a}]nus[bar{a}]reneti.

(21.) VP 1.27: [acute{s}]istebhya [bar{a}]gam[bar{a}]t siddh[bar{a}]h s[bar{a}]dhavo dharmas[bar{a}]dhanam arthapraty[bar{a}]yan[bar{a}]bhede viparit[bar{a}]s tv as[bar{a}]dhavah. See below, [ss]4.4.5.

(22.) VP 1.29: n[bar{a}]narthik[bar{a}]m im[bar{a}][dot{m}] ka[acute{s}]cid vyavasth[bar{a}][dot{m}] kartum arhati \ tasm[bar{a}]n nibadhyate [acute{s}]istaih s[bar{a}]dhutvavisay[bar{a}] smrtih.

(23.) VP 1.43: tasm[bar{a}]d akrtaka[dot{m}] [acute{s}]astra[dot{m}] smrti[dot{m}] ca sanibandhan[bar{a}]m [bar{a}][acute{s}]rity[bar{a}]rabhyate [acute{s}]sistaih s[bar{a}]dhutvavisay[bar{a}] smrtih.

(24.) For the present discussion, it does not matter whether this referent is an individual or an individual qua member of a class delimited by a defining generic property, an external existent, or a mental entity.

(25.) See Houben, p. 20.

(26.) In the Vaiy[bar{a}]karanasiddh[bar{a}]ntak[bar{a}]rik[bar{a}], Vaiy[bar{a}]karanabh[bar{a}]sana, and Vaiy[bar{a}]karanasiddh[bar{a}]ntama[tilde{n}]j[bar{u}]s[bar{a}], as well as abbreviated versions of the last two, the Vaiy[bar{a}]karanabh[bar{u}]sanas[bar{a}]ra and Vaiy[bar{a}]karanasiddh[bar{a}]natalaghuma[tilde{n}]j[bar{u}]s[bar{a} ].

(27.) In their M[bar{i}]m[bar{a}][dot{m}]s[bar{a}]s[bar{a}]tra and Bh[bar{a}]van[bar{a}]viveka or Vidhiviveka.

(28.) In their Ny[bar{a}]yama[tilde{n}]jar[bar{i}] and Tattvacint[bar{a}]mani.

(29.) VP 2.87: iti v[bar{a}]kyesu ye dharm[bar{a}]h pad[bar{a}]rthopanibandhan[bar{a}]h \ te sarve na prakalperan pada[dot{m}] cet sy[bar{a}]d av[bar{a}]cakam.

(30.) VP 2.88: avibhakte 'pi v[bar{a}]ky[bar{a}]rthe [acute{s}]aktibhedad apoddhrte v[bar{a}]ky[bar{a}]ntaravibh[bar{a}]gena yathokta[dot{m}] na virudhyate.

(31.) VP 2.72: nirj[tilde{n}][bar{a}]t[bar{a}]rtha[dot{m}] pada[dot{m}] yac ca tadarthe pratip[bar{a}]dite \ pik[bar{a}]di yad avij[tilde{n}]ata[dot{m}] tat kim ity anuyujyate "There are words whose meanings are understood, and when their meanings have been understood, one asks 'what is[ldots]' concerning words whose meanings have not been understood." I have translated with plural forms under the assumption that padam and so on are generic singulars. The issue of how to interpret words like pika--which are used among mlecchas but not among [acute{s}]istas--is taken up in the [bar{A}]ryamlecch[bar{a}]dhikarana of Mim[bar{a}][dot{m}]s[bar{a}]s[bar{u}]tras (JS codita[dot{m}] tu prat[bar{a}]yet[bar{a}]virod[bar{a}]t pram[bar{a}]nena), where the siddh[bar{a}]nta is that such terms are to be understood in the meanings authorized by mleccha usage and not on the basis of etymological or grammatical analysis.

(32.) VP 2.90-92: gavaye narasi[dot{m}]he c[bar{a}]py ekaj[tilde{n}][bar{a}]n[bar{a}]vrte yath[bar{a}] \ bh[bar{a}]ga[dot{m}] j[bar{a}]tyantarasyaiva sadr[acute{s}]a[dot{m}] pratipadyate \aprasiddha[dot{m}] tu ya[dot{m}] bh[bar{a}]gam adrstam anupa[acute{s}]yati \ t[bar{a}]vaty asa[dot{m}]vida[dot{m}] m[bar{a}]dhah sarvatra pratipadyate \\ tath[bar{a}] pik[bar{a}]diyogena v[bar{a}]kye 'tyantavilaksane \ sadr[acute{s}]asyaiva sa[dot{m}]j[tilde{n}][bar{a}]nam asato rthasya manyate. For 2.90b, I have accepted the reading ekaj[tilde{n}][bar{a}]n[bar{a}]vrte in accordance with Punyar[bar{a}]ja's commentary, where he paraphrases this with ekaj[tilde{n}][bar{a}]nena abhinnena [bar{a}]vrte visay[bar{i}]krte paricchinne sati "by one knowledge: one that is not split up, when [ldots] is covered: when [ldots] has been made the object of [ldots], has been determined." That is, a gayal or Narasi[dot{m}]ha is the object of a single undifferentiated knowledge. The reading shown appears also in K. A. Subramania Iyer's and Raghun[bar{a}]tha [acute{S}]arm[bar{a}]'s editions, as well as in A. N. Aklujkar's unpublished edition. Rau's edition, based on the k[bar{a}]rik[bar{a}] manuscripts alone, reads ekaj[tilede{n}][bar{a}]n[bar{a}]d rte "without. [ldots]" Under this reading, the verse speaks of one's understanding a part that is similar to what pertains to a totally different generic class without actually having a cognition of such a part in a gayal or Narasi[dot{m}]ha. In Aklujkar's edition, 2.90 has v[bar{a}]py instead of c[bar{a}]py and 2.91 is: aprasiddha[dot{m}] tu yad bh[bar{a}]gam adrstam anupa[acute{s}]yati \ t[bar{a}]vat tv asa[dot{m}]vidan m[bar{u}]dhah sarva[dot{m}] na pratipadyate.

(33.) VP 2.16: a[acute{s}]abdo yadi v[bar{a}]ky[bar{a}]rthah pad[bar{a}]rtho 'pi tath[bar{a}] bhavet \ eva[dot{m}] ca sati sambandhah [acute{s}]abdasy[bar{a}]rthena h[bar{i}]yate. I have adopted the reading a[acute{s}]abdo instead of Rau's a[acute{s}][bar{a}]bdo on the basis of the Vrtti. The argument advanced in this verse is obviously against Bh[bar{a}]tta M[bar{i}]m[bar{a}][dot{m}]sakas, who let individual words signify their meanings and then have these word meanings come into relation to yield a meaning that is not signified by any speech unit. See Cardona 1983: 148-51.

(34.) VP 2.73: s[bar{a}]marthyapr[bar{a}]pita[dot{m}] yac ca vyaktyartham anusajyate \ [acute{s}]rutir ev[bar{a}]nusa[dot{n}]gena b[bar{a}]dhik[bar{a}] li[dot{n}]gav[bar{a}]kyayoh. This alludes to what is said in JS [acute{s}]rutili[dot{n}]gav[bar{a}]kyaprakaranasth[bar{a}]nasam[bar{a }]khy[bar{a}]n[bar{a}]n[bar{a}]m samav[bar{a}]ye p[bar{a}]radaurbalyam arthaviprakars[bar{a}]t, which is actually cited in the Vrtti on VP 2.75. Direct expression, indirect inference due to the capacity of something said to serve as as indication that something else must obtain, use of a term with another in a single utterance, the mutual expectation between the meanings of terms, collocation, and the use of a derived term are placed on a scale such that each later factor has less weight than an earlier one in case of both coming into play for interpreting a given statement. The Vrtti on VP 2.73 (ita[acute{s}] c[bar{a}]vibh[bar{a}]gapakso no yuktah [acute{s}]rutiv[bar{a}]kyasamav[bar{a}]ye [acute{s}]rutitv[bar{a}]vi[acute{s}]esena paradaurbaly[bar{a}]sambhav[bar{a}]t) begins by noting that this gives another reason (ita[acute{s}] ca) why it is not proper to consider an utterance as a single indivisible whole (avibh[bar{a}]gapakso no yuktah). If both [acute{s}]ruti and v[bar{a}]kya come into play ([acute{s}]rutiv[bar{a}]kyasamav[bar{a}]ye), under this view it is impossible to make a decision, because it is not possible that one be less strong than the other, since there is no distinction in that there is simply [acute{s}]ruti. Bhartrhari takes up the same issues in the J[bar{a}]tisamudde[acute{s}]a (VP 3.1.75-76) and considers also the possibility that the referents of both terms are directly linked to the action. These issues cannot be discussed here.

(35.) This example is given in Punyar[bar{a}]ja's commentary on VP 2.95. Punyar[bar{a}]ja here also emphasizes that the argument is made against those who assume that only padas are real: yadi pad[bar{a}]ny eva saty[bar{a}]ni tad[bar{a}] dadhy [bar{a}]nayety[bar{a}]disa[dot{m}]hit[bar{a}]y[bar{a}]m r[bar{u}]pavin[bar{a}][acute{s}][bar{a}]t padasya niyatasy[bar{a}]bh[bar{a}]ve kam avadhi[dot{m}] grhitv[bar{a}] tadartho vivicyat[bar{a}]m [ldots]

(36.) VP 2.95: r[bar{u}]pan[bar{a}][acute{s}]e pad[bar{a}]n[bar{a}][dot{m}] sy[bar{a}]t katha[dot{m}] c[bar{a}]vadhikalpan[bar{a}] \ agrh[bar{i}]t[bar{a}]vadhau [acute{s}]abde katha[dot{m}] c[bar{a}]rtho vivicyate.

(37.) [acute{S}]lV, V[bar{a}]ky[bar{a}]dhikarana 178: tath[bar{a}] r[bar{a}]j[bar{a}]rthav[bar{a}]n drsto r[bar{a}]j[tilde{n}]ety atra co n[bar{a}]sty asau \ dodhi gour iti n[bar{a}]pimau vidmo dadhy atra g[bar{a}]m iti.

(38.) [acute{S}]lV, V[bar{a}]ky[bar{a}]dhikarana 230: a[acute{s}][bar{a}]bde c[bar{a}]pi v[bar{a}]ky[bar{a}]rthe na pad[bar{a}]rthesv a[acute{s}][bar{a}]bdat[bar{a}] v[bar{a}]ky[bar{a}]rthasyeva naites[bar{a}][dot{m}] nimitt[bar{a}]ntarasambhavah.

(39.) Houben (1993: 160) remarks, "It should be pointed out that even according to the view that the sentence is the primary unit, it is acceptable to divide the sentence secondarily into words and these into smaller parts." I cannot enter here into details concerning passages from the Dipik[bar{a}].

(40.) Punyar[bar{a}]ja introduces this, saying sphotapaksam apy upasamhartum [bar{a}]ha "He says [ldots] to summarize the position that an utterance is sphota."

(41.) K. A. Subramania Iyer's edition reads ekasyaiv[bar{a}]rthat[bar{a}]m "only one has meaning," which is found in the Vrtti on VP 2.75.

(42.) VP 2.56: nityatve samud[bar{a}]y[bar{a}]n[bar{a}]m j[bar{a}]ter v[bar{a}] parikalpane \ ekasyaik[bar{a}]rthat[bar{a}]m [bar{a}]hur v[bar{a}]kyasy[bar{a}]vyabhic[bar{a}]rin[bar{a}]m. This follows k[bar{a}]rik[bar{a}]s (VP 2.54-55) summarizing the position that a sentence is a composite of words whose meanings enter into relations with each other and the view that the sequence of words itself constitutes the utterance.

(43.) VPT[bar{i}]k[bar{a}] 2.57.

(44.) The contrasting views are set forth in VP 2.58: padaprakrtibh[bar{a}]va[acute{s}] ca vrttibhedena varnyate pad[bar{a}]n[bar{a}][dot{m}] sa[dot{m}]hit[bar{a}] yonih sa[dot{m}]hit[bar{a}] v[bar{a}] pad[bar{a}][acute{s}]ray[bar{a}]. Bhartrhari goes on to discuss places in the Mah[bar{a}]bh[bar{a}]sya where Pata[tilde{n}]jali speaks of padas as products which authors of padap[bar{a}]thas (padak[bar{a}]r[bar{a}]h) produce.

(45.) VPVr 2.58: tatra kes[bar{a}][tilde{n}]cit paurusey[bar{a}]ny [bar{a}]mn[bar{a}]yapad[bar{a}]ni [acute{s}]abdesu smrtipaksasya v[bar{a}] \ kes[bar{a}][tilde{n}]cit tu padar[bar{a}]pa ev[bar{a}]mn[bar{a}]yah sa[dot{m}]hit[bar{a}] paurusey[bar{i}] smrtipaksasya v[bar{a}] \ kes[bar{a}][tilde{n}]cit tu nity[bar{a}]v ubh[bar{a}]v apy etau sam[bar{a}]mn[bar{a}]yau padasam[bar{a}]mn[bar{a}]yas tu pratip[bar{a}]dakatvena nityah itaras tu pratip[bar{a}]dyatvena nityah \ kes[bar{a}][tilde{n}]cin nityasyaikasy[bar{a}]mn[bar{a}]yasya dve ete nitye vibh[bar{a}]g[bar{a}]vibh[bar{a}]ga[acute{s}akt[bar{i}] pratip[bar{a}]dakapratipattavyar[bar{a}]pena vartete.

(46.) Hel[bar{a}]r[bar{a}]ja, introduction to VP 3.1.1: iha pad[bar{a}]rth[bar{a}]stakaparatv[bar{a}]d v[bar{a}]kyapad[bar{a}]yasya prathamak[bar{a}]ndena prayojan[bar{a}]dipad[bar{a}]rthe nirn[bar{a}]te 'nantarak[bar{a}]ndopap[bar{a}]ditopapattibhir v[bar{a}]kyatadarthayor anv[bar{a}]khyeyasthitalaksanayoh pad[bar{a}]rthayor nirn[bar{a}]tatv[bar{a}]t tadaupayik[bar{a}]poddh[bar{a}]rapadavic[bar{a}]rah prakramyate.

(47.) VPVr 1.24-26 (65.1-3): tatr[bar{a}]poddh[bar{a}]rapad[bar{a}]rtho n[bar{a}]m[bar{a}]tyantasa[dot{m}]srstah samsarg[bar{a}]d anumeyena parikalpitena r[bar{u}]pena prakrtavivekah sannapoddhriyate praviviktasya hi tasya vastuno vyavah[bar{a}]r[bar{a}]tita[dot{m}] r[bar{u}]pam.

(48.) VPVr 1.24-26 (65: 3-4): tat tu svapratyay[bar{a}]nuk[bar{a}]rena yath[bar{a}]gama[dot{m}] bh[bar{a}]van[bar{a}]bhy[bar{a}]sava[acute{s}][bar{a}]d utpreksay[bar{a}] pr[bar{a}]yena vyavasth[bar{a}]pyate.

(49.) VPVr 1.24-26 (65: 4-6): tathaiva c[bar{a}]pravibh[bar{a}]ge [acute{s}]abd[bar{a}]tmani k[bar{a}]ry[bar{a}]rtham anvayavyatirek[bar{a}]bhy[bar{a}]m r[bar{a}]pasamanugama-kalpanay[bar{a}] samud[bar{a}]y[bar{a}]d apoddhrt[bar{a}]n[bar{a}]m [acute{s}]abd[bar{a}]n[bar{a}]m abhidheyat-ven[bar{a}][acute{s}r[bar{i}]yate. This point is also emphasized elsewhere. VP 2.12 states that vrsabha 'bull' (cf. rsabha 'bull'), udaka 'water' (cf. udan 'water'), y[bar{a}]vaka 'food made from barley' (cf. y[bar{a}]va 'food made from barley') contain meaningless parts; they are analyzed, to be sure, and this is done by reasoning through anvaya and vyatireka, but anvaya and vyatireka are merely a means allowing one to speak of such items in terms of grammatical operations describing them through derivation (bh[bar{a}]gair anarthakair yukt[bar{a}] vrsabhodakay[bar{a}]vak[bar{a}]h anvayavyatirekau tu vyavah[bar{a}]ranibandhanam). Commenting on this k[bar{a}]rika, Punyar[bar{a}]ja remarks as follows: There is no understanding of the meaning of an element rsabha separately in vrsabha or that of udan, y[bar{a}]va in udaka, y[bar{a}]vaka. However, how would one who does not know be instructed in the grammar except through reasoning by anvaya and vy atireka with respect to bases and affixes? Thus also, the only view that is appropriate is that an impartite sentence alone is a signifier (nahi vrsabhe rsabhasyodakay[bar{a}]vakayor voday[bar{a}]va[acute{s}]abdayor arthanugamah ka[acute{s}]cid asti api tu [acute{s}][bar{a}]stre padasya prakrtipratyayayor anvayavyatirek[bar{a}]bhy[bar{a}][dot{m}] vin[bar{a}] katham aj[tilde{n}]o vyutp[bar{a}]dyeta \ eva[bar{a}]ca nira[dot{m}][acute{s}]sam eva v[bar{a}]kya[dot{m}] v[bar{a}]cakam ity eva yuktam).

(50.) VPVr 1.24-26 (65: 6-7): so 'yam apoddh[bar{a}]rapad[bar{a}]rthah [acute{s}][bar{a}]stravyavah[bar{a}]ram anupatati [acute{s}][bar{a}]stravyavah[bar{a}]rasadr[acute{s}]a[dot{m}] ca laukikabhedavyavah[bar{a}]ram.

(51.) Paddhati 64.26-65.1: apoddh[bar{a}]rana[dot{m}] sa[dot{m}]sargap[bar{u}]rvakam iti p[bar{u}]rv[bar{a}]vastham [bar{a}]ha atyantasa[bar{a}]srsta iti vibh[bar{a}]gap[bar{u}]rvakasa[dot{m}]-sarganisedh[bar{a}]yaty[bar{a} ]ntagrahanam.

(52.) Paddhati 65.10-12: katha[dot{m}] niravayavatv[bar{a}]t prthakkriyety [bar{a}]ha anumeyeneti anvayavyatirek[bar{a}]bhy[bar{a}][dot{m}] bh[bar{a}]g[bar{a}]num[bar{a}]nt \ t[bar{a}]v eva niravayave katham ity [bar{a}]ha kalpiteneti \ asatyat[bar{a}]m anvayavyatirekayor [bar{a}]ha. Much later, N[bar{a}]ge[acute{s}]a again makes the same point when he says (LM 14): [ldots] prativ[bar{a}]kye sanketagrah[bar{a}]sambhav[bar{a}]t tadanv[bar{a}]khy[bar{a}]nasya lagh[bar{u}]p[bar{a}]yen[bar{a}]sambhav[bar{a}]c ca kalpanay[bar{a}] pad[bar{a}]ni pravibhajya pade prakrtipratyayabh[bar{a}]gakalpanena kalpit[bar{a}]bhy[bar{a}]m anvayavyatirek[bar{a}]bhy[bar{a}][dot{m}] tattadarthavibh[bar{a}]ga[dot{m}] [acute{s}]astram[bar{a}]travisaya[dot{m}] parikalpayanti sm[bar{a}]c[bar{a}]ry[bar{a}]h "Since it is not possible to grasp a conventional relation with respect to each sentence and also because it is not possible to describe each sentence in a brief manner, teachers fictitiously divided up words and, by assuming bases and affixes as separate parts in a word, posited such and such meaning parts--whose domain is solely the grammar--through assumed anvaya and vyatireka." The assumption that one can thus carry out reasoning from anvaya and vyatireka with respect to putative parts in turn rests on the assumption that certain things that look similar are in fact the same. Bhartrhari makes this point frequently, as in VP 2.92 (see [ss]3.5, with note 32); see also above with note 49.

(53.) Paddhati 65.15-16: yatah pravibhaktaih pad[bar{a}]rthair na pravrttinivrttilaksano vyavah[bar{a}]rah[ldots]samsrstair eveti.

(54.) VPVr 1.24-26 (68: 5-6): anv[bar{a}]khyey[bar{a}][acute{s}] ca ye [acute{s}]abd[bar{a}]h : kes[bar{a}][tilde{n}]cit pad[bar{a}]vadhikam anv[bar{a}]khy[bar{a}]na[dot{m}] v[bar{a}]ky[bar{a}]vadhikam ekes[bar{a}]m. My earlier wording (Cardona 1976: 301: "[ldots] in addition to the sentence, some take syntactic units (pada) as the units to be analysed") was poor, since one could misunderstand the intent. I should not have used 'analysed'.

(55.) The Vrtti goes on to give examples such that P[bar{a}]nini could be understood to work under the padasa[dot{m}]sk[bar{a}]rapaksa in addition to the v[bar{a}]kyasa[dot{m}]sk[bar{a}]rapaksa. I cannot discuss this point here.

(56.) VPVr 1.24-26 (77.1): sthitalaksanas tu [acute{s}][bar{a}]stre pad[bar{a}]rtho v[bar{a}]ky[bar{a}]rtho v[bar{a}]. The Vrtti goes on to explain just what this meaning is. I do not think it is necessary to consider this here.

(57.) VPVr 1.24-26 (66.4-67.1): tath[bar{a}] p[bar{u}]rvapadartha uttarapad[bar{a}]rtho 'nyapad[bar{a}]rthah pr[bar{a}]tipadik[bar{a}]rtho dh[bar{a}]tvarthah pratyay[bar{a}]rtha ity ekapadav[bar{a}]cyo py aniyat[bar{a}]vadhir bahudh[bar{a}] pravibhajya kai[acute{s}]cit katha[tilde{n}]cid apoddhriyate.

(58.) VP 2.7-9: yathaika eva sarv[bar{a}]rthaprak[bar{a}][acute{s}]ah pravibhajyate dr[acute{s}]yabhed[bar{a}]nuk[bar{a}]rena v[bar{a}]ky[bar{a}]rthanugamas tath[bar{a}] \\ citrasyaikasya r[bar{u}]pasy yath[bar{a}] bhedanidar[acute{s}]anaih \\ nil[bar{a}]dibhih sam[bar{a}]khy[bar{a}]na[dot{m}] kriyate bhinnalaksanaih \tathaivaikasya v[bar{a}]kyasya nir[bar{a}]k[bar{a}][dot{n}]ksasya sarvatah \ [acute{s}]abd[bar{a}]ntaraih sam[bar{a}]khy[bar{a}]na[dot{m}] s[bar{a}]k[bar{a}][dot{n}]ksair anugamyate. In 2.7d, I have adopted the reading v[bar{a}]ky[bar{a}]rth[bar{a}]nugamas tath[bar{a}], found in the editions of Raghun[bar{a}]tha [acute{S}]arm[bar{a}] K. A. Subramania Iyer, and Aklujkar, instead of Rau's v[bar{a}]ky[bar{a}]rth[bar{a}]vagamas tath[bar{a}] for two reasons: this accords with 2.9d anugamyate and it is the reading reflected in Punyar[bar{a}]j's commentary.

(59.) VP 2.10: yath[bar{a}] pade vibhjyante prakrtipratyay[bar{a}]dayah \ apoddh[bar{a}]ras tath[bar{a}] v[bar{a}]kye pad[bar{a}]n[bar{a}]m upapadyate. K. A. Subramania Iyer's and Raghun[bar{a}]tha [acute{S}]arm[bar{a}]'s editions have upavarnyate 'is described', which occurs also in Punyar[bar{a}]ja's Tik[bar{a}].

(60.) VP 3.1.1-2: dvidh[bar{a}] kai[acute{s}]cit pada[dot{m}] bhinna[dot{m}] caturdh[bar{a}] pa[tilde{n}]cadh[bar{a}]pi v[bar{a}] apoddhrtyaiva v[bar{a}]kyebhyah prakrtipratyay[bar{a}] divat \pad[bar{a}]rth[bar{a}]n[bar{a}]m apoddh[bar{a}]re j[bar{a}]tir v[bar{a}] dravyam eva v[bar{a}] \ pad[bar{a}]rthau sarva[acute{s}]abd[bar{a}]n[bar{a}][dot{m}] nity[bar{a}]v evopavarnitau.

(61.) Due to space limitations, I have omitted discussing evidence from the Mah[bar{a}]bh[bar{a}]syad[bar{i}]pik[bar{a]] which, though understandably scantier than the evidence from the V[bar{a}]kyapad[bar{i]]ya and its Vrtti nevertheless is in harmony with it.

(62.) Hel[bar{a}]r[bar{a}]ja 3.1.2 (8.4--5): tath[bar{a}] hi sarves[bar{a}]m api [acute{s}]abd[bar{a}]n[bar{a}][dot{m}] padar[bar{u}]p[bar{a}]n[bar{a}]m n[bar{a}]m[bar{a}]khy[bar{a}]t[bar{a}]disv[bar{a}]bh[bar{a}]v[bar{a}] n[bar{a}]m j[bar{a}]tiv[bar{a}]dimate j[bar{a}tir ev[bar{a}]rtho na dravyam / dravyav[bar{a}]dimate tu dravyam eva na j[bar{a}]tih. Hel[bar{a}]r[bar{a}]ja goes on to say that the use of v[bar{a}] 'or' twice in the k[bar{a}]ri[bar{a}]k indicates a third view: that a term signifies an individual qualified by a generic property. This need not he discussed here.

(63.) What is more, as is pointed out in the vrtti on VP 1.24-26 (67: 1-2: sthialaksanas tu v[bar{a}]kayar[bar{u}]popagrahah kalpitodee [acute{s}]avibh[bar{a}]go vi[acute{s}]ista ekah kriy[bar{a}]tm[bar{a}] vicchinnapad[bar{a}]rthagrahano-p[bar{a}]yapratip[bar{a}]dyah), in the P[bar{a}]nin[bar{i}]ya scheme of things, the fixed meaning that is linked to a sentence is an action (kriy[bar{a}]tm[bar{a}]), qualified by the karakas that participate in bringing this to completion, and this meaning is conveyed using the meanings of abstracted words as means. One such abstracted word is a verb form. Accordingly, a separate samudde[acute{s}]a, is devoted to considering what an action is.

(64.) In what follows, I will use s[bar{a}]dhu[acute{s]]abda (or simply [acute{s}]abda) and apa[acute{s}]abda (or apabhra[dot{m}][acute{s}]a where appropriate) to refer to terms of the types (3) and (4), respectively. In accordance with usage in various schools of thought, I will also use [acute{s}]akti ('power, capacity') as a cover term to refer to several things: the capacity that a given term has to produce a verbal cognition (bodhakatva) of a given meaning and a relation that holds directly between a given term and a meaning. The latter is considered to be a wish (icch[bar{a}]) that a given meaning be understood from a term or that a term produce the cognition of a given meaning as well as a signifier-significand relation (v[bar{a}]cyav[bar{a}]cakabh[bar{a]]vasambandha). These positions are most prominently associated with kaundabhatta the Naiy[bar{a}]yikas and N[bar{a}ge[acute{s}]a.

(65.) Conversely, if (3) is treated as derived from (4), the latter signify directly and the former signify indirectly, through the intermediary of (4). See [ss]4.4.4.

(66.) Bh I.5,20--22: lagh[bar{i}]y[bar{a}][bar{n}] [acute{s}]abdaopade[acute{s}]o gar[bar{i}]y[bar{a}]n apa[acute{s}]abdopade[acute{s}]ah / ekaiskasya [acute{s}]abdasya bahavo' pabhra[dot{m}][acute{s}][bar{a}]h / tad yath[bar{a}] gaur ity asya [bar{s}]abdasya g[bar{a}]v[bar{i}] gon[bar{i}] gopotalik[bar{a}]dayo 'pabhra[dot{m}][acute{s}][bar{a}]h.

(67.) Bh. I.10.6-8: j[bar{n}][bar{a}]ne dharma iti cet tath[bar{a}]dharmah pr[bar{a}]pnoti / yo hi [acute{s}]abd[bar{a}][bar{n}] j[bar{a}]n[bar{a}]ty apa[acute{s}]abd[bar{a}]n apy asau j[bar{a}]n[bar{a}]ti / yathaiva [acute{s}]abdaj[bar{n}][bar{a}]ne dharma evam apa[acute{s}]abdaj[bar{n}][bar{a}]ne 'py adharmah / athav[bar{a}] bh[bar{u}]y[bar{a}]n adharmah pr[bar{a}]pnoti / bh[bar{u}]y[bar{a}]mso pa[acute{s}]abd[bar{a}] alp[bar{i}]y[bar{a}][dot{m}]sah [acute{s}]abd[bar{a}]h / ekaikasya [acute{s}]abdasya bahavo 'pabhra[dot{m}][acute{s}][bar{a}]h / tad yath[bar{a}] gaur ity asya [acute{s}]abdasya g[bar{a}]v[bar{i}] gon[bar{i}] gopotalik[bar{a}]dayo' pabhra[dot{m}][acute{s}][bar{a}]h. / See Cardona 1997: 549 ([ss] 833).

(68.) Bh. I.8.20-22: evam ih[bar{a}]pi sam[bar{a}]n[bar{a}]y[bar{a}]m arthagatau [acute{s}]abdena c[bar{a}]pa[acute{s}]abdena ca dharmaniyamah kriyate : [acute{s}]abdenaiv[bar{a}]rtho 'bhidheyo n[bar{a}]pa[bar{a}]abdeneti, See Cardona 1997: 547 ([ss]830).

(69.) Such items are not just nominals. They are also verb forms. Thus, in 1.3.1 vt. 12, K[bar{a}]ty[bar{a}]yana says that one reason for listing verb bases in the Dh[bar{a}]tup[bar{a}]tha is to prevent the class name dh[bar{a}]tu from applying to a set of terms [bar{a}]napayati and so on (bh[bar{a}]v[bar{a}]dip[bar{a}]thah pr[bar{a}]tipadik[bar{a}]napayaty[bar{a}]dinivrttyarthah). Pata[tilde{n}]jali gives as examples also vattati and vaddhati. These are clearly Middle Indic equivalents of Sanskrit [bar{a}]j[bar{n}][bar{a}]payati 'commands', vartate 'occurs', and vardhate 'grows', with active rather than middle endings and phonological developments characteristic of Middle Indic.

(70.) From earliest known times, P[bar{a}]nin[bar{i}]yas maintained this, as is clear from K[bar{a}]ty[bar{a}]yana's first v[bar{a}]rttika (siddhe [acute{s}]abd[bar{a}]rthasambandhe lokato 'rthaprayukte [acute{s}]a]bdaprayoge [acute{s}][bar{a}]strena dharmaniyamah). In accordance with this, as interpreted in the Mah[bar{a}]bh[bar{a}]sya, Bhartrhari says that great rsis--authors of s[bar{u}]tras such as P[bar{a}]nini, authors of v[bar{a}]rttikas and authors of bh[bar{a}]syas--have handed down the tradition that linguistic units, their meanings, and the relations between the two are eternal (VP 1.23: nity[bar{a}]h [acute{s}]abd[bar{a}]rthasambandh[bar{a}]s tatr[bar{a}]mn[bar{a}]t[bar{a}] maharsibhih s[bar{u}]tr[bar{a}]n[bar{a}][dot{m}] s[bar{a}]nutantr[bar{a}]n[bar{a}]m bhasyanam ca pranetrbhih).

(71.) Bh. I.19.21-23: a[acute{s}]akti[bar{a}]anukaran[bar{a}]rthah \ a[acute{s}]akty[bar{a}] kay[bar{a}]cid br[bar{a}]hmany[bar{a}] rtaka iti proyoktavya ltaka iti prayuktam \ tasy[bar{a}]nukarana[dot{m}] br[bar{a}]hmany ltaka ity [bar{a}]ha kum[bar{a}]ry ltaka ity [bar{a}]heti.

(72.) JS prayogotpattya[acute{s}][bar{a}]stratvac chabdesu na vyavasth[bar{a}] sy[bar{a}]t.

(73.) [acute{S}]Bh (II.182-83): gaur g[bar{a}]v[bar{i}] gon[bar{i}] gopotalikety evam[bar{a}]dayah [acute{s}]abd[bar{a}] ud[bar{a}]haranam \ go[acute{s}]abdo yath[bar{a}] s[bar{a}]sn[bar{a}]dimati pram[bar{a}]nam ki[dot{m}] tath[bar{a}] g[bar{a}]vy[bar{a}]dayo 'py uta neti sandehah \ kim atraikah [acute{s}]abdo 'vicchinnap[bar{a}]ramparyo 'rth[bar{a}]tbhidh[bar{a}]yi itare 'pabhram[acute{s}][bar{a}] uta sarve 'n[bar{a}]dayah.

(74.) [acute{S}]Bh (II.183): sarva iti br[bar{u}]mah kutah \ pratyay[bar{a}]t \ prat[bar{i}]yate hi g[bar{a}]vy[bar{a}]dibhyah s[bar{a}]sn[bar{a}]dim[bar{a}]n arthah tasm[bar{a}]d ito varsa[acute{s}]ate 'py asy[bar{a}]rthasya sambandha [bar{a}]sid eva tatah parena tata[acute{s}] ca parenety an[bar{a}]dit[bar{a}] \ kart[bar{a}] c[bar{a}]sya sambandhasya n[bar{a}]stiti vyavasthitam eva. Cf. JS autpattikas tu [acute{s}]abdasy[bar{a}]rthena sambandhah[ldots]

(75.) [acute{S}]Bh (II.183): tasm[bar{a}]t sarve s[bar{a}]dhavah sarvair bh[bar{a}]sitavyam \ sarve hi s[bar{a}]dhayanty artham yath[bar{a}] hastah karah p[bar{a}]nir iti.

(76.) [acute{S}]Bh (II.183): arth[bar{a}]ya hy eta ucc[bar{a}]ryante n[bar{a}]drst[bar{a}]ya \ na hy es[bar{a}]m ucc[bar{a}]rane s[bar{a}]stram asti \ tasm[bar{a}]n na vyavatistheta ka[acute{s}]cid eka eva s[bar{a}]dhur itare's s[bar{a}]dhava iti.

(77.) JS [acute{s}]abde prayatnanispatter apar[bar{a}]dhasya bh[bar{a}]gitvam. I have taken prayatnanispatteh as an ablative stating a reason. This is the first interpretation Kum[bar{a}]rila gives (TV 11.211: prayatnanispatter iti p[bar{u}]rvoktany[bar{a}]y[bar{a}]vadh[bar{a}]rirtaprayatn[bar{a}]bhi vyaktir eva hetutvenopadi[acute{s}]yate), although he goes on to give alternative interpretations, which cannot be considered here.

(78.) [acute{S}]Bh (II.210-13): tatr[bar{a}]par[bar{a}]dhyet[bar{a}]py ucc[bar{a}]rayit[bar{a}] yath[bar{a}] [acute{s}]uske patisy[bar{a}]m[bar{i}]ti kardame patati sakrd upaspraksy[bar{a}]m[bar{i}]ti dvir upaspr[acute{s}]ati \ tato par[bar{a}]dh[bar{a}]t pravrtt[bar{a}] g[bar{a}]vy[bar{a}]dayo bhaveyur na niyogaro 'vicchinnap[bar{a}]rampary[bar{a}] eveti.

(79.) JS any[bar{a}]ya[acute{s}] c[bar{a}]neka[acute{s}]abdatvam. Accepting many terms would require also accepting many separate relations between these terms and the single meaning, thus resulting in prolixity.

(80.) JS tatra tattvam abhiyogavi[acute{s}]es[bar{a}]t. tatra can refer to the items under discussion or to the doubt concerning them. tattvam ('being that') is best interpreted to refer to the properties of being correct or incorrect. The property of being learned (abhiyoga) characterizes the authoritative learned (abhiyukta) speakers who institute restrictions, that is, the tradition of grammarians. Kum[bar{a}]rila (TV II.215) and Some[acute{s}]vara (Ny[bar{a}]yasudh[bar{a}] 310) explicitly bring in the connection with grammar.

(81.) [acute{S}]Bh (II.215):[ldots]yam abhiyukt[bar{a}] upadi[acute{s}]anty esa eva s[bar{a}]dhur iti s[bar{a}]dhur ity avagantavyah

(82.) JS tada[acute{s}]akti[acute{s}] c[bar{a}]nur[bar{a}]patv[bar{a}]t. See note 84.

(83.) TV (II.214): yath[bar{a}] ca prakrtis[bar{a}]r[bar{u}]pyadv[bar{a}]ren[bar{a}]pabhra[dot{m}][acute {s}][bar{a}]h pr[bar{a}]krt[bar{a}]m eva [acute{s}]aktim [bar{a}]virbh[bar{a}]vayanto 'rthapratipatt[bar{a}]v upayoga[dot{m}] gacchanti tath[bar{a}] tada[acute{s}]akti[acute{s}] c[bar{a}]nur[bar{u}]py[bar{a}]d ity atra varnayisyate. See below concerning [acute{S}]abara's description of what occurs.

(84.) [acute{S}]Bh (II.228): atha yad uktam artho 'vagamyate g[bar{a}]vy[bar{a}]dibhyah ata es[bar{a}]m apy an[bar{a}]dir arthena sambandha iti tada[acute{s}]aktir es[bar{a}][dot{m}] gamyate \ go[acute{s}]abdam ucc[bar{a}]rayituk[bar{a}]mena kenacid a[acute{s}]akty[bar{a}] g[bar{a}]vity ucc[bar{a}]ritam \ aparena j[bar{n}][bar{a}]ta[dot{m}] s[bar{a}]sn[bar{a}]dim[bar{a}]n asya vivaksitah tadartham gaur ity ucc[bar{a}]rayituk[bar{a}]mo g[bar{a}]vity ucc[bar{a}]rayati \ tatah [acute{s}]iksitv[bar{a}] 'pare 'pi s[bar{a}]sn[bar{a}]dimati vivaksite g[bar{a}]v[bar{i}]ty ucc[bar{a}]rayanti \ tena g[bar{a}]vy[bar{a}]dibhyah s[bar{a}]sn[bar{a}]dim[bar{a}]n avagamyate anur[bar{u}]po hi g[bar{a}]vy[bar{a}]dir go[acute{s}]abdasya. As shown, [acute{S}]abara says tada[acute{s}]aktir es[bar{a}][dot{m}] gamyate, and es[bar{a}]m must refer contextually to the apabhra[dot{m}][acute{s}]a terms g[bar{a}]vi and so on, so that tada[acute{s}]aktih has here to be interpreted as a sasthitatpurusa (= tes[bar{a}]m a[acute{s}]aktih) in which tad refers to s[bar{a}]dhu terms like go, and a[acute{s}]akti denotes the absence of [acute{s}]akti which pertains to them: apabhra[dot{m}][acute{s}]a terms lack the [acute{s}]akit which s[bar{a}]dhu terms have.

(85.) JS ekade[acute{s}]atv[bar{a}]c ca vibhaktivyatyaye sy[bar{a}]t.

(86.) [acute{S}]Bh (II.228): ata eva vibhaktivyatyaye 'pi pratyayo bhavati \ asmakair [bar{a}]gacch[bar{a}]m[bar{i}]ty a[acute{s}]makaikade[acute{s}]a upalabhyate \ a[acute{s}]makebhya ity eva [acute{s}]abdah smaryate \ tato '[acute{s}]makebhyaity eso 'rtha upalabhyate \ eva[dot{m}] g[bar{a}]vy[bar{a}]didar[acute{s}]an[bar{a}]d gos[acute{a}]bdasmarana[dot{m}] tatah s[bar{a}]sn[bar{a}]dim[bar{a}]n avagamyate. The examples are well chosen: g[bar{a}]v[bar{i}] and so on are Middle Indic, and in Middle Indic the instrumental plural and ablative plural merge, with the form etymologically equivalent to the Old Indo-Aryan instrumental serving both functions.

(87.) The passage in question concerns the issue noted above. The claim is advanced that there is no occasion to assume that the use of terms like g[bar{a}]v[bar{i}] in signifying what is also signified by go had a beginning, since it is not recorded in any smrti that such terms had their relation with their referent created by someone. This claim is refuted by showing how there can be a beginning for the use of g[bar{a}]v[bar{i}] and so on. This occurs in the way that a child pronounces while wishing to utter something else; e.g., katt[bar{a}] and vinnuh instead of kart[bar{a}] and visnuh. Brhat[bar{i}] III.143:[ldots]na ca g[bar{a}]vy[bar{a}]din[bar{a}][dot{m}] ghat[bar{a}]d[bar{i}]n[bar{a}]m iva sambandhasya kart[bar{a}] smaryate tasm[bar{a}]d [bar{a}]dimatt[bar{a}]y[bar{a}] avasara eva n[bar{a}]sti satya[dot{m}] yady avasaro na asti tv asau \ dr[acute{s}]yate hi b[bar{a}]l[bar{a}]n[bar{a}]m anya[acute{s}]abdocc[bar{a}]ranecch[bar{a}]y[bar{a}]m api [acute{s}]abd[bar{a}]ntarocc[bar{a}]rana[dot{m}] yath[bar{a}] kartecch[bar{a}]y[bar{a}][dot{m}] katteti visnur vinnur iti).

(88.) RjuP III.143: idam atr[bar{a}]bhipretam : kenacid go[acute{s}]abdam ucc[bar{a}]rayituk[bar{a}]mena apatukaranena pram[bar{a}]d[bar{a}]din[bar{a}] v[bar{a}] g[bar{a}]vity ucc[bar{a}]ritam \ tatra vrddh[bar{a}]ntarena kenacit prakaran[bar{a}]din[bar{a}] tad[bar{i}]y[bar{a}][dot{m}] vivaks[bar{a}]m avagamya go[acute{s}]abd[bar{a}]rtha[dot{m}] prat[bar{i}]tya tena vyavah[bar{a}]rah pravartitah \ tatr[bar{a}]ny[bar{a}]bhy[bar{a}]m avyutpann[bar{a}]bhy[bar{a}][dot{m}] bhr[bar{a}]ntyaivam avadh[bar{a}]ritam : g[bar{a}]v[bar{i}][acute{s}]abd[bar{a}]d ev[bar{a}]nen[bar{a}]yam artho 'vyavadh[bar{a}]nena pratipanna iti \ tata[acute{s}] ca t[bar{a}]bhy[bar{a}][dot{m}] s[bar{a}]sn[bar{a}]dimati v[bar{a}]cakatva[dot{m}] g[bar{a}]v[bar{i}][acute{s}]abdasya bhr[bar{a}]ntyaiv[bar{a}]vadh[bar{a}]ritam tath[bar{a}]vas[bar{a}]yinau tena [acute{s}]abdena vyavahrtavantau tadvyavah[bar{a}]r[bar{a}]d b[bar{a}]l[bar{a}] apare 'vyutpatty[bar{a}] vyavaharant[bar{i}]ty [bar{a}]dimattay[bar{a}] vyavah[bar{a}]rasiddhih.

(89.) I say this because Ga[dot{n}]ge[acute{s}]a uses the perfect babh[bar{u}]va (see note 90).

(90.) TC IV.2.642-43: nanu mlecch[bar{a}]d[bar{i}]n[bar{a}][dot{m}] sa[dot{m}]skrtam aj[bar{a}]nat[bar{a}][dot{m}] katha[dot{m}] tacchakty[bar{a}]ropah \ ucyate \ kenacid gaur iti [acute{s}]abde prayoktavye pram[bar{a}]d[bar{a}]d g[bar{a}]v[bar{i}][acute{s}]adbe prayukte vyutpannas tena go[acute{s}]abdam unn[bar{i}]ya tato g[bar{a}][dot{m}] prat[bar{i}]tya vyavahrtav[bar{a}]n \ yath[bar{a}]huh : amb[bar{a}]mbeti yad[bar{a}] b[bar{a}]lah [acute{s}]iksyam[bar{a}]nah prabh[bar{a}]sate \ avyakta[dot{m}] tadvid[bar{a}][dot{m}] tena vyakte bhavati nirnaya iti \ p[bar{a}]r[acute{s}]vastha[acute{s}] ca vyutpitsur g[bar{a}]v[bar{i}][acute{s}]abd[bar{a}]d ev[bar{a}]ya[dot{m}] g[bar{a}][dot{m}] prat[bar{i}]tav[bar{a}]n ity avagamya g[bar{a}]vi[acute{s}]abdam eva go[acute{s}]aktatvena prat[bar{i}]ty[bar{a}]nyes[bar{a}][dot{m}] vyutp[bar{a}]dako babh[bar{u}]veti \ tatah prabhrty apabhra[dot{m}][acute{s}]e [acute{s}]aktavabhramah. The reference is to VP 1.179 (see [ss]4.4.4). Earlier (TC IV.641), Ga[dot{n}]ge[acute{s}]a cites also JS (see note 79) to buttress one of the arguments against granting aprabhra[dot{m}][acute{s}]a terms the [acute{s}]akti relation. N[bar{a}]ge[acute{s}]a portrays the same situation; see [ss]4.6.4.

(91.) In order to shorten the presentation, I am taking the Vaiy[bar{a}]karanabh[bar{u}]sanas[bar{a}]ra as my basic source, giving abbreviated references to the Vaiy[bar{a}]karanabh[bar{u}]sana, as appropriate. For the same reason, I do not deal with what Bhattoji says in his [acute{S}]abdakaustubha concerning what constitutes s[bar{a}]dhutva and whether apabhra[dot{m}][acute{s}]a terms signify--directly or indirectly--as well as what is said on the same topic in texts such as the Padama[tilde{n}]jar[bar{i}] and Uddyotana.

(92.) VBh 218 (end of k[bar{a}]rik[bar{a}] 37): tasm[bar{a}]d bodhakatvam [acute{s}]aktir iti mate na ka[acute{s}]cid dosa iti siddham. Although this [acute{s}]akti is indeed a capacity that resides in words as signifiers, it is not strictly speaking a relation.

(93.) VBhS 296: nanv evam bh[bar{a}]s[bar{a}]dito bodhadar[acute{s}]an[bar{a}]d bodhakat[bar{a}]r[bar{u}]p[bar{a}] [acute{s}]aktis tatr[bar{a}]pi sy[bar{a}]t \ tath[bar{a}] ca s[bar{a}]dhut[bar{a}]pi sy[bar{a}]t \ [acute{s}]aktatvasyaiva s[bar{a}]dhut[bar{a}]y[bar{a}] vy[bar{a}]karan[bar{a}]dhikarane pratip[bar{a}]dan[bar{a}]t[ldots]. VBh. 218 (introduction to k[bar{a}]rik[bar{a}] 38): nanv eva[dot{m}] bh[bar{a}]s[bar{a}]dito 'pi bodhadar[acute{s}]an[bar{a}]t tatr[bar{a}]pi [acute{s}]aktisv[bar{i}]k[bar{a}]ra [bar{a}]va[acute{s}]yakah[ldots]tath[bar{a}] ca [acute{s}]aktimattv[bar{a}]vi[acute{s}]es[bar{a}]d g[bar{a}]vy[bar{a}]d[bar{i}]n[bar{a}][dot{m}] s[bar{a}]dhut[bar{a}]pattih iti ced[ldots]. The allusion in the Vaiy[bar{a}]karanabh[bar{u}]sanas[bar{a}]ra is to the section of the Jaimins[bar{u}]tras discussed above in [ss]4.2.1.

(94.) v[bar{a}]cakah, glossed (VBhs 296) as bodhakah.

(95.) anum[bar{a}]nena. In the Vaiy[bar{a}]karanabh[bar{u}]sana, Kaundabhatta says this means 'due to remembering a s[bar{a}]dhu term' or 'due to erroneously attributing [acute{s}]akti' (VBh. 218: anum[bar{a}]nena s[bar{a}]dhusmaran[bar{a}]t [acute{s}]aktibhram[bar{a}]d v[bar{a}]). In his shorter work, he explains this with a parallel: apabhra[dot{m}][acute{s}]a terms serve for recalling s[bar{a}]dhu terms (s[bar{a}]dhu[acute{s}]abdam anum[bar{a}]ya "after recalling the s[bar{a}]dhu term") just as written symbols serve to recall spoken sounds; they do not directly signify, so that they are not s[bar{a}]dhu. VBhS 296: as[bar{a}]dhur g[bar{a}]vy[bar{a}]dir anum[bar{a}]nena [acute{s}]adbam anum[bar{a}]ya v[bar{a}]cako bodhakah kai[acute{s}]cid isyate \ tath[bar{a}] ca lipivat tes[bar{a}][dot{m}] s[bar{a}]dhusmarana evopayogo na tu s[bar{a}]ks[bar{a}]ttaddv[bar{a}]cakatvam ato no s[bar{a}]dhutvam iti bh[bar{a}]vah. Harivallabha (Darpana 296: anum[bar{a}]nam atra smrtih anu pa[acute{s}]c[bar{a}]n m[bar{a}]nam iti vyutpatteh na tu vy[bar{a}]ptij[tilde{n}][bar{a}]nam tacch[bar{u}]ny[bar{a}]n[bar{a}]m api s[bar{a}]bdabodhadar[acute{s}]an[bar{a}]t) notes that anum[bar{a}]na here refers to recollection and that the anum[bar{a}]na in question is not inference through knowledge of pervasion, since one sees that even those who do not have a knowledge of pervasion as applicable to inference have a verbal cognition from as[bar{a}]dhu terms. The Vrtti on VP 1.180 invokes the example of fire and smoke (see [ss]4.4.4, with note 135), but this can be used as a parallel: an as[bar{a}]dhu term calls to mind a s[bar{a}]dhu term as smoke calls fire to mind.

(96.) VBh 218: atra naiy[bar{a}]yik[bar{a}]din[bar{a}][dot{m}] sam[bar{a}]dhim [bar{a}]ha pratham[bar{a}]rdhena : as[bar{a}]dhur anum[bar{a}]nena v[bar{a}]cakah kai[acute{s}]cid isyate v[bar{a}]cakatv[bar{a}]vi[acute{s}]ese v[bar{a}] niyamah punyap[bar{a}]payoh. This is k[bar{a}]rik[bar{a}] 38 of Bhattoji's vaiy[bar{a}]karanasiddh[bar{a}]ntak[bar{a}]rik[bar{a}]. Although the verse is taken from the V[bar{a}]kyapad[bar{i}]ya (3.3.30), Kaundabhatta treats it simply as part of Bhattoji's text upon which he comments. Given that Kaundabhatta was Bhattoji's nephew, there is no reason to doubt that this was intended. In the Vaiy[bar{a}]karanabh[bar{a}]sanas[bar{a}]ra, Kaundabhatta introduces the k[bar{a}]rik[bar{a}] saying that Bhattoji answers in two ways the doubt set forth earlier (ity [bar{a}][acute{s}]a[grave{n}]k[bar{a}][dot{m}] dvidh[bar{a}] sam[bar{a}]dhatte). However, he later (VBhS 298: id[bar{a}]n[bar{i}][dot{m}] svamatam [bar{a}]ha) introduces the second half of the verse saying that now Bhattoji states his own view.

(97.) The interpretation preferred by commentators is that apabhra[dot{m}][acute{s}]a terms are apparently identified with s[bar{a}]dhu terms: Paddhati on VP 1.177 (141) (230.20-21: t[bar{a}]d[bar{a}]tmyam iti : s[bar{a}]dhu[acute{s}]abd[bar{a}]tmat[bar{a}][dot{m}] pratipadya; Darpana 297: gopada ucc[bar{a}]ran[bar{i}]ye karan[bar{a}]p[bar{a}]tavena g[bar{a}]vity ucc[bar{a}]ritam \ vastuto gopadam evedam iti t[bar{a}]d[bar{a}]tmyena bh[bar{a}]sam[bar{a}]n[bar{a}] g[bar{a}]vy[bar{a}]di[acute{s}]abd[bar{a}] gav[bar{a}]dipad[bar{a}]rthasya prak[bar{a}][acute{s}]ak[bar{a}] ity arthah; Pariks[bar{a}] 297: t[bar{a}]d[bar{a}]tmyeti : s[bar{a}]dhut[bar{a}]d[bar{a}]tm-yaprak[bar{a}]rakaj[tilde{n}][bar{a} ]navi[acute{s}]esyat[bar{a}][dot{m}] prapya very arthah. There is also another interpretation, under which apabhra[dot{m}][acute{s}]a terms are identified with meanings just as s[bar{a}]dhu terms are; e.g., VBhS-K[bar{a}][acute{s}]ik[bar{a}] 422: t[bar{a}]d[bar{a}]tmyam upagamyeveti : arthat[bar{a}]d[bar{a}]tmya[dot{m}] gagary[bar{a}]di[acute{s}]abde grh[bar{i}]tvety arthah. In view of the comparison the Vrtti draws between this identification and the way of communicating through gestures (see [ss]4.4.4, with note 128), I consider the first interpretation preferable.

(98.) I assume that this has to do with a child learning to address its mother, so that the mother uses the vocative; cf. R[bar{a}]mapras[bar{a}]da Trip[bar{a}]th[bar{i}] 1990: 129. Of course, amb[bar{a}]mbeti can also refer to the nominative amba repeated.

(99.) The reading bambeti yath[bar{a}] b[bar{a}]lah in VP 1.179a is known from commentators (e.g., VBhS-K[bar{a}][acute{s}]ik[bar{a}] 423). See [ss]4.4.4, with note 134.

(100.) smrti[acute{s}][bar{a}]strena. In the first instance this encompasses grammars like P[bar{a}]nini's, but also includes other works, such as dictionaries, which have to do with the transmission of speech, and other authoritative works. There is also a reading smrtim[bar{a}]trena.

(101.) VBhS 296-97: ukta[dot{m}] hi v[bar{a}]kyapad[bar{i}]ye : te s[bar{a}]dhusv anum[bar{a}]nena pratyayotpattihetavah t[bar{a}]d[bar{a}]tmyam upagamyeva [acute{s}]abd[bar{a}]rthasya prak[bar{a}][acute{s}]ak[bar{a}]h \\ na [acute{s}]istair anugamyante pary[bar{a}]y[bar{a}] iva s[bar{a}]dhavah \ te yatah smrti[acute{s}][bar{a}]strena tasm[bar{a}]t s[bar{a}]ks[bar{a}]d av[bar{a}]cak[bar{a}]h \\ amb[bar{a}]mbeti yad[bar{a}] b[bar{a}]lah [acute{s}]iksam[bar{a}]nah prabh[bar{a}]sate \ avyakta[dot{m}] tadvid[bar{a}][dot{m}] tena vyakte bhavati nirnayah \ eva[dot{m}] s[bar{a}]dhau prayoktavye yo 'pabhra[dot{m}]sah prayujyate \ tena s[bar{a}]dhuvyavahitah ka[acute{s}]cid artho 'bhidh[bar{i}]yate. These verses are cited in VBh 219 also after a short series of arguments that ends with na c[bar{a}]pabhra[dot{m}][acute{s}][bar{a}]d bodho na sy[bar{a}]d iti v[bar{a}]cyam vyutpannasya s[bar{a}]dhusmaran[bar{a}]d bodhopapatteh \ ukta[dot{m}] hi v[bar{a}]kyapad[bar{i}]ya [bar{a}]gamasamuccayak[bar{a}]nda[acute{s}]ese "Nor should one say that there would be no cognition from an apabhra[dot{m}][acute{s}]a item, since a cognition is accounted for from remembering a s[bar{a}]dhu item. It has been said in the supplement to the [bar{A}]gamasamuccayak[bar{a}]nda in the V[bar{a}]kyapad[bar{i}]ya." The verses are VP 1.177-80. I have given them as they appear in Aklujkar's edition, which differs slightly from Rau's.

(102.) VBhS 297-98: nanv apabhra[dot{m}][acute{s}][bar{a}]n[bar{a}][dot{m}] s[bar{a}]ks[bar{a}]d av[bar{a}]cakatve ki[dot{m}] m[bar{a}]na[dot{m}] [acute{s}]aktikalpakavyavah[bar{a}]r[bar{a}]des tulyatv[bar{a}]d iti cet \ satyam \ tattadde[acute{s}]abhinnesu tesu tesu [acute{s}]aktikalpane gaurav[bar{a}]t \ na ca pary[bar{a}]yatulyat[bar{a}] [acute{s}]a[dot{n}]ky[bar{a}] \ tes[bar{a}][dot{m}] sarvade[acute{s}]esv ekatv[bar{a}]d vinigaman[bar{a}]virahena sarvatra [acute{s}]aktikalpan[bar{a}] \ na hy apabhra[dot{m}][acute{s}]e tath[bar{a}] \ anyath[bar{a}] bh[bar{a}]s[bar{a}]n[bar{a}][dot{m}] pary[bar{a}]yatay[bar{a}] ganan[bar{a}]patteh \ eva[tilde{n}]ca [acute{s}]aktatvam ev[bar{a}]stu s[bar{a}]dhutvam iti naiy[bar{a}]yikam[bar{a}]m[bar{a}][dot{m}]s[bar{a}]d[bar{i}]n[bar{a}] [dot{m}] matam. Kaundabhatta speaks of verbal communication and so on (vyavah[bar{a}]r[bar{a}]deh), since with respect to sadhu terms not only usage but also grammar and lexicon have authority. As can be seen the arguments are those considered earlier ([ss]4.2.1). The claim that the same s[bar{a}]dhu and apabhra[dot{m}][acute{s}]a terms are differentiated in that the former are used everywhere is not acceptable even from P[bar{a}]ninian tradition, as N[bar{a}]ge[acute{s}]a later notes (see [ss]4.3.2, with note 118).

(103.) VBhS 298-99: id[bar{a}]n[bar{i}][dot{m}] svamatam [bar{a}]ha v[bar{a}]cakatv[bar{a}]vi[acute{s}]ese ceti \ aya[dot{m}] bh[bar{a}]vah : apabhra[dot{m}][acute{s}][bar{a}]n[bar{a}]m a[acute{s}]aktatve tato bodha eva na sy[bar{a}]t \ na ca s[bar{a}]dhusmaran[bar{a}]t tato bodhah \ t[bar{a}]n avidus[bar{a}][dot{m}] p[bar{a}]mar[bar{a}]n[bar{a}]m api bodh[bar{a}]t tes[bar{a}][dot{m}] s[bar{a}]dhor abodh[bar{a}]c ca \ na ca [acute{s}]aktibhram[bar{a}]t tebhyo bodhah bodhakatvasy[bar{a}]b[bar{a}]dhena tadgrahasy[bar{a}]bhramatv[bar{a}]t. Cf. VBh 220: atrocyate apahra[dot{m}][acute{s}][bar{a}]n[bar{a}]m abodhakatve tato bodha eva na sy[bar{a}]t \ na ca s[bar{a}]dhusmaran[bar{a}]d bodhah \ tam avidus[bar{a}][dot{m}] j[bar{a}]yam[bar{a}]natv[bar{a}]t tasm[bar{a}]d aj[bar{a}]yam[bar{a}]natv[bar{a}]c ca[ldots]n[bar{a}]pi [acute{s}]aktibhram[bar{a}]t tato bodhah bodhakatvasy[bar{a}]b[bar{a}]dhena tajj[tilde{n}][bar{a}]nasy[bar{a}]bhramatv[bar{a}]t.

(104.) These speakers are usually viewed as incapable of pronouncing in certain ways viguna 'lacking a quality') due to some fault in their articulatory organs. E.g., Darpana, Par[bar{i}]ks[bar{a}] 300: vaigunya[dot{m}] ca Karan[bar{a}]pa[bar{a}]avar[bar{u}]pam. Vrsabha, on the other hand, says that the speakers can be faulty also due to inattention, laziness, and deprivations such as thirst, and hunger: Paddhati 232.25-233.1: vigunesv iti : [a][acute{s}]akti[h] pram[bar{a}]d[bar{a}]lasyatrtksudh[bar{a}]di.

(105.) VbhS 300: ukta[dot{m}] ca v[bar{a}]kyapadiye : p[bar{a}]rampary[bar{a}]d apabhra[dot{m}][acute{s}][bar{a}] vigunesv abhidh[bar{a}]trsu \ prasiddhim [bar{a}]gat[bar{a}] yesu tes[bar{a}][dot{m}] s[bar{a}]dhur av[bar{a}]cakah \ daiv[bar{i}] v[bar{a}]g vyavak[bar{i}]rneyam a[acute{s}]aktair abhidh[bar{a}]trbhih anityadar[acute{s}]in[bar{a}][dot{m}] tv asmin v[bar{a}]de buddhiviparyayah iti. The k[bar{a}]rik[bar{a}]s cited are the next-but-final verses of the first k[bar{a}]nda of the V[bar{a}]kyapad[bar{a}]ya (1.181-82). VBh 220 introduces the same verses with ukta[dot{m}] hi v[bar{a}]kyapad[bar{a}]ye. Kaundabhatta does not give an exegesis of the first k[bar{a}]rik[bar{a}] cited but does say that by av[bar{a}]cakah is meant 'which does not produce a verbal cognition of a meaning' (abodhakah), and he notes that this is based on the position that the capacity in question consists only in producing a cognition. VBh 220: av[bar{a}]cakah abodhakah bodhakatvasyaivoktar[bar{i}]ty[bar{a}] [acute{s}]aktitv[bar{a}]d iti bh[bar{a}]vah; VBhS 300: av[bar{a}]cakah abodhakah. I will take up the second verse in connection with the V[bar{a}]kyapad[bar{i}]yavrtti interpretation (see [ss]4.4.5).

(106.) VBhs 301-2: nanv eva[dot{m}] s[bar{a}]dhut[bar{a}] tes[bar{a}][dot{m}] sy[bar{a}]d ity ata [bar{a}]ha niyama iti punyajananabodhan[bar{a}]ya s[bar{a}]dh[bar{a}]n[bar{a}][dot{m}] s[bar{a}]dhubhir bh[bar{a}]sitavyam iti vidhih p[bar{a}]pajananabodhan[bar{a}]ya n[bar{a}]s[bar{a}]dubhir iti nisedhah \ tath[bar{a}] ca punyajananayogyatva[dot{m}] s[bar{a}]dhutvam \ tatra p[bar{a}]pajananayogyatvam as[bar{a}]dhutvam.

(107.) LM 125: s[bar{a}] ca [acute{s}]aktih s[bar{a}]dhanesv iv[bar{a}]pabhra[dot{m}][acute{s}]esv api [acute{s}]aktigr[bar{a}]haka[acute{s}]iromaner vyavah[bar{a}]rasya tulyatv[bar{a}]t.

(108.) LM 125-26: na ca s[bar{a}]dhusmaran[bar{a}]t tato bodhah[ldots]iti v[bar{a}]cyam.

(109.) LM 125-26: te s[bar{a}]dhusv anum[bar{a}]nena pratyayotpattihetavah \ amb[bar{a}]mbeti yad[bar{a}] b[bar{a}]lah [acute{s}]iksyam[bar{a}]nah \ avyakta[dot{m}] tadvid[bar{a}][dot{m}] tena vyakte bhavti ni[acute{s}]cayah \\ iti haryukteh \ anum[bar{a}]nam atra j[tilde{n}][bar{a}]nam \ s[bar{a}]dhuvisayasmaranenety arthah tadvid[bar{a}][dot{m}] s[bar{a}]dhuvid[bar{a}][dot{m}] \ as[bar{a}]dhoh s[bar{a}]dhuprakrtikatv[bar{a}]t s[bar{a}]dr[acute{s}]yena tajj[tilde{n}][bar{a}]nam.

(110.) LM 126: s[bar{a}]dhusmaranam, vin[bar{a}]pi bodh[bar{a}]nubhav[bar{a}]t tadv[bar{a}]cakas[bar{a}]dhu[acute{s}]abdam aj[bar{a}]nat[bar{a}][dot{m}] bodh[bar{a}]n[bar{a}]patte[acute{s}] ca. The second objection could equally apply with respect to persons who know only apabhra[dot{m}][acute{s}]a usage. In my presentation, I have followed Vaidyan[bar{a}]tha's Kal[bar{a}] commentary on the assumption that, being N[bar{a}]ge[acute{s}]a's direct student, he reflects the author's intent. The Kal[bar{a}] notes that there are two sorts of persons who are learned in Sanskrit usage (vyutpann[bar{a}]h): those who know each particular Sanskrit word and those who are lacking in such detailed knowledge. The second objection is intended for the second type of speaker. Kal[bar{a}] 127: nanu vyutpann[bar{a}] dvividh[bar{a}]h tattadv[bar{a}]ckasa[dot{m}]skrative[acute{s}]aj[tilde{n}][bar{a}]nav antah tadvikal[bar{a}][acute{s}] ca tatr[bar{a}]dy[bar{a}]n[bar{a}][dot{m}] tath[bar{a}]bodhe 'pi dvitiy[bar{a}]nupapattir datt[bar{a}] \ s[bar{a}] na yukt[bar{a}] s[bar{a}]m[bar{a}]nyaj[tilde{n}][bar{a}]pakatvena r[bar{u}]pena tats[bar{a}]dhusmaran[bar{a}]d bodhasambhav[bar{a}]t ata [bar{a}]ha : tadartheti.

(111.) Ku[tilde{n}]jik[bar{a}] 125: etadarhabodhaka[dot{m}] ki[tilde{n}]cit s[bar{a}]dhupada[dot{m}] bhavisyat[bar{i}]ty anum[bar{a}]n[bar{a}]d bodha iti mata[dot{m}] nir[bar{a}]karoti : n[bar{a}]rtheti

(112.) LM 126: tadarthaj[tilde{n}][bar{a}]pakatvena smarana[dot{m}] tu n[bar{a}]rthopasth[bar{a}]paka[dot{m}] [acute{s}]aktat[bar{a}]vacchedak[bar{a}]nup[bar{u}]rvyagrah[bar{a}]t.

(113.) LM 126: tadv[bar{a}]cakasarvan[bar{a}]masmaran[bar{a}]nanubhav[bar{a}]c ca.

(114.) In the present context, N[bar{a}]ge[acute{s}]a brings into play gagar[bar{i}] used in the same sense as ghata 'jug, water pot, jar'. It is also possible in other contexts to consider this an apabhra[dot{m}][acute{s}]a for gargar[bar{i}], known as a s[bar{a}]dhu term synonymous with manthan[bar{i}], referring to a vessel in which curds are put and churned; cf. Amarako[acute{s}]a 2.9.74: manthan[bar{i}] gargar[bar{i}] same.

(115.) LM 126: ucc[bar{a}]ritasyaiva bodhakatvena smrtas[bar{a}]dhuto bodh[bar{a}]sambhav[bar{a}]c ca.

(116.) LM 126-27: na ca [acute{s}]aktibhram[bar{a}]d bodhah p[bar{u}]rvap[bar{u}]rvabhram[bar{a}]c cottarottarabharama iti p[bar{a}]mar[bar{a}]n[bar{a}][dot{m}] [acute{s}]aktyagrahe 'pi tadbhramopapattir iti v[bar{a}]cyam ghatatvavi[acute{s}]istaghat[bar{a}]dir[bar{a}]p[bar{a}]rthanir[bar{a }]pit[bar{a}]y[bar{a}] ghat[bar{a}]dipadavrttitvena grh[bar{i}]t[bar{a}]y[bar{a}] bhinn[bar{a}]nup[bar{u}]rv[bar{i}]katvar[bar{u}]pavi[acute{s}]esadar[ acute{s}]anasattvena s[bar{a}]dh[bar{a}]ranadharmadar[acute{s}]an[bar{a}]bh[bar{a}]vena ca gagar[bar{i}]pad[bar{a}]dau bhram[bar{a}]nupapatteh.

(117.) LM 127-28: yadi tu a[acute{s}]akty[bar{a}] kenacid gagar[bar{i}]ti prayukte 'mb[bar{a}]mbety[bar{a}]d[bar{a}]v iva ghata iti s[bar{a}]dhu[acute{s}]abdasmaran[bar{a}]t prayojyasya bodhe tatasthasya gagar[bar{i}][acute{s}]abd[bar{a}]d ev[bar{a}]sya bodha iti bhrame[bar{a}]adyasya [acute{s}]aktibhramas tanm[bar{a}]laka[acute{s}] c[bar{a}]nyes[bar{a}]m api \ tad ukta[dot{m}] harin[bar{a}] : t[bar{a}]d[bar{a}]tmyam upagamyeva [acute{s}]abd[bar{a}]rthasya prak[bar{a}][acute{s}]ak[bar{a}]h \ iti \ iva[acute{s}]abdena tadupagamasya bhramatva[dot{m}] s[bar{u}]citam. N[tilde{a}]ge[acute{s}]a also has the defender of this view invoke JS and [acute{S}]abara (see [ss]4.2.1), LM 130: tad ukta[dot{m}] jaimini[bar{a}]: tada[acute{s}]akti[acute{s}] c[bar{a}]nur[bar{u}]patv[bar{a}]d iti tadbh[bar{a}]syakrt[bar{a}] ca tatroktar[bar{i}]ty[bar{a}] [acute{s}]aktibhrama ity ucyate.[ldots] The section cited here, beginning with yadi tu[ldots]ity ucyate ("But if the following is said[ldots]" ) gives the argument refuted immediately afterwards (see next note).

(118.) LM 130-31:[ldots]tad[bar{a}] pratyeka[dot{m}] tattatsa[dot{m}]skrtasya tattadapabhra[dot{m}][acute{s}]ena vinigaman[bar{a}]virah[bar{a}]t tesu [acute{s}]aktih \ na ca sakalade[acute{s}]a{acute{s}]istaparigrh[bar{i}]tatva[dot{m}] vinigamakam \ [acute{s}]avatir gatikarm[bar{a}] kambojesu vik[bar{a}]ra evainam [bar{a}]ry[bar{a}] bh[bar{a}]santa ity[bar{a}]dibh[bar{a}]syar[bar{i}]ty[bar{a}] tattadde[acute{s}]aniyatasa[dot{m}]skrtesu [acute{s}]aktisiddhyan[bar{a}]patteh. The reference is to a Mah[bar{a}]bh[bar{a}]sya passage (I.9.24-10.1), where Pata[tilde{n}]jali illustrates dialectal usage. N[bar{a}]ge[acute{s}]a goes on to give an additional argument, involving Prakrit poetry, which I omit.

(119.) LM 131: ata eva str[bar{i}][acute{s}][bar{u}]drab[bar{a}]l[bar{a}]n[bar{a}][dot{m}] prayukte s[bar{a}]dh[bar{a}]v arthasa[dot{m}][acute{s}]aye tadapabhra[dot{m}][acute{s}]ena nirnayah. Presumably N[bar{a}]ge[acute{s}]a reflects the kind of situation found nowadays among panditas, who speak and write Sanskrit with what amounts to native control but normally interact in a vernacular. In Varanasi, even rickshaw drivers control a register of Hindi and Bhojpuri that is fairly Sanskritized, so that they can understand a great many Sanskrit words. On the other hand, even the best pandita sometimes cannot make clear in Sanskrit alone something he wishes to explain and then resorts to invoking a term from the vernacular ([ldots] iti bh[bar{a}]s[bar{a}]y[bar{a}]m). The situation N[tilde{a}]ge[acute{s}]a speaks of was known much earlier, since the Vrtti on VP 1.181 speaks of this: see note 136. See Aklujkar 1996, Deshpande 1979, Hock and Pandharipande 1976 for recent discussions of issues concerning the use of Sanskrit at various times in Indian culture.

(120.) LM 139: s[bar{a}]dhutva[dot{m}] ca vy[bar{a}]karanavya[dot{n}]gyo 'rthavi[acute{s}]ista[acute}]sabdanistah punyajanakat[bar{a}]vacchedako j[bar{a}]tivi[acute{s}]esah.

(121.) PLM 49: s[bar{a}]dhutva[dot{m}] ca vy[bar{a}]karan[bar{a}]nv[bar{a}]khyeyatva[dot{m}] punyajanakat[bar{a}]vacchedakadharmavattva[dot{m}] v[bar{a}] tadbhinnatvam as[bar{a}]dhutvam. In accordance with manuscript evidence, I have emended tadbhinnam of Kapil Deva's edition tadbhinnatvam, which appears in my critical edition ([ss]19).

(122.) VP 1.175-176): [acute{s}]abdah sa[dot{m}]sk[bar{a}]rah[bar{i}]no yo gaur iti prayuyuksite \ tam apabhra[dot{m}][acute{s}]am icchanti vi[acute{s}]ist[bar{a}]rthanive[acute{s}]inam, \\ asvagony[bar{a}]dayah [acute{s}]abd[bar{a}]h s[bar{a}]dhavo visay[bar{a}]ntare nimittabhed[bar{a}]t sarvatra s[bar{a}]dhutva[dot{m}] ca vyavasthitam. I have kept the reading prayuyuksite, which appears in editions other than Rau's. This has support from manuscripts as well as commentaries and is also syntactically preferable. The reading prayuyuksyate is Rau's compromise accounting for prayuyuksate, prayuyu[dot{m}]ksate, and prayuyuksite found in k[bar{a}]rik[bar{a}] manuscripts.

(123.) VPVr 1.176/140 (229.9): [bar{a}]vapone gon[bar{i}]ti svaviprayog[bar{a}]bhidh[bar{a}]ne c[bar{a}]sva ity etayor avasthita[dot{m}] s[bar{a}]dhutvam.

(124.) VPVr 1.176/140 (230.1-3): tath[bar{a}] s[bar{a}]sn[bar{a}]dimati hres[bar{a}]dili[dot{n}]ge ca nimitt[bar{a}]ntar[bar{a}]t pravrttayor anyatra visaye labdhasa[dot{m}]sk[bar{a}]rayoh s[bar{a}]dhutvam eva vij[tilde{n}][bar{a}]yate \ gon[bar{i}]va gaur iti bahuks[bar{i}]radh[bar{a}]ran[bar{a}]divisay[bar{a}]d [bar{a}]vapanatvas[bar{a}]m[bar{a}]ny[bar{a}]d abhidh[bar{i}]yate tath[bar{a}]vidyam[bar{a}]na[dot{m}] svam asya so 'yam asva iti. The same is said in the D[bar{i}]pik[bar{a}] (I.9.26-10.3) on the Mah[bar{a}]bh[bar{a}]sya passage (I.2.19) where the verse yas tu prayu[dot{n}]kte ku[acute{s}]alovi[acute{s}]ese[ldots]is cited: sa eva [acute{s}]abdo 'rthavi[acute{s}]ese kasmim[acute{s}]cit s[bar{a}]dhuh kasmi[dot{m}][acute{s}]cid as[bar{a}]dhur ity [bar{a}]khy[bar{a}]yate \ yath[bar{a}] goni[acute{s}]abdah s[bar{a}]sn[bar{a}]dimaty as[bar{a}]dhur tath[bar{a}][acute{s}]va[acute{s}]abdah kesar[bar{a}]dimati s[bar{a}]dhur na nihsva iti \ asva iti nirdh[bar{a}]ne s[bar{a}]dhur naika[acute{s}]aph[bar{a}]dilaksane \ yadi tu gon[bar{i}][acute{s}]abdo 'pi nimitt[bar{a}]ntar[bar{a}]t s[bar{a}]sn[bar{a}]dimati prayujyate--gon[bar{i}]va gon[bar{i}]ti--s[bar{a}]dhur eva sy[bar{a}]t a[acute{s}]ve v[bar{a}]sva[acute{s}]abda[dot{m}] dhan[bar{a}]bh[bar{a}]vadv[bar{a}]raka[dot{m}] prayu[tilde{n}]j[bar{i}]ta sa s[bar{a}]dhur eva. The Vrtti on VP 1.175/139 (229.5-6: te ca s[bar{a}]sn[bar{a}]dimaty eva labdhasvar[bar{a}]p[bar{a}]h s[bar{a}]dhutva[dot{m}] vijahati arth[bar{a}]ntare tu prayujyam[bar{a}]n[bar{a}]h s[bar{a}]dhava eva vij[tilde{n}][bar{a}]yante \ na hy etes[bar{a}][dot{m}] r[bar{a}]pam[bar{a}]trapratibaddham as[bar{a}]dhutvam) ends on the same note: terms considered corrup tions of others lose their status as s[bar{a}]dhu terms only when they have a particular form used with respect to a particular meaning, and the very same forms are recognized as s[bar{a}]dhu when used in some other meaning; their not being sadhu is not linked solely to form.

(125.) VPVr 1.175/139 (229.1-5): [acute{s}]abdaprakrtir apabhra[dot{m}][acute{s}]a iti sa[dot{n}]grahak[bar{a}]rah n[bar{a}]prakrtir apabhra[dot{m}][acute{s}]ah svatantrah ka[acute{s}]cid vidyate \ sarvasyaiva hi s[bar{a}]dhur ev[bar{a}]pabhra[dot{m}][acute{s}]asya prakrtih \ prasiddhes tu r[bar{u}]dhit[bar{a}]m [bar{a}]padyam[bar{a}]n[bar{a}]h sv[bar{a}]tantryam eva kecid apabhra[dot{m}][acute{s}][bar{a}] labhante \ tatra gaur iti prayoktavye '[acute{s}]akty[bar{a}] pram[bar{a}]d[bar{a}]dibhir v[bar{a}] g[bar{a}]vy[bar{a}]dayas tatprakrtayo 'pabhra[dot{m}][acute{s}][bar{a}]h prayujyante.

(126.) VP 1.177: te s[bar{a}]dhusv anum[bar{a}]nena pratyayotpattihetavah \ t[bar{a}]d[bar{a}]tmyam upagamyeva [acute{s}]abd[bar{a}]rthasya prak[bar{a}][acute{s}]ak[bar{a}]h. As Vrsabha notes, the k[bar{a}]rik[bar{a}] is susceptible to different interpretations that depend on different syntactic connections. I consider the interpretation which takes anum[bar{a}]nena in construction with s[bar{a}]dhusu to be preferable both in that it maintains the integrity of the half-verse and in that the Vrtti on VP 1.180 agrees with this (see note 135).

(127.) Such gestures are considered to convey certain meanings without one's having to use words. In the Mah[bar{a}]bh[bar{a}]sya on 2.1.1 (I.363.25-26) and 2.1.34-35 (I.388.3-4), Pata[tilde{n}]jali remarks that many meanings are understood without the use of words: antarena khalv api [acute{s}]abdaprayoga[dot{m}] bahavo 'rth[bar{a}]h gamyante 'ksinikocaih p[bar{a}]nivih[bar{a}]rai[acute{s}] ca[ldots]

(128.) VPVr 1.177/141 (230.8--10): apabhra[dot{m}][acute{s}][bar{a}] hi s[bar{a}]dh[bar{u}]n[bar{a}][dot{m}] [acute{s}]abd[bar{a}]n[bar{a}][dot{m}] visaye prayujyam[bar{a}]n[bar{a}] yathaiv[bar{a}]ksinikoc[bar{a}]dayah paricay[bar{a}]d upagrh[bar{i}]tasvar[bar{u}]p[bar{a}] iva prasiddh[bar{a}]s tath[bar{a}] s[bar{a}]dhupran[bar{a}]dikay[bar{a}]rtha[dot{m}] praty[bar{a}]yayanti. The same parallel is invoked in the Vrtti on VP 1.24--26 (72.3), 1.183 (235.5).

(129.) Paddhati 230.25--231.1: yath[bar{a}]ksinikoc[bar{a}]dayo [na] s[bar{a}]ks[bar{a}]c chabd[bar{a}]rtha[dot{m}] praty[bar{a}]yayanti api tu p[bar{a}]rva[dot{m}] sa[dot{n}]ketav[bar{a}]kyam[ldots]upagrh[bar{i}]tasvar[bar{a}]p[bar{a }] iveti : tasya sa[dot{n}]ketav[bar{a}]kyasya svar[bar{a}]pam [bar{a}]tmani nive[acute{s}]ya tadr[bar{u}]pat[bar{a}]m [bar{a}]padya s[bar{a}]ks[bar{a}]d iva pratip[bar{a}]dayanti.

(130.) VPVr 1.177/141 (230,9--10): tatra s[bar{a}]ks[bar{a}]d abhidh[bar{a}]na[dot{m}] neti [acute{s}]lok[bar{a}]ntaropany[bar{a}]sah.

(131.) Vrtti, introduction to VP 1.178/142 (231.1--3): atha kasm[bar{a}]d ete go[acute{s}]abdasya g[bar{a}]vy[bar{a}]dayah pary[bar{a}]y[bar{a}] na vij[tilde{n}][bar{a}]yante \ na hi [acute{s}]istasam[bar{a}]c[bar{a}]raprasiddher anyad etatprak[bar{a}]resu smrtinibandhanesv arthesu nimittam abhidh[bar{i}]yate \ g[bar{a}]vy[bar{a}]daya[acute{s}] cet pary[bar{a}]y[bar{a}]h syur ete 'pi [acute{s}]istair laksanair anugamyeran prayujyera[dot{m}][acute{s}] ca.

(132.) Vrtti, introduction to VP 1.178/142 (231.3-5): ya[acute{s}] ca pratyakspaksena prayojakesv arthesv abhidheyesu pravartate sa s[bar{a}]dhuh \ s[bar{a}]ks[bar{a}]t tu prayojaka[dot{m}] v[bar{a}]cyam arthar[bar{a}]pa[dot{m}] s[bar{a}]dhubhih praty[bar{a}]yyate. Aklujkar suggests emending to as[bar{a}]ks[bar{a}]t and as[bar{a}]dhubhih. On the other hand, tu can be justified, since there is a contrast in that the second sentence deals with the other side of the coin: the meaning that is to be signified in contrast to the term which signifies.

(133.) Vrtti, introduction to VP 1.178/142 (231.5), VP 1.178: tasm[bar{a}]d [bar{a}]ha : na [acute{s}]istair anugamyante pary[bar{a}]y[bar{a}] iva s[bar{a}]dhavah \ te yatah smrti[acute{s}][bar{a}]strena tasm[bar{a}]t s[bar{a}]ks[bar{a}]d av[bar{a}]cak[bar{a}]h.

(134.) VP 1.179--181: amb[bar{a}]mbeti yad[bar{a}] b[bar{a}]lah [acute{s}]iksam[bar{a}]nah prabh[bar{a}]sate \ avyakta[dot{m}] tadvid[bar{a}][dot{m}] tena vyakte bhavati ni[acute{s}]cayah \eva[dot{m}] s[bar{a}]dhu prayoktavye yo 'pabhra[dot{m}]sah prayujyate \ tena s[bar{a}]dhuvyavahitah ka[acute{s}]cid artho 'bhidh[bar{i}]yate \\ p[bar{a}]rampary[bar{a}]d apabhra[dot{m}][acute{s}][bar{a}] vigunesv abhidh[bar{a}]trsu prasiddhim [bar{a}]gat[bar{a}] yena tes[bar{a}][dot{m}] s[bar{a}]dhur av[bar{a}]cakah. I have accepted the readings amb[bar{a}]mbeti and prabh[bar{a}]sate found in most editions, including Aklujkar's. Rau has ambvambv iti and [acute{s}]iksam[bar{a}]no 'pabh[bar{a}]sate. See [ss]4.3.1, with notes 98--99.

(135.) VPVr 1.180/144 (232.8-10): sa[dot{n}]k[bar{i}]rn[bar{a}]y[bar{a}][dot{m}] v[bar{a}]ci s[bar{a}]dhuvisaye 'pa[acute{s}]abd[bar{a}]h prayujyante \ taih [acute{s}]ist[bar{a}] laksanavidah s[bar{a}]dh[bar{u}]n pratipadyante tair eva s[bar{a}]dhubhis tadartham abhidh[bar{i}]yam[bar{a}]na[dot{m}] pa[acute{s}]yanti \ anum[bar{a}]na[dot{m}] tu dh[bar{u}]ma iv[bar{a}]gner as[bar{a}]dhur itares[bar{a}]m. K. A. Subramania Iyer and Raghun[bar{a}]tha [acute{S}]arm[bar{a}] both have anum[bar{a}]nas tu in the Vrtti text, but this is an error. The Paddhati correctly has anum[bar{a}]nam, which appears also in Aklujkar's edition. I have interpreted anum[bar{a}]nam here as a derivate signifying an instrument.

(136.) VPVr 1.181/145 (233.1-4): ih[bar{a}]bhy[bar{a}]s[bar{a}]t str[bar{i}][acute{s}][bar{u}]drac[bar{a}]nd[bar{a}]l[bar{a}]dibhir apabhra[dot{m}][acute{s}][bar{a}]h prayujyam[bar{a}]n[bar{a}]h tath[bar{a}] pram[bar{a}]dyatsu vaktrsu r[bar{u}]dhim up[bar{a}]gat[bar{a}] yena tair eva prasiddhataro vyavah[bar{a}]rah sati ca s[bar{a}]dhuprayog[bar{a}]t sa[dot{m}][acute{s}]aye yas tasy[bar{a}]pabhra[dot{m}][acute{s}]as tena samprati nirnayah kriyate tam eva c[bar{a}]s[bar{a}]dhu[dot{m}] v[bar{a}]caka[dot{m}] pratyaksapakse manyante s[bar{a}]dhu[dot{m}] c[bar{a}]num[bar{a}]napakse vyavasth[bar{a}]payanti. The k[ddot{a}]rik[bar{a}] directly states that since apabhra[dot{m}][acute{s}]as had become established among such speakers, for them a sadhu term does not signify (s[bar{a}]dhur av[bar{a}]cakah).

(137.) Vrsabha (Paddhati 233.20: vyatik[bar{i}]rneti : sambhinn[bar{a}] k[bar{a}]lusyam upan[bar{i}]t[bar{a}]) first glosses vyatik[bar{i}]rn[bar{a}] with sambhinn[bar{a}] ('mixed, identical'), then adds k[ddot{a}]lusyam upan[bar{i}]t[bar{a}] 'brought to the state of being dirty'. The Vrtti also interprets vyatik[bar{i}]rn[bar{a}] to mean 'mixed', since it uses sank[bar{i}]ryam[bar{a}]n[bar{a}] (see note 140) in speaking of speech becoming mixed with apabhra[dot{m}][acute{s}]as. Elsewhere, Bhartrhari refers to the impurities that affect speech and for which grammar is the cure (VP 1.14ab: tad dv[bar{a}]ram apavargasya v[bar{a}][dot{n}]mal[bar{a}]na[dot{m}] cikitsitam) as well as of the impurities affecting the body, speech and the mind, which are purified through teachings of medicine, grammar, and those teachings that concern the inner self (VP 1.174: k[bar{a}]yav[bar{a}]gbuddhivisay[bar{a}] ye mal[bar{a}]h samavasthit[bar{a}]h cikits[bar{a}]laksan[bar{a}]dhy[bar{a}]tma[acute{s}][bar{a}]strais tes[bar{a}][dot{m}] vi[acute{s}]uddhayah). These impurities of speech are apabhra[dot{m}][acute{s}]as. Accordingly, I have used "defiled" here.

(138.) VP 1.182: daiv[bar{i}] v[bar{a}]g vyatik[bar{i}]rneyam a[acute{s}]aktair abhidh[bar{a}]trbhih anityadar[acute{s}]in[bar{a}][dot{m}] tv asmin v[bar{a}]de buddhiviparyayah.

(139.) VP 1.183: ubhayes[bar{a}]m avicched[bar{a}]d anya[acute{s}]abdavivaksay[bar{a}] \ yo 'nyah prayujyate [acute{s}]abdo na so 'rthasy[bar{a}]bhidh[bar{a}]yakah. The Paddhati (234.20-21: avicched[bar{a}]d iti : s[bar{a}]dhvas[bar{a}]dhuvibh[bar{a}]gasmaranasy[bar{a}]vicched[bar{a }]t) relates the continuity to the distinction made between s[bar{a}]dhu and as[bar{a}]dhu terms: the recollection of these being distinct is without interruption. Raghun[bar{a}]tha [acute{S}]arm[bar{a}] (Amb[bar{a}]kartri 1.183/155: ubhayes[bar{a}][dot{m}] [acute{s}]abd[bar{a}]n[bar{a}]m apa[acute{s}]abd[bar{a}]n[bar{a}]n c[bar{a}]n[bar{a}]dau sa[dot{m}]s[bar{a}]re 'vicched[bar{a}]t) relates the continuity to [acute{s}]abda and apa[acute{s}]abda.

(140.) VPVr 1.182/146 (233.7-234.2): [acute{s}]r[bar{u}]yate : pur[bar{a}]kalpe sva[acute{s}]ar[bar{i}]rajyotis[bar{a}][dot{m}] manusy[bar{a}]n[bar{a}][dot{m}] yathaiv[bar{a}]nrt[bar{a}]dibhir asa[dot{n}]k[bar{i}]rn[bar{a}] v[bar{a}]g [bar{a}]sit tath[bar{a}] sarvair apabhra[dot{m}][acute{s}]aih \ s[bar{a}] tu sa[dot{n}]k[bar{i}]ryam[bar{a}]n[bar{a}] p[bar{u}]rvados[bar{a}]bhy[bar{a}]sabh[bar{a}]van[bar{a}]nusa[dot{n}] g[bar{a}]t k[bar{a}]lena prakrtir iva tes[bar{a}][dot{m}] prayokt[bar{r}]n[bar{a}][dot{m}] r[bar{u}]dhim up[bar{a}]gat[bar{a}].

(141.) The Vrtti says simply anityav[bar{a}]dinah. In view of what it says subsequently, the view in question is that the s[bar{a}]dhu terms are not eternal. This is a view explicitly mentioned in the D[bar{i}]pik[bar{a}] on Mah[bar{a}]bh[bar{a}]sya 1.6.12 (ki[dot{m}] punar nityah [acute{s}]abda [bar{a}]hosvit k[bar{a}]ryah), where Bhartrhari contrasts two views (D[bar{i}]pik[bar{a}] I.16.28-17.2): kecid eva[dot{m}] manyante : ya evaite pr[bar{a}]krt[bar{a}]h [acute{s}]abd[bar{a}]s ta evaite nity[bar{a}]h \ prakrtau bhav[bar{a}]h pr[bar{a}]krt[bar{a}]h \ anye manyante: iya[dot{m}] daivi v[bar{a}]k s[bar{a}] tu purus[bar{a}][acute{s}]akter [bar{a}]lasy[bar{a}]d v[bar{a}] prak[bar{i}]rn[bar{a}] yath[bar{a}] svast[bar{i}]ti [acute{s}]iksam[bar{a}]no b[bar{a}]lo 'nyathocc[bar{a}]rayat[bar{i}]ti "Some say that pr[bar{a}]krta terms alone are eternal. These are pr[bar{a}]krta in that they have their source in the original. Others maintain that divine speech has become mixed with impurities due to the incapacity or laziness of men, as when a child learning to say svasti pronounces this otherwise."

(142.) Commenting on VP 1.12, the Vrtti says that what is meant by v[bar{a}]cah paramo rasah 'the highest essence of speech' is the mass of words whose status as s[bar{a}]dhu is established because they signify and are sources of felicity (VPVr 42.6-7: paramo rasah : v[bar{a}]cakatv[bar{a}]bhyudayahetutv[bar{a}]c ca vyavasthitas[bar{a}]dhubh[bar{a}]vah [acute{s}]abdasam[bar{u}]ho 'bhidh[bar{i}]yate). In the present context, this very set of words is viewed differently, by those who do consider s[bar{a}]dhu terms neither eternal nor sources of merit. For them, s[bar{a}]dhu terms are not original but derivate, and the original stuff (prakrti) from which they derive as modifications (vik[bar{a}]ra) is the speech forms others call apa[acute{s}]abda and apabhra[dot{m}][acute{s}]a. Vrsabha (Paddhati 234.15-16: prakrtau bhavam iti : prakrtih svabh[bar{a}]vah apabhra[dot{m}][acute{s}][bar{a}]h sv[bar{a}]bh[bar{a}]vik[bar{a}]h) notes that prakrti signifies something's nature svabh[bar{a}]va), but immediately goes on to say that apabhra[dot{m}][acute{s}]as are natural. That is, although from the point of view of those who maintain that s[bar{a}]dhu terms are eternal and original apabhra[dot{m}][acute{s}]as are corruptions and only apparently have the status of original stuff (prakrtir iva), from another point of view these very terms are treated as established norms that are sources of artificial s[bar{a}]dhu usage.

(143.) This is a view found elsewhere, as is well known. Thus, in his commentary on K[bar{a}]vy[bar{a}]la[dot{n}]k[bar{a}]ra 2.12, Namis[bar{a}]dhu accounts for pr[bar{a}]krta in several ways. He first describes it in the usual manner, as a derivate with the taddhita suffix an added to a pada N-7, with a seventh-triplet ending, to form a derivate meaning 'located in X' (tatra bhavah: Ast[bar{a}]dhy[bar{a}]y[bar{a}] 4.3.53). He notes that pr[bar{a}]krtam thus derives from prakrti-i and means prakrtau bhavam, and that the source denoted by prakrti here is the inherent speech activity of all creatures, without the adornment endowed by grammar and such. He then says that alternatively pr[bar{a}]krta is that speech itself (pr[bar{a}]krteti : sakalajant[bar{u}]n[bar{a}][dot{m}] vy[bar{a}]karan[bar{a}]dibhir an[bar{a}]hitasa[dot{m}]sk[bar{a}]rah sahajo vacanavy[bar{a}]p[bar{a}]rah prakrtih \ tatra bhava[dot{m}] saiva v[bar{a}] pr[bar{a}]krtam), then goes on to give other explanations; see Pischel and Jh[bar{a}] 1965: 14 ([ss]16). In the Ga[ddot{u}]davaho (93ab: sayal[bar{a}]o ima[dot{m}] v[bar{a}]y[bar{a}] visanti etto ya nenti v[bar{a}]y[bar{a}]o) V[bar{a}]kpati similarly says that all languages emanate from and go back to Pr[bar{a}]krta. On tadbhava, see Kahrs 1992.

(144.) Vrsabha (Paddhati 234.16-18) remarks that sambhinnabuddhibhih refers to men who cannot discriminate between what women may be approached or not, what may be said or not, and so on, and that this amounts to speaking of heretics (n[bar{a}]stika). He is not alone in considering that sambhinnabuddhi refers to a special type of person, a n[bar{a}]stika; Kaiyata does too (Prad[bar{i}]pa II.389).

(145.) VPVr 1.182/146 (234.2-5): antiyav[bar{a}]dinas tu ye s[bar{a}]dh[bar{u}]n[bar{a}][dot{m}] dharmahetutva[dot{m}] na pratipadyante mallasamay[bar{a}]disadr[acute{s}][bar{i}]m [s[bar{a}]dhva]s[bar{a}]dhuvyavasth[bar{a}][dot{m}] manyante te prakrtau bhava[dot{m}] pr[bar{a}]krta[dot{m}] [acute{s}]abd[bar{a}]n[bar{a}][dot{m}] sam[bar{u}]ham [bar{a}]caksate vik[bar{a}]ras tu pa[acute{s}]c[bar{a}]d vyavasth[bar{a}]pitah yah sambhinnabuddhibhih purusaih svarasa[dot{m}]sk[bar{a}]r[bar{a}]dibhir nirn[bar{i}]yate.

(146.) VPVr 183/147 (235.1-5): yes[bar{a}]m api ca naiva pur[bar{a}]kalpo na ca daiv[bar{i}] v[bar{a}]g asa[dot{n}]k[bar{i}]rn[bar{a}] kad[bar{a}]cid [bar{a}]s[bar{i}]t tes[bar{a}]m api gamy[bar{a}]-gamy[bar{a}]divyavasth[bar{a}]vad iya[dot{m}] s[bar{a}]dhvas[bar{a}]dhuvyavasth[bar{a}] nityam avicchedena [acute{s}]istaih smaryate tatr[bar{a}]nya[acute{s}]abdavivaksay[bar{a}] b[bar{a}]lapral[bar{a}]pavad arthesu prayujyam[bar{a}]no yah [acute{s}]abdo r[bar{u}]dho ya[acute{s}] ca na r[bar{u}]dhas t[bar{a}]v ubh[bar{a}]v apy arthasya na v[bar{a}]cakau bhavatah \ tatra tu s[bar{a}]dhuvyavahit[bar{a}] v[bar{a}] bhavaty arthapratipattir abhy[bar{a}]s[bar{a}]d v[bar{a}] pramatt[bar{a}]n[bar{a}]m aksinikoc[bar{a}]divat sampratyayam[bar{a}]tra[ddot{m}] j[bar{a}]yate. Aklujkar's edition differs slightly from that of K. A. Subramania Iyer, which I have followed above.

(147.) For Houben's position on VP 1.183 see [ss]4.6.1.

(148.) VPVr 1.27 (82.1-4): yathaiv[bar{a}]ny[bar{a}]ni dharmas[bar{a}]dhan[bar{a}]ni vi[acute{s}]istopade[acute{s}]ap[bar{a}]rampary[bar{a}]gam[bar{a}]vic cheden[bar{a}]gat[bar{a}]ny anabhi[acute{s}]a[dot{n}]kan[bar{i}]y[bar{a}]ni vyavasthit[bar{a}]ni yath[bar{a}] ca pratisiddh[bar{a}]ni hi[dot{m}]s[bar{a}]nrtastey[bar{a}]d[bar{i}]ni a[acute{s}]ist[bar{a}]pratisiddh[bar{a}]ni ca hikkitahasitakand[bar{u}]yit[bar{a}]d[bar{i}]ni tath[bar{a}] s[bar{a}]dhvas[bar{a}]dhuvyavasth[bar{a}]nam apy anavacchinnap[bar{a}]ramparyam anabhi[acute{s}]a[dot{n}]kan[bar{i}]ya[dot{m}] yath[bar{a}]gam[bar{a}]d eva siddham iti. I have adopted here the text that agrees best with the Mah[bar{a}]bh[bar{a}]sya and the Paddhati.

(149.) s[bar{a}]dhutvaj[bar{n}][bar{a}]navisay[bar{a}] seya[dot{m}] vy[bar{a}]karanasmrtih \ avicchedena [acute{s}]ist[bar{a}]n[bar{a}]m ida[dot{m}] smrtinibandhanam. The second half of the verse is subject to different syntactic interpretations. If idam is coreferential with smrtinibandhanam and the latter is a sasth[bar{i}]tatpurusa, both refer to a composition relative to a smrti, a tradition. Under this interpretation, [acute{s}]ist[bar{a}]n[bar{a}][dot{m}] smrtinibandhanam is a construction of the type devadattasya gurukulam ("Devadatta's teacher's home") such that [acute{s}]ist[bar{a}]n[bar{a}]m is construed with smrti, referring to the recollection (smrtih = smaranam) of [acute{s}]istas. This recollection is carried on without interruption (avicchedena). That is, [acute{s}]istas continue to carry on a tradition of correct usage, which is memorialized in the grammar. On the other hand, if idam is considered to refer to the grammar (vy[bar{a}]karanam)--in accordance with vy[bar{a}]karanasmrtih of the preceding half-verse--and smrtinibandhanam is interpreted as a bahuvr[bar{i}]hi, then this refers to the grammar as a work whose cause is the continuous recollection of [acute{s}]istas.

(150.) VPVr 1.158/133 (212.8-213.2): yathaiv bhaksy[bar{a}]bhaksyagamy[bar{a}]gamyav[bar{a}]cy[bar{a}]v[bar{a}]cy[ bar{a}]divisay[bar{a}] vyavasthit[bar{a}] smrtayah y[bar{a}]su nibaddha[dot{m}] sam[bar{a}]c[bar{a}]ra[dot{m}] [acute{s}]ist[bar{a}] na vyatikr[bar{a}]manti tatheyam api v[bar{a}]cy[bar{a}]v[bar{a}]cyavi[acute{s}]esavisay[bar{a}] vy[bar{a}]karanasmrtih \ smrto hy arthah p[bar{a}]rampary[bar{a}]d avicchedena punah punar nibadhyate prasiddhasam[bar{a}]c[bar{a}]r[bar{a}]y[bar{a}][dot{m}] ca smrt[bar{a}]v anibandhana[acute{s}]abd[bar{a}]y[bar{a}][dot{m}] [acute{s}]istasam[bar{a}]c[bar{a}]r[bar{a}]vicchedenaiva smaryate. I have adopted Aklujkar's reading instead of Subramania Iyer's prasiddhasamay[bar{a}]c[bar{a}]r[bar{a}]y[bar{a}][dot{m}].

(151.) This is made also a theoretical issue: must one grant [acute{s}]akti to apa[acute{s}]abdas at the risk of prolixity or is there some way to avoid this? However, Bhartrhari does not go into this debate.

(152.) Hel[bar{a}]r[bar{a}]ja 3.3.30 (143.12-13): daiv[bar{i}] v[bar{a}]g a[acute{s}]aktair vyavak[bar{i}]rn[bar{a}] [ldots] anumitam[bar{u}]laprakrtir vidus[bar{a}][dot{m}] v[bar{a}]cik[bar{a}]. I have shown a lacuna where K. A. Subramania Iyer's edition has b[bar{a}]lavad andh[bar{a}]divad ("as that of a child, as that of a blind person"). Aklujkar suggests the emendation b[bar{a}]lapad[bar{a}]mb[bar{a}]divad ("as a child's word amba and such"). Houben says (p. 364 note 687), "Emend to b[bar{a}]lavad amb[bar{a}]divad" and translates (p. 364), "like [the word] 'amb[bar{a}]' etc. [of incompetent persons] like children." I suggest one should also consider b[bar{a}]labamb[bar{a}]divad, which has support from testimonia (see notes 99, 134).

(153.) vidus[bar{a}]m. These are clearly the [acute{s}]istas, speakers who know correct speech. As Virendra Sharma suggests (1977: 235) these same learned persons can be referred to by kai[acute{s}]cit in the k[bar{a}]rik[bar{a}], though Kaundabhatta and Bhattoji consider kai[acute{s}]cit to refer to other upholders of a particular view.

(154.) Hel. 3.3.30(143.13-14): na hi vidv[bar{a}][dot{m}]so 'rtham apa[acute{s}]abd[bar{a}]t s[bar{a}]ks[bar{a}]d avasyant[bar{i}]ti n[bar{a}]pa[acute{s}]abd[bar{a}]n[bar{a}]m arthena ka[acute{s}]cit sambandhah.

(155.) Hel. 3.3.30 (143.14-18): ata eva pur[bar{a}]kalpe 'nrt[bar{a}]dibhir iv[bar{a}]pabhra[dot{m}][acute{s}]air api rahit[bar{a}] v[bar{a}]g [bar{a}]s[bar{i}]d iti brahmak[bar{a}]nda uktam \ av[bar{a}]cak[bar{a}] apabhra[dot{m}][acute{s}][bar{a}]h \ te tu s[bar{a}]dr[acute{s}]y[bar{a}]t s[bar{a}]dhu[acute{s}]abdam anum[bar{a}]payanti tebhyo 'rthasampratyaya iti \ tath[bar{a}] ca sa[dot{n}]grahak[bar{a}]rah [acute{s}]abdaprakrtir apabhra[dot{m}]sa iti. See notes 125, 140.

(156.) Hel. 3.3.30 (143.18-144.1): adyatve tv adharmab[bar{a}]huly[bar{a}]d anrt[bar{a}]divad v[bar{a}][dot{n}]mal[bar{a}]n[bar{a}][dot{m}] r[bar{u}]dhih tath[bar{a}] c[bar{a}]vyavadh[bar{a}]ne naiv[bar{a}]pa[acute{s}]abdebhyo 'rthaprat[bar{i}]tau na mlecchitavai n[bar{a}]pabh[bar{a}]sitavai s[bar{a}]dhubhir bh[bar{a}]sitavyam iti [acute{s}][bar{a}]strapr[bar{a}]m[bar{a}]ny[bar{a}]t s[bar{a}]dh[bar{u}]n[bar{a}]m eva dharm[bar{a}][dot{n}]gatvam iti tadanusrtih [acute{s}][bar{a}]strena. "na mlecchitavai [ldots]" is an implicit reference to Mah[bar{a}]bh[bar{a}]sya I.2.7-8. The phrase tadanusrtih [acute{s}][bar{a}]strena "following after them by the grammar" is like [acute{s}]istair anugamyante in VP 1.178 (see [ss][ss]4.3.1, 4.4.3).

(157.) eva 'only': "that apa[acute{s}]abdas are preceded by an unsplit s[bar{a}]dhu[acute{s}]abda is alone concluded." The view expressed accords with Bhartrhari's emphasis on granting primacy to indivisible units from which parts are abstracted; see [ss]3.2. Similarly, in the Dravyasamudde[acute{s}]a (VP 3.2.16: v[bar{a}]cy[bar{a}] s[bar{a}] sarvap[acute{s}]abd[bar{a}]n[bar{a}][dot{m}] [acute{s}]abd[bar{a}][acute{s}] ca na prthak tatah \ aprthaktve ca sambandhas tayor n[bar{a}]n[bar{a}]tmanor iva), Bhartrari stresses the unity in the ultimate being (satt[bar{a}]), that is, in Brahman, of all words and all things signified: that ultimate original source (par[bar{a}] prakrtih) spoken of in the preceding k[bar{a}]rik[bar{a}] is what all words signify and these words themselves are not distinct from it, so that there is a relation between them only as though it were between two separate entities, although there is no true distinctness.

(158.) Hel. 3.3.30 (144.1-3): tath[bar{a}] ca bhedasy[bar{a}]bhedap[bar{u}]rvakatv[bar{a}]t purusavikalp[bar{a}]niyamen[bar{a}]nantatv[bar{a}]d apa[acute{s}]abd[bar{a}]n[bar{a}]m abhinnas[bar{a}]dhu[acute{s}]abdap[bar{u}]rvakatvam eva ni[acute{s}]ciyata iti [acute{s}]abdaprakrtir apa[acute{s}]abdah siddha iti [acute{s}]abda eva vidy[bar{a}]. Hel[bar{a}]r[bar{a}]ja literally says that the [acute{s}]abda alone is knowledge ([acute{s}]abda eva vidy[bar{a}]), but this is surely to be understood as based on a [acute{s}]abda's being the object of knowledge (vidy[bar{a}]visayah), as noted by Raghun[bar{a}]tha [acute{S}]arm[bar{a}] (Amb[bar{a}]kartr[bar{i}] 3.3.30, p. 260). The usage is comparable to Bhartrhari's speaking of a sentence meaning as a flash of knowledge (pratibh[bar{a}]), since it is the object of a single cognition.

(159.) The ultimate level of speech in Bhartrhari's system is identical with Brahman; see the literature referred to in Cardona 1976: 302 with note 359.

(160.) Hel. 3.3.30 (144.3-5): tad atra yath[bar{a}] vidy[bar{a}]vasth[bar{a}]bhinnabrahm[bar{a}]tmik[bar{a}] tath[bar{a}] s[bar{a}]dhu[acute{s}]abd[bar{a}]vasth[bar{a}] vidy[bar{a}] yath[bar{a}] ca vidy[bar{a}]y[bar{a}][dot{m}] bhedo vitathah tath[bar{a}]pabhra[dot{m}][acute{s}][bar{a}]vasth[bar{a}] v[bar{a}][dot{n}]malar[bar{u}]peti param[bar{a}]rthataditar[bar{a}]vasth[bar{a}]pekso vikalpah.

(161.) He does not also say arth[bar{a}]bhidh[bar{a}]ne "there being a signifying of a meaning."

(162.) avakarnayanti. Cf. avakarnya in [acute{S}]i[acute{s}]up[bar{a}]lavadha 15.67: abhidh[bar{a}]ya r[bar{u}]ksmam iti m[bar{a}] sma gama iti prth[bar{a}]suter [bar{i}]rit[bar{a}]m \ v[bar{a}]cam anunayapar[bar{a}][dot{m}] sa tatah sahas[bar{a}]varkarnya niriy[bar{a}]ya sa[dot{m}]sadah. "After delivering himself of the harsh statement and hearing with scorn Yudhisthira's conciliatory 'Don't go', ([acute{S}]i[acute{s}]up[bar{a}]la) quickly left the assembly." Mallin[bar{a}]tha notes that avakarnya means "after hearing without respect" (an[bar{a}]darena [acute{s}]rutv[bar{a}]). I take avarkarnayanti as an impersonal third plural of the type [bar{a}]huh '[ldots] say'. See [ss]4.6.3.

(163.) Hel. 3.3.30 (144.5-9): avidy[bar{a}]da[acute{s}][bar{a}]peksam eva sam[bar{a}]n[bar{a}]y[bar{a}]m arthagatau [acute{s}]abdena c[bar{a}]pa[acute{s}]abdena ceti bha[acute{s}]yam \ arthagat[bar{a}]v iti vacan[bar{a}]d arth[bar{a}]bhidh[bar{a}]nam apa[acute{s}]abd[bar{a}]n[bar{a}]m avakarnayanti avidy[bar{a}]y[bar{a}][dot{m}] bh[bar{u}]yas[bar{a}]pa[acute{s}]abdair vyavah[bar{a}]r[bar{a}]d arthaprat[bar{i}]tim[bar{a}]tra[dot{m}] bhaven n[bar{a}]mety arthah \ r[bar{u}]dhatv[bar{a}]t tu vyavah[bar{a}]rasy[bar{a}]vi[acute{s}]esa[dot{m}] granthak[bar{a}]ra [bar{a}]ha \ tatra ca [acute{s}][bar{a}]stra[dot{m}] niy[bar{a}]makam : s[bar{a}]dhu[acute{s}]abdair ev[bar{a}]rtho vaktavyo n[bar{a}]pa[acute{s}]abdaih eva[dot{m}] kriyam[bar{a}]nam abhyudayak[bar{a}]ri bhavat[bar{i}]ti.

(164.) This is accompanied by a note (373) in which Houben refers the reader to "'Bhartrhari and the ancient Vrtti' (forthcoming, c)" for additional discussion of the problem of authorship. His bibliography lists (p. 437) under "Houben" an entry "forthc., b 'Bhartrhari's V[bar{a}]kyapad[bar{i}]ya and the ancient Vrtti'", which one must assume is meant. An article entitled "Bhartrhari's V[bar{a}]kyapad[bar{i}]ya and the ancient Vrtti (1): The Vrtti and Vrsabhadeva's Paddhati on V[bar{a}]kyapad[bar{i}]ya 1.46a [bar{a}]tmabheda[dot{m}] / [bar{a}]tmabhedas [ldots]" and an addendum to this entitled "Postscript: A note on Pt. Raghun[bar{a}]tha [acute{S}]arm[bar{a}]'s interpretation of VP 1.46 and Vrtti" are scheduled to appear in the Annals of the Bhandarkar Oriental Research Institute. Houben kindly sent me copies of these papers, and, with the author's permission, I have discussed them in [ss]4.2.3 of Cardona, forthcoming.

(165.) Houben's translation of VP 1.182 is (p. 239), "Divine speech is nowadays (iyam) mixed up by incompetent speakers. But the propounders of the impermanent have with regard to this doctrine the opposite view."

(166.) In view of the content and the citation from the Vrtti given at the end of this paragraph, one must conclude that "182" and "181" are errors for "183" and "182."

(167.) [acute{s}]istebhya [bar{a}]gam[bar{a}]t siddh[bar{a}]h s[bar{a}]dhavo dharmas[bar{a}]dhanam arthapraty[bar{a}]yan[bar{a}]bhede viparit[bar{a}]s tv as[bar{a}]dhavah. As the Paddhati points out, as[bar{a}]dhu terms can he contrary by virtue of either of these (81.22-23: viparit[bar{a}] iti : [acute{s}]istopde[acute{s}]ap[bar{a}]ramparyen[bar{a}]navasthit[bar{a }] adharmas[bar{a}]dhana[dot{m}] v[bar{a}]).

(168.) The Paddhati (81.21--22: yady api s[bar{a}]ks[bar{a}]d anum[bar{a}]neneti ca bhedas tath[bar{a}]py arthapraty[bar{a}]yanam abhinnam s[bar{a}]dh[bar{u}]n[bar{a}]m as[bar{a}]dh[bar{u}]n[bar{a}][dot{m}] ca tulyam) points this out. Similarly, commenting on Bh I.8.21--22 (see note 68), the Pradipa notes that, although under one thesis apabhra[dot{m}][acute{s}]as do not signify directly, they nevertheless still signify indirectly, through the intermediary of s[bar{a}]dhu terms. Kaiyata goes on to note that some apabhra[dot{m}][acute{s}]as have gained established status through continuous transmission, so that they convey meanings directly, without calling s[bar{a}]dhu terms to mind. He also remarks that some maintain apabhra[dot{m}][acute{s}]as signify directly just as do s[bar{a}]dhu terms. Pr I.35: yady api s[bar{a}]ks[bar{a}]d apabhra[dot{m}][acute{s}][bar{a}] na v[bar{a}]cak[bar{a}]s tath[bar{a}]pi smaryam[bar{a}]nas[bar{a}]dhu[acute{s}]abdavyavadh[bar{a}]nen[bar{a}] rtha[dot{m}] praty[bar{a}]yayanti \ kecic c[bar{a}]pabhra[dot{m}][acute{s}][bar{a}]h paramparay[bar{a}] nir[bar{u}]dhim [bar{a}]gat[bar{a}]h s[bar{a}]dhu[acute{s}]abd[bar{a}]n asm[bar{a}]rayanta ev[bar{a}]rtha[dot{m}] praty[bar{a}]yayanti \ anye tu manyante : s[bar{a}]dhu[acute{s}]abdavad apabhra[dot{m}][acute{s}][bar{a}] api s[bar{a}]ks[bar{a}]d arthasya v[bar{a}]cak[bar{a}] iti. See [ss]4.6.4.

(169.) bh[bar{a}]syak[bar{a}]rah: III.1: 235.16 ([bar{a}]ha), 236.21-22 (manyate), 279.12, 296.18 ([bar{a}]ha), 299.4-5 (praty[bar{a}]caste), 305.18-19 (codayati), 336.23 (na praty[bar{a}]caks[bar{i}]ta), 338.25-339.1 (manyate); III.2: 30.11 (necchati), 99.22-24 ([bar{a}]ha), 104.14-15 (samarthayate), 301.20 ([bar{a}]ha), 353.11, 361.17-18 (n[bar{a}]tra niradiksat), 374.18 (pr[bar{a}]ha), 374.25 (pr[bar{a}]ha), 402.27 ([bar{a}]dide[acute{s}]a), 408.9-10 (praty[bar{a}]cakhyau); bh[bar{a}]syak[bar{a}]rena: III.1: 107.14, 201.14, 202.6-7, 328.21; III.2: 3.2, 20.2, 124.3, 127.12, 127.16-17, 138.19, 166.4, 167.20, 314.27, 336.24, 356.15, 360.21, 386.7, 410.5; bh[bar{a}]syakrt[bar{a}]: III.1: 235.17-18; III.2: 22.7-8, 287.6; bh[bar{a}]syak[bar{a}]rasya: III.1: 352.11, III.2: 285.7, 340.14-15, 349.5, 371.23-24. References are to pages and lines of K. A. Subramania Iyer 1963, 1973. These references are available from the indices to these volumes, although only the stem forms bh[bar{a}]syak[bar{a}]ra, bh[bar{a}]syakrt appear in the indices. I have given references only to passages where Hel[bar{a}]r[bar{a}]ja uses case forms of these terms, omitting places where these appear in derivates. I have al so cited in parentheses verb forms, where these occur.

(170.) It is probable that Bhartrhari used [acute{s}]akti not merely in the sense of power or capacity but more specifically in the sense of the capacity inherent in a word to signify its meaning. I say this because a passage such as the Vrtti on VP 2.226 speaks of extracting [acute{s}]akti from the meaning of a whole compound, which is associated with many [acute{s}]aktis (VPVr 2.266 [p. 247]: samud[bar{a}]y[bar{a}]rth[bar{a}]d aneka[acute{s}]akteh [acute{s}]aktyapoddh[bar{a}]rena[ldots]). That is, avyay[bar{i}]bh[bar{a}]va, tatpurusa, bahuvr[bar{i}]hi, and dvandva compounds are described by some in terms of semantics, such that they are respectively compounds whose principal meaning is that of the prior term, the last term, neither term, and both terms. The whole is associated with a single meaning but one can extract partial meanings.

(171.) Ud 1.35: nir[bar{u}]dhim [bar{a}]gat[bar{a}] iti : te ca [acute{s}]aktibhramena bodhak[bar{a}] in bh[bar{a}]vah [acute{s}]aktibhrama[acute{s}]. cettham : kenacid g[bar{a}]viti prayukte gaur iti sadhu iti s[bar{a}]dhu[acute{s}]abdasmaran[bar{a}]t prayojyasya bodhe 'pi tatasthasya g[bar{a}]vi[acute{s}]abd[bar{a}]d ev[bar{a}]sya gobodha iti bhramah tanm[bar{u}]lako 'nyes[bar{a}]m api bhrama iti.

(172.) Ud 1.35: vastuto vinigamak[bar{a}]virah[bar{a}]d bh[bar{a}]s[bar{a}][acute{s}]bdesv api [acute{s}]aktir evety [bar{a}]ha : anye tv iti. Similarly, the Ratnaprak[bar{a}][acute{s}]a on the Mah[bar{a}]bh[bar{a}]sya passage under discussion directly remarks that Pata[tilde{n}]jali's saying sam[bar{a}]n[bar{a}]y[bar{a}]m arthagatau serves to refute those pseudo-scholars who maintain that apabhra[dot{m}][acute{s}]as do not signify directly, as do s[bar{a}]dhu terms, and instead do so indirectly, through serving to recall s[bar{a}]dhu terms. This is because it conflicts with experience and with what the Bh[bar{a}]sya passage in question, based on this, has to say and also because the understanding of meaning which arises from apa[acute{s}]abdas for p[bar{a}]maras, who do not know s[bar{a}]dhu terms, cannot have the recollection of s[bar{a}]dhu terms as intermediary. RaPr 1.91: sam[bar{a}]n[bar{a}]y[bar{a}]m arthagatav iti: etena apabhra[dot{m}][acute{s}][bar{a}]n[bar{a}][dot{m}] s[bar{a}]dhu[acute{s}]abdasm[bar{a}]rakatven[bar{a}]rthabodhakatva[do t{m}] na tu s[bar{a}]ks[bar{a}]t s[bar{a}]dhu[acute{s}]abdavad iti vadantah panditammany[bar{a}] nirast[bar{a}]h anubhavena tanm[bar{u}]lakaprakrtabh[bar{a}]syena ca virodh[bar{a}]t sadhu[acute{s}]abd[bar{a}]n aj[bar{a}]nat[bar{a}][dot{m}] p[bar{a}]mar[bar{a}]n[bar{a}]m apa[acute{s}]abdair j[bar{a}]yam[bar{a}]nasy[bar{a}]rthabodhasya s[bar{a}]dhu[acute{s}]abdasmaranadv[bar{a}]rakatv[bar{a}]sambhav[bar{ a}]c ca.

(173.) Page 341, note 567: "[ldots] According to Hel[bar{a}]r[bar{a}]ja tat in taddharmanos refers to tasya sa[dot{m}]bandhasya; in my interpretation it refers to dharma in 5 and to atyantaparatantratva in 4."

(174.) n[bar{a}]bhidh[bar{a}]na[dot{m}] svadharmena sambandhasy[bar{a}]sti v[bar{a}]cakam \ atyantaparatantratv[bar{a}]d r[bar{u}]pa[dot{m}] n[bar{a}]sy[bar{a}]padi[acute{s}]yate. That is, as Hel[bar{a}]r[bar{a}]ja notes, only a genitive ending is used in conveying a relation qua relation.

(175.) atyantaparatantratv[bar{a}]t. Houben (pp. 170, 340) translates the second half of VP 3.3.4, "Because it is extremely dependent, its form cannot be pointed out"; earlier, K. A. Subramania Iyer (1971: 80, "Being extremely dependent, its own form is never cognized") also translated using "extremely" for the Sanskrit atyanta. Both would have done better using "absolutely." For, as Hel[bar{a}]r[bar{a}]ja points out, what is at issue is that a relation is absolutely dependent, so that one can never refer to it separately as a relation by means of a term other than a genitive ending. In this respect, a relation differs from a quality (guna), which, though dependent, can be referred to as a quality by means of a distinct term. For example, one can say [acute{s}]ukla[dot{m}] r[bar{u}]pam 'the color white'.

(176.) Bh II.218.14-19: katha[dot{m}] punar atasmin sa ity etad bhavati \ caturbhih prak[bar{a}]rair atasmin sa ity etad bhavati : t[bar{a}]tsthy[bar{a}]t t[bar{a}]ddharmy[bar{a}]t tats[bar{a}]m[bar{i}]py[bar{a}]t tats[bar{a}]hacary[bar{a}]d iti t[bar{a}]tsthy[bar{a}]t t[bar{a}]vat : ma[tilde{n}]c[bar{a}] hasanti girir dahyate \ t[bar{a}]ddharmy[bar{a}]t : jatina[dot{m}] y[bar{a}]nta[dot{m}] brahmadatta ity [bar{a}]ha \ brahmadatte y[bar{a}]ni k[bar{a}]ry[bar{a}]ni jatiny api t[bar{a}]ni kriyanta ity ato jat[bar{i}] brahmadatta ity ucyate tats[bar{a}]m[bar{i}]py[bar{a}]t : ga[dot{n}]g[bar{a}]y[bar{a}]m ghosah k[bar{u}]pe gargakulam \ tats[bar{a}]hacary[bar{a}]t : kunt[bar{a}]n prave[acute{s}]aya yast[bar{i}]h prave[acute{s}]ayeti. The Bh[bar{a}]sya on 6.1.37 (111.32.12) also mentions such extended usage due to Y being intended for X (t[bar{a}]darthy[bar{a}]t), and this relation too is used frequently in interpreting terms in s[bar{u}]tras. Cf. also Ny[bar{a}]yas[bar{u}]tra 2.2.62.

(177.) Hel. 3.3.6 (129.10-11): p[bar{a}]ratantrya[dot{m}] sambandhalaksanam iti sa[dot{m}]yogasamav[bar{a}]yayor dravyagun[bar{a}]disu tattv[bar{a}]t sambandha[acute{s}]abdapravrttih.

(178.) Hel. 3.3.6 (129.11): tasya sambandhasyeva dharmah p[bar{a}]ratantryalaksano yayos tau taddharm[bar{a}]nau.

(179.) Cf. Mah[bar{a}]bh[bar{a}]sya on 1.1.70(I.180.18-19): uttarapadalopo 'tra drastavyah \ tad yath[bar{a}] ustramukham iva mukha[dot{m}] yasya so 'yam ustramukhah kharamukhah eva[dot{m}] tatk[bar{a}]lak[bar{a}]las tatk[bar{a}]lah tatk[bar{a}]lasyeti. It is neither possible nor necessary to discuss here how such compounds are obtained and interpreted without assuming deletion of a term.


Abhyankar, Kashinatha Vasudev. 1962-72. The Vy[bar{a}]karana-Mah[bar{a}]bh[bar{a}]sya of Pata[tilde{n}]jali, edited by F. Kielhorn, third edition [ldots] Poona: Bhandarkar Oriental Research Institute.

Abhyankar, Kashinatha Vasudev, and Ganesh Ambadas Joshi. 1976. [Acute{S}]rijaiminipran[bar{i}]ta[dot{m}] M[bar{i}]m[bar{a}][dot{m}]s[bar{a}]dar[acute{s}]anam[ldots]([Bar{A}]n and[bar{a}][acute{s}]rama Sanskrit Series 97). Pune: [Bar{A}]nand[bar{a}][acute{s}]rama.

Aklujkar, Ashok N. 1996. "The Early History of Sanskrit as a Supreme Language." In Houben (1996a), pp. 59-85.

_____. n.d. Critical edition of the V[bar{a}]kyapad[bar{i}]ya with the Vrtti and Hel[bar{a}]r[bar{a}]ja's Prak[bar{i}]rnaprak[bar{a}][acute{s}]a on the Sambandhasamudde[acute{s}]a. [I have a computer file of this through Aklujkar's kindness.]

Bhadkamkar, H. M. 1985. The Nirukta of Y[bar{a}]ska (with Nighantu) Edited with Durga's Commentary, by [ldots] assisted by R. G.

Bhadkamkar, volume I (Bombay Sanskrit and Prakrit Series 73). Poona: Bhandarkar Oriental Research Institute. [Reprint of 1918 edition.]

Bhate, Saroja, and Johannes Bronkhorst. 1993. Proceedings of the First International Conference on Bhartrhari, University of Poona, January 6-8, 1992. Asiatische Studien 47.1. Bern: Peter Lang, 1993. (Reprint 1994, 1997: Bhartrhari Philosopher and Grammarian, Proceedings of the First International Conference on Bhartrhari, (University of Poona, January 6-8, 1992). Delhi: Motilal Barnarsidass.

Bhatt[bar{a}]ch[bar{a}]rya, Manudeva. 1985. Brhadvaiy[bar{a}]karanabh[bar{u}]sanam of [Acute{S}]ri Kaund Bhatta [A Commentary on Bhattojidikshita's Vaiy[bar{a}]karanamatonmajjanam], Edited with 'R[bar{u}]p[bar{a}]l[bar{i}]' Notes and Appendix. Varanasi: Chaukhamba Amarbharati Prakashan.

Bronkhorst, Johannes. 1987. Mah[bar{a}]bh[bar{a}]syad[bar{i}]pik[bar{a}] of Bhartrhari, fascicule IV: [Bar{A}]hnika 1. Post-Graduate and Research Department Series 28. Pune: Bhandarkar Oriental Research Institute.

Cardona, George. 1976. P[bar{a}]nini, a Survey of Research. The Hague: Mouton. [Delhi: Motilal Banarsidass, 1980.]

_____. 1983. Linguistic Analysis and Some Indian Traditions. Pandit Shripad Shastri Deodhar Memorial Lectures, first series. Pune: Bhandarkar Oriental Research Institute.

_____. 1997. P[bar{a}]nini, his Work and its Traditions, volume I: Background and Introduction, second edition. Delhi: Motilal Banarsidass.

_____. Forthcoming. Recent Research in P[bar{a}]ninian Studies. Delhi:

Motilal Banarsidass.

Deshpande, Madhav M. 1979. Sociolinguistic Attitudes in India: An Historical Reconstruction. Ann Arbor: Karoma.

Durg[hat{a}]pras[hat{a}]d and W[hat{a}]sudev Laxman [acute{S}][hat{a}]str[hat{i}] Pan[acute{s}][jhat{i}]kar. 1928. The K[hat{a}]vy[hat{a}]lank[hat{a}]ra (A Treatise on Rhetoric) of Rudrata with the Commentary of Namis[hat{a}]dhu. K[bar{a}]vyam[bar{a}]l[bar{a}], 2. Bombay: Nirnaya S[bar{a}]gar Press. [Reprinted 1983: Delhi: Motilal Banarsidass.]

Dv[bar{a}]rik[bar{a}]d[bar{a}]s [acute{S}][bar{a}]str[bar{i}], Sv[bar{a}]m[bar{i}]. 1978. [acute{S}]lokav[bar{a}]rttika of [acute{S}]ri Kum[bar{a}]rila Bhatta with the Commentary Ny[bar{a}]yaratn[bar{a}]kara of [acute{S}]ri P[bar{a}]rthas[bar{a}]rathi Mi[acute{s}]ra. Pr[bar{a}]chyabh[bar{a}]rat Series 10. Varanasi: Tara Publications.

Hock, Hans Henrich, and Rajeshwari Pandharipande. 1976. "The Sociolinguistic Position of Sanskrit in Pre-Muslim India." Studies in Language Learning 1.2: 105-38.

Houben, Jan E. M. 1992-93: "Bhartrhari's Perspectivism (3): On the Structure of the Third K[bar{a}]nda of the V[bar{a}]kyapad[bar{i}]ya." Sambodhi 18: 1-32.

_____. 1993. "Who are the Padadarsins?" In Bhate and Bronkhorst 1993. Pp. 155-69.

_____. 1996. Ideology and Status of Sanskrit: Contributions to the History of the Sanskrit Language. Brill's Indological Library, volume 13. Leiden: E. J. Brill.

Joshi, Sad[bar{a}][acute{s}]iva [acute{S}]astr[bar{i}]. 1939. The Vaiy[bar{a}]karanabh[bar{u}]sanas[bar{a}]ra by M. M. [acute{S}]r[bar{i}] Kaunda Bhatta with the Darpana Commentary by [acute{S}]ri Harivallabha, the Par[bar{i}]ks[bar{a}] Commentary by Bhairava Mi[acute{s}]ra, and a short Commentary by [acute{S}]r[bar{i}] Kri[acute{s}]na Mitra, with Ti[dot{n}]arthav[bar{a}]das[bar{a}]ra by [acute{S}]r[bar{i}] Khudd[bar{i}] Jh[bar{a}] [acute{S}]arm[bar{a}]. K[bar{a}]sh[bar{i}] Sanskrit Series, 133. Benares: Chowkhamba.

Kahrs, Eivind, 1992. "What is a tadbhava word?" IIJ 35: 225-49.

K[bar{a}]m[bar{a}]khy[bar{a}]n[bar{a}]tha Tarkav[bar{a}]g[bar{i}]ca. 1901. The Tattva-Cint[bar{a}]mani by Gangeca Up[bar{a}]dhy[bar{a}]ya, part IV-volume II: Cabda-Khanda from Vidhiv[bar{a}]da to Pram[bar{a}]na-catustaya-pr[bar{a}]m[bar{a}]nyav[bar{a}][bar{a}]da, with the Commentaries of Mathur[bar{a}]n[bar{a}]tha Tarkav[bar{a}]g[bar{i}]ca and Jayadeva Micra; and with the [bar{A}]khy[bar{a}]tav[bar{a}]da and Nabar{n}]v[bar{a}]da of Mah[bar{a}]mahop[bar{a}]dhy[bar{a}]ya Raghun[bar{a}]tha Ciromani. Bibliotheca Indica, new series, nos. 900, 908, 915, 918, 921, 927, 935, 943, 955, 960 & 977. Calcutta: The Asiatic Society of Bengal. [Reprinted 1990: Delhi: Chaukhamba Sanskrit Pratishthan.]

Kapil Dev Shastri. 1975. N[bar{a}]ge[acute{s}]abhatta-krta Vaiy[bar{a}]karana-siddh[bar{a}]nta-paramalaghuma[bar{n}]j[bar{u}]s[b ar{a}] (m[bar{u}]la grantha anuv[bar{a}]da evam sam[bar{i}]ks[bar{a}]tmakavy[bar{a}]khy[bar{a}]). Kurukshetra: Kurukshetra University.

Mukund Shastri. 1901-4. Ny[hat{a}]yasudh[hat{a}], a Commentary on Tantrav[hat{a}]rtika by Pandit S[hat{o}]meshwara Bhatta. Benares: Chowkhamba.

Nandkishore Shastri. 1934. Vaiyakarana Bhushan Sar of M. M. Kaund Bhatt with Kashika and Darpana Commentaries by Pt. Hari Shastri and Pt. Hari Vallabha Shastri, Explanatory Note by Pt. Guru Prasad Shastri. Rajasthan Sanskrit College Granthamala, 10. Benares: Bhargava Pustakalaya.

Narasimhacharya. 1973-83. Commentaires sur le Mah[bar{a}]bh[bar{a}]sya de Pata[tilde{n}]jali et le Pradipa de Kaiyata: Mah[bar{a}]bh[bar{a}]syaprad[bar{i}]pavy[bar{a}]khy[bar{a}]n[bar{a}]n i. Publications de l'Institut francais d'Indologie, 51.1-10. Pondich[acute{e}]ry Institut francais d'Indologie.

Narayana Pillai, P. K. 1951. Jaimin[bar{i}]yas[bar{u}]tr[bar{a}]rthasa[dot{n}]graha of Rsiputra Parame[bar{s}]vara, part I: Adhy[bar{a}]ya 1 P[bar{a}]da i to Adhy[bar{a}]ya III P[bar{a}]da iii. Trivandrum Sanskrit Series, 156. Trivandrum: University of Travancore.

Pischel, Richard. 1965. Comparative Grammar of the Pr[bar{a}]krt Languages by R. Pischel. Translated from the German by Subhadra Jh[bar{a}]. Delhi: Motilal Banarsidass.

Raghun[bar{a}]tha [acute{S}]arm[bar{a}]. 1963. V[bar{a}]kyapad[bar{i}]yam, Part I (Brahma-K[bar{a}]ndam), with the Commentary Svopaj[tilde{n}]avrtti by Harivrsabha and Amb[bar{a}]kartri by Pt. Raghun[bar{a}]tha [acute{S}]arm[bar{a}]. Sarasvat[bar{i}] Bhavana Grantha-m[bar{a}]l[bar{a}], 91. Varanasi: Sampurnanand Sanskrit University. [Reissued with new pagination, 1976, 1988.]

_____. 1974. V[bar{a}]kyapad[bar{i}]yam, Part III (Pada K[bar{a}]nda, J[bar{a}]ti, Dravya and Sambandha Samudde[acute{s}]a) with the Commentary Prak[bar{a}]s[bar{a}] by Helaraja and Amb[bar{a}]kartri by Pt. Raghun[bar{a}]tha [acute{S}]arm[bar{a}]. Sarasvat[bar{i}] Bhavana Grantha-m[bar{a}]l[bar{a}], 91. Varanasi: Sampurnanand Sanskrit University.

Ramseier, Yves. 1993. "Bibliography on Bhartrhari." In Bhate and Bronkhorst, pp. 235-67.

Rau, Wilhelm. 1977. Bhartrharis V[bar{a}]kyapad[bar{i}]ya: Die M[bar{u}]lak[bar{a}]rik[bar{a}]s nach den Handschriften herausgegeben und mit einem P[bar{a}]da-Index versehen. AKML XLII. 4. Wiesbaden: Franz Steiner Verlag.

Sharma, Virendra. 1977. V[bar{a}]kyapad[bar{i}]ya-sambandhasamudde[acute{s}]a Hel[bar{a}]r[bar{a}]j[bar{i}]ya vy[bar{a}]khy[bar{a}] ke prak[bar{a}][acute{s}]a mem eka vivecan[bar{a}]tmaka adhyayana (V[bar{a}]kyapad[bar{i}]ya Sambandha-Samudde[acute{s}]a: A Critical Study with Special Reference to the Commentary of Hel[bar{a}]r[bar{a}]ja). Panjab University Indological Series, 9. Hoshiarpur: Vishveshvaranand Vishva Bandhu Institute of Sanskrit and Indological Studies, Panjab University.

Subrahmanya Sastri. S. 1962. Brhat[bar{i}] of Prabh[bar{a}]kara Mi[acute{s}]ra [on the Mim[bar{a}][dot{m}]s[bar{a}] S[bar{u}]trabh[bar{a}]sya of [acute{S}]abara] with the Rjuvimal[bar{a}] Pa[bar{n}]cik[bar{a}] of [acute{S}][bar{a}]lik[bar{a}]nvtha, part III: Adhy[bar{a}]ya I P[bar{a}]das 2, 3, 4 and Adhy[bar{a}]ya II, Complete. Madras University Sanskrit Series, 24. Madras: University of Madras.

Subramania Iyer, K. A. 1963. V[bar{a}]kyapad[bar{i}]ya of Bhartrhari with the Commentary of Hel[bar{a}]r[bar{a}]ja, K[bar{a}]nda III, Part 1. Deccan College Monograph Series, 21. Poona: Deccan College.

_____. 1966. V[bar{a}]kyapad[bar{i}]ya of Bhartrhari with the Commentaries Vrtti and Paddhati of Vrsabhadeva. Deccan College Monograph Series, 32. Poona: Deccan College.

_____. 1971. The V[bar{a}]kyapad[bar{i}]ya of Bhartrhari: Chapter III, Part i. English translation. Poona: Deccan College.

_____. 1973. V[bar{a}]kyapad[bar{i}]ya of Bhartrhari with the Commentary of Hel[bar{a}]r[bar{a}]ja, K[bar{a}]nda III, Part II. Poona: Deccan College.

_____. 1983. The V[bar{a}]kyapad[bar{i}]ya of Bhartrhari, An ancient Treatise on the Philosophy of Sanskrit Grammar, Containing the Tik[bar{a}] of Punyar[bar{a}]ja and the Ancient Vrtti, K[bar{a}]nda II, with a Foreword by Ashok Aklujkar. Delhi: Motilal Banarsidass.

T[bar{a}]r[bar{a}]n[bar{a}]tha Ny[bar{a}]ya-Tarkat[bar{i}]rtha, Amarendramohan Tarkat[bar{i}]rtha and Hemantakum[bar{a}]r Tarkat[bar{i}]rtha. 1936--1944. Ny[bar{a}]yadar[acute{s}]anam with V[bar{a}]tsy[bar{a}]yana's Bh[bar{a}]sya, Uddyotakara's V[bar{a}]rttika, V[bar{a}]caspatimi[acute{s}}ra's T[bar{a}]tparyat[bar{i}]k[bar{a}] & Vi[acute{s}]van[bar{a}]tha's Vrtti. Calcutta Sanskrit Series, 18, 29. [Reprinted 1982: Kyoto: Rinsen Book Co.]

Trip[bar{a}]th[bar{i}], R[bar{a}]mapras[bar{a}]da. 1990. Vaiy[bar{a}]karanasiddh[bar{a}]ntalaghuma[tilde{n}]j[bar{u}]s[bar{a}] of [acute{S}]r[bar{i}] N[bar{a}]ge[acute{s}]abhatta [Part One] with Three Commentaries: 'Ku[tilde{n}]jik[bar{a}]' by [acute{S}]r[bar{i}] Durbal[bar{a}]c[bar{a}]rya, 'Kal[bar{a}]' by [acute{S}]r[bar{i}] B[bar{a}]lambhatta and 'Saralv' (Hindi Commentary) by [acute{S}]r[bar{i}] R[bar{a}]mapras[bar{a}]da Trip[bar{a}]thi. Ga[dot{n}]g[bar{a}]n[bar{a}]thajh[bar{a}]-Grantham[bar{a}]la, 12. Varanasi: Sampurnanand Sanskrit University.

Vedavrata. 1962--63. [acute{S}]r[bar{i}]bhagavatpata[tilde{n}]jaliviracitam Vy[bar{a}]karana-Mah[bar{a}]bh[bar{a}]syam ([acute{S}]r[bar{i}]kaiyyatakrtapradipena-nvgoj[bar{i}]bhatta-krtena- bh[bar{a}]syaprad[bar{i}]podyotena ca vibh[bar{u}]sitam). 5 volumes. Gurukul Jhajjar (Rohatak): Hariy[bar{a}]nv-S[bar{a}]hitya-Sa[dot{m}]sth[bar{a}]nam.


Amb[bar{a}]kartr[bar{i}]: see Raghun[bar{a}]tha [acute{S}]arm[bar{a}].

Bh: Mah[bar{a}]bh[bar{a}]sya: see Abhyankar.

Darpana: see Joshi, Sad[bar{a}][acute{s}]iva [acute{S}][bar{a}]str[bar{i}].

D[bar{i}]pik[bar{a}] I: see Bronkhorst.

Hel([bar{a}]rvja): see Subramania Iyer 1963, 1973.

Jaimin[bar{i}]yas[bar{u}]tr[bar{a}]wrthasangraha: see Narayana Pillai.

JS: M[bar{i}]m[bar{a}][dot{m}]s[bar{a}]s[bar{u}]tra of Jaimini: see Abhyankar and Joshi.

Kal[bar{a}]: see Trip[bar{a}]th[bar{i}].

K[bar{a}]vy[bar{a}]la[dot{n}]k[bar{a}]ra: see Durg[hat{a}]pras[hat{a}]d and Pan[acute{s}][hat{i}]kar.

Ku[tilde{n}]jik[bar{a}]: see Trip[bar{a}]th[bar{i}].

LM: Laghuma[tilde{n}]j[bar{u}]s[bar{a}]: see Trip[bar{a}]th[bar{i}].

Mbh: Mah[bar{a}]bh[bar{a}]sya: see Abhyankar.

Nirukta: see Bhadkamkar.

Ny[bar{a}]yasudh[bar{a}]: see Mukund Sh[hat{a}]stri.

Ny[bar{a}]yas[bar{u}]tra: see T[bar{a}]r[bar{a}]n[bar{a}]tha Ny[bar{a}]ya-Tarkat[bar{i}]rtha et al.

Paddhati: see Subramania Iyer 1966.

Par[bar{i}]ks[bar{a}]: see Joshi.

PLM: Paramalaghuma[tilde{n}]j[bar{u}]s[bar{a}]: see Kapil Dev Shastri.

RaPr: Ratnaprak[bar{a}][acute{s}]a on Mah[bar{a}]bh[bar{a}]sya: see Narasimhacharya.

RjuP: Rjuvimal[bar{a}] Pa[tilde{n}]cik[bar{a}]: see Subrahmanya Sastri.

[acute{S}]Bh: [acute{S}]abara's Bh[bar{a}]sya on Jaiminis[bar{u}]tras: see Abhyankar and Joshi.

[acute{S}]IV: [acute{S}]lokav[bar{a}]rttika on [acute{S}][bar{a}]barabh[bar{a}]sya: see Dv[bar{a}]rik[bar{a}]d[bar{a}]s [acute{S}][bar{a}]str[acute{i}].

TC IV2: Tattvacint[bar{a}]mani volume IV.2: see K[bar{a}]m[bar{a}]khy[bar{a}]n[bar{a}]tha Tarkav[bar{a}]g[bar{i}]ca.

TV: Tantrav[bar{a}]rttika: see Dv[bar{a}]rik[bar{a}]d[bar{a}]s [acute{S}][bar{a}]str[bar{i}].

Uddyotana: Annambhatta's Uddyotana on Kaiyata's Prad[bar{i}]pa: see Narasimhacharya.

Ud: N[bar{a}]ge[acute{s}]a's Uddyota on Kaiyata's Prad[bar{i}]pa: see Vedavrata.

VBh: Vaiy[bar{a}]karanabh[bar{u}]sana: see Bhatt[bar{a}]ch[bar{a}]rya.

VBhS: Vaiy[bar{a}]karanabh[bar{u}]sanas[bar{a}]ra: see Joshi.

VBhS-K[bar{a}][acute{s}]ik[bar{a}]: see Nandkishore Shastri.

VP: V[bar{a}]kyapad[bar{i}]ya: see Aklujkar, Rau, Subramania Iyer. [K[bar{a}]rik[bar{a}] numbers are first given according to Rau's edition, then according to Subramania Iyer's edition, with page and line numbers from the latter.]

VPTik[bar{a}]: Punyar[bar{a}]ja's commentary on V[bar{a}]kyapad[bar{i}]ya, k[bar{a}]nda 2: see Subramania Iyer 1983.

VPVr: V[bar{a}]kyapad[bar{i}]ya Vrtti: see Subramania Iyer 1966, 1983.
COPYRIGHT 1999 American Oriental Society
No portion of this article can be reproduced without the express written permission from the copyright holder.
Copyright 1999 Gale, Cengage Learning. All rights reserved.

Article Details
Printer friendly Cite/link Email Feedback
Publication:The Journal of the American Oriental Society
Date:Jan 1, 1999
Previous Article:Hommage a Kwong Hing Foon; Erudes d'histoire culturelle de la Chine.

Terms of use | Privacy policy | Copyright © 2018 Farlex, Inc. | Feedback | For webmasters