The Bloomsbury Research Handbook of Indian Philosophy of Language 9781350049161, 9781350049154, 9781350049147

Citation preview

The Bloomsbury Research Handbook of Indian Philosophy of Language

BLOOMSBURY RESEARCH HANDBOOKS IN ASIAN PHILOSOPHY Series Editors Chakravarthi Ram-Prasad, Lancaster University, UK Sor-hoon Tan, National University of Singapore Editorial Advisory Board Roger Ames, Professor of Philosophy, University of Hawai‘i, USA; Doug Berger, Associate Professor of Philosophy, Southern Illinois University, USA; Carine Defoort, Professor of Philosophy, KU Leuven, Belgium; Owen Flanagan, James B. Duke Professor of Philosophy, Duke University, USA; Jessica Frazier, Lecturer in Religious Studies, University of Kent, UK; Chenyang Li, Professor of Chinese Philosophy, Nanyang Technological University, Singapore; Ronnie Littlejohn, Professor of Philosophy, Director of Asian Studies, Belmont University, USA; Evan Thompson, Professor of Philosophy, University of British Columbia, Canada Bringing together established academics and rising stars, Bloomsbury Research Handbooks in Asian Philosophy survey philosophical topics across all the main schools of Asian thought. Each volume focuses on the history and development of a core subject in a single tradition, asking how the field has changed, highlighting current disputes, anticipating new directions of study, illustrating the Western philosophical significance of a subject and demonstrating why a topic is important for understanding Asian thought. From knowledge, being, gender and ethics to methodology, language and art, these research handbooks provide up-to-date and authoritative overviews of Asian philosophy in the twenty-first century. Available Titles The Bloomsbury Research Handbook of Chinese Philosophy and Gender, edited by Ann A. Pang White The Bloomsbury Research Handbook of Chinese Philosophy Methodologies, edited by Sor-hoon Tan The Bloomsbury Research Handbook of Contemporary Japanese Philosophy, edited by Michiko Yusa The Bloomsbury Research Handbook of Early Chinese Ethics and Political Philosophy, edited by Alexus McLeod The Bloomsbury Research Handbook of Indian Aesthetics and the Philosophy of Art, edited by Arindam Chakrabarti The Bloomsbury Research Handbook of Indian Ethics, edited by Shyam Ranganathan The Bloomsbury Research Handbook of Indian Philosophy and Gender, edited by Veena R. Howard

The Bloomsbury Research Handbook of Indian Philosophy of Language Edited by Alessandro Graheli

BLOOMSBURY ACADEMIC Bloomsbury Publishing Plc 50 Bedford Square, London, WC1B 3DP, UK 1385 Broadway, New York, NY 10018, USA 29 Earlsfort Terrace, Dublin 2, Ireland BLOOMSBURY, BLOOMSBURY ACADEMIC and the Diana logo are trademarks of Bloomsbury Publishing Plc First published in Great Britain 2020 This edition published in 2022 Copyright © Alessandro Graheli and Contributors, 2020 Alessandro Graheli has asserted his right under the Copyright, Designs and Patents Act, 1988, to be identified as Editor of this work. Cover image: Illuminated frontispiece of the Dasam Granth © British Library Board / Bridgeman Images Series design by Clare Turner All rights reserved. No part of this publication may be reproduced or transmitted in any form or by any means, electronic or mechanical, including photocopying, recording, or any information storage or retrieval system, without prior permission in writing from the publishers. Bloomsbury Publishing Plc does not have any control over, or responsibility for, any third-party websites referred to or in this book. All internet addresses given in this book were correct at the time of going to press. The author and publisher regret any inconvenience caused if addresses have changed or sites have ceased to exist, but can accept no responsibility for any such changes. A catalogue record for this book is available from the British Library. A catalog record for this book is available from the Library of Congress. ISBN: HB: 978-1-3500-4916-1 PB: 978-1-3503-5579-8 ePDF: 978-1-3500-4914-7 eBook: 978-1-3500-4913-0 Series: Bloomsbury Research Handbooks in Asian Philosophy Typeset by Deanta Global Publishing Services, Chennai, India

To find out more about our authors and books visit www.bloomsbury.com and sign up for our newsletters.

Contents List of Contributors Introduction

viii 1

Part One  Of Speech Units: Phonemes, Words, and Sentences Introduction

11

1

Linguistic Segmentation in Early Vyākaraṇa  13 Maria Piera Candotti

2

From Permanent Phonemes to Words  39 Monika Nowakowska

3

Ontology and Epistemology of Speech in Nyāya  56 Alessandro Graheli

4

The Theory of the Sphoṭa  76 Akane Saito

5

A Buddhist Refutation of Sphoṭa  108 Sara McClintock

6

The Place of Language in the Philosophy of the Recognition  135 Marco Ferrante

Part Two  Of Word Meanings Introduction

153

7

Early Nyāya on the Meaning of Common Nouns  155 Matthew R. Dasti

8

Dignāga on Relation  165 Kei Kataoka

9

Meanings of Words and Sentences in Mīmāṃsā  181 Elisa Freschi

vi

Contents

10 Śabdārtha as Sense or Reference: Dharmakīrti on Synonymy  202 Patrick McAllister 11 Semantic Relations and Causation of Verbal Knowledge  226 Alessandro Graheli 12 Human Intellect and God’s Will in Navya Nyāya Semantics  239 Yoichi Iwasaki Part Three  Of Sentence Meanings Introduction

249

13 Śālikanātha’s “Introduction” to His “Fundamentals of Sentence Meaning”  251 Andrew Ollett 14 Śālikanātha on Language Acquisition: A Study of “Vākyārthamātṛkā” II  278 Kei Kataoka 15 The Deontic Nature of Language in Mīmāṃsā and Vedānta Schools  295 Elisa Freschi 16 Speaking of the Individual: Prakāśātman’s Akhaṇḍārthavāda and the Beginnings of a Theory of Language in Classical Advaita-Vedānta  313 Hugo David 17 The Role of Intention in Gaṅgeśa’s Non-Communicational Model of Verbal Understanding  340 Yoichi Iwasaki Part Four  Of Implicatures and Figurative Meanings Introduction

351

18 Kumārila on the Role of Implicature in Sentence-Signification  353 Lawrence McCrea 19 The Intentionality of Words: Jayanta’s Syncretism of Nyāya and Mīmāṃsā  361 Alessandro Graheli 20 Rasa as Sentence Meaning  371 Andrew Ollett

Contents

vii

21 Meaning beyond Words: The Implicature Wars  396 Daniele Cuneo Chronology Glossary Annotated Bibliography of Indian Philosophy of Language Index

423 427 453 464

Contributors

Maria Piera Candotti is Lecturer at the Department “Civiltà e Forme del Sapere” at Pisa University, Italy. Daniele Cuneo is Lecturer (maître de conférences) of Sanskrit and Indian Civilization at the Department of Oriental Studies, University Sorbonne Nouvelle, Paris 3, Paris, France. Matthew R. Dasti is Associate Professor of Philosophy at Bridgewater State University, USA. Hugo David is Lecturer (maître de conférences) of Philosophy and Indian Studies at the Indology Center of the École française d’Extrême-Orient (EFEO), Pondicherry, India. Marco Ferrante is Berggruen Research Fellow in Comparative Philosophy, Wolfson College, University of Oxford, UK. Elisa Freschi is Assistant at the Department of South Asian, Tibetan, and Buddhist Studies, University of Vienna, Austria, and Principal Investigator of two projects at the Institute for the Cultural and Intellectual History of Asia, Austrian Academy of Science, Vienna, Austria: one on the theology in the Viśiṣṭādvaita Vedānta school and the other on deontic logic and Mīmāṃsā. Alessandro Graheli is Assistant at the Department of South Asian, Tibetan, and Buddhist Studies, University of Vienna, Austria, and Principal Investigator of the project “The Meaning of the Sentence in Indian Philosophy” at the Institute for the Cultural and Intellectual History of Asia, Austrian Academy of Science, Vienna. Yoichi Iwasaki is Associate Professor at the Department of Indian Philosophy, Nagoya University, Japan. Kei Kataoka is Associate Professor at the Department of Indian Philosophy, Kyushu University, Fukuoka, Japan.

Contributors

ix

Patrick McAllister is Research Fellow in Buddhist Studies at the Institute for the Cultural and Intellectual History of Asia, Austrian Academy of Science, Vienna. Sara McClintock is Associate Professor in the Department of Religion at Emory University, USA. Lawrence McCrea is Professor of Sanskrit Studies at the Department of Asian Studies, Cornell University, USA. Monika Nowakowska is Assistant Professor at the Chair of South Asian Studies, Faculty of Oriental Studies, University of Warsaw, Poland. Andrew Ollett is Assistant Professor in the Department of South Asian Languages and Civilizations at the University of Chicago, USA. Akane Saito is Overseas Research Fellow of the Japan Society for the Promotion of Science (JSPS), affiliated to the Pondicherry Centre of the École française d’Extrême-Orient, Pondicherry, India.

x

Introduction

As for an original work, how could we contrive it! The diverse position of voices is what needs to be appraised. Those very same flowers, worn in former chaplets, if composed afresh on a thread will definitely find new appeal. — Bhaṭṭa Jayanta

The Themes of This Handbook Let’s take a simple Sanskrit sentence uttered in rural India: “gām ānaya!” (“Bring the cow!”). Let’s now envision the reaction of Baby Doe (Devadatta) who observes the reaction of his father John Doe (Yajñadatta), whenever he hears “Bring the cow!” uttered by Richard Doe (Caitra), the grandfather. Witnessing the phenomenon of the appearance of a cow produced by John, whenever that particular sound is uttered, Baby Doe wonders: Why does the utterance “Bring the cow!” cause the action of bringing a cow? How do the sound “cow” and the cow relate to each other? What are the sufficient and necessary causes for the appearance of the cow after this sound is uttered? What are the causal roles of Richard and John? What are the psychological and cognitive processes at work in their respective minds? How do we learn to use such sounds to trigger similar events?

Apparently trivial sentences such as “Bring the cow!” and their related circumstances have generated centuries and millennia of philosophical debates in South Asia, mostly in Sanskrit language and about the usage of Sanskrit language. Authors have addressed these and many other related questions around

2

Introduction

language, pertaining to the heterogeneous domains of ontology, linguistics, and epistemology. Ontologically, Sanskrit authors have asked themselves about the physical and causal nature of linguistic sounds and also about this animal to which “cow” somehow relates. Is the referent “cow” an external reality, or an image in the speaker’s or hearer’s mind, or both? Is there a difference between the cow out there in the world and the image generated in the listener’s mind? What sort of relation links such a sound to the denoted animal? Is this denotandum an individual, a universal, or a combination of both? Linguistically, is /gāmānaya/ an indivisible unit? Or is it a composite whole? In this case, what are the minimal parts of this whole? Are they the words /gām/ and /ānaya/? Or, rather, the morphemes /go/, /am/, /ā/, /nī/, and /ya/? Or even the phones /g/, /ā/, /m/, etc.? In this case, does each part function in the same way? Or are there even smaller units, such as voicedness, velarness, plosiveness, and so on? Epistemically, what is the quality of the meaning of “Bring the cow!” in respect to that of “Bring the unicorn!”? Does the cognitive output of words count as justified and true belief? Is it a sui generis belief, or is it just a perceptive and inferential output? In South Asian philosophy of language such lines of research intersect and overlap, according to the traditions and authors who sought answers to the above questions, so much so that it is not always easy to distinguish whether the focus is mainly epistemological, metaphysical, or linguistic, and perhaps it is even misleading to isolate any one among the three. Hence in this book there will be occasional trespasses, though the main philosophical focus shall remain on language.

Jayanta, the Guiding Spirit This handbook is in many ways influenced by a monument of Sanskrit scholarship, the Nyāyamañjarī of Bhaṭṭa Jayanta (ninth century CE). In this work, Jayanta represents the worldview of Sanskrit authors who thought of the study of language in terms of the interdisciplinary trivium of grammar (Vyākaraṇa), hermeneutics (Mīmāṃsā), and epistemology (Nyāya). At Jayanta’s time, indeed, the grammarians were credited as the authorities on words (pada), the hermeneuts as the experts on sentences (vākya), and the epistemologists as the masters on the correct use of language for the sake of knowing reality (pramāṇa). By the end of the first millennium, the fourth

Introduction

3

discipline of poetics (Alaṅkāraśāstra) also gained mileage on themes around pragmatics, implicatures, and the many nuances of figurative usages. This handbook, which is largely devoted to researches on these four disciplines’ tenets, is thus a photograph of Sanskrit philosophy of language in its golden period, during which the most groundbreaking ideas were developed on the strength of a dialectic interaction among authors of these four traditions. The handbook will also represent the fifth voice, most impactful in these dialectic developments, provided by the Buddhist authors who often challenged the brahmanical dogmatic tenets and thus contributed to raise the philosophical quality of the debate. Furthermore, other dramatis personae deserved roles in this book: Jaina authors. For various reasons, however, none of the four specialists invited to submit a paper could deliver a paper in time for publication. As far as this volume is concerned, therefore, Jaina’s philosophy of language shall remain a desideratum. Jayanta is in various ways an embodiment of this dialectic evolution. He was a recognized Nyāya author, but also a reputed authority in Mīmāṃsā and Vyākaraṇa, and was well-versed on the tenets of prominent Buddhist authors such as Dignāga and Dharmakīrti. As for poetics, Jayanta displayed mixed feelings. He certainly was a gifted writer, a lover of belles lettres and even the author of a witty philosophical drama. And yet he thought little of the opinion of “poets” on epistemological issues, dismissing the reputed theorist Ānandavardhana as “a self-fancied scholar.” Contrary to the common view at Jayanta’s time, however, it is now clear how impactful the reflections of these poeticians were on the history of ideas around language (śabda). Jayanta was a self-critical thinker, capable of admitting his limits and uncharacteristically transparent in crediting previous authors, even those from rival traditions. The attentive reader of Jayanta will find in his philosophical dialogues echoes of the great hermeneutic insight that, in contrast to the common idea, the quest for the right philosophical question requires more awareness than the attempt of a possible answer (Gadamer 2000, II,3,c; p. 746, […] das Fragen—ganz im Gegensatz zu der allgemeinen Meinung—schwerer ist als das Antworten).

“Indian Philosophy of Language”? But was there ever real philosophy, and more specifically philosophy of language, in South Asia?

4

Introduction

Two centuries have passed since G. W. F. Hegel announced that philosophy was born in ancient Greece and that it could not exist in India in its mature form. Though Hegel himself was less apodictic in later days and may by now be excused, his enduring inheritance has been a condemnation of Indian rationality in the academic discourse. As noted by Halbfass (1988, 146), “The willingness to concede India an ‘actual’ philosophy as well, an attitude which Hegel occasionally gave utterance to during his later years, generally received little notice.” Hegel’s opinions about India have already been amply documented (see Halbfass 1988, Chapter 7; Tola and Dragonetti 2002), so here it suffices to remind the reader, with Halbfass (1988, 146), how “an essentially restrictive view of the history of philosophy emerged which was to eventually dominate nineteenth and early twentieth century thinking and which explicitly excluded the Orient, and thus India, from the historical record of philosophy.” Hegel’s prejudices have often been reflected and magnified into a Eurocentric stance toward “Eastern philosophy,” for historical reasons documented and discussed in Halbfass (1988, Chapter 9), that only in the last decades has begun to be doubted and challenged. Hegel’s stance was a product of his time, rooted in ideological, religious, political, and dogmatic idiosyncrasies which are not going to be addressed here. Nevertheless, a major reason behind his uncharitable assessment of “Indian” philosophy was the pervasive ignorance of the sources and of the language. After accessing the first studies and translations of some Sanskrit sources done by early Indologists, Hegel partially revised his assessment, though his more informed stance never overcame the mainstream idea of the crippled Geist of Indian philosophy. Nowadays, we certainly have improved our knowledge of India, as compared to Hegel’s times, and yet the bias of a spiritual or even superstitious India is still the predominant paradigm, with limited scope for rationality and philosophy. Moreover, there is no equivalent of “philosophy” in Sanskrit, and the closest equivalents such as darśana, ānvīkṣikī, śāstra, and vāda are certainly more specific than the universal “philosophy” (Halbfass 1992, chapter II.15): darśana suggests a system of thought, ānvīkṣikī points to the rational aspect of enquiry, śāstra evokes a discipline of thought, and vāda implies a dialectic process. Without attempting a daunting definition of the term, by “philosophy” here I modestly mean critical reasoning, that is, reasoning about reasoning. There are at least three activities shared by “philosophers of language” and the Sanskrit

Introduction

5

authors presented in this handbook, regardless of the obvious discrepancy in goals, methods, age, and cultural background: 1. a rational inquiry into the nature of language (śabda-parīkṣā); 2. a critical reflection about our own use of language and metalanguage (śabdavicāra); and 3. a disposition to dialectically engage in debates with competing views (śabdavāda), which works through a dialectic movement of theses (pakṣa) challenged by antitheses (pratipakṣa) and results in syntheses (siddhānta). As for the significance of philosophy of language in South Asian thought, several ideas around language were developed comparatively early and were destined to become building blocks of most subsequent philosophical speculations. In comparison to the evolution of modern philosophy, there was never a “linguistic turn,” as the awareness of language and metalanguage, and of their importance in philosophical thought, dates back at least to Pāṇini’s time.

The Structure The topics of this handbook have been selected and arranged from an emic and thematic perspective, as well as on the basis of the interest they have generated in the history of philosophy in South Asia. Since I am trying to highlight the historical context of the ideas developed by the original authors, as well as their reception in South Asia and beyond, the inner structure of each of this handbook’s parts reflects the assumed chronology of the authors and their works. The volume unfolds in four sections: 1. 2. 3. 4.

Of Speech Units (vācaka); Of Word Meanings (padārtha); Of Sentence Meanings (vākyārtha); Of Implicatures and Figurative Meanings (lakṣaṇā, etc.).

The first part concerns the quest for the minimal unit of speech. The second is about the nature of the things denoted by language. The third part is about sentences and their meanings—the core of actual linguistic usage. This threefold division is directly inspired by the concept of Jayanta’s sections on language in his Blossoms of Reason (Nyāyamañjarī). The unspoken, hidden, and yet performative aspects of speech are the topic of the fourth and conclusive part of the handbook.

6

Introduction

Readers who wish to deepen their understanding on the historical, lexical, and philosophical facets will find further guidance and insight in the Chronology and especially the Glossary. Inspired by Umberto Eco’s discussion on the semantic models of dictionaries and encyclopedias (Eco 1996, 72 et passim), the Glossary is a preliminary attempt to build a local encyclopedia (Eco 1996, 113ff.) of technical terms, by means of definitions, examples, and other contextual applications found in the universe of reference of the original Sanskrit authors. Advice for further reading can be found in the Annotated Bibliography, a team effort of this volume’s contributors, with short annotations on influential books and essays on South Asian philosophy of language. This Bibliography also reveals how some of the authors and themes that emerge as prominent in this handbook have been previously largely neglected in English-language literature, most noticeably Mīmāṃsā authors and the influence of their ideas in subsequent developments in pragmatics and poetics. In present-day words, the first two parts of the handbook concern semantics (and, partly, phonemics), while the last one is clearly within the sphere of pragmatics. The third part, however, stands on the threshold, testing the limits of such disciplinary divisions and perhaps even of this handbook’s very structure. For instance, any informed description of the signifier necessarily involves the signified, as well as the relation between signifier and signified. Otherwise said, how can one even start discussing meaningful speech units without addressing meaning first? The proposed structure, therefore, should not be hypostasized, because it chiefly has a didactic and hermeneutic intent. Nevertheless, it offers the advantage of being the same structure found in some influential Sanskrit sources, thus revealing the predominant philosophical questions that were asked by the original authors. Furthermore, through this structure, we are trying to display the dialectical propensity found in most Sanskrit philosophical literature, which contrary to common assumptions is one of its most striking features. The South Asian ideas around speech units, word referents, sentence meanings, and implicatures evolved precisely through the challenges and tests presented by opposing philosophical views, a history of effects that we have tried to actualize again in the dialectic exchanges among this handbook’s chapters.

The Projected Audience In the last two centuries, Sanskrit scholars have produced a massive corpus of highly specialized literature on the philosophy contained in Sanskrit sources,

Introduction

7

so the reader may wonder about the necessity of a new volume. The main reason for a new volume is that most indological studies are hardly accessible to non-Sanskritists, even if written in modern languages, with the result that the many philosophical gems are just taken from one hidden casket and moved to a new one, encoded in a new language for initiates. There are specific and often unavoidable reasons behind this state of affairs, such as the intrinsic challenges in the transposition of Sanskrit thought into English—due to the linguistic, chronological, geographical, and cultural distance—that the interpreter is confronted with. There are choices to be made, in between the extremes of adhering to the letter of the source language or betraying it for the sake of intelligibility in the target language. This handbook treats advanced research topics and questions of general importance, with a rigor that complies with the needs of a specialized audience. At the same time, it intends to reach a more generalist audience of philosophically minded readers, beyond the precincts of South Asian studies and without assuming a prior competence in Sanskrit language and literature. In the main text we tried our best to limit the use of Sanskrit technical terms and to provide an English equivalent for each of these. For this purpose, inquisitive readers may also refer to the Glossary, where they will find further clarification. Experts of South Asian studies, conversely, will find in the notes at the chapters’ ends many details useful to their own research, including Sanskrit passages, critical and exegetical notes, and bibliographical details. The Glossary addresses their needs as well, since it provides several technical definitions, from different angles, of important or ambiguous terms.

Acknowledgments Research on this book was generously supported in Vienna by the FWF (Austrian Science Fund), Project P28069 (2015–20), as well as by the IKGA (Institut für Kultur-und Geistesgeschichte Asiens) at the Austrian Academy of Sciences, Vienna, and the ISTB (Institut für Südasien-, Tibet-und Buddhismuskunde) at the University of Vienna. I’m grateful to the contributors of this volume, who responded on time to my solicitations and should all be credited for helping with the efficient production during the editing stages. Many of them also helped me by correcting mistakes and providing new information over the course of private conversations as well as public debates during the several reading sessions and workshops we

8

Introduction

held in Vienna in the last few years. I also want to thank Dania Huber, who painstakingly helped with the Bibliographies and the Chronology. Most of all, however, I’m indebted to Elisa Freschi, a true philosopher whose devotion to sophia is a constant source of inspiration, and to Francesco and Patrizia, who lovingly train me in the subtle art of dialectics.

References Eco, Umberto. 1996. Semiotica e Filosofia del Linguaggio. Or. ed. 1984. Torino: Einaudi. Gadamer, H.G. 2000. Verità e metodo. Ed. and tr. by Gianni Vattimo, with facing German text, and with Gadamer’s own post-scripta and preface to the Italian edition. Milano: Bompiani. Halbfass, Wilhelm. 1988. India and Europe: An Essay in Philosophical Understanding. Or. ed. in German, Indien und Europa, Schwabe & Co., Basel. Delhi: Motilal Banarsidass. Halbfass, Wilhelm. 1992. On Being and What There Is: Classical Vaiśesika and the History of Indian Ontology. Albany: State University of New York Press. Tola, Fernando, and Carmen Dragonetti. 2002. “What Indian Philosophy Owes Hegel.” Annals of the Bhandarkar Oriental Research Institute 83: 1–36.

Part One

Of Speech Units: Phonemes, Words, and Sentences

10

Introduction

In Sanskrit semantics, authors distinguished between a signifier (vācaka) and a signified (vācya), both derived from the verb “to signify” (vac-). This first part of the handbook will try to illustrate on the nature of this signifier. More specifically, the ensuing debate revolves around the concepts of phone or phoneme (varṇa), word (pada), and sentence (vākya). The debate around speech units is traced back to the early grammatical theories, where phonemes and words are defined in purely phono-morphological terms and without recourse to meaning. This early stage is followed by further theories with a focus on the radical difference between meaningless phonemes and meaningful words, and the account of different concepts or levels of reality accessible through language (Chapter 1). Mīmāṃsā authors argued in favor of phones (or phonemes), as the building blocks of speech. They thought it necessary to accept the existence of permanent speech units to account for the observed phenomenon of intersubjective communication, which is indeed possible exactly because there are permanent and immutable speech units that can be learned, remembered, and reused in constant patterns. Once the real and permanent existence of such speech units is accepted, it follows that words and sentences can be nothing else but strings of speech units, arranged in specific sequences (Chapter 2). Nyāya authors disputed the character of permanence attributed to speech units by Mīmāṃsā authors. How can sound be considered permanent, when all empirical evidence shows exactly the opposite? The permanence of speech also interfered with their whole epistemological project of establishing the veracity of speech on the basis of the authority of the speaker (Chapter 3). Later grammarians coined the usage of the Sanskrit term sphoṭa to denote the minimal unit of speech, which, in stark contrast to the linguistic atomism of Mīmāṃsā and Nyāya, was held to be the sentence. This sphoṭa was considered as the permanent, indivisible, and immutable core of speech, manifested by sound, and as the true morpheme, that is, the true vehicle of signification. According to them, phonemes and even words are mere abstractions that, while certainly

12

Introduction

useful for didactic purposes, do not play any role in real-world signification (Chapter 4). The theory of the sphoṭa, however, was criticized from different angles not only by Mīmāṃsā and Nyāya authors but also by Buddhist ones (Chapter 5), whose views were incompatible with the concept of an indivisible sphoṭa. Buddhists concurred with Nyāya authors on the impermanence of sound, but also joined Bhartṛhari and the grammarians in a relativistic account of speech segmentation, which occurs at the lower level of reality that we call human communication and interaction (vyavahāra). Chapter 6 is an assessment of the linguistic aspects of the influential theory of recognition (pratyabhijñā) of the Kashmiri theologian and philosopher Abhinavagupta. Ferrante shows how Abhinavagupta’s theory of “phonemic consciousness” is a testament to the enduring importance of the other theories of meaning presented in this section of the handbook: the influence of the ideas of Bhartṛhari, of Mīmāṃsā and Buddhist authors, and even of Jayanta, is visible throughout Abhinavagupta’s writings.

1

Linguistic Segmentation in Early Vyākaraṇa Maria Piera Candotti

1  Grammar before Grammar: The Origins of a Vedic Science Since the oldest testimonies of Vedic culture, we are confronted with reflections on the intrinsically segmental nature of language. Human word is divided (vyākṛta), distinct (vyakta), and measured (parimita); and even pada, the term which will later be used to signify “word,” is in fact first and above all a footstep (and hence a foot), a trace, a mark in a trail and then the metrical measure, the foot of the verse. Such a segmental dimension of language is not only crucial when it is used as a handy conceptual device to analyze a complex phenomenon, but, at least when considering early Vedic ideas on language and its functions, seems connected with the very nature of language itself. A famous yet extremely obscure riddle hymn of the Ṛg Veda (ca. fifteenth century BCE) states that language, like a cow, is measured in four footprints, well known to the wise brahmins: three of them, which are imprinted secretly, they do not divide (neṅgayanti), but they speak the fourth part of the language as men do (Ṛg Veda, I.164.45–46).1 Some lines before the poet had extolled the “syllable” (akṣara) as the highest heaven where all the gods live and without knowing which the knowledge of the hymn itself would have no value. The knowledge of the part is thus at least as important as that of the whole, if not more. If the action of “dividing language” is not so indisputably expressed in this passage,2 among the most ancient texts of Vedic literature we find a number of occurrences that address this same issue of linguistic segmentation with clearer terms. As Deshpande (1997, 76) puts it, “A general tendency of explaining the behavior or treatment of wholes in terms of their components is indeed widespread in the middle Vedic literature of the Yajurveda Saṃhitās and Brāhmaṇas.”

14

Maria Piera Candotti

1.1  Segmenting Language and the World A later passage in Vedic literature recollects the story of the origin of language: once upon a time the voice (vāc) was not separated (a-vy-ā-kṛta). The gods resorted to Indra’s help: “Please, do divide (vy-ā-kuru) this voice for us.” He answered: “let me choose a boon.” […] Then, having stepped down in the middle, Indra divided it: for this reason this same voice speaks distinctly (vy-ākṛta) (Taittirīya Saṃ., VI.4.7.3).3

What exactly Indra did, what he separated, is subject to debate, but it is important to highlight the consequence of his action, namely that the voice from that moment on spoke separately or distinctly. Even though different interpretations of this passage have been suggested, it seems probable that the division of the linguistic continuum in different linguistic signs was hinted at.4 In any case, the passage tells us that language as a tool for communication is intrinsically articulated, divided, and discrete; however, only gods (and learned brahmins) have the power to perceive it in that way.5 Let us notice that division or segmentation pertains to the true nature of language, even though it is not perceptible as such by humans, which amounts to say, as already underlined by Deshpande (1997, 82), that in these segmented units there is “no hint of unreality.” Significantly, the illustrated concept of entering language in order to articulate it evokes the Vedic concept of nāmarūpa (lit. “name and form”), the individuating principle through which the indistinct primordial reality is divided.6 There also, the divinity enters the indistinct reality in order to separate it, even though in the nāmarūpa contexts there is a definite shift from a metalinguistic to an ontological plane. In the Chandogya Upaniṣad, 6.3.2 (Olivelle 1998), we are told that the primeval principle, the One without second wished to become many; from it, thus, three divine principles evolved in succession: heat, water, and food. The One then resolved to divide/analyze (vy-ā-kṛ) name and form by entering them with its living soul.7 The division of language is somehow mirrored by the segmentation, division, of reality;8 a topic which we will have ample opportunity to scrutinize later on. Besides the relationship with the nāmarūpa-based cosmological speculations, another element of great interest in the Taittirīya Saṃhitā passage mentioned at the beginning of the paragraph is the relationship with grammar (vy-ā-karaṇa, lit. “segmentation, instrument of segmentation”) itself, as the very name easily shows. Indra is not only a refiner of human language, but reputed to also be one of the first grammarians, the first creator of the grammatical method as such,

Linguistic Segmentation in Early Vyākaraṇa

15

that is, a method based on a finite set of elements to which a finite set of rules are applied in order to account for the infinite potential set of correct sentences. There has been a lot of discussion on the fact that the term vy-ā-karaṇa is supposedly not fit to represent what grammar in Indian tradition actually does, since grammar does not analyze complex forms, reducing them to their minimal constituents; on the contrary, it starts from minimal units and builds up larger ones in a compositional way. Nevertheless, this does not necessarily imply that this operation cannot be interpreted as analytical, just as a formula such as 2 + 1 = 3 may be considered as a correct analysis of the number three (or, to say things differently, as a test of the soundness of the concept of unit). There is yet another dimension which, even though not evident in the Taittirīya passage itself, is for sure in the background of all Vedic linguistic speculations, namely the existence of dissonance factors between language and reality. The transparent relationship between the segmentation of reality and the segmentation of language may be clouded by a number of facts, first of all that the gods themselves love “what is obscure,” what is not directly perceptible to the senses (parokṣa). The Śatapatha Brāhmaṇa, VI 1.1.2, while discussing the usage of the well-known theonym Indra to denote the vital breath, says, “Inasmuch as it kindled (indh) it is Indha; the gods call Indha ‘Indra,’ in an obscure way, because the gods love what is obscure.”9 The “Indra” of the revealed text is thus a kind of mystic name that conceals and at the same time preserves another name that gives access to the true nature of the god, that is, “Indha.” The language of the text of the Ṛgvedic hymns, increate and eternal, obscure and rich in creative ambiguity, allows a glimpse of a superior level—of language and, consequently, of reality. Since the beginning, we thus find a theme which will be crucial in the centuries to come: the precarious balance between the supposed capacity of language to mirror reality in a one-to-one relationship—which will be later identified as the implicit “correspondence principle” accepted by many classical philosophical schools—and the awareness of the creative potentiality of opaqueness and ambiguity.

2  Segmenting the Linguistic Flux: Meaning and Form Different dynamics of part-whole relationship are at the base also of all the most ancient linguistic traditions in Vedic culture, namely phonetics (śikṣā), semantic analysis (nirvacana), and grammar (vyākaraṇa). All these schools aim

16

Maria Piera Candotti

at preserving the form and meaning of the revealed texts, by offering an analysis that makes them comprehensible and manageable by human beings. Their focus was principally practical, but this by no means entails (as, in fact, it almost never does) that they lacked abstract speculation. Irrespective of the difficulty in establishing a firm relative chronology of the first preserved texts of these different philological sciences, it can be stated with some confidence that their apex largely overlapped, as they all contributed to the construction the Vedic canon between the sixth and fifth century BCE. Each of them also claimed to be the culminating, perfecting point of this same process: Yāska’s Nirukta, the only surviving text of the tradition called nirvacana, “semantic analysis,” claimed to be “the fulfillment of grammar” (vyākaraṇasya kārtsnyam), while later on Patañjali (ca. second century BCE) asserted that grammar was an “ultimate science” (uttarā vidyā) grounded on the achievements of other sciences, while at the same time bringing them to completion.

2.1  The Functional Segmentation of the Vedic Text The first segmentation we are aware of is that of the so-called padapāṭha, a word-for-word recitation of the Vedic Hymns where each word is enounced without the phono-syntactic rules that normally apply in Sanskrit between two words in contact. The padapāṭha works as an analysis of the continuous recitation (saṃhitapāṭha), actually used in the ritual context. We have used this formula “word for word” for the sake of simplicity, but the target of the analysis of the padapāṭha is not the “word” in the common sense, that is, it is not a natural minimum semantic unit. Rather, the word in the padapāṭha is recognized on purely functional grounds through the specific set of phono-syntactic rules applying between “external” linguistic units in contact as opposed to another set of rules applying to “internal” units (e.g., between a verbal base and a suffix). Thus, due to the presence of external-sandhi rules, inflected words in a sentence but also members of a compound, secondary bases, and other bases triggering external sandhi before specific suffixes are all identified as “words.”10 From the beginning it is thus clear that segmentation occurs even below the word level. As the history of the word pada itself shows, the origin of segmentation is first to be found, to use the Saussurian terminology, at the level of the signifier (signifiant) and not of the signifié: a pada is first of all the metrical foot, a unit identified in terms of form and rhythm and completely detached from meaning, and only later it will be employed to signify a full-fledged word. Similarly based on purely formal grounds are the concepts of syllable (akṣara), the basis of the

Linguistic Segmentation in Early Vyākaraṇa

17

metrical foot, a term which occurs already in the Ṛg Veda as well as, somehow later, of single sound (varṇa). Later Prātiśākhyas, treatises that teach how to reconstruct continuous recitation starting from these “words,” will call these segments “pure, original” (śuddha). Again: what is analyzed and depurated from surface interferences is ontologically truer to the essence of language— it is pratyakṣa “direct perception”—even though, or better just because, not directly perceptible by mere humans.

2.2 The Role of Meaning in Segmentation and the Semantic Analysis Yet, in the domain of Vedic philological sciences, the very distinction on which the first two parts of the this book are organized—namely, one dedicated to the signifier (vācaka) and another to the signified (vācya)—was subject to close scrutiny and debate, and questioned from the beginning. The possibility of segmenting the linguistic flux on purely phonic and functional grounds was seriously challenged. It was clear, among the authors of those traditions, that it is not always possible to identify a linguistic unit without referring to the meaning it conveys. In the abovementioned Nirukta, 1.17, Yāska also considers the purpose and advantages of studying the science of semantic analysis (nirvacana) in comparison with its strongest competitors, such as grammar itself. Among such purposes he mentions that the continuous recitation is said to be “founded on words” (padaprakṛti-) and that without semantic analysis such a segmentation is not possible.11 It is quite clear that the possibility of bringing about segmentation on purely functional grounds is here forcefully rejected. Meaning, states Yāska, prevails on form when it comes to the analysis of words. This crucial role of meaning becomes absolutely evident when Yāska sets up three rules to be followed when carrying out the analysis of words (Nirukta, 2.1): Those words in which accent and grammatical formation are coherent with meaning and followed by a modification that is specifically taught must be analyzed accordingly. When, on the other hand, the meaning does not follow [the form] and the modification is not specifically taught one must remain focused on meaning through similitude with some [existing] analytical formation (vṛtti).12 If one can neither find a similitude [in existing formations] he should make the analysis on the grounds of the similitude of a syllable or of a single sound. He must never give up the analysis. And one must not focus on the correct grammatical form (saṃskāra): in fact there are existing formations that are dubious. The different segmented units must be interpreted following the meaning.13

18

Maria Piera Candotti

This is perhaps the first explicit reflection on the dimension of the linguistic sign as intrinsically biplanar, that is, specified both in terms of form and in terms of meaning. At the same time, there is the definite awareness of the limits of analysis, the difficulty in identifying perfectly matching blocks of form and meaning, and the explicit mention of the prevalence of meaning in case of conflict. This is forcefully restated in Nirukta, 2.7, where Yāska says that one should provide a single analysis for words with different forms but one and the same meaning, and as many analyses as there are meanings for phonically identical words: which amounts to say that udaka and jāla, both meaning “water,” should be analyzed together, while akṣa (dice) and akṣa (pole) will obtain two different analyses. This centrality of meaning (and, in particular, verbal meaning) in derivation seems to be one of the major points of difference between the specialists of semantic analysis (nairukta) and the grammarians, as discussed in a famous dialogue between Śākaṭayāṇa and nairuktas, on the one hand, and Gārgya with the grammarians, on the other.14 Without discussing the intricacies of the dialogue, it is sufficient to mention here the crucial point that Yāska evokes many times in the course of the debate: meaning and function never perfectly coincide, not only in those words whose meaning is not transparent from their form and which are the principal focus of the Nirukta, but even in the clear-cut grammatical formations—such as dātṛ (donor,), analyzed through the verbal base dā- (to give), and the suffix -tṛ (agent). The “method” proposed to extract meaningful units is that of comparison between the more opaque unit and the more transparent one, always relying on meaning over form.15 As Kahrs (1998) convincingly showed, this amounts to reasoning in terms of substitution: the more opaque unit may be used in the place of the more transparent one and is thus equivalent to it. The more transparent formation is often (even though not necessarily) constructed around an action expressed by a verb with which the word to be analyzed is intimately connected. To stay with the abovementioned example of Indra, commented in Nirukta, 10.8–9, more than one potential analysis is active for each name, and one does not rule out the others: Indra: “he divides (dṝ-) the refreshing fluid (irā-) or he gives (dā-) the refreshing fluid or he establishes (dhā-) the refreshing fluid or he scatters (caus. of dṝ-) the refreshing fluid or he holds (dhṛ-) the refreshing fluid.” Or: “the drop (indu-) flows (dru- > pres. dravati) or the drop pleases (ram-).” Or: “he kindles (indh-) the living beings”; inasmuch as he kindled him with the breaths, this is the indraic essence of Indra; this is specifically known.16

Linguistic Segmentation in Early Vyākaraṇa

19

This way of reasoning has for a long time puzzled modern scholars who have considered it bluntly nonscientific, disregarding nevertheless the specific aim of the school of semantic analysis. Its goal is not to give a linguistic analysis of words but, rather, to use analysis to account for the multifarious semantic power of words: Indra may be segmented in various ways, each of which accounts for a specific semantic feature of this name. Different parts may be distinguished inside the same whole, and this amounts to say that there is no predictable relationship between the whole and the parts. The role of meaning in segmentation might thus vary considerably from one school to another, and it will thus be crucial to understand the specific positions of grammarians on this point.

3  Pāṇini: Wholes as Contexts for the Interpretation of Parts 3.1  Grammar as a Calculus or Grammar as a Map The most accomplished achievement of this analytical trend of thought is to be found in the first text of the grammatical tradition (vyākaraṇa) that has been preserved, that is, the Eight Lessons (Aṣṭādhyāyī) of Pāṇini (ca. fifth/ fourth century BCE). But before tackling the specific topic of Pāṇini’s theory, it is necessary to describe briefly what the Aṣṭādhyāyī is, as the form in which Pāṇini’s thought had been handed down deeply influenced its interpretation in later centuries. The grammar of Pāṇini is not a unitary text, rather it is a macrotext composed of (at least) four mutually dependent texts: 1. 2. 3. 4.

a list of the basic sounds of Sanskrit language; a list of verbal bases (presently followed by a general meaning); a list of nominal bases necessary for further derivations; a text—the Eight Lessons stricto sensu—of around four thousand rules teaching how to combine the lower rank linguistic units in higher rank linguistic formations.

The rules are brief, technical aphorisms, a fact that has made it quite easy to interpret them as another list instead of as a true text, thus favoring an interpretation of each rule out of its original context. The meaning of Pāṇini’s text was thus sought in each single rule, as if the context played no semantic role. But, on the contrary, the linear order of enunciation—or, in other words,

20

Maria Piera Candotti

the textual dimension of the rules—may be somehow counterintuitive, albeit not devoid of some basic organizing meta-principles: ●●

●●

●●

Rules teaching similar operations (e.g., composition, affixation, and substitution rules)17 are grouped together. Each group of rules teaching a given operation is internally organized on the basis of a general rule-versus-exceptions pattern Each group of rules is linearly organized in order to maximize concision in using information deriving from previous rules (anuvṛtti).

Just to have a taste of what kind of text these rules construct, we may summarize sequentially the beginning of the Aṣṭādhyāyī, 1.1.1–6, which teaches the two apophonic “elongated” (vṛddhi) and “full” (guṇa) grades.18 The name vṛddhi is attributed to the sounds a, ai, and au [1] and the name guṇa to sounds a, e, and o [2], and both of these apophonic grades occur in the place of i, u, ṛ, and ḷ [3]. Nevertheless, there are some occurrences where this does not apply, listed in the rules [4–6]. The text shows us that Pāṇini is reasoning in terms of functions and places where these functions occur, for example, i, u, ṛ, and ḷ in the place of which a, ai, and au occur. Nevertheless, the dismissal of this textual dimension and of the meaning it conveys is quite common both in ancient and modern scholarship. Such an approach has without doubt been favored by a radical change in the target of grammar itself, as the linguistic context required more and more to teach Sanskrit from scratch. The meaning of the text is then found in its teaching of derivations (prakriyā) of specific linguistic forms. These derivations are obtained by applying relevant rules, which are taken from all over the grammar, in a new, nonlinear, order that guarantees the desired output: at first, base + affix, followed by inflectional affixes, effects on the pre-affixal base, variational substitution rules, and so on. The rules are interpreted as gears of a mechanism, rather than as threads of a web weaving an image. In recent times, Roodbergen (2008, 273) lucidly synthesized this approach: “Pāṇini’s grammar is not just an analysis of a language and the formulation of rules based on the analysis. It is also a generative calculus, which is actually the main thrust of his grammar.” S. D. Joshi (2009, 3) forcefully restated that “the Aṣṭādhyāyī is not a grammar. […] It is a device, a derivational word-generating device. It presupposes knowledge of phonetics and it is based on morphemic analysis.” And, in truth, such an approach to the text has very old roots, as we also see in the cultural sixteenth- to eighteenth-century phenomenon of the “grammars by derivations” (prakriyā grammars) such as the Siddhānta Kaumudī of Bhaṭṭoji

Linguistic Segmentation in Early Vyākaraṇa

21

Dikṣita or the Prakriyā Sarvasva of Nārāyaṇa Bhaṭṭa. These texts maintain word for word the original text of the Aṣṭādhyāyī, but reshuffle the single rules imposing a new order of enunciation, mirroring as far as possible the desired order of application of the rules in order to account for different classes of wordforms: nominal stems ending in -a, in ā, and so on; verbal present stem in -a or -ya; and so on. These new texts convey a linguistic description which is thus generative or transformational in nature.19 Modern scholarship on Pāṇini is deeply indebted to this later trend of the school and has the tendency to read the ancient grammarians, Pāṇini included, through the lens of these highly learned yet more recent commentators. The generative model, which is the commonest frame of interpretation of Pāṇini’s texts, derives both from the mediation of these traditional commentators and from the influence of modern transformational and generative linguistic paradigms. And yet there have been critical voices and doubts raised about such an interpretation of the text and its capacity to deliver a credible historical picture of the text and of its cultural role in late Vedic times. Already, Buiskool (1939, 3) was wondering “whether Pāṇini really aims at shortness of expression only or, though making unmistakable concessions to this, starts nevertheless from a thorough mental planning of the subject-matter.” If we reconsider the original structure of the Aṣṭādhyāyī, focusing on the text as we received it, following its original organization principles, we can better appreciate the strong relationship of Pāṇini with the overall domain of lateVedic thought. Such a structure is not peculiar to grammar alone: Vedic technical literature, in particular liturgical literature, follows a schema of presentation that begins with what is common (sāmānya) and is followed by all the phenomena which are special or specific (viśeṣa). This pattern has been vividly described by Kahrs (1998, 184–85): The methodology employed revolves around such concepts as prakṛti “prototype” and vikṛti “modification.” This is also referred to in terms of an image from the art of weaving as tantra “warp” and āvāpa “woof” denoting respectively the basic model which is the constant part of a ritual and the special features which differ from one ritual to another.

Coming back to the description of language, such a pattern highlights the similarities and differences between linguistic phenomena and draws a horizontal map of the linguistic uses where the specificities (variations, specific usages, etc.) are delimited by borders traced by the rules themselves.

22

Maria Piera Candotti

This is particularly evident in case of substitution, where rules deal mostly with allomorphy phenomena that establish variants of a given form in given contexts. Yet, a predominantly spatial, synchronic, descriptive pattern is evident also in the description of complex formations: the rules tend to teach where to apply a certain operation and not when. We can use as an example Pāṇini’s set of rules (Aṣṭādhyāyī, 3.1.68 and following) describing the ten present-class stem formations: the most general rule teaches to add the modifying suffix -a- to any verbal base whatsoever to form a present stem [68].20 This accounts for the first class, which is numerically the most representative of the ten classes. The rules that follow specify the other ten classes as exceptions of this way of forming the present stem: specifically after the group of verbal bases beginning with div- there is (the modifying suffix) -ya- [69]21 which, preferably but not exclusively, occurs also after a specified set of other verbal bases [70],22 and so on. Significantly, among the various conditions that specify a given operation (what should precede and what should follow), we find also meaning itself as a condition that may specify a right or left context but also the output form, for instance (Aṣṭādhyāyī, 3.1.76), tanūkaraṇe takṣaḥ: “After the verbal base takṣ-, if used in the meaning of ‘making slender,’ the present suffix -nu-/-no- occurs” (e.g., takṣṇoti, “slendering,” but takṣati, “skinning”). A metaphor involving the concept of a geographical map was proposed for the first time by Kahrs (1998, 184) to account for this descriptive pattern: If we have a map—and I think it is justified to call the ritual and linguistic descriptions of the ancient Indians a map—[…] we may […] ask what features of the territory are represented on the map. If the territory is absolutely uniform, nothing would be represented on the map except the borders of the territory. Otherwise, what will be represented is really differences of various kinds— differences in height, vegetation, surface, population structures, etc.

We are thus confronted with two divergent interpretations of the Pāṇinian text: one, more influenced by classical commentators, that focuses on the single rules, taken independently from their context, as gears in a mechanism or operations within a calculus; the other, which gives more importance to the late Vedic tradition contemporaneous to it and interprets the text as a description of linguistic facts based on the basic opposition between what is common and what is specific. It goes without saying that such a difference reverberates strongly on other aspects of Pāṇini’s linguistic theory, such as the status and function of the linguistic units obtained by segmentation.

Linguistic Segmentation in Early Vyākaraṇa

23

3.2  Linguistic Units as Places Where Things Happen Pāṇini’s approach to the wide range of linguistic units he uses—phonemes, phonic strings, augments, suffixes, primary verbal and nominal bases, complex nominal and verbal bases, pre-suffixal bases, words, compounds, and syntagms—is extremely fluid: a given linguistic unit, such as i- may be quoted as a pure phonic string as well as the meaningful verbal base “to go” or the corresponding pre-suffixal base,23 depending on the needs of each single rule. Segmentation is thus strictly functional to the operation that is specifically taught: in an affixation rule, for example, what precedes the affix is either a nominal or a verbal base, but a substitution rule before that same affix will trigger a pre-suffixal base (aṅga) as preceding segment. Upon such a multiform set of units Pāṇini does not impose a strict categorization apart from what is needed for the functioning of the rules themselves. This is true also of some categories which could seem crucial from a purely abstract point of view—and which, nevertheless, are not meant to be exhaustive from a descriptive one. Let us consider the case of units characterized by upadeśa, “direct teaching,”24 that is, units identified by the specific phonic form they show either in the lists of constituents which are parts of the Pāṇinian macro-text or in the rules in which they are specifically enjoined. One could assume that we are here dealing with the abstract identification of a set of starting and telic elements, the building blocks or the minimum units of language. Nevertheless, the set of units directly taught is not co-extensive with the set of all the telic elements: there is no exhaustive list of nominal bases, just to quote a macroscopic example. Even though nominal bases are for sure telic elements, not all of them are taught by direct teaching. Direct teaching cannot, in other words, be equated to an enumeration of the building blocks of grammar. Moreover, the direction in the construction of units is not univocal: it is true, most of the operations are compositional (from smaller units to larger ones) but some are analytical (from larger units to smaller one, from inflected words to nominal bases) and numerous others, variational (substitution of one unit to another in given contexts). In fact, complex linguistic units may be used as the base to obtain lower rank linguistic units: through operations of zerosubstitution of endings, for example, inflected words may be the base to obtain new, complex, nominal, or verbal bases. Another example of the fluid status of segments in Pāṇini is the distinction between units quoted as phonic (i.e. involving all the sounds quoted in the list of sounds) and all the others. As it can be shown with a fresh analysis of the

24

Maria Piera Candotti

rule 1.1.56, which deals with a crucial specificity of phonic rules,25 Pāṇini does not work with a strict opposition between phonic, monoplanar units, on the one hand, and morphological, biplanar, on the other. On the contrary, he works with a continuum of segments which has two ends of the spectrum: ●●

●●

pure sounds which are not specified, neither as per meaning nor as per function, and even not as per position;26 and prototypical morphological units, specified in terms of meaning, function and position, such as some affixes and verbal bases.

Between these two extremes, all the other linguistic units find their place. In this model even pure sounds do not seem to have a rigidly distinct status. While single sounds seem to be completely devoid of linguistic information (position, function, meaning, etc.), strings of sounds are already specified in terms of boundaries of the unit they substitute,27 and other meaningless strings are specified in terms of position.28 There are no traces in Pāṇini of a concept of double articulation where the same utterance can be at the same time analyzed as a meaningless sequence of sounds and a meaningful sequence of morphs. Linguistic units in Pāṇini are strictly functional to the specific linguistic phenomenon contingently described in order to unequivocally identify the locus of the operation. Meaning is just one among the many restricting conditions that may be activated when needed.

3.3  Is There a Basic Linguistic Unit for Pāṇini? A crucial question for modern scholarship has been that of the status of basic meaningful units against the higher rank units of which the first ones are parts. What is, in other words, the degree of “reality” of these subunits? From what we have just explained, the question itself may seem somehow inappropriate. The role of meaning in Pāṇini’s grammar has for a long time been interpreted in a strictly compositional approach: the minimum linguistic units are the building blocks that construct the higher rank units in terms both of form and meaning. Such an approach is intertwined with the global interpretation of Pāṇini’s explanatory model as basically generative or transformational. But we have seen that such an interpretation seems to better suit later developments of grammar, rather than the Vedic cultural context in which the grammar was composed. We have seen that in the Vedic tradition contemporaneous and precedent to Pāṇini, there were schools that favored a purely functionalist approach to segmentation through which they identified some basic items such as syllable (akṣara), word

Linguistic Segmentation in Early Vyākaraṇa

25

(pada), and metrical foot (pada or pāda). Significantly, Pāṇini is indebted to these schools, for instance, when it comes to his very concept of word: thus, even though a word is for sure a meaningful unit in Pāṇini’s system, it is nevertheless defined on purely functional grounds, almost identical with those of the Prātiśākhyas.29 Moreover, Pāṇini’s fluid usage of these “building blocks” ill suits such a descriptive approach. Be that as it may, a sort of common knowledge has been gradually crystallizing over the years, imposing a linear development in the history of ideas among early grammarians on what is supposed to be the “principal” meaning unit in language, beginning with Pāṇini and ending with Bhartṛhari (ca. AD 6). We can illustrate this with Bronkhorst (1998, 35), to whom we owe some crucial reflections on this topic: “If for Pāṇini the principal meaning-bearer is the constituent of the word and for Patañjali the word itself, Bhartṛhari goes one step further and says that it is the sentence which is the real meaning-bearer.”30 In this schematic representation, there is more than one point that deserves further scrutiny. First of all, as we already noticed, it implicitly assumes as a fact that Pāṇini was working with strictly biplanar units, that is, with units specified in terms of form and meaning, and that he was committed to a strictly compositional approach. To the elements suggesting caution that we have highlighted in the previous sections, we can here add another element that goes against such a position. In fact, Pāṇini has recourse to the concept of “unit endowed with meaning” (arthavat) only once, that is, to identify the third telic element that cannot be defined by enumeration, the nominal base (pratipādika): “a meaningful element which is not a verbal base nor an affix is a nominal base” (Aṣṭādhyāyī, 1.2.45).31 Nevertheless, Pāṇini’s aim is not to identify the three minimum, irreducible, meaningful units of his grammar, so much so that he completes his definition by another rule (Aṣṭādhyāyī, 1.2.46) teaching that also complex bases—primary, secondary, and compounded—get that same name of pratipādika. Secondly, the very notion of a “principal” or “real” meaning bearer is problematic: does it refer to a kind of “natural” meaning-bearer unit, something which is commonly perceived as the minimum semantic element? Such an interpretation would be far from satisfactory. It seems, on the one hand, quite clear that Pāṇini is not interested in any “natural” linguistic unit whatsoever, as he dismisses with nonchalance one of the best candidates, the word itself. Moreover, some of his building blocks are taught without meaning or with abstract, non-lexical meanings, such as the primary affixes, which are taught through thematic relations.32 Another possibility is to consider that by “principal”

26

Maria Piera Candotti

a “really existing” linguistic unit—irrespective of its being meaningful or not— is meant. In this second option, what is “real” from the linguistic point of view? This second option deserves attentive scrutiny: in Vedic culture we have seen more than one trace of the concept that parts are just as real as wholes. On the other hand, no part seems more real than any other. The linguistic elements below the word level (verbal bases, suffixes, nominal bases, etc.) have no claim, in a map-like perspective, to a higher level of reality: they are without doubt the minimum meaningful segments, but are in no way more “real” than the others. Pāṇini seems extremely conscious of the limits of his analysis and his aim isn’t in any way that of matching linguistic analysis with ontology. The reality of linguistic parts in Pāṇini (meaningful and not) comes from the very regularity of the linguistic system, which makes the presence of the units predictable (prasakta) and their absence significant: it is on this regularity which is based Pāṇini’s device of linguistic zero to account for cases where a linguistic unit present elsewhere in the system (such as the -s of nominative singular seen in many forms like rāmas “Rāma NOM. M>”) is expected also when not directly perceived, such as in sarit “river NOM. M.,” and considered as realized with phonic zero.

4  After Pāṇini: A New Approach to the Parts-and-Whole Relationship The first commentators following Pāṇini are Kātyāyana (ca. third century BCE), author of the “Critical Aphorisms” (Vārttika), and Patañjali (second century BCE) to whom we owe the “Great Commentary” (Mahābhāṣya) that collects and comments on a number of rules of the Aṣṭādhyāyī and of Kātyāyana’s aphorisms. Somewhere in between these two authors, we can recognize two major shifts that go far beyond the position of the single authors and are testimony to a deeper change of cultural paradigm: 1. the emergence of two natural meaning-units, namely the inflected word (pada) and the sentence (vākya), as contrasted with more “abstract” ones; and 2. the search for permanent (nitya) biplanar units extracted through a contrastive process of comparison. The second shift is entailed by the first: once the equal status of differently segmented parts is questioned by individuating “natural” segments, it

Linguistic Segmentation in Early Vyākaraṇa

27

becomes necessary to prove the permanence of all the linguistic signs as units consisting of form and meaning to guarantee language’s capacity to represent reality. Tradition reads this position precisely in what is considered the first of Kātyāyana’s aphorisms, but in all probability this interpretation is forced into the text. We will concentrate here on its first part, which commentaries, beginning with Patañjali, have analyzed extracting it form the original context. The syntagm siddhe śabdārthasambandhe was probably simply to be interpreted as “the link between the linguistic sound and the meaning being already established,” that is, given by usage outside grammar, but was reinterpreted forcing the term siddha, “established, already given, ready,” toward the meaning of “immutable, permanent” (nitya). Later on this nitya will be further specified as meaning “eternal,” but the grammatical tradition for a long time did not want to load this concept of too constricting ontological presuppositions.33 Nevertheless, the grammatical tradition, starting from Patañjali onward, has undoubtedly considered the permanence of linguistic forms in their double component of form and meaning as a fundamental—and, over time, more and more compelling—trait of its own conception of language.

4.1  Kātyāyana and the Natural Level of Signification In Kātyāyana’s aphorisms, the implicit hierarchy among the different linguistic units and the privileged status of the “natural” meaning bearers are already in place. It is well known that we owe to him the identification and definition of the level of “sentence” (vākya),34 which was left undefined by Pāṇini. But what concerns us more at present is his understanding of the “word” (pada) as a whole (saṅghāta) and of the relationship with its subunits. On these two points, the discussion developed by Kātyāyana while commenting on Pāṇini’s definition of nominal base (Aṣṭādhyāyī, 1.2.45) as “a meaningful element which is not a verbal base nor an affix” is crucial. The statement is problematic under two different points of view. First of all, the definition cannot stand alone, as it is not desired that either the inflected word or the sentence—the two natural meaningful units—be defined as “nominal bases” (pratipādika).35 It is thus necessary to somehow restrict the application of the rule, which will be done by the rule that follows (Aṣṭādhyāyī, 1.2.46). On the other hand, the very possibility to be meaningful for a “non-natural” semantic unit, such as the nominal base, is also to be questioned. To follow Kātyāyana’s words, “The nominal base cannot be meaningful, as it does not express meaning when used alone” (Aṣṭādhyāyī, vārttika 7 ad 1.2.45).36

28

Maria Piera Candotti

In other words, a base such as vṛkṣa- “tree,” which is never employed alone, cannot convey its meaning unless used in language as an inflected word: vṛkṣaḥ, “a single tree,” vṛkṣau, “two trees,” and vṛkṣāḥ, “many trees.” There is no abstract comprehension of the semantic content “tree” without at least the identification of number (and, sometimes, gender). There is thus a difference in status between the meaning “tree” of the nominal base, never experienced as such in actual usage, and the concrete meaning each time conveyed by the word (pada) employed in a sentence. The natural meaning of words is ontologically different from the meaning of subunits, however we want to define them. The final answer to the issue is then given in these terms: “But this is solved because of what remains (anvaya) and what exceeds (vyatireka)” (Aṣṭādhyāyī, vārttika 9 ad 1.2.45, siddhaṃ tv anvayavyatirekābhyām). The process of analysis through anvaya and vyatireka will become central in Patañjali’s theory of segmentation of words,37 but it is not traceable directly to him. In fact not only, as we have seen, does Kātyāyana quote it, but he even takes it for granted so much so that he does not explain here what this device precisely is. Nevertheless, we may have a glimpse at it in a parallel discussion (Aṣṭādhyāyī, vārttika 4 and 6 ad 1.3.1), which significantly concerns the parallel concept of verbal base (dhātu), and the difficulty of selecting it from a semantic point of view because the meaning is expressed by the whole (saṅghātenārthagateḥ) and thus cannot be expressed by the verbal base alone.38 The possibility of identifying a meaning in the verbal base is again to be found in the comparison between continuum and discontinuum in linguistic forms, between what remains and what exceeds, which here Kātyāyana explains through a concrete example: given a full verbal form, by changing the verbal base and not the suffix one cannot detect any change in the meaning of the suffix, and conversely when the sole affix is changed no change in meaning is detected in the base.39 Using here the examples offered by Patañjali, the meaning of “agent of the action” of the verbal ending -ti, third-person singular, does not change when it follows the verbal base pac-, “to cook,” or paṭh, to recite: paca-ti (he cooks) versus paṭh-a-ti (he recites). Similarly, the meaning of the verbal base does not change between pac-ana (neuter action noun, “cooking”) and pāk-a, masculine, with the same meaning. The reasoning based on the comparison between continuum and discontinuum allows to assure the existence of a specific meaning also for units inferior to the word level, a meaning which cannot be apprehended in everyday communication. But what is crucial for Kātyāyana is simply to identify subunits such as verbal base (dhātu-) and nominal base (prātipadika-)

Linguistic Segmentation in Early Vyākaraṇa

29

as being “meaningful,” not to identify the specific contribution of each of them to the compositional construction of the meaning. Such a specific contribution in fact does not exist with reference to the meaning of full words, which have an independent reality in speakers’ perception; strict semantic compositionality is, in fact, discarded by Kātyāyana: “Moreover, denotation is autonomous from the constituent linguistic forms.”40 This strong assertion is stated twice by the author and it concerns formations based upon full words (pada) such as compounds (e.g., mata-pitarau “mother and fathers,” with dual ending) and other complex formations (e.g., pitarau lit. “the two fathers,” that is, the parents, always in the dual ending).41 This is also why the focus is not on generic “meaning” (artha) but specifically on “denotation” (abhidhāna), that is, the specific, contextualized meaning of a full word in an utterance: such a denotation is not reducible even to the single meanings of the full words in different contexts. Segmentation, an essential method in Pāṇini’s grammar, is not denied by Kātyāyana, but it is limited to the domain of form; meaning, though recognized as a crucial element of the form itself, may only be segmented on a purely abstract level. Only the inflected words and the sentence obtain the status of “natural” segments inasmuch as they are natural biplanar segments.

4.2  Kātyāyana and Patañjali on the Status of Phonemes among Linguistic Units and Double Articulation as a Feature of Language Once this double-level segmentation is put in place, the specific status of pure phonic units becomes something to be discussed and defined, although this necessity was not felt by Pāṇini. If only some segments (words, sentences, etc.) are naturally meaningful while others are meaningful only by abstraction, it is necessary to understand the specific status of the minimum linguistic unit, the single sound. It is extremely significant that the long discussion on Aṣṭādhyāyī, 1.2.45, concerning the status of the units inferior to pada that led to the elicitation of the anvaya-vyatireka device, continues precisely with the questioning on the status of single sounds. Are the sounds (varṇa) meaningful or not? May there be linguistic segments which are nevertheless not biplanar? Kātyāyana and Patañjali first review some elements in favor of the meaningfulness of sounds, namely the fact that (1) there are nominal and verbal bases that are formed by a single sound and that are meaningful (e.g., the verbal base i-, “to go”), (2) if a change (vyatyaya) or a loss in the sounds of a word occurs, the meaning of the

30

Maria Piera Candotti

word itself changes (e.g., kūpa, “well”; yūpa, “post”; and sūpa, “soup”), and (3) the set of linguistic sounds has meaning and, consequently, its parts must also be meaningful. Point (2), without naming it directly, polemically calls into question the process of anvaya and vyatireka. With slightly different arguments, both Kātyāyana and Patañjali will prove the fallacy of the abovementioned arguments; Patañjali in particular points out that if in kūpa, yūpa, and sūpa every single sound were meaningful, then the three words, having three sounds in common, would have a large part of the meaning in common and the change of meaning would be predictable. But the three words have completely unrelated meanings. From this Patañjali concluded that kūpa, yūpa, and sūpa are different, completely unrelated, sets (saṅghaṭa) of sounds.42 The third point is addressed by Kātyāyana in (Aṣṭādhyāyī, vārttika 11 ad 1.2.45): “To the assertion that individual sounds are meaningful because the whole is endowed with meaning, one responds that one observes the attainment of a goal (artha) of an object through an accessory that does not have the same purpose.”43 Paraphrasing the extremely elliptical language of his aphorisms, we may state that Kātyāyana recognizes at least two ontologically different kinds of wholes, that is, those that are a mere sum of their constituent parts (such as a heap of rice) and those that cannot be reduced to the sum of their parts, which are structures, that is, arrangements of elements. Such wholes may “attain a goal” even though their parts can’t. To this second type belong the words viewed as collections of sounds: sounds, though meaningless, may create a meaningful system. Patañjali strives to find a good example for this second type of wholes and in the end he compares them to a chariot: “As the different parts of a chariot, taken one at a time, are not able to complete the action of advancing while the collection of these same parts is, in the same way the collections (samudāya) of sounds are significant and the parts (avayava) are meaningless.” Sounds are thus integrant monoplanar parts of a linguistic system which is basically biplanar. This special status of sounds puts forward the question of the degree of reality of such segments. Are sounds natural, permanent linguistic units or pure abstractions? And if they are permanent, that is, not affected by individual phenomenic changes, on what is this permanence grounded? Both Kātyāyana and Patañjali feel the need to stress the stability or permanence of linguistic sounds, above the phenomenological variety of their possible realizations. The linguistic sounds are fixed (avasthita-) and the differences in length we may perceive among different realizations of the

Linguistic Segmentation in Early Vyākaraṇa

31

same sounds are pure phenomenic accidents (śabdaguṇa).44 Patañjali will further elaborate on this distinction calling the first, with a term that will have a great role to play in the centuries to come, “bursting sound” (sphoṭa) and the second, “resounding” (dhvaṇi). The “bursting” sound is thus at the origin of all the different resonances of sounds in actual usage and accounts for its permanent form.45

4.3  Patañjali’s Tightening of the Linguistic System: Assessing the Reality of Wholes Patañjali’s example of the chariot is too philosophically connoted to be casual. As it happens, in this early author we have glimpses of the formation of philosophical questions that will occupy the following centuries. At least two schools of thought have deeply questioned the relationship between the whole and the parts: the philosophical school of Brahmanical tradition of the Vaiśeṣika and that of the Buddhists. Especially the latter school attacked the possibility of existence of complex objects, the Ego among and above all: ultimate and indivisible constituents (dharmas)—physical and psychic—alone possess ontological reality. In a text of the Buddhist school, the “Questions of King Milinda” (Milindapañha), in some of its parts quite archaic, we find exactly the example of the cart. The text narrates of the Buddhist monk Nāgasena who receives a visit from King Milinda, probably an Indo-Greek king named Menander, eager to discuss Buddhist philosophical positions. The discussion abruptly revolves around the cart the king used to come and visit the monk. Nāgasena then asks that he be told what this chariot is. The king is forced to admit that the chariot is neither the rudder nor the axle, wheels, frame, flagpole, yoke, reins, or goad; it is also not the collection of all these parts or something outside of them. From the Buddhist point of view this is explained by the fact the chariot as a whole exists only as an appellation, common notion, designation (nāmadheya). The grammarians’ reply to this is that, on the contrary, there exist some wholes, some collections—full words among them—which are irreducible to their constituent parts and thus can claim an independent status of their own: the word is meaningful and ontologically irreducible to its parts, the sounds. As Bronkhorst has pointed out in a number of occasions,46 this concern for preserving the “reality” of full words was shared with grammarians by some Buddhists as well. Bronkhorst emphasizes how the Buddhists needed to prevent the word, given its function of justifying the illusion of complex objects, from

32

Maria Piera Candotti

falling victim to the atomistic deconstruction applied to reality. The problem is already dealt with in the texts of the Sarvāstivāda school—contemporaneous if not precedent Patañjali—which identify at least two irreducible constituent of words, that is, “the phoneme” (vyañjanakāya) and “the word” (padakāya). It is almost impossible to overestimate the influence that Buddhist challenges had on the development of Patañjali’s linguistic thought. Significantly, it is to Patañjali that we owe the further step toward a perfect biplanarity of linguistic units, a step that entails a strict compositional approach to linguistic analysis and which further tightens the bonds between inferior and superior linguistic units. If the segmented units perfectly account for both the form and meaning of linguistic utterances, composition and segmentation become biunivocal operations. A specific meaning entails a specific form, and vice versa. Precisely while discussing one of the phenomena for which Kātyāyana had invoked the irreducibility, the autonomy (svabhāva) of meaning from form, Patañjali, on the contrary, states that “without the usage of the linguistic unit (śabda) which expresses a given meaning there is no comprehension of that meaning” (Mahābhāṣya, ad A 1.2.64 vt. 29, vol. 1, p. 241,1).47 Thus form and meaning are at all levels strictly bound together. The greater rigidity in the compositional approach becomes evident in Patañjali’s treatment of the exceptions to that same principle, such as linguistic zero. Patañjali’s aim is thus to save the permanence of the link between form and meaning—and thus, indirectly, the authoritativeness of Vedic texts—without discarding segmentation which is at the basis of grammatical method to preserve the texts themselves. This is not the right place to account for Bhartṛhari’s linguistic thought, which will be amply analyzed later on. Yet, it seems important to close this chapter with a glimpse of the view of this author who closed Bronkhorst’s triad featuring Pāṇini as an advocate of sub-word units; Patañjali, of the word; and Bhartṛhari, of the sentence. As we have shown, the development has been less linear, and Bhartṛhari’s specific innovations are better evaluated when contextualized against the background of the grammatical system of which Bhartṛhari was an eminent exponent. Bhartṛhari’s theory, innovative as it surely is, is also a recovery of a more Pāṇinian-oriented view on segmentation after and beyond Patañjali’s rigid biplanarity. Bhartṛhari’s aim is to free grammar from the burden of “deriving words and sentences starting from meaning inputs” and to establish the gnoseologic priority of wholes from which each time different parts may be extracted, based on the specific needs.

Linguistic Segmentation in Early Vyākaraṇa

33

Notes 1 catvā́ ri vā́ k párimitā padā́ ni tā́ ni vidur brāhmaṇā́ yé manīṣíṇaḥ / gúhā trī́ṇi níhitā néṅgayanti turī́yaṃ vācó manuṣyā vadanti // 2 The meaning of the verb iṅg-, we translated as “divide,” is still debated. Its causative form (iṅgaya-) used here is generally interpreted as “making move, propel” (cf. the translation in Jamison and Brereton 2014, vol. 1, 359, “they do not set in motion the three which are imprinted in secret”). Yet a suggestive alternative translation, grounded on the technical meaning of that same verbal base in the Pratiśākhyas, has been proposed by Maggi (2001, 311–23) and it is followed here. In those texts, focused on the phonetics and phonosyntax of Vedic hymns, the verb is used to signify the operation of moving, separating, for example, two members of a compound, and the deverbal iṅgya- “movable” is predicated of such elements. 3 imā́ ṃ no vā́ caṃ vyā́ kurv íti / sò ’bravīt / váraṃ vṛṇai máhyaṃ caiváiṣá vāyáve ca sahá gṛhyātā íti / tásmād aindravāyaváḥ sahá gṛhyate / tā́ m índro madhyatò ’vakrámya vyā́ karot / tásmād iyáṃ vyā́ kṛtā vā́ g udyate / 4 Different translations of this passage (and specifically of the verbal base vy-ā-kṛ, “divide” hence “analyze”) have been suggested by Scharfe (1977, 80, 83), Thieme (1982, 23–34) and Bronkhorst (1999, 14–18) in connection with the meaning of the term vyākaraṇa, used to denote the science of grammar. Such efforts generally originate from an uneasiness in attributing the term vyākaraṇa (act of analyzing) to the ancient grammatical practice as we know it, that is, to Pāṇini’s grammar which is supposed to proceed not through analysis but, on the contrary, through composition. Nevertheless in the present passage the role of Indra is not that of the first, divine, grammarian, but rather that of the creator of human, articulate, language. 5 I follow here Deshpande’s (1997, 82) interpretation of the passage. 6 Literature on nāmarūpa is almost endless. For a recent discussion on the Vedic attestation of this important philosophical term, see Olalde (2014). 7 hantāham imās tisro devatā anena jīvenātmanānupraviśya nāmarūpe vyākaravāṇīti. 8 And must conversely match the structure of the ritual action (Deshpande 1997, 78). 9 yad ainddha / tasmād indhaḥ / indho ha vai tam indra ity ācakṣate parokṣaṃ parokṣakāmā hi devāḥ / 10 There are thus some nominal bases which are called “word” (pada) before some suffixes. 11 The composite padaprakṛti- may be analyzed either as a subordinative compound meaning “the base of the word” or as an exocentric compound meaning “having

34

Maria Piera Candotti

the word as its base.” In the first case the continuous recitation would be declared to be the base from which the words are extracted. See Deshpande (1997, 88). 12 The word vṛtti is here translated with the technical meaning of “grammatical derivation” or “formation,” akin to the meaning of this same term some lines before this passage. The majority of scholars tend to interpret it as “action” or “course of action,” a meaning not attested as such in technical literature. The common interpretation had already been questioned by Mehendale (1978). 13 tad yeṣu padeṣu svarasaṃskārau samarthau prādeśikena vikāreṇānvitau syātām tathā tāni nirbrūyāt / athānanvite ’rthe ’prādeśike vikāre ’rthanityaḥ parīkṣeta /   kenacid vṛttisāmānyena / avidyamāne sāmānye ’py akṣaravarṇasāmānyān nirbrūyāt / na tv eva na nirbrūyāt / na saṃskāram ādriyeta / viśayavatyo hi vṛttayo bhavanti / yathārthaṃ vibhaktīḥ sannamayet / 14 See Nirukta, 1.12–14. 15 Yāska explicitly claims that one may never give up the analysis, should one rely only on the similitude of one syllable between the word to be analyzed and the word with similar meaning. 16 indra irāṃ dṛṇātīti vā / irāṃ dadātīti vā / irāṃ dadhātīti vā / irāṃ dārayata iti vā / irāṃ dhārayata iti vā / indave dravatīti vā / indau ramata iti vā / indhe bhūtānīti vā / “tad yad enam prānaih samaindhaṃs tad indrasyendratvam” iti vijñāyate / 17 Composition rules teach the combination of inflected words; these procedures start from a syntagm of inflected words and build from it new synthetic bases with an equivalent meaning. Affixation rules teach the addition of items, specified in terms of position, function and (when necessary) meaning, to other items likewise specified in terms of position, function, and meaning: for example, the addition of a suffix to a verbal base. Substitution rules teach the replacement of a verbal base, a nominal base, an affix, an inflected word, or a speech sound with another base, affix, inflected word, or speech sound in specified contexts. 18 Apophony accounts for vowel alternation such as may be seen in series like Śiva (PN) versus Śaiva (devotee of Śiva) (vṛddhi) or śruta- (ppp. “heard”) versus śrotum inf. “to hear” versus a-śrausit (aorist “he heard”). 19 A crucial discussion on the role of these new grammars has been first proposed by Bali (1976), a new perspective will be found in Candotti and Pontillo, forthcoming. A presentation of the difference between the two approaches in the domain of metalinguistics may be found in Candotti (2012). 20 Aṣṭādhyāyī, 3.1.68: kartari śap, for example, paṭh-a-ti (first class). 21 Aṣṭādhyāyī, 3.1.69: divādibhyaḥ śyan, for example, div-ya-ti (fourth class). 22 Aṣṭādhyāyī, 3.1.70: vā bhrāś​abhlā​śabhr​amukr​amukl​amutr​asitr​uṭila​ṣaḥ, for example, bhrāśyate/bhrāśate (both the first and the fourth classes are acceptable, but the fourth is preferred). 23 That is, the verbal base in the specific form it assumes before a suffix.

Linguistic Segmentation in Early Vyākaraṇa

35

24 The interpretation of upadeśa marking the “first teaching” comes much later: Bhaṭṭoji is the first author who states that upadeśa is ādyoccāraṇa. 25 The traditional translation of the rule is found in Cardona (1997, 58). A different translation and interpretation has been advanced in Candotti and Pontillo (2004, 2). 26 A given sound has no predictable position (it cannot, for example, predictably occur after a verbal base, like some suffixes do) unless it is taught as a substitute of another sounds that already occupies a specific place in a whole. 27 Some metarules specify, for example, that if the substitute is a string of more than one sound, it will replace the place holder in its totality while single sounds replace only the last sound of the place holder. 28 For example, units specified in terms of position and function but not meaning (the so-called svārtha-affixes) or the increments specified only in terms of position. 29 In fact “word” is defined in the Aṣṭādhyāyī as a unit ending with nominal or verbal endings, or as a nominal base triggering external phono-syntactic rules instead of internal ones. 30 “Si pour Pāṇini le principal porteur de sense est le constituant du mot, et pour Patañjali le mot lui-même, Bhartṛhari va plus loin et dit que c’est la phrase qui est le vrai porteur de sens.” 31 arthavad adhātur apratyayaḥ pratipādikam. 32 Primary derivatives such as khanitra (spade) are described through a limited set of six syntactic-semantic standard roles expressing the relationship with the action expressed by the verbal base: khan (digging) + itra (instrument for). 33 A well-known passage from Bhartṛhari (Mahābhāṣyadīpikā, 1, p. 17, ll. 22–26 ad vārttika 1) distinguishes between a vyāvahārā nityatā, “permanence in usage,” that is, a generic unavailability of language to the manipulation by the speaking subject, and a paramā nityatā, “absolute permanence,” and affirms that the former is more than sufficient to explain the linguistic mechanism, without the grammarian being obliged to take a position regarding the latter. 34 In fact, two definitions of sentence (vākya) are given while commenting on Aṣṭādhyāyī, 2.1.1. Irrespective of what definition is chosen it is clear that the sentence is considered by Kātyāyana as an independent linguistic unit that cannot be reduced to a pure accumulation (samudāya) of meaningful words. 35 For the phrase, see the discussion developed in Aṣṭādhyāyī, vārttika 1–6 ad 1.2.45. 36 arthavattā nopapapadyate kevalenāvacanāt. 37 Patañjali, commenting on this aphorisms, gives the following example: “In common parlance if someone says vṛkṣaḥ ‘tree’ (singular, masculine), a certain linguistic sound is heard: the linguistic sound vṛkṣa- ending with a and the ending -s. It also conveys a certain meaning, that is the object with roots, trunk, fruits and leaves, and the singular number. If someone then says vṛkṣau ‘two trees,’ a sound is lost, one arises and one remains. The sound -s is lost, the sound -au rises and

36

Maria Piera Candotti

vṛkṣa- ending with a is maintained. And at the same time, a certain meaning is lost, one arises and another is maintained. The singular is lost, the dual arises and the object endowed with roots, trunk, fruits and leaves is maintained. Consequently we think that the meaning that disappears is the meaning of the sound that disappears, the meaning that arises is the meaning of the sound that arises and the meaning that remains is the meaning of the sound that remains.” 38 More precisely, Kātyāyana is here discussing the possibility of giving a purely semantic definition of dhātu (verbal base) by defining it as an element which signifies an action (kriyāvacana). 39 pratyayārthasya avyatirekāt prakṛtyantareṣu […] dhātoś ca arthābhedāt pratyayāntareṣu […] siddhaṃ tu anvayavyatirekābhyām. 40 Aṣṭādhyāyī, 2.2.29, vārttika 15 (Mahābhāṣya, vol. 1, p. 433,25): “abhidhānaṃ ca punaḥ svābhāvikam.” Similarly, see also Mahābhāṣya, vol. 1, p. 242,33. 41 For a discussion of the philosophical implications of the term svabhāva, see Candotti and Pontillo (2015). 42 Crucial here also the terminological shift from the grammatical concept of pada (word) to the more general concept of whole or set of elements, expressed with a word, saṅghata, with definite Buddhist background. 43 saṅghātārthavattvāt ceti ced dṛṣṭa hy atadarthena guṇena guṇino ’rthabhāvaḥ. 44 See Aṣṭādhyāyī, vārttika 5 ad 1.1.70. 45 For a recent fresh discussion of Patañjali’s sphoṭa, see Bronkhorst (2016, 312–16). 46 See Mahābhāṣyadīpikā and Bronkhorst (1998). A fresh discussion may be found in Bronkhorst (2016, 302–59). 47 na hy antareṇa tadvācinaḥ śabdasya prayogaṃ tasyārthasya gatir bhavati.

Primary Sources Aṣṭādhyāyī

Sharma, Ram Nath, ed. 1987–2003. The Aṣṭādhyāyī of Pāṇini. Edited and translated, 6 vols. New Delhi: Munshiram Manoharlal.

Mahābhāṣya

Kielhorn, Franz, ed. 1880–85. The Vyākaraṇa-Mahābhāṣya of Patañjali. 3 vols. Third edition, revised and furnished with additional readings, references and select critical notes by K. V. Abhyankar, Pune: Bhandarkar Oriental Research Institute, 1962–72. Bombay: Government Central Book Depôt.

Mahābhāṣyadīpikā

Bronkhorst, Johannes, ed. 1987. Mahābhāṣyadīpikā of Bhartṛhari. Fascicule IV, Āhnika I. Pune: Bhandarkar Oriental Research Institute.

Linguistic Segmentation in Early Vyākaraṇa

37

Nirukta

Sarup, Lakshman, ed. 1984. The Nighaṇṭu and the Nirukta. Original edition 1927. Delhi: Motilal Banarsidass.

Ṛg Veda

Sontakke, N.S, and C. G. Kashikar, eds. 1933–51. ṚgvedaSaṃhitā. With the Commentary of Sāyaṇācārya. Pune: Vaidika Samshodana Mandala.

Śatapatha Br.

Weber, Albrecht, ed. 1964. The Śatapatha-Brāhmaṇa in the Mādhyandinaśākhā with Extracts from the Commentaries of Sāyaṇa, Harisvāmin and Dvivedagaṅga. Original edition 1855. Varanasi: Chowkhamba Sanskrit Series Office.

Taittirīya Saṃ.

Kashyap, Rangasami L., ed. 2004. Kṛṣṇa Yajur Vedīya Taittirīya Saṃhitā (Samagramūlam). Bangalore: Sri Aurobindo Kapali Sastry Institute of Vedic Culture.

Secondary Sources Bali, Suryakant. 1976. Bhaṭṭoji Dikṣita. His Contribution to Sanskrit Grammar. New Delhi: Munshiram Manoharlal. Bronkhorst, Johannes. 1998. “Les éléments linguistiques porteurs de sens dans la tradition grammaticale du Sanscrit.” Histoire Épistémologie Langage 20 (1): 29–38. Bronkhorst, Johannes. 1999. Language et réalité: sur un épisode de la pensée indienne. English translation 2011, Language and Reality. On an episode in Indian Thought. Leiden, Brill. Turnhout: Brepols. Bronkhorst, Johannes. 2016. How the Brahmins Won: From Alexander to the Guptas. Leiden: Brill. Buiskool, Herman Eildert. 1939. The Tripādī, Being an Abridged English Recast of Pūrvatrāsiddham (An Analytical-Synthetical Inquiry into the System of the Last Three Chapters of Pāṇini’s Aṣṭādhyāyī). Leiden: Brill. Candotti, Maria Piera. 2012. “The Role and Import of the Metalinguistic Chapters in the ‘New’ Pāṇinian Grammars.” In Saṃskṛta-sādhutā, “Goodness of Sanskrit.” Studies in Honour of Professor Ashok N. Aklujkar, ed. by Chikafumi Watanabe, Michele Desmarais, and Yoshichika Honda, 86–99. New Delhi: D.K. Printworld. Candotti, Maria Piera, and Tiziana Pontillo. 2004. “Substitution as a Descriptive Model in Pāṇini’s Grammar: Towards an Opposition between Phonological and Morphological Levels.” In Atti del secondo incontro genovese di studî Vedici e Pāṇiniani: Genova, 23 luglio 2003; 15 ottobre 2003, ed. by Rosa Ronzitti, and Guido Borghi, 1–45. Recco: Le Mani Microart’s Edizioni. Candotti, Maria Piera, and Tiziana Pontillo. 2015. “Svabhāva in Grammar. Notes on the Early History of a Philosophical Term.” Supplemento, Rivista Studi Orientali 2: 85–110.

38

Maria Piera Candotti

Candotti, Maria Piera, and Tiziana Pontillo. Forthcoming. “In Favour of a Linear Reading of the Aṣṭādhyāyī: Materials from the Discussion on Substitution.” In Proceedings of the 15th World Sanskrit Conference, Delhi: Linguistic Section, ed. by Jan Houben, 1–24. Provisional pagination. Cardona, George. 1997. Pāṇini: His Work and Its Traditions. Delhi: Motilal Banarsidass. Deshpande, Madhav Murlidhar. 1997. “Building Blocks or Useful Fictions: Changing View of Morphology in Ancient Indian Thought.” In India and Beyond: Aspects of Literature, Meaning, Ritual and Thought: Essays in Honour of Frits Staal, ed. by Dick van der Meij, 71–127. London: Kegan Paul International. Jamison, Stephanie W., and Joel P. Brereton, eds. 2014. The Rigveda. The Earliest Religious Poetry of India. 3 vols. Translated into English. New York: Oxford University Press. Joshi, Shivram Dattatray. 2009. “The Background of the Aṣṭādhyāyī.” In Sanskrit Computational Linguistics: Third International Symposium, Hyderabad, India, January 15–17, 2009. Proceedings, ed. by Amba Kulkarni and Gérard Huet, 1–5. Lecture Notes in Computer Science 5406. Berlin/Heidelberg: Springer. Kahrs, Eivind. 1998. Indian Semantic Analysis: The “Nirvacana” Tradition. Cambridge: Cambridge University Press. Maggi, Daniele. 2001. “Dividere la lingua: dal Rigveda ai più antichi trattati indiani di fonetica.” In Dal “paradigma” alla parola. Riflessioni sul metalinguaggio della linguistica. Atti del convegno Udine-Gorizia, 10–11 febbraio 1999, ed. by Cristina Vallini and Vincenzo Orioles, 311–23. Lingue, linguaggi, metalinguaggio 2. Roma: Il calamo. Mehendale, Madhukar Anant. 1978. “Nirukta Notes XX: Yāska’s Second Rule of Derivation.” In Nirukta Notes. Series II, 57–62. Pune: Deccan College Postgraduate and Research Institute. Olalde, Liudmila. 2014. Zum Begriff “Nāmarūpa”: Das Individuum im Pāli-Kanon. Lumbini/Wiesbaden: Lumbini International Research Institute/Reichert Verlag. Olivelle, Patrick. 1998. The Early Upaniṣads, Annotated Text and Translation. Oxford: Oxford University Press. Roodbergen, Jouthe A. F. 2008. Dictionary of Pāṇinian Grammatical Terminology. Pune: Bhandarkar Oriental Research Institute. Scharfe, Hartmut. 1977. Grammatical Literature. A History of Indian Literature 5. Wiesbaden: Otto Harrassowitz. Thieme, Paul. 1982. “Meaning and Form of the ‘Grammar’ of Pāṇini.” Studien zur Indologie und Iranistik 8/9: 1–34.

2

From Permanent Phonemes to Words Monika Nowakowska

1 Introduction This contribution concerns the view on the nature of linguistic sounds and their recognition held by early representatives of the Vedic exegetical tradition of Mīmāṃsā. We will have a look at the foundations of Mīmāṃsā, that is, the aphorisms (Mīmāṃsāsūtra, PMS) of Jaimini (ca. 200 BCE?), and the earliest complete commentary on the aphorisms known to us, the Śabarabhāṣya (ŚBh, ca. fifth century CE).1 The main focus of Mīmāṃsā authors was the correct interpretation of the Vedic ritual as taught in the corpus of Veda, which was regarded as the elevated and epistemologically reliable source to know “ritual duty” (dharma). In Mīmāṃsā, this Vedic speech was considered authorless and impersonal (apauruṣeya), and thus free from possible defects of common speech, such as untruth, falsity, wrong intentions, and so on. It was also considered permanent, transmitted, and communicated by successive generations of Vedic experts and practitioners from times immemorial. Vedic speech was also believed to be forever meaningful, bound with its meaning by a fixed and permanent relation. Vedic language, however, also had traits in common with the vernacular (laukika). Particularly, like vernacular, it had to be understood in order to be the effective medium of ritual injunctions. As such, in both its two Vedic and vernacular renditions, language was considered in Mīmāṃsā as a given and forever meaningful entity. What is speech? What is a meaningful linguistic unit and its articulated representation? What is the process of meaning? Such issues were the center of the debate already in the early tradition of Mīmāṃsā. For “speech,” authors mainly used the term śabda. On the one hand, śabda could refer either to a singular, particular instantiation of a linguistic expression, like a phoneme or a word, or

40

Monika Nowakowska

to speech or language in general. On the other hand, it could be conceptualized either as some ideal, unchanging entity, or as its articulation, a physical sound.2 As we will see below, in the light of Mīmāṃsā’s considerations, śabda is usually and paradigmatically understood as “word,” and thus as a meaningful, permanent, ideal linguistic item, regardless of its physical articulation. It is permanent, because it is “invariable,” “constant,” or “fixed” (nitya), even while being physically articulated at specific moments in time. As such, a word was to be bound by an equally fixed and invariable semantic relation with its fixed and invariable meaning. This conceptualization of śabda met with a wide range of criticism, some of which will concisely be reported below. But it was exactly through their opponents’ criticism that Mīmāṃsā thinkers were able to develop more refined theories to support their position. Another term that plays a very important role in the discussion is varṇa, sometimes substituted synonymically with the word akṣara, which is understood in our context as an atomic speech sound unit. In the light of the above, we need to interpret this term more like a phoneme than a phone. This is because in Mīmāṃsā, as we will see below, varṇas are fixed, permanent, single, and idealized speech units. They are not identified with their vocalizations but are rather manifested by them. The distinction between the idealized speech unit and its vocalization is referred to and emphatically stressed by Mīmāṃsā authors, though often intentionally dismissed or not admitted by their opponents. Consequently, in this contribution we will mostly use “phoneme” for the term varṇa. However, whenever the counterarguments specifically use varṇa for “vocalization of phoneme,” we will rather use the more neutral, unspecified expression “speech sound” to signal the argumentative oscillation of the term interpretation. In Sanskrit philosophical literature, the discussion was typically staged as an exchange of arguments between at least two contrasting theses, sometimes among a larger number of disputants. The debate was presented in a dialogic structure, giving voice first to the opponent’s standpoint, a prima facie view, which was followed by the rejoinder from the author’s standpoint. Language, especially in its Vedic realization and in relation to the exegesis of the ritual, is clearly the focus of Mīmāṃsā investigation. Nevertheless, this does not mean that this tradition analyzes grammatical or lexical categories on a regular basis. It is rather interested in larger structures, that is, sentences and texts, their pragmatic and deontic capacity (see Chapter 15 in this volume), and their interpretation. However, when confronted with objections to its axioms of the Veda’s validity, impersonality, and necessary, timeless, original

From Permanent Phonemes to Words

41

meaningfulness, it is also sometimes drawn into discussions of ontological issues. One of Mīmāṃsā’s fundamental tenets is the naturality and fixedness of the relation between words and meanings. Thus, it has also to deliberate on the nature of the members of this relation.

2  Śabda is Phonemes 2.1  The Definition of śabda The topic of phonemes as the building bricks of words is first raised in the ŚBh, more specifically in the “Commentator’s Section” (Vṛttikāragrantha, VKG, 38–40), a fragment of some earlier commentary on the PMS by the Vṛttikāra (Commentator), which is included by Śabara in his text. This section of the ŚBh introduces some strikingly novel material on various topics of great philosophical significance and consequence, which are not much discussed elsewhere by Jaimini and Śabara. The broader context is epistemological: Jaimini distinguishes speech in the form of ritual injunction from other means of valid knowledge (pramāṇa), particularly perception. Ritual injunction is the only source of information on imperceptible things (PMS, 1.1.5). This special epistemic status of śabda, however, requires the solid grounding of its various aspects, one of which is the word-meaning relation. The question of what a word is appears precisely in the discussion of the original and fixed relation among words and meanings, just after the charge from the opponent of an artificial (kṛtaka) character of such relation in the context of the language acquisition process, when it is pointed out that words and meanings are irreconcilably different and distinct. When someone utters the word “cow” (gauḥ), the listener understands an object (artha) possessing a dewlap or hooves. The word and its meaning are two unrelated entities, cognized and categorized in a different way and in different loci: “śabda is grasped in the mouth, artha is grasped on the ground.”3 The prima facie view interprets the denotative situation and its members literally, expressing referential realism, but this naïve, simplified approach is in fact discursively needed as it prompts more subtle differentiations and qualifications on the route desired by Mīmāṃsā authors. The exchange on the word-meaning relation leads the discussion toward the fundamental question of the exact identification of śabda, in the utterance of a word such as “gauḥ.”4 The Vṛttikāra refers to a definition of śabda by

42

Monika Nowakowska

a former authority (Upavarṣa),5 who described it quite simply as phonemes, enumerating them; so in the example of a word “gauḥ,” it consists of the voiced velar consonant /g/, the diphthong vowel /au/ and the voiceless fricative /ḥ/ (gakāraukāravisarjanīyā iti). The Vṛttikāra in addition points to a common conceptualization of a word, or any meaningful linguistic unit, as an “object of auditory perception.” And such an object precisely consists of individual speech sounds, phonemes.6

2.2  Critique of the Definition 2.2.1  The Impermanence of Speech Sounds The definition meets with criticism from those who claim that in order to understand the meaning of a word, and to identify speech sounds as a word, one has to grasp not just a group of sounds but rather a different whole, which is distinct from phonemes. The reason for such a claim is the impermanence of speech sounds: they resound and then fall silent, one after the other and never simultaneously. Therefore they cannot support the cognition of a whole word.7 One of the possible solutions to this problem is found by resorting to memory, analyzing the process of recollecting the spoken sounds, which could lead to the understanding of a given word. However, according to opponents such as the Buddhists, recollections as memory contents are as impermanent as speech sounds, and therefore they would not be a solution to the problem of grasping a whole word.8 The second explanation, supported by the Vṛttikāra, resorts to the unseen effects (saṃskāras, literally “arrangements”) that each and every speech sound leaves behind, while the cognition of a word happens with the resounding of the last speech sound accompanied by the effects (saṃskāras) produced by previous speech sounds.9 These saṃskāras might eventually be interpreted as mental impressions, which is how the term is understood in the context of cognitive processes more common in other currents of Indian philosophy as well as in later Mīmāṃsā. But as Kataoka (1999) demonstrated, in the ŚBh the term saṃskāra still had the ritual connotation of the effect of a preparatory, purificatory, and subservient action, which enriches the ritual material with the capacity required for a particular purpose. In the context of grasping a meaningful word, the Vṛttikāra proposes that successive speech sounds would give rise to unseen effects and capacities, which by accompanying the last speech sound would contribute to the cognition of the word.

From Permanent Phonemes to Words

43

2.2.2  The Popular Understanding The above explanation is rather complex and stands against the popular notion, expressed in everyday speech (laukikaṃ vacanam), that “we know the meaning of a word from the very word,” rather than from its parts, the phonemes.10 This may seem, on the surface, a rather naïve observation, but it refers to a general intuitive recognition that words are identified and distinguished as whole units. The objection, however, doesn’t bother the Vṛttikāra, because vernacular speech has a limited reliability. Statements of improbable or meaningless character (anupapannārtha), after all, are not that uncommon.11 People do utter various things that do not necessarily make sense. An unconditional dependence on common and otherwise ungrounded notions based on everyday human speech would be unreasonable. Mīmāṃsā authors clearly distinguish the directly valid verbal knowledge embodied by the Veda (vaidikaṃ vacanam) from humanly made statements (laukikaṃ vacanam), whose validity depends on other factors. In this way, the Vṛttikāra deflects the objection onto the general ground of unreliability of human speech.

2.2.3  The Language Experts’ Understanding A more serious argument, built after dismissing the popular understanding of the process of word or speech recognition, is based on the recognized and scientific approach to language. Here, the opponent refers to the authority of Yāska’s etymological treatise, the Nirukta (cf. Houben 1997, 71–74), which defines a word, for example a verb, as an entity (bhāva) having a sequence (pūr vāparībhūta).12 Vṛttikāra’s rejoinder is that even the opinion of an expert is not sufficient to prove something that is not grasped by an accepted means of valid knowledge (pramāṇa).13 It is safer to stand by the hypothesis of phonemes and their unseen effects (saṃskāras).

2.2.4  The Semantic Indirectness of Speech Sounds This hypothesis, however, raises the question of the primacy of speech in the process of cognizing a meaning: if the cognition is caused by the unseen effects, then speech itself, that is, the phonemes, must be considered secondary (gauṇa) in the process, and not as directly contributing to the meaning comprehension. This charge cannot be accepted, however, as speech sounds are indeed the necessary causes (nimitta) of the cognition of a word meaning. The cognitive process happens only when speech sounds are there, when they resound. If they do not, there is no comprehension of the meaning.14

44

Monika Nowakowska

Furthermore, even the possibility of a secondary role of speech cannot outweigh the difficulty of the lack of any cognitive basis for assuming some word entity distinct from the phonemes. The Vṛttikāra adds that words do not always mean straight away what they express; sometimes they indirectly suggest something else. One should not be afraid of figurativeness.15 This remark sheds a little more light on this problem of indirectness, as it stresses its semantic aspect: if a word consists in its phonemes, and yet the phonemes do not deliver the word meaning directly, but rather through their unseen effects, then phonemes do not immediately denote something, but rather they point to something else.

2.3  The Conclusion—Word Is Phonemes The Vṛttikāra concludes that the meaning of a word is grasped only on the condition that the phonemes building the word are grasped. There is no real entity, apart from them (tebhyo na vyatirikta), which is called “word” (pada). Otherwise said, “a word is nothing but phonemes” (padam akṣarāṇy eva).16 Although a theory involving unseen effects (saṃskāras) requires the assumption of something not perceptible (adṛṣṭakalpanā),17 this remains a more economic theory, because the opponent’s theory would require the double assumption of a word, an unseen entity which is different from phonemes, and of the unseen effects of phonemes. Therefore, a word is nothing but phonemes.18 What phonemes exactly are, however, remains to be explained, as the Vṛttikāra does not elaborate on their nature.

3  Speech Units Are Permanent For Mīmāṃsā authors the fundamental aspect of language was its timeless continuance, because the speech embodied by the Veda was the only source of valid information on dharma. Therefore, word permanency is already found as a topic at the beginning of the Mīmāṃsā aphorisms (PMS, 1.1.6–23) and was destined to become one of the primary linguistic theses of the tradition, as well as the later subject of extensive discussions by Śabara and other commentators.19 Since it was already found in Jaimini’s aphorisms, textually and historically the debate sketched below was initiated in Mīmāṃsā prior to and independently of the discussion on phonemes. Thus, some of the arguments in both exchanges draw from the same domains, although the claims put forward in the following

From Permanent Phonemes to Words

45

addressed a higher level of speech analysis, not of the phonemes of a word but rather of the meaningful words of a language.

3.1  Arguments against the Permanence of Words Śabara begins his analysis with the prima facie view, objecting that the wordmeaning relation cannot be fixed or permanent, if words themselves are not permanent. And words are indeed ephemeral. A permanent relation between an impermanent word and a meaning is impossible. This is corroborated by the fact that nobody can know the meaning of a word he hears for the first time.20 Most of the arguments raised by Jaimini and Śabara refer to the everyday experience of speech and to its perception. As one would expect, they are focused on the sonic realization of speech sounds, and not on ideal and unchanging linguistic units. This is because these Mīmāṃsā authors aim at a conclusive distinction of the two aspects of language, which allows them to refute most of the objections to the permanence of speech as inaccurate and off topic.

3.1.1  Articulatory Effort First, the opponent raises the issue that words are impermanent because from observation one can infer a strong correlation between the utterer’s articulatory effort (prayatna) and the articulated word, which implies that words are produced by this effort. If one argues that the articulatory effort does not produce phonemes, but rather manifests them, this is also not possible: entities are said to be manifested only when they exist prior to their manifestation, and yet there is no proof of an existence of words prior to their utterance.21

3.1.2  Momentariness of Speech Sounds Moreover, words resound only for a moment, when articulated, and then they fall silent. As for this silence, it cannot be explained as non-perceived presence, because when something is present while not perceived, the obstacle impeding the perception is perceived. And one cannot say that words are detached from their substratum, because their substratum is ether (ākāśa), a notoriously omnipresent element, which is consequently present in the hearing organ while no sound is grasped.22 This objection points out an incoherence in the Mīmāṃsā thesis: if the locus of speech sounds is ether, an ubiquitous element, speech sounds—again interpreted as vocalizations—must also be ubiquitous and thus perceivable at any time.

46

Monika Nowakowska

3.1.3  Common Usage A further objection, which might be called an argument from linguistic usage, is inherited from earlier grammatical literature. Speakers commonly use expressions such as “Do a speech!” or “Do not make a sound!,” using the verbal root “to do,” which also means “to make” or “to produce,” alongside the term “śabda.”23 The argument, based on the semantic value of the root “to do” (kṛ-), indirectly refers to the daily human experience of generating speech sounds, and emphasizes the potentially endless and new articulations of words uttered every day in various contexts, without any feeling of these words’ sameness or permanent existence.

3.1.4  Simultaneous Multilocation Another reason why speech is impermanent is that one and the same word can be heard at the same time in different places, which implies again that there is not one eternal and permanent specimen of a given word with one permanent meaning. This plurality of loci would be impossible if a word is supposed not to be particularized, but rather single and permanent.24

3.1.5  Acoustic Variability Furthermore, words tend to undergo modifications and changes, for example, the last phoneme of the word dadhi, because of the euphonic requirements dictated by the adjacent word “atra,” is transformed from /i/ into /y/ (dadhyatra). Something that undergoes change is by definition not permanent.25 Moreover, words do change not only in form but also in volume, for instance, when uttered by many people in unison. Changes in volume show that when words are articulated they cannot just be manifested—in which case there would not be any change of volume of sound—but they are rather produced, while every additional speaker adds to their overall loudness.26 These objections heavily rely on the human experience of language, on its physical articulation, and on its perception. They emphasize various aspects of speech in terms of verbalization as a time-constrained and productive activity of human beings.

3.2  The Defense of the Permanence of Speech In answer to the above arguments, Jaimini and Śabara restate that words are permanent and are not produced but merely manifested through contingent

From Permanent Phonemes to Words

47

aspects such as their articulation (see Section 3.1.1). Their rejoinders draw the crucial distinction between the unchanging speech and its changing vocalization. They explain that the silence of words is simply their non-perception, resulting from not coming into one’s auditory range (see Section 3.1.2), while the process of manifestation of speech (śabda) is explained by conjunctions and disjunctions (saṃyogavibhāga) of air particles, which themselves are not perceptible.27 Conjunction and disjunction are two of the qualities of space and time, known from the repertoire of ontological categories of Vaiśeṣika (see Halbfass 1992, 71). They are recalled here by both sides of the discussion to explain the process of hearing speech. Śabara’s position is that air particles impelled by articulation collide with yet unmoved adjacent air particles on all sides, which gives rise to conjunctions and disjunctions among the particles as long as the momentum carries them. These conjunctions and disjunctions are imperceptible, as the substance of air is not perceptible. Speech is heard until they stop. This also explains why śabda can be heard further away with a favorable wind.28 In reply to the argument on producing words attested by common verbal phrases (see Section 3.1.3), our authors explain how any reliance on popular usage as a source of knowledge is rather risky. Furthermore, speech can often be interpreted nonliterally. Śabara reminds his opponents of verbal expressions such as “Produce cow dung!” (gomayāni kuru), which is not meant to order anyone to (literally) make cow dung but just to gather or bring some. As for the problem of simultaneously grasping the same word in various and distinct places (see Section 3.1.4), Jaimini proposes a comparison to the sun, which is one and the same and yet observable in multiple places.29 And yet, just as with a word, the exact locus of the sun cannot be determined. As for the sensation that śabda is perceived at the same time in different locations, this is illusory, for its real locus is ether (ākāśa), which is one and single. Thus, when one and the same word, which is single, is grasped in many places, this is because of the plurality of these places and of speakers.30 More importantly, Jaimini and Śabara reply to the argument of words undergoing modifications (see Section 3.1.5) that no speech unit is actually transformed in euphonic procedures. What instead happens in the case of “dadhyatra” the phoneme /i/ is replaced by another phoneme, /y/.31 This is not the change of an original form, because a new phoneme is simply another linguistic unit. There is no material change here.32

48

Monika Nowakowska

As for the difference in volume of the sound of a phoneme,33 speech units do not have parts, so a higher volume of speech units uttered by several people cannot consist of multiplications of its supposed parts. It is rather the result of the combined effort of multiple speakers of single speech units, and then we talk about its uttering or voicing (nāda). One can differentiate and physically modify only the voice with which a word is pronounced, its physical instantiation, not its type. The difference of volume refers then to a degree of voicing, while it does not change anything in the linguistic unit itself.34

3.3  Further Arguments in Favor of Permanence Having unpacked their interpretation of speech nature, Jaimini and Śabara come to a number of more general, programmatic arguments for the permanence of words, which firstly are needed in the system and provide the only correct explanation of the functioning of the language, and secondly are epistemologically established by means of perception and testimony.

3.3.1  Valid Testimony on dharma In principle, speech units must be permanent because they have the paramount epistemological purpose of communicating to others, in the sense that they are means of valid cognition (pramāṇa), and particularly so in the case of Vedic words on ritual duty (dharma). The validity of the Veda depends on them.35 The independent durability of speech is indispensable for the coherence of the system.

3.3.2 Meaningfulness Furthermore, language is a given: it already works and enables meaningful communication. The claim that word is produced or reproduced every time it is spoken is absurd, according to Śabara, for one could not explain the repetitions of one word as just any successive reproductions of this word every time anew—identified as specimens of the same word, thanks to the similarity among the reproductions—because likeness of sounds cannot be dependable. First, there are different words with very approximate pronunciation, not at all semantically related, like mālā (garland) and śālā (house). Second, semantically related and similarly sounding terms as gauḥ (cow) and gāvī (cow) constitute totally different cases; the latter dialectical version is treated by Mīmāṃsā as an incorrect or corrupted form, which is understood only via the correct, original version.36 Therefore, speech is permanent, and it must be so because its meaning

From Permanent Phonemes to Words

49

is permanent as well. The meaning of a word, in fact, is a universal (ākṛti; PMS, 1.1.19). In his final step, Śabara recapitulates the ontological qualities of a speech unit: it never perishes, it is permanent, it is denotative, and it is single and remains the same; it may sound differently under various articulatory conditions, but it is still the same linguistic item in its semantic capacity, constantly related to its meaning. Having established all this, Śabara further strengthens the linguistic foundations of Mīmāṃsā with the assistance of the two primary means of valid cognition—namely, perception and verbal testimony.

3.3.3 Perception Śabara stresses that people normally use and recognize particular words as the very same words, not just similar ones. Furthermore, the identity of words is perceived, while similarity could be only inferred or reasoned out otherwise, and perception is always stronger than any reasoning. The argument of non-perception—if words are not always perceived they cannot be permanent—stands refuted on the strength of human experience: the nonperception of a thing that was previously perceived does not necessarily imply that it ceased to exist. Non-perception does not equal nonexistence, because the latter is determined only when all the valid means of knowledge fail. In the case of speech, this is certainly not the case, since it is cognizable by perception. Moreover, there is no possibility of perceiving or experiencing any destruction or final termination of speech, either.37 It can never be destroyed, nor are any material cause for its destruction found.38

3.3.4  The Veda’s Authority Last but not least, even the authoritative sources of verbal knowledge—namely, the Veda—speak of speech being permanent. They do not declare the fact outright, but indicate it indirectly (liṅga). For example, according to Śabara, the Veda encourages a poet to praise the god Agni with “voice that ceases not” (vacā virūpa nityayā).39 Śabara does not delve into the point further and does not offer more examples. Neither does he introduce any prima facie objection with quotations from the Veda that would indicate to the contrary. Clearly, according to him this is not the most important argument in favor of speech permanence but, nevertheless, one of a strong corroborative value. Thus, since the valid means of knowledge support the Mīmāṃsā thesis, it has to be concluded that word is permanent.

50

Monika Nowakowska

4  Conclusions—Speech Is Meaningful Sounds Arisen in the oral tradition of the Veda and Vedic ritual and responding to the requirements of Vedic exegesis, Mīmāṃsā focused on the capacity of language to communicate knowledge and to enjoin ritual activity. In this sense, Mīmāṃsā authors needed to analyze what the characteristics, processes, and conditions are that render speech such a unique means of cognition, especially in the form of the Veda. They claimed that speech is really and precisely phonemes (varṇas or akṣaras), and that these, taken in specific sequences—as subsequent Mīmāṃsā representatives would stress— are permanently meaningful. The ranges of the word sequences are learned from senior speakers, that is, from the linguistic community, in the process of language acquisition. And the phonemes which build words remain meaningful, regardless of the contingencies of specific utterers and their articulatory abilities. In this way, speech is eternal, fixed, and not touched by the impermanency and failings of the individual human medium. In all these discussions, there is one noticeable absence: Mīmāṃsā authors, in their analyses, did not consider it relevant to contemplate the medium of thought. What was thought for them? Was thought linguistic? Was the language of thought same as that of speech? It seems that it was, inferring from the Mīmāṃsā discussions on perception, where the categorization stage of perception is entangled with language. But this aspect does not appear in the discussion about śabda’s nature and qualities, which is rather focused on the more fundamental level of language acquisition and speech comprehension. Their focus of investigation is speech, and more precisely meaningful linguistic units. It is spoken or recited language and not, we should add, written or silently read language, and not even a mental or internal language. Even though Kumārila in his commentaries acknowledges that it is possible, albeit not commendable, to read the Veda from a script,40 for Mīmāṃsā authors the primary medium of the Vedic teaching as the most important source of information on ritual duty was typically vocalized and memorized speech. That is why Mīmāṃsā devoted so much attention to the processes of learning and comprehending word meanings, which required, first of all, resounding and articulated words. According to Mīmāṃsā authors, in the beginning of an individual human life there is indeed speech: phonemes, fixed and meaningful linguistic units, and the uncountable though defined potential of their communicative vocalizations.

From Permanent Phonemes to Words

51

Notes 1 2 3 4

5 6 7

8 9

10 11

12

13

On these datings, see Verpoorten 1987, 3, 8; Kataoka 2011, 14–20. Cf. also Houben (1997, 60). ŚBh, ad PMS 1.1.5 (VKG, 36): mukhe hi śabdam upalabhāmahe, bhūmāv artham. “atha gaur ity atra kaḥ śabdaḥ”? The question is echoed from Patañjali’s Mahābhāṣya (cf. Paspaśāhnika, Text, p. 4; Coward and Kunjunni Raja 2001, 68–69). On this portion of the ŚBh see also Biardeau (1964, 178–81). Quoted also later by Śaṃkara and Bhāskara, see Nakamura (2004, 47). Cf. VKG, 110; Kataoka (2011, 14, 18). ŚBh, ad PMS 1.1.5 (VKG, 38): śrotragrahaṇe hy arthe loke śabdaśabdaḥ prasiddhaḥ. te ca śrotragrahaṇāḥ. ŚBh, ad PMS 1.1.5 (VKG, 38): yady evam, arthapratyayo nopapadyate— katham?—ekaikākṣaravijñāne hy artho nopalabhyate. na cākṣaravyatirikto ’nyaḥ kaścid asti samudāyo nāma, yato ’rthapratipattiḥ syāt. yadā hi gakāro, na tadā aukārādivisarjanīyau. yadā aukāravisarjanīyau, na tadā gakāraḥ. ato gakārādivyatirikto ’nyo gośabdo ’sti, yato ’rthapratipattiḥ syāt. ŚBh, ad PMS 1.1.5 (VKG, 38): antarhite ’pi śabde smaraṇād arthapratyaya iti cen,— na. smṛter api kṣaṇikatvād akṣarais tulyatā. ŚBh, ad PMS 1.1.5 (VKG, 38): pūrvavarṇajanitasaṃskārasahito ’ntyo varṇaḥ pratyāyaka ity adoṣaḥ. See Kataoka (1999). An earlier interpretative translation of the term “saṃskāra” is “residual effect” (Jha 1983). ŚBh, ad PMS 1.1.5 (VKG, 38): nanv evam api “śabdād arthaṃ pratipadyāmaha” iti laukikaṃ vacanam anupapannaṃ syāt. Here the ŚBh contrasts a simple meaningful sentence with the standard example of a series of unrelated expressions, resulting in a nonsense utterance, known from Paspaśāhnika, 1.1.3. ŚBh, ad PMS 1.1.5 (VKG, 38): ucyate: yadi nopapadyate, anupapannaṃ nāma. na hi laukikaṃ vacanam anupapannam ity etāvatā pratyakṣādibhir anavagamyamāno ’rthaḥ śakyo ’bhyupagantum. laukikāni vacanāny upapannārthāny anupapannārthāni ca dṛśyante, yathā “devadatta, gām abhyāja” ity evamādīni “daśa dāḍimāni ṣaḍ apūpā” ity evamādīni ca. ŚBh, ad PMS 1.1.5 (VKG, 38): nanu ca śāstrakārā apy evam āhuḥ “pūrvāparībhūtaṃ bhāvam ākhyātenācaṣṭe ’vrajati, pacati’ ity upakramaprabhṛty apavargaparyantam” iti yathā. Cf. Nirukta, 1.1, p. 28: tad yatrobhe bhāvapradhāne bhavataḥ pūrvāparībhūtaṃ bhāvam ākhyātenācaṣṭe. vrajati pacatīti (Sarup: 28); “But where both have bhāva ‘being, becoming’ as the main thing, the bhāva which has a sequence is denoted by a verb, e.g. ‘he goes,’ ‘he cooks’” (Houben 1997, 72). ŚBh, ad PMS 1.1.5 (VKG, 38): na śāstrakāravacanam apy alam imam artham apramāṇakam upapādayitum.

52

Monika Nowakowska

14 ŚBh, ad PMS 1.1.5 (VKG, 40): akṣarebhyaḥ saṃskāraḥ, saṃskārād arthapratipattir iti bhavanty arthapratipattāv akṣarāṇi nimittam—gauṇa eṣa śabda iti cet,—na gauṇo ’kṣareṣu nimittabhāvaḥ, tadbhāve bhāvāt tadabhāve cābhāvāt. 15 ŚBh, ad PMS 1.1.5 (VKG, 40): athāpi gauṇaḥ syāt, na “gauṇaḥ śabdo mā bhūd” ity etāvatā pratyakṣādibhir anavagamyamāno ’rthaḥ śakyaḥ parikalpayitum. na hy “agnir māṇavaka” ity ukte ’gniśabdo gauṇo mā bhūd iti “jvalana eva māṇavaka” ity adhyavasīyate. 16 ŚBh, ad PMS 1.1.5 (VKG, 40): na ca pratyakṣo gakārādibhyo ’nyo gośabda iti, bhedadarśanābhāvād abhedadarśanāc ca. gakārādīni hi pratyakṣāṇi. tasmād “gaur” iti gakārādivisarjanīyāntaṃ padam akṣarāṇy eva. na tebhyo vyatiriktam anyat padaṃ nāma iti. 17 ŚBh, ad PMS 1.1.5 (VKG, 40): nanu saṃskārakalpanāyām apy adṛṣṭakalpanā. 18 ŚBh, ad PMS 1.1.5 (VKG, 40): ucyate: śābdakalpanāyāṃ sā ca śabdakalpanā ca. tasmād akṣarāṇy eva padam. 19 Cf. Frauwallner (1961); ŚNA. On this portion of the ŚBh, see also Biardeau (1964, 181 ff.). 20 ŚBh, ad PMS 1.1.6 (ŚNA, 51): uktaṃ nityaḥ śabdārthayoḥ saṃbandha iti. tad anupapannam, śabdasyānityatvāt. vinaṣṭaḥ śabdaḥ, punar anyasya kriyamāṇasyārthenākṛtakaḥ saṃbandho nopapadyate. na hi prathamaśrutāt kutaścana śabdāt kaścid arthaṃ pratyeti. 21 ŚBh, ad PMS 1.1.6 (ŚNA, 51–52): kathaṃ punar anityaḥ śabdaḥ. prayatnād uttarakāle dṛśyate. ataḥ prayatnānantaryāt tena kriyata iti gamyate. nanv abhivyañyāt sa enam. neti brūmaḥ. na hy asya prāg abhivyañjanāt sadbhāve kiṃcana pramāṇam asti. saṃś cābhivyajyate nāsan. 22 ŚBh, ad PMS 1.1.7 (ŚNA, 52): na cainam uccaritaṃ muhūrtam apy upalabhāmahe. ato vinaṣṭa ity avagacchāmaḥ. na ca san nopalabhyate. anupalaṃ bhakāraṇānāṃ vyavadhānādīnām abhāve ’py anupalambhanāt. na cāsau viṣayam aprāptaḥ, ākāśaviṣayatvāt, karṇacchidre ’py anupalambhanāt. 23 Cf. Paspaśāhnika, 1.1.15, pp. 6,23. PMS, 1.1.8: karotiśabdāt. ŚBh, ad PMS 1.1.8 (ŚNA, 53): “śabdaṃ kuru,” “mā śabdaṃ kārṣīḥ,” “śabdakāry ayaṃ māṇavakaḥ” iti vyavahartāraḥ prayuñjate. tena nūnam avagacchanty “asann evāyaṃ śabdaḥ” iti. Cf. also ŚNA, 569–568 (40–41). 24 ŚBh, ad PMS 1.1.9 (ŚNA, 54): nānādeśeṣu hi yugapac chabdam upalabhāmahe. tad ekasya nityasyānupapannam. asati viśeṣe nityasya nānekatvam. kāryāṇāṃ tu bahūnāṃ nānādeśeṣu kriyamāṇānām upapadyate ’nekadeśasaṃbandhaḥ. tasmād anityaḥ. 25 ŚBh, ad PMS 1.1.10 (ŚNA, 54): api ca “dadhyatra” ity atrekāraḥ prakṛtir, yakāro vikṛtir ity upapadiśanti. yad vikriyate, tad anityam. This is a recurring argument that appears in many discussions on permanence. See, for example, the debate on ātman by Kumārila in the very ŚV (ŚV, 5.(ātmavāda).20–21). 26 PMS, 1.1.11 (ŚNA, 55): vṛddhiś ca kartṛbhūmnā syāt.

From Permanent Phonemes to Words

53

27 PMS, 1.1.13 (ŚNA, 56): sataḥ param adarśanaṃ viṣayānāgamāt. See also ŚBh, ad PMS 1.1.13 (ŚNA, 56–57): atrāpi yadi śakṣyāmo nityatām asya vispaṣṭaṃ vaktum, tato nityapratyayasā marthyāt kadācid upalambhaṃ kadācid anupalambhaṃ dṛṣṭvā kiṃcid upalambhane nimittaṃ kalpayiṣyāmaḥ. tac ca saṃyogavibhāgasadbhāve sati bhavatīti saṃyogavibhāgāv evābhivyañjakāv iti vakṣyāmaḥ. 28 ŚBh, ad PMS 1.1.13 (ŚNA, 60): abhighātena hi preritā vāyavaḥ stimitāni vāyvantarāṇi pratibādhamānāḥ sarvatodikkān saṃyogavibhāgān utpādayanto yāvad vegam abhipratiṣṭhante. te ca vāyor apratyakṣatvāt saṃyogavibhāgā nopalabhyante. 29 PMS, 1.1.15 (ŚNA, 61): ādityavad yaugapadyam. Cf. ŚNA, 15. 30 ŚBh, ad PMS 1.1.15 (ŚNA, 64): ata eva vyāmoho yan nānādeśeṣu śabda iti. ākāśadeśaś ca śabdaḥ. ekaḥ punar ākāśaḥ. ato ’pi na nānādeśeṣu śabdaḥ. api caikarūpye sati deśabhede ca sati kāmaṃ deśā eva bhinnāḥ syuḥ, na śabdaḥ. tasmād ayam adoṣaḥ. 31 PMS, 1.1.16 (ŚNA, 65): śabdāntaraṃ vikāraḥ. 32 ŚBh, ad PMS 1.1.16 (ŚNA, 65): na ca “dadhyatra” ity atra prakṛtivikārabhāvaḥ. śabdāntaram ikārād yakāraḥ. 33 PMS, 1.1.17 (ŚNA, 66): nādavṛddhi parā. 34 ŚBh, ad PMS 1.1.17 (ŚNA, 66–68): niravayavo hi śabdaḥ avayavabhedānavagamāt. niravayavatvān mahattvānupapattiḥ. ato na vardhate. […] saṃyogavibhāgāś ca nairantaryeṇa kriyamāṇāḥ śabdam abhivyañjanto nādaśabdavācyāḥ. tena nādasyaiṣā vṛddhir na śabdasyeti. 35 PMS, 1.1.18 (ŚNA, 68): nityas tu syād darśanasya parārthatvāt. 36 ŚBh, ad PMS 1.1.18 (ŚNA, 70): na catraikenoccāraṇayatnena saṃvyavahāraś cārthasaṃbandhaś ca śakyate kartum. 37 ŚBh, ad PMS 1.1.20 (ŚNA, 73): na ca śabdasyānto na kṣayo lakṣyate. 38 ŚBh, ad PMS 1.1.21 (ŚNA, 76–77): naivaṃ śabdasya kiṃcit kāraṇam upalabhyate yad vināśād vinaṅkṣyatīty avagamyate. 39 Cf. ŚNA, 78. Ṛg Veda, v. 8.75.6, p. 179: tasmai nūnam abhidyave vācā virūpa nityayā, vṛṣṇe codasva suṣṭutim. The English translation, as more suiting the Mīmāṃsā interpretation, is by Griffith (1891, 299), hymn VIII.LXIV.Agni: “6 Now, O Virūpa, rouse for him, Strong God who shines at early morn, Fair praise with voice that ceases not.” The indication is not so apparent in the latest translation of the verse by S. W. Jamison (Jamison and Brereton 2014, 1172): “6. Now for him, for the heaven-bound bull, o Virupa [=poet], with your very own speech rouse your lovely praise hymn.” More on this quotation in Garge (1952, 72, 155). 40 For example, in the Tantravārttika, ad PMS 1.3.4; p. 187.

54

Monika Nowakowska

Primary Sources Mīmāṃsādarśana

Subbāśāstrī, ed. 1929–34. Śrīmajjaiminipraṇītaṃ Mīmāṃsādarśanam. Vol. 6. Anandashrama Sanskrit Series 97. Pune: Anandasramamudranalayah.

Nirukta

Sarup, Lakshman, ed. 1984. The Nighaṇṭu and the Nirukta. Original edition 1927. Delhi: Motilal Banarsidass.

Paspaśāhnika

Joshi, S. D., and J. A. F. Roodbergen, eds. 1968–90. Patañjali’s Vyākaraṇa-Mahābhāṣya. Introduction, Text, Translation and Notes. Pune: University of Poona.

PMS

Pūrvamīmāṃsāsūtra. In Mīmāṃsādarśanam.

Ṛg Veda

Aufrecht, Th., ed. 1877. Die Hymnen des Rigveda. 2. Aufl., Theil 2: Maṇḍala 7–10. Bonn: Marcus.

ŚBh

Śābarabhāṣya of Śabara. In Mīmāṃsādarśana.

ŚNA

Kataoka, Kei, ed. Śābarabhāṣya ad 1.1.6–23: Śabdanityatvādhikaraṇa. In Kataoka 2007.

ŚV

Rai, Ganga Sagar, ed. 1993. Ślokavārttika of Śrī Kumārila Bhaṭta with the Commentary Nyāyaratnākara of Śrī Pārthasārathi Miśra. 2nd edition. Varanasi: Ratna Publications.

Tantravārttika

Tantravārttikam of Kumārila. 1929–34. In Mīmāṃsādarśana.

VKG

Frauwallner, Erich, ed. Vṛttikāragrantha. In Frauwallner 1968.

Secondary Sources Biardeau, Madeleine. 1964. Théorie de la connaissance et philosophie de la parole dans le brahmanisme classique. Paris: Mouton. Coward, Harold G., and K. Kunjunni Raja, eds. 2001. Encyclopedia of Indian Philosophies. Vol. 5. First edition 1990. New Delhi: Motilal Banarsidass. Frauwallner, Erich. 1961. “Mīmāṃsāsūtram I,1,6–23.” Wiener Zeitschrift für die Kunde Süd- und Ostasiens 5: 113–24. Frauwallner, Erich. 1968. Materialien zur ältesten Erkenntnislehre der Karmamīmāṃsā. Veröffentlichungen der Kommission für Sprachen und Kulturen Süd- und Ostasiens 6. Wien: Österreichische Akademie der Wissenschaften. Garge, Damodar Vishnu. 1952. Citations in Śabara-bhāṣya: A Study. Deccan College Dissertion Series 8. Pune: Deccan College.

From Permanent Phonemes to Words

55

Griffith, Ralph T.H. 1891. The Hymns of the Rig Veda: Translated with a Popular Commentary. Vol. 3. Benares: E.J. Lazarus & Co. Halbfass, Wilhelm. 1992. On Being and What There Is: Classical Vaiśesika and the History of Indian Ontology. Albany: State University of New York Press. Houben, Jan E.M. 1997. “Part II: The Sanskrit Tradition.” In The Emergence of Semantics in Four Linguistic Traditions: Hebrew, Sanskrit, Greek, Arabic, ed. by Wout van Bekkum et al., 51–145. Amsterdam Studies in the Theory and History of Linguistic Science. Series III—Studies in the History of the Language Sciences 82. Amsterdam/Philadephia: John Benjamins Publishing Company. Jamison, Stephanie W., and Joel P. Brereton, eds. 2014. The Rigveda. The Earliest Religious Poetry of India. 3 vols. Translated into English. New York: Oxford University Press. Kataoka, Kei. 1999. “The Mīmāṃsā Concept of saṃskāra and the saṃskāra in the Process of Cognizing a Word-Meaning: pūrva-varṇa-janita-saṃskāra.” Journal of the Japanese Association for South Asian Studies 11: 1–24. Kataoka, Kei. 2007. “A Critical Edition of Śābarabhāṣya ad 1.1.6–23: Śabdanityatvādhikaraṇa.” Toyo Bunka Kenkyusho Kiyo (The Memoirs of the Institute of Oriental Culture) 152: 580–30 (29–79). Kataoka, Kei. 2011. Kumārila on Truth, Omniscience, and Killing. Part 2. An Annotated Translation of Mīmāṃsā-Ślokavārttika ad 1.1.2 (Codanāsūtra). Beiträge zur Kultur-und Geistesgeschichte Asiens 68. Wien: Verlag der Österreichischen Akademie der Wissenschaften. Nakamura, Hajime. 2004. A History of Early Vedānta Philosophy. Part Two. Translated into English by Hajime Nakamura, Trevor Leggett and others. Delhi: Motilal Banarsidass. Verpoorten, Jean-Marie. 1987. Mīmāṃsā Literature. Vol. 6. A History of Indian Literature edited by Jan Gonda 5. Wiesbaden: Otto Harrassowitz.

3

Ontology and Epistemology of Speech in Nyāya Alessandro Graheli

1 Introduction In the discipline of Nyāya, the main fields of investigation have traditionally been epistemology and logic, with a focus on the study of the causes of knowledge. Language is among such causes, so one of the central themes of most Nyāya works is its role in the acquisition of knowledge. Nyāya is also rooted in a complex system of metaphysics an idiosyncratic brand of “naive” realism developed through a symbiosis with another ancient school of thought, Vaiśeṣika, from which it inherited most of its metaphysical apparatus and with which it shared its epistemology and logic. Together, the Nyāya and Vaiśeṣika traditions have been said to have offered “one of the most vigorous efforts at the construction of a substantialist, realist ontology that the world has ever seen” (Potter 1977, 1). The oldest known work of Nyāya is an aphoristic work, called Nyāyasūtra and ascribed to Akṣapāda, also known as Gautama. It was probably composed earlier than the second century CE.1 The oldest and extant version of the Nyāyasūtra is preserved in manuscripts of the seminal commentary of Vātsyāyana (end of fifth century CE according to Potter 1977, 239), the Nyāyabhāṣya. After Vātsyāyana, the tradition of Nyāya was further promoted by another influential commentator, Uddyotakara (sixth century), who refined and expanded the Nyāyabhāṣya in his Nyāyavārttika. In the last two decades of the ninth century, the Kashmirian Bhaṭṭa Jayanta gathered the important tenets of these and other lost Nyāya authors on language in his masterpiece, the Nyāyamañjarī (henceforth NM). Although after Jayanta the Nyāya discipline has kept evolving, mostly in the area of logic, in this chapter the historical lens is focused on the tradition of Nyāya up to Jayanta’s time, particularly because this is the most prolific

Ontology and Epistemology of Speech in Nyāya

57

era on matters related to philosophy of language. Jayanta is a pivotal figure in the history of South Asian philosophy. In his influential NM, he discusses the mainstream theories of language developed until his time, and he presents a mature and lucid view on the epistemology of testimony, as well as his views on semantics. The paradigm of Nyāya and of Jayanta, in brief, is that speech is a specific type of sound, which in the ontology of Nyāya-Vaiśeṣika is the quality inhering in the fifth element, “ether,” and it is by definition ephemeral, not permanent. Epistemologically, the primary purpose of speech is the communication of knowledge, that is, testimonial knowledge. As such, speech is an epistemological category, to be studied along with perception and inferential processes.

2  The Ontology of the Ephemeral Speech In Nyāya, the principle of the authority (āpta) applies to speech in general, so that it is considered the ground for the reliability of both common and Vedic language. Even the authority of the Veda is founded on this principle, unlike in Mīmāṃsā. Furthermore, the foundation of paradigmatic speech on its trustworthy speaker necessarily implies that speech is not permanent, since any teacher’s utterance must occur at some definite point in time. If speech is impermanent, moreover, it is difficult to explain how an utterance and its referent can have a natural, fixed, or permanent relation. In regard to this problem, Nyāya authors need to find a coherent justification of the impermanence of metaphysical sound, since in their ontology speech is a specific type of sound. The metaphysics of Nyāya is largely inherited from the atomistic tradition of early Vaiśeṣika, with minor variations. The world, reality, is constituted of different entities, all having specific and general characteristics, independently of our subjective thinking. The phenomenic world of positive differences is narrowed down into a map of six most basic categories, which are in principle all real, knowable, and nameable.2 The six categories are substance, quality, action, generality, specificity, and inherence. Each of these six categories has further subcategories. In this scheme, sound is listed as one of the twenty-four qualities (shown in uppercase in Figure 3.1). Moreover, it is a quality that archetypically inheres in ether (ākāśa), one of the nine substances. A peculiar view shared by Vaiśeṣika and Nyāya, therefore, is that sound is a quality,3 and not a substance, that it inheres in ether (ākāśa), and, most relevantly for the present purposes, that it is held to be impermanent. The question of

58

Alessandro Graheli

Figure 3.1  Sound in the Metaphysical Map of Vaiśeṣika (3) Activity

(2) Quality

(1) Substance earth

color

weight

going up

water

taste

fludity

going down

fire

smell

viscosity

contracting

air

touch

knowledge

expanding

ether

*SOUND*

pleasure

moving

time

number

pain

space

dimension

desire

(4) Universal

self

distinction

hatred

upper lower

mind (6) Inherence

contact

effort

disjunction

merit

farness

demerit

nearness

disposition

(5) Specificity

impermanence of sound is scrutinized in Nyāyasūtra, 2.2.13–39. Though this is not the occasion for getting into the minutiae of the argumentation,4 in sum three major reasons are adduced in favor of the impermanence of sound: (1) Sound has a beginning, and it is generated by specific causes, such as the contact between an axe and the wood, or the tongue and the palate. What has a beginning is by definition impermanent. (2) Sound is the object of perception, and what is perceived is necessarily impermanent. (3) Sound is generally described as possessing the typical characteristics of products, for instance, when it is characterized as “loud” or “soft.” Vidyabhusana (1921, 109–10) explains how (2) and (3) are vulnerable to rebuttals, while (1) seems indeed a better and more widely acceptable argument, consistent with the metaphysical tenets of Nyāya and yet less bound to the dogmatic assumption of the tradition. In any case, once it is established that sound in general is impermanent, it follows that speech is impermanent, too, because it is included into this broader category of sound, which comprises both articulate and inarticulate sounds.5 A more mature stage of the debate is found in Jayanta’s defense of the impermanence of speech (NM, I 513–72). Jayanta corrals the Nyāya stance against the arguments of the Bhāṭṭa Mīmāṃsā authors, among others. These

Ontology and Epistemology of Speech in Nyāya

59

authors maintain that the permanence of phonemes is proven on the strength of their recognition (pratyabhijñā), always in constant patterns and at different times, and through an appeal to postulation (arthāpatti), because otherwise the constant character of signification and of intersubjective communication could not be explained. The two arguments, however, collapse if we can establish a universal property for each phoneme. When a speaker utters gagana (sky), is there any difference between the two “g” sounds produced and recognized? If we say that there is one single “g,” we end up with some sort of linguistic monism that is not acceptable for both Naiyāyikas and Mīmāṃsakas. If we say that these are two different “g”s, we must admit that they have a property in common, their g-ness, that is, the universal of the phoneme “g.” Jayanta’s conclusion is that phonemes are a product, and if they are a product we must admit a creating agency prior to them. If phonemes are originated, then even words and sentences, which are made of phonemes, cannot be considered permanent. Not only that, but even if words are believed to have fixed referents, sentences are compositions of words, so a sentient agent must have been the author of such compositions. Jayanta concludes his defense of impermanence by quoting the following inferential argument from Nyāyavārttika, ad 1.1.5, p. 43,8, based on the implicit major premiss that if something possesses a universal, while being perceivable through external senses, then it is necessarily impermanent: Sound possesses impermanence. Because it is something perceived through human external senses and is endowed with a universal. Like a pot.

The specification “being endowed with a universal” is needed in order to exclude the universal of sound, sound-ness, as a possible probandum. In Vaiśeṣika ontology, universals are by definition permanent, so obviously if soundness were the probandum, the inference would misfire. The specification “through human external senses” is needed to exclude the self as the probandum, because the self is also permanent by definition. Kumārila sarcastically criticized this inference as inconclusive, because one does perceive a universal, say cow-ness, which possesses another universal, say hump-ness: Who would use logical reasons such as “possessing a universal” or “being perceivable,” which are linked to any existent thing and are adduced to prove the impermanence of sound? Certainly not a logician!6

60

Alessandro Graheli

From Jayanta’s standpoint, however, this criticism entails the absurd tenet that a universal can inhere in another universal,7 while among Nyāya scholars it is widely known that universals do not inhere in other universals.8 Therefore, “it is the statement of someone utterly ignorant of the logicians’ ways.”9 Leaving aside the merit of the respective beliefs in Mīmāṃsā and Nyāya, Jayanta’s argument reveals the ambiguity in the use of the terms jāti (genus) and sāmānya (universal) as synonyms, something that Kumārila would probably not accept. Incidentally, in the economy of this volume the exact acceptation of the two terms becomes relevant in the semantics of word meaning and sentence meaning, which are discussed in terms of universals and particularities in Mīmāṃsā (see Chapter 13 in this volume) and of possessors of universals in Nyāya (cf. Chapters 7 and 11 in this volume).

3  Segments of Speech: From Phonemes to Words As in Mīmāṃsā, also in the ontology of Nyāya the discrete speech units (varṇa) composing words are treated as real entities, with the additional problem that in Nyāya such speech segments are identified as the actual sounds perceived by the hearer. The distinction between such levels of parole and langue levels, to use the Saussurian terms, must be differently interpreted when dealing with the views of the two traditions. In Nyāya, particularly, the sounds contingently uttered by each speaker are even considered ephemeral events by definition, so authors like Jayanta need some explaining to ensure some stable character of speech units (varṇa) that guarantees communication. In earlier Nyāya works, one cannot find an articulated explanation of the process that leads from the minimal speech units to the formation of words, and to the understanding of their meaning, but the process is discussed in great detail by Jayanta. His ideas on meaning have clearly evolved from centuries of disputes between Nyāya, Vyākaraṇa, Mīmāṃsā, and Buddhist authors. After earnestly presenting several such views, in his NM, 143–84, Jayanta concludes that the phonemes are the causes of the knowledge of the meant object, in disagreement with the proponent of the sphoṭa (cf. Chapter 4 in this volume) and largely sharing the views of Śabara and Kumārila: a string of phonemic sounds in a fixed sequence, which forms a word through the complex psychological activity of the hearer, generates knowledge of the meant object. The process involves a sequence of perceptual moments (anubhava), which generate mental dispositions (saṃskāra), and which in turn trigger a sequence of recollections

Ontology and Epistemology of Speech in Nyāya

61

(smaraṇa). Jayanta’s version of the theory is an elaboration of the principle set by the Mīmāṃsā commentator in the Vṛttikāragrantha, 38,13: “The last phoneme, assisted by the dispositions generated by the previous phonemes, is the cause of knowledge of a word-meaning.”10 If a word is composed of atomic phonemes, indeed, one needs to explain how such atomic units are perceived and synthetized to form complete and meaningful words. Śabara’s model works as follows: While listening the word gauḥ (“cow”), the listener first hears “g.” The result is a “g” experience. This experience is embedded in the mind in the form of a mental trace, a disposition “g.” Then she hears the next phoneme, “au,” which results in an “au” experience, and in a disposition “au.” Then she hears the last phoneme, ḥ, which generates the analogous “ḥ” experience. According to Śabara’s version, this “ḥ” experience, assisted by the recollection generated by the dispositions “g” and “au,” is enough to explain the hearing cognition of a complete word, which then causes the knowledge of the cow. From the Nyāya point of view, this explanation presents several problems. First, since a phoneme is a sound, and since sound is impermanent, the three phonemes are also impermanent, that is, momentary. The three momentary sounds, therefore, must occur in sequence. The main difficulty is to explain the cognitive process without transgressing the crucial principle of the impossibility of simultaneous cognitions (jñānayaugapadya), a Nyāya-Vaiśeṣika tenet based on the idea that the mind, one of the nine substances (see Figure 3.1), is atomic in nature, and that a perception is nothing but the combined effort of the atomic mind and an organ of sense. For these reasons, the perception of the three momentary sounds cannot occur at the same moment. Furthermore, even a recollection requires the activity of the mind, so it cannot occur simultaneously with the perception of “ḥ.” The problem of simultaneous cognitions is further exacerbated by the need to account for the cognition of the meaning of gauḥ, and since words are used and are truly meaningful only in sentences, one needs to put together an even more problematic string of perceptual moments, dispositions, recollections, known meanings, dispositions of meanings, and recollected meanings. Before Jayanta, to my knowledge, no Nyāya author ventured into this conund­ rum, and Jayanta himself seems not to be fully satisfied with his own solution. Eventually, after exploring several models leading from the single phonemes to the understanding of the sentence meanings, he proposes the following theory: At first, according to the adequate phonemic sequence, there is the cognition of the first word [according to the principles explained above]. Then, simultaneously, the recollection of the conventional meaning of the word (saṅketa) and the

62

Alessandro Graheli disposition of the first word occur, because the prohibition of simultaneity pertains to two cognitive acts and not to a disposition together with a cognitive act. Then there is the cognition of the word meaning, which in turn generates a disposition. Then, again, according to the phonemic sequence, there is the cognition of the second word. Then there is the recollection of the meaning of the second word. Then an intensified disposition occurs, i.e., a disposition of the second word assisted by the disposition of the first. Then there is knowledge of this second word meaning. Then there is an intensified disposition of the second word meaning, with the assistance of the disposition of the first word meaning. Then, again, according to the phonemic sequence, there are the perception of the third word, the recollection of the word meaning, a more intense disposition assisted by the previous dispositions [i.e., a disposition including the first three words together]. Accordingly, there is knowledge of the third word meaning, and from that, accordingly, there is a most intense disposition [of the three word meanings]. This is how—with a standing and enriched disposition generated by the cognition of the words, and a further standing and enriched disposition generated by the knowledge of the word meaning, two recollections in succession are generated, because of the original succession of the two enriched disposition: (1) the recollection of the previous words caused by the dispositions of the three words, occurring right after the knowledge of the third word meaning, and (2) the recollection of the word meanings, caused by the dispositions of the three word meanings. Of these two, the sentence is the group of the words carried by the first recollection, and the sentence meaning is the group of the word meanings carried by the other recollection.11

This description of the process, however, exposes a crucial vulnerability of speech as testimony, because memory is not in the list of four epistemic processes and it is notoriously unreliable; this mediating role of memory may undermine the veracity of the acquired knowledge. Jayanta (NM, II 201,4 and ff.) seems to have been aware of this problem, and as a tentative correction he proposes an alternative explanation of the psychological process, replacing the moments of recollection with instances of apperceptive cognition (anuvyavasāya). The merits and flaws of this alternative model, however, will have to be discussed elsewhere.

4  Epistemology of Speech 4.1  The Four Epistemic Conditions The main interest of Nyāya is made clear at the very beginning of the Nyāyasūtra, whose very first word is pramāṇa (epistemic instrument). This term is glossed as

Ontology and Epistemology of Speech in Nyāya

63

Figure 3.2  The Four Factors of Knowledge epistemic instrument epistemic agent

epistemic object

episteme

designates a causal relation

“that through which one knows” by Vātsyāyana, where “knowing” is predicated on an objective and external reality.12 Vātsyāyana lists two further conditions along with the epistemic instrument, the epistemic agent (pramātṛ) and the epistemic object (prameya). When all these conditions satisfactorily meet, the ensuing result is knowledge (pramiti). An act of knowledge thus involves four interrelated items (Figure 3.2), traditionally referred to as the “pramāṇaquadruplet” (pramāṇacatuṣṭaya). The three terms listed by Vātsyāyana as conditions for knowledge are all construed from the verbal root pra-mā- with the addition of primary suffixes that indicate precise functions in the act of knowing: an agent, an object, and an instrument. The terminology used to describe this epistemological model is clearly inspired by the semantic-syntactic paradigm of Pāṇinian grammar. In his Aṣṭādhyāyī, 1.4.23–54 Pāṇini defines six such semanticsyntactic relations (kāraka): (1) independent agent, (2) direct object, (3) instrumental cause, (4) indirect object, (5) point of departure, and (6) substrate, which are denoted by the nominative, accusative, instrumental, dative, ablative, and locative, that is, the nominal endings of Sanskrit language, respectively.13

4.2  Instrument, Complex of Causes, or Process? Let’s take visual perception as a standard of comparison. According to Vātsyāyana’s grammatical explanation of the term pramāṇa, what exactly is this instrument? Is it the eye organ? The light? The mind? The self? Or the external object?14 All of these are instrumental in the production of knowledge.

64

Alessandro Graheli

After scrutinizing the issue, Jayanta concludes that a pramāṇa is not just a single epistemic instrument but, rather, a complex causal system (sāmagrī), and elaborates a more detailed definition: A pramāṇa is a system of causes—that may be cognitions as well as noncognitions, and are distinct from the agent and the object—that produces nonambiguous and non-erroneous knowledge of an object.15

An act of knowledge is thus conditioned by multiple factors of action (kārakas), and not just by a single instrument denoted by an instrumental case: just as “the cook is cooking rice” actually entails an agent who is boiling rice grains in a pot by means of fire, in a cognitive operation culminating in knowledge there are multiple causal factors at play: Pre-determined factors of action such as rice grains, water, fire, pot, etc., all contribute to the realization of the unitary action of cooking. Similarly, the activity called “knowing” occurs on the precondition of a connection of the self, the senses, the mind, and the external object.16

For instance, when at night an agent becomes aware of seeing an object through the light of a lamp, the pramāṇa consists of multiple factors that include the light of the lamp, the eyes, and the mind, rather than a single instrument (Figure 3.3). Moreover, even the perception of the light of the lamp is needed, and therefore we have a case of a piece of knowledge (the perception of light) that becomes instrumental for a second act of knowledge (the perception of the object seen through the light of the lamp).17

Figure 3.3  The Causal Apparatus epistemic complex of necessary conditions epistemic agent

epistemic object

episteme

designates a causal relation

Ontology and Epistemology of Speech in Nyāya

65

4.3  Speech as One of the Four Epistemic Processes The debate on the taxonomy of epistemic instruments constitutes a significant portion of the epistemological debate in Sanskrit literature, because of the many views about the nature, number, quality, reliability, hierarchy, and interdependence of such epistemic instruments. Nyāya authors, on the authority of Nyāyasūtra, 1.1.4, accept and defend four epistemic processes: ●●

●●

●●

●●

Perception Inference Comparison Speech, intended as reliable testimony

Thus speech is understood as having the same epistemological status of perception, inference, and comparison, insofar as it can yield knowledge of reality. In the Nyāyasūtra it is specifically defined as “the instruction (upadeśa) of an authority (āpta).”18 Several objections have been raised against this sui generis role of speech, which are presented, discussed, and refuted extensively by Jayanta and other Nyāya authors. These cannot be discussed here, as they fall beyond the scope of this chapter, but they all fall in a spectrum with two extremes. In sum, on the one hand there are reductionist views, such as the Buddhist’s, according to which knowledge from words is nothing but a specific type of inferential knowledge;19 on the other hand, a way to argue for an indistinct role of speech is by subsuming perception and inference under linguistic knowledge: speech is everything that truly exists. Bhartṛhari and many other authors inspired by him entertain the notion that every knowledge is necessarily linguistic (see Chapters 4 and 6 in this volume), irregardless of its cause and irrespective of whether it is an instance of perception, inference, or speech. This is because the phenomenal world of differences is nothing but a transformation of the one and only supreme linguistic principle that transmogrifies into words, objects, and any other apparent reality. There is no knowledge without language, because every knowledge is permeated by language.20 According to Jayanta, however, the view that every cognition is permeated by language is mainly a consequence of the usage of words: When asked “what object did you see,” one may answer “cow,” and, “which type of knowledge did you get,” again “cow,” and “which word did you use,” again “cow.” This confusion is just caused by usage. Actually, however, the word, the cognitive content, and the object of the cognition are quite distinct things.21

66

Alessandro Graheli

More specifically, Jayanta justifies the distinction between perception, inference, and speech on the basis of the epistemic instrument used to obtain them, that is, the sense organs, inferential marks, and an audible instruction, respectively, and on the basis of the different terminology commonly used to denote these distinct epistemic acts (NM, I 399,12–13). The word “sound” has a generic and non-epistemological acceptation,22 but this does not disprove the distinct role of speech in epistemology. In sum, knowledge from words is caused by an idiosyncratic epistemic process, namely speech, and like perception, inference, and comparison it necessarily entails four primary factors (Figure 3.4): ●●

●●

●●

●●

an epistemic agent, in this case a listener; an epistemic process involving several necessary conditions, which include the instruction of an authority, the auditory perception of the words, the recollection of the semantic relation between words and referents, and so on; an epistemic object, a referent of the utterance; an episteme, a resulting knowledge of the referent.

4.4  The Role of the Speaker Unlike perceptual and inferential knowledge, speech requires the additional factor of an active speaker—an authority who, according to Vātsyāyana, qualifies as one on the basis of two properties:23 (1) “direct experience of the true essence of the artha” and (2) “desire to describe the directly experienced artha as it is or it is not” Figure 3.4  The Factors of Knowledge from Words instruction of authority, auditory perception, etc. referent

listener

knowledge of the referent

designates causality

Ontology and Epistemology of Speech in Nyāya

67

Once these two criteria are satisfied, a speaker is authoritative, regardless the social status, because teachers can be found “among ancient sages, cultivated people and even barbarians.”24 On the surface, the first criterion seems to be confined to the instructor’s sense perception. If understood as such, however, this condition narrows down the scope of speech to very specific cases, even while preventing an undesirable petitio principii caused by the lack of a foundational principle. This is probably why Jayanta feels the need to specify that this “direct experience,” or perhaps “first-hand experience,” does not limit the teacher to his perceptual knowledge: “The authority is not undermined if the true nature of the taught object has been ascertained through inference, etc.”25 Here the “etc.” used by Jayanta after “inference” may be interpreted as inclusive of speech as well. In this case, however, then a chain of acts of speaking and listening is contemplated, and such an undetermined epistemological foundation may then clash with the original idea of a “first-hand experience” of the authority and perhaps undermine the whole agenda of authoritative testimony.

4.5  Speech Is Instruction Fully embracing the letter of the original aphorism (Nyāyasūtra, 1.1.7-8), Jayanta stresses the synonymity of “speaking” and “instructing.” In the following dialogue he describes what “instructing” exactly means: [Opponent:] What is “instructing”? [Proponent:] It is “doing an action of naming” (abhidhānakriyā). [Opponent:] What exactly is “doing an action of naming”? [Proponent:] It is “causing knowledge of an object.” [Opponent:] Then the eyes must also be instructors, because they do cause knowledge of an object. [Proponent:] One must specify that speech causes knowledge of an object under the condition of being itself known. [Opponent:] Then even smoke [inferentially] instructs [about fire] after it is itself known, and even the eyes instruct. [Proponent:] [In the case of speech] the resulting cognition is caused by an association with itself. [Opponent:] Then a face [e.g., reflected in a mirror] instructs by inference the presence of the associated parts of the body of a person. Besides, then even speech does not instruct, because it does not have any real association with the object.

68

Alessandro Graheli [Proponent:] Here the condition “being itself known” is restricted to speech. [Opponent:] Then even the hearing organ instructs when it causes the knowledge of speech. And speech then is not instructing anymore, because it is the restricted object of knowledge, not the cause of knowledge of an object. The action of naming is not even simply knowledge from words, because in this case “instructing” would also be “causing knowledge of the speaker’s intention (vivakṣā)” and “causing inferential knowledge of ether,” since both cognitions can be caused by speech. Therefore, since the nature of this naming action remains uncertain, you cannot define instructing as an action of naming.26 [Jayanta:] “Naming” is better characterized by “knowledge of an object caused by audible matter (śrotragrāhyavastu),” because the expression “naming” is used in this sense in common parlance. The object of knowledge (artha) that is “named,” or “told,” is the same object that becomes the content of such a [spoken] piece of knowledge. This is why the cause of such a cognition of an object must be a succession of speech units grasped by the hearing organ. This is also why the alleged sphoṭa cannot be speech, since it is not audible matter. The word “speech” is attested in the sense of audible content, and speech units are exactly what is heard. An incomplete definition of speech such as “something that causes the cognition of the object” would be faulty, on the one hand because it would include inferential marks such as smoke, and on the other because it would not include words of an unknown name-named relation, which would then be divested of their character of speech just because they do not produce the cognition of the object.27

According to Jayanta, therefore, speech—that is, instruction—is best defined by the criteria of both (1) being audible and (2) causing knowledge of the object. Elsewhere, Jayanta explicitly links referentiality to the very definition of speech in Nyāyasūtra, 1.1.7, “speech is the instruction of an authority,” where he glosses “instruction” with “act of reference.”28 Since he does not qualify this statement, we can assume that he means to say that every act of reference is an instruction, and every instruction is an act of reference.

5 Conclusion In Nyāya, speech is chiefly studied as an epistemological phenomenon, and only secondarily as a linguistic one. Consequently, even the object of speechderived knowledge is an epistemic object. When understood in the metaphysical

Ontology and Epistemology of Speech in Nyāya

69

frame of reference of Nyāya-Vaiśeṣika, it is a thing, or state of affairs, really existing out there in the world, and not a theoretical or psychologistic “sense,” “meaning,” or “image” in the mind of the speaker or the hearer. The positive description of the phenomenic world through Vaiśeṣika metaphysics, the assertive epistemology, and the fascination for the grammatical categories of the scientific lingua franca of the time, Sanskrit, can be seen as strategies to ensure scholarly exchanges and debate, in tune with the importance assigned to dialectics in the early stages of the school. But beyond epistemology, what is then the status of other common uses of speech, such as in poetry, rhetorics, or in performative or deontic aspects of language? Do Jayanta and other Nyāya authors even admit the possibility of “understanding” a sentence without committing to its value of truth? Whenever we hear congruous descriptive speech, our reaction may be either committal or noncommittal, in regard to a truth value expressed by it. In both cases we understand the expression, but we happen to believe or disbelieve its content. But is a neutral, purely “linguistic” understanding of words possible at all? According to Matilal (1994, 355), this possibility is not endorsed in Nyāya: “The Naiyāyikas were against the deployment of such a basic attitude prior to the belief-claim or knowledge claim that arises in the hearer.” Taber (1996, 20), however, disagrees. While studying arguments in favor and against the reduction of verbal testimony to inference, he noticed that the distinction between linguistic understanding and epistemic commitment is indeed found in some Nyāya texts: “While it is indeed the case that Nyāya, especially later Nyāya, rejects an initial grasp of the meaning of a statement as the author’s thought or intention, it nevertheless does make a distinction between apprehending the meaning of a statement and apprehending its truth.” Taber also cautions that his criticism of Matilal’s characterization of Nyāya is specifically based on his study of Jayanta’s point of view and may not be applicable elsewhere (Taber 1996, 20). Actually, the two views are easily reconciled. Taber is right because Jayanta is certainly a peculiar Naiyāyika who freely harvests from Mīmāṃsā sources on every linguistic aspect that is not satisfactorily dealt with in his own tradition (see Chapter 19 in this volume), and consequently there are features of his linguistic theory that are not fully adhering to the Nyāya tenets. The possibility of a linguistic understanding without an epistemic commitment, related to the theory of signification of the sentence, is one of such features. Matilal is right because his representation is faithful to the epistemological paradigm of Nyāya.

70

Alessandro Graheli

Furthermore, the issue of distinguishing purely linguistic comprehension and testimonial knowledge might have been a pseudo-problem from Jayanta’s perspective, as he would have probably concurred in that “if the ability to use language meaningfully is connected with the making of true reports then it is surely the consistent making of true reports that matters” (Coady 1994, 245). In other words, if the main use of language is to communicate beliefs, in principle there are no “neutral” statements, and false statements can indeed be explained in terms of an inappropriate use of language.

Notes 1 The date of the Nyāyasūtra is inferred from its chronological relation to other authors and works, particularly Nāgārjuna’s (see Potter 1977, 221). What is often overlooked, however, is how the issue of its recension and date is complicated by the fact that the oldest existent version of the sūtras is the one embedded in manuscripts of the NBh: the oldest extant manuscripts of the NBh were written in the thirteenth century, and the extant manuscripts of the uncommented sūtras are later by-products, offshoots of the main NBh transmission. Furthermore, in this main transmission it is not always easy to discriminate between sūtra and commentary, so the number of sūtras and the exact reading of some of them remains a controversial issue, subject to interpretation. Judgments of a relative chronology based on quotations of the sūtras by other authors should take this scenario into account. 2 Praśastapādabhāṣya, 16,1–2: ṣaṇṇām api padārthānām astitvābhidheyatvajñeyatvāni. 3 In the Nyāyakalikā (Nyāyakalikā, 20,5–6), Jayanta shows an argument to infer that sound is a quality as an example of an inference by subtraction pariśeṣānumāna. 4 Further details can be found in Vidyabhusana (1921, 109–11) and Potter (1977, 254–55). 5 NBh, ad 2.2.40: dvividhaś cāyaṃ śabdo varṇātmako dhvanirūpaś ca. 6 ŚV, anumāna 21cd–22ab, cited in NM, I.571,2–3, NM4: jātimattvendriyatvādi vastusanmātrabandhanam / śabdānityatvasiddhyarthaṃ ko vaded yo na tārkikaḥ // 7 ŚV, śabdanityatā 339cd: ekārthasamavāyena jātir jātimatī hi vaḥ / 8 NM, I 570,9: nissāmānyāni sāmānyādīnī. See Praśastapādabhāṣya, 17,5–7, dravyādīnāṃ trayāṇām […] sāmānyaviśeṣavattvam, and Praśastapādabhāṣya, 19,1–2, sāmānyādīnām trayāṇāṃ […] asāmānyaviśeṣavattvam. 9 NM, I 571,4: tadaviditatārkikaparispandasya vyāhṛtam. 10 pūrvavarṇajanitasaṃskārasahito ’ntyo varṇaḥ pratyāyakaḥ.

Ontology and Epistemology of Speech in Nyāya

71

11 NM, II 199,18–200,8: tatra ceyaṃ kalpanā. varṇakrameṇa tāvat prathamapadajñānam. tataḥ saṅketasmaraṇaṃ. saṃskāraś ca yugapad bhavataḥ. jñānayor hi yaugapadyaṃ śāstre pratiṣiddham, na saṃskārajñānayoḥ. tataḥ padārthajñānam. tenāpi saṃskāraḥ. punar varṇakrameṇa dvitīyapadajñānam. tataḥ saṅketasmaraṇam. pūrvasaṃskārasahitena ca (ca] ca tena NM) paṭutaraḥ saṃskāraḥ. tataḥ padārthajñānam. tena pūrvapadārthajñānasaṃskāreṇa ca paṭutaraḥ saṃskāraḥ. (tataḥ […] saṃskāraḥ] om. NM) punaḥ pūrva (pūrva-] pūrvavarṇa- NM) krameṇa tṛtīyapadajñānam. saṅketasmaraṇam. pūrvasaṃskārāpekṣaḥ paṭutamaḥ (paṭutamaḥ] paṭutaraḥ MDUC 2606; NM) saṃskāraḥ. tad anu padārthajñānaṃ. tato’pi tathaiva paṭutamaḥ (tato’pi tathaiva paṭutamaḥ] tathaiva tato pi paṭutaraḥ MDUC 2606; om. NM) saṃskāra (tad anu […] saṃskāra] om. NM) iti. evaṃ padajñānapīvare saṃskāre padārthajñānajanite ca tādṛśi saṃskāre sthite antyapadārtha- (-padārtha-] -pada- MDUC 2606) -jñānānantaraṃ padasaṃskārāt pūrvapadaviṣayā (-viṣayā] -viṣaya- NM) smṛtiḥ padārthasaṃskārāc ca padārthaviṣayā smṛtir iti saṃskārakramāt (-kramāt] -trayāt NM) krameṇa dve smṛtīsmṛtī bhavataḥ. tatraikasyāṃ smṛtāv upārūḍhaḥ padasamūho vākyam itarasyām upārūḍhaḥ padārthasamūho vākyārthaḥ. 12 NBh, p. 1,12–15: pramīyate’nena iti pramāṇam. 13 When using the term kāraka, Jayanta walks the line between ontology and linguistic descriptions, so I’m here using the ambiguous equivalent “factor.” For a grammatical evaluation of the term, see Cardona (1997, 138) and Cardona (1976, 215). Cardona renders kāraka as “”participant” of an action. Remarkably, the genitive case is treated separately because the referent of a genitive is not directly related to the action, and as such it does not qualify as a kāraka. For a brief survey and a deeper study of kārakas as semantic-syntactic relations as compared to the deep-structure cases of Fillmore 1968, see respectively Blake (2004, 63–66) and Keidan (2011). For a more philosophical account of the kāraka system, see Deshpande (2017). 14 The accepted version, in later Nyāya, is that the sense organ is itself the pramāṇa. See, for example,, Tarkasaṃgraha, 23: tasmād indriyaṃ pratyakṣapramāṇam iti siddham. 15 NM, 1,I.38,10–11: tasmāt kartṛkarmavilakṣaṇā saṃśa​yavip​aryay​arahi​tārth​āvabo­ dhavi​dhāyi​nī bodhābodhasvabhāvā sāmagrī pramāṇam iti yuktam / 16 (NM, 1, I.43,7–10): tasmād yathā hi kārakāṇi taṇḍulasalilānalasthālyādīni siddhasvabhāvāni sādhyaṃ dhātvartham ekaṃ pākalakṣaṇam urarīkṛtya saṃsṛjyante, saṃsṛṣṭāni ca kriyām utpādayanti, tathā ātmendriyamano ’rthasannikarṣe sati jñānākhyo vyāpāra upajāyate / 17 Jayanta’s explanation seems to imply that the epistemic agent (pramātṛ) and the epistemic object (prameya) should theoretically be considered as part of the causal apparatus as well, even though Jayanta’s commentator Cakradhara

72

18 19

20

21

22

23

24

25 26

Alessandro Graheli states that the pramāṇa is distinct from agent and object (Granthibhaṅga, 7,12: kartṛkarmavyatiriktaṃ janakaṃ yat tat pramāṇam).) Notably, however, Shah (Granthibhaṅga, 7fn) quotes a string of verses quoted in the Syādvādaratnākara and ascribed by its author, Vādidevasūri, to a lost work attributed to Jayanta, the Nyāyapallava. In these verses the epistemic agent and the epistemic object are included in the causal apparatus, though the distinction of the epistemic agent, epistemic object, and epistemic instrument (pramāṇa) remains valid because eventually the object is reflected in the awareness of the agent as the cognitive result (sāmagryantaḥpraveśe’pi svarūpaṃ kartṛkarmaṇoḥ / phalavat pratibhātīti na catuṣṭvaṃ vinaṅkṣyati //). Nyāyasūtra, 1.1.7-8: āptopadeśaḥ śabdaḥ. The distinction of testimonial knowledge from other forms of knowledge is discussed from many angles in Matilal and Chakrabarti (1994). For the Vaiśeṣika and Buddhist arguments in favor of the reduction of testimony to inference, and for the Nyāya defense of the specificity of speech as presented in the NM, see Graheli (2017). Vākyapadīya, I.131, p. 49, quoted in NM, II.477,14, NM9: na so ’sti pratyayo loke yaḥ śabdānugamād ṛte / On Jayanta’s refutation of linguistic monism (śabdādvaita), see NM, I 209–25 and NM, II 476–85. See also in this volume Saito (Chapter 4) and McClintock (Chapter 5). NM, II 480,12–15: kas tvayā dṛṣṭo ’rthaḥ iti pṛṣṭo vakti gaur iti. kīdṛśaṃ te jñānam utpannam gaur iti. kaṃ (kaṃ] kīdṛśaṃ NM) śabdaṃ prayuktavān (prayuktavān] prayuktavān asi NM) gaur iti. tata eṣā bhrantiḥ. vastutas tu viviktā evaite śabdajñānārthāḥ. For instance, the presence of sound is used as an inferential mark to prove the existence of the fifth element, ether (ākāśa). See Praśastapādabhāṣya, 58,8, where the inference is explained: śabdaḥ […] na sparśavad viśeṣaguṇaḥ. bāhyendriyapratyakṣatvād […] nātmaguṇaḥ. śrotragrāhyatvād vaiśeṣikaguṇabhāvāc ca na dikkālamanasām. pāriśeṣād guṇo bhūtvā ākāśasyādhigame liṅgam. NBh, ad 1.1.7: āptaḥ khalu sākṣātkṛtadharmā yathādṛṣṭasyārthasya cikhyāpayiṣayā prayukta upadeṣṭā sākṣātkaraṇam arthasyāptiḥ tayā pravartata ity āptaḥ. NM, I 400,11: ṛṣyāryamlecchasāmānyaṃ vaktavyaṃ cāptalakṣaṇam. In this connection, see also Chakrabarti (1994, 103) with the vivid example of a thief or a murderer confessing in a court of law. NM, I 399,12–13: na tu pratyakṣeṇaiva grahaṇam iti niyamaḥ, anumānādiniścitārthopadeśino ’py āptatvānapāyāt. NM, I 398,2–12: upadiśyata iti ko ’rthaḥ? abhidhānakriyā kriyate. keyam abhidhānakriyā nāma? artha- (artha] om. NM MDUC 2606) -pratītir iti cet,

Ontology and Epistemology of Speech in Nyāya

73

cakṣurāderapi tatkaraṇatvād upadeśatvaprasaṅgaḥ. svāvagatipūrvikā pratītir iti cet dhūmāder apy upadeśatāprasaṅgaḥ. cakṣurāder api tadupadeśatvaprasaṅgaḥ. svasādṛśyena pratītir iti cet, bimbasyāpi pādādyanumitāv upadeśatvaprasaṅgaḥ. śabde ca tadabhāvād anupadeśatvaṃ syāt. śabdāvacchinnā pratītir iti cet, śrotrasya tajjanakatvād upadeśatvaprasaṅgaḥ. śabdasya ca svāvacchedena pratītijanakatvaniṣedhād anupadeśatvaṃ bhavet. nāpi śabdakaraṇikāpratītiḥ abhidhānakriyā, vivakṣāyāṃ ākāśānumāne vā tasyopadeśatvaprasaṅgād ity abhidhānakriyāsvarūpāniścayāt na tasyāḥ (na tasyāḥ] kasyāḥ BORI 390 a.c.) karaṇam upadeśaḥ. 27 NM, I 399,1–8: ucyat​e—śro​tragr​āhyav​astuk​araṇi​kātad​artha​pratī​tir abhidhānakriyā, itthaṃ loke vyavahārāt. uktaḥ abhihitaś ca sa evārtho loke vyapadiśyate, yas (yas] yas tu NM MDUC 2606) tathāvidhapratītiviṣayatāṃ pratipannaḥ, śrotragrāhyasya varṇarāśer evārthapratītikaraṇatvāt. na tu śrotrapratyayaviṣayaḥ sphoṭātmā śabdaḥ. śrotragrahaṇe hy arthe śabdaśabdaḥ prasiddhaḥ. varṇā eva ca śrotragrahaṇāḥ. yato ’rthapratītiḥ sa śabda iti tūcyamāne dhūmādir api śabdaḥ syāt. agṛhītasambandhaś ca śabdaḥ śabdatvaṃ jahyāt, arthapratipatter akaraṇāt. 28 NM, I.399.1–2: abhidhānakriyā.

Primary Sources Aṣṭādhyāyī

Sharma, Ram Nath, ed. 1987–2003. The Aṣṭādhyāyī of Pāṇini. Edited and translated, 6 vols. New Delhi: Munshiram Manoharlal.

BORI 390

Pune, Bhandarkar Oriental Research Institute, Manuscript no. 390/1875–76. 432 foll. birch-bark, bound, Śāradā script.

Nyāyadarśana

Thakur, Anantalal, ed. 1997a. Gautamīyanyāyadarśana with Bhāṣya of Vātsyāyana. Nyāyacaturgranthikā 1. New Delhi: Indian Council of Philosophical Research.

MDUC 2606

Thenjipalam (Malappuram District), Malayalam Department of the University of Calicut, Manuscript no. 2606. 188 foll. palm-leaf, Grantha-Malāyālam script.

NBh

Nyāyabhāṣya of Vātsyāyana. In Nyāyadarśana.

Nyāyakalikā

Kataoka, Kei, ed. 2013. “A Critical Edition of Bhaṭṭa Jayanta’s Nyāyakalikā. First Part.” The Memoirs of the Institute for Advanced Studies on Asia 163: 1–53.

Nyāyasūtra

Nyāyasūtra of Gautama. In Nyāyadarśana.

74

Alessandro Graheli

Nyāyavārttika

Thakur, Anantalal, ed. 1997b. Nyāyabhāṣyavārttika of Bhāradvāja Uddyotakara. Nyāyacaturgranthikā 2. New Delhi: Indian Council of Philosophical Research.

Praśastapādabhāṣya

Dvivedin, Vindhyesvariprasada, ed. 1895. The Bhāṣya of Praśastapāda together with the Nyāyakandalī of Śrīdhara. Vizianagram Sanskrit Series 5. Benares: E.J. Lazarus & Co.

ŚV

Dvārikādāsa Śāstrī, ed. 1978. Ślokavārttika of Śrī Kumārila Bhaṭṭa with the Commentary Nyāyaratnākara of Śrī Pārthasārathi Miśra. Prāchyabhārati Series 10. Varanasi: Tara Publications.

Tarkasaṃgraha

Sharma, Mahadev, ed. 1919. Tarkasaṃgraha of Annambhaṭṭa with Nyāyabodhinī and Padakṛtya. Sixth edition. Bombay: Nirnaya Sagar.

Vākyapadīya

Rau, Wilhelm, ed. 1977. Bhartṛharis Vākyapadīya, die mūlakārikās nach den Handschriften herausgegeben und mit einem pāda-Index versehen. Wiesbaden: Franz Steiner Verlag.

Vṛttikāragrantha

Frauwallner, Erich, ed. Vṛttikāragrantha. In Frauwallner 1968.

Secondary Sources Blake, Barry J. 2004. Case. Second edition. Cambridge: Cambridge University Press. Cardona, George. 1976. Pāṇini, A Survey of Research. The Hague, Paris: Mouton. Cardona, George. 1997. Pāṇini: His Work and Its Traditions. Delhi: Motilal Banarsidass. Chakrabarti, Arindam. 1994. “Telling as Letting Know.” In Matilal and Chakrabarti, 99–124. Coady, C.A.J. 1994. “Testimony, Observation and ‘Autonomous Knowledge.’” In Matilal and Chakrabarti, 225–50. Deshpande, Madhav M. 2017. “Semantics of Kārakas in Pāṇini: An Exploration of Philosophical and Linguistic Issues.” In Sanskrit and Related Studies: Contemporary Researches and Reflections, ed. by Bimal K. Matilal and Purusottama Bilimoria, 33–57. Delhi: Sri Satguru Publications. Fillmore, Charles J. 1968. “The Case of Case.” In Universals in Linguistic Theory, ed. by E. Bach and R.T. Harms, 1–88. New York: Holt, Rinehart / Winston. Frauwallner, Erich. 1968. Materialien zur ältesten Erkenntnislehre der Karmamīmāṃsā. Veröffentlichungen der Kommission für Sprachen und Kulturen Süd- und Ostasiens 6. Wien: Österreichische Akademie der Wissenschaften.

Ontology and Epistemology of Speech in Nyāya

75

Graheli, Alessandro. 2017. “Bhaṭṭa Jayanta: Comprehension, Knowledge, and the Reduction of Testimony to Inference.” Kervan: International Journal of AfroAsiatic Studies 21: 175–226. Keidan, Artemij. 2011. “The ‘Kāraka-vibhakti’ Device as a Heuristic Tool for the Compositional History of Pāṇini’s Aṣṭādhyāyī.” Nuova serie, Rivista degli Studi Orientali 84: 204–21. Matilal, Bimal Krishna. 1994. “Understanding, Knowing and Justification.” In Matilal and Chakrabarti, 347–57. Matilal, Bimal Krishna, and Arindam Chakrabarti, eds. 1994. Knowing from Words: Western and Indian Philosophical Analysis of Understanding and Testimony. Dordrecht: Kluwer. Potter, Karl H., ed. 1977. Indian Metaphysics and Epistemology: The Tradition of Nyāya-Vaiśeṣika up to Gaṅgeśa. Encyclopedia of Indian Philosophies, vol. 2. Delhi: Motilal Banarsidass. Taber, John A. 1996. “Is Verbal Testimony a Form of Inference?” Studies in Humanities and Social Sciences 3 (2): 19–31. Vidyabhusana, Satis Chandra, ed. 1921. A History of Indian Logic. Calcutta: Calcutta University.

4

The Theory of the Sphoṭa Akane Saito

1  Introduction: Five Views on the Sphoṭa This contribution traces the history of the concept of sphoṭa, which I will leave untranslated. Oversimplifying, this Sanskrit word serves the purpose of describing the minimal and meaningful unit of speech. When we study the history of the arguments concerning sentence meaning or word meaning in Indian philosophy, we necessarily need to account for the theory of the sphoṭa. This theory was propounded by the Grammarians (Vaiyākaraṇas) and taken up for discussion over and over again by several authors. Throughout its long history, dominated by the fifth-century philosopher and Grammarian Bhartṛhari, the concept of sphoṭa differs in each system and period. Beginning from the ancient times when the Grammarians first referred to it, I differentiate the following five variants: 1. The concept of sphoṭa as inseparably connected to heard or uttered sound, held by Patañjali (second century BCE) and elaborated by Bhartṛhari. 2. The view of sphoṭa as a baseless postulation for the understanding of the meaning, in contrast with the theory of the phoneme (varṇa). This view was held by various anti-sphoṭa philosophers, such as Kumārila Bhaṭṭa and Dharmakīrti. 3. A theory held by Maṇḍana Miśra (eighth century CE), who following Bhartṛhari developed this theory to explain how one perceives a wordform. 4. A theory that equates sphoṭa with the highest reality of both speech and the universe, held by Śaiva scholars in medieval Kashmir.

The Theory of the Sphoṭa

77

5. A theory held by premodern Grammarians that focuses on “meaningbearingness” (arthapratyāyakatva) and is ultimately connected with the Vedāntic concept of the “Undivided absolute” (brahman). The basic understanding of the sphoṭa is that it is the nature of the word or speech (śabdasvarūpa), whose distinguishing character is emphatically considered its indivisibility. A reason for changes in the theory is the aspect or level of speech authors have focused on. A collateral effect of the views on the sphoṭa is the understanding of what exactly are sound and phonemes. In this chapter I shall briefly illustrate how each of the five different theories of sphoṭa is discussed, followed, or rejected.

2  Patañjali’s Ideas on Śabda and Sound In the earliest occurrences of the sphoṭa in the Grammarians’ texts, the terms śabda (“word” or “linguistic unit”) and sphoṭa share overlapping semantic fields. The key to understanding this complexity is the ongoing discussion on the principle of the fixed relation between a word and its referent, which has been conflated with the discussion on sphoṭa by later philosophers and by the Grammarians. First of all, let us differentiate these two concepts of śabda and sphoṭa.

2.1  What Is Śabda? The grammatical rules in Pāṇini’s Aṣṭādhyāyī show us his strict formulation of the relation between the word and its referent. In his commentary to the Aṣṭādhyāyī, the Mahābhāṣya, Patañjali discusses the philosophical issues of language using the Grammarians’ analytic approach. For example, in the Aṣṭādhyāyī, 1.2.45–46 the nominal stem (prātipadika) is defined as the meaningful unit of language: “A nominal stem is a meaningful element (arthavat), that which is neither a verbal root, nor an affix; nor anything ending in any affix other than a primary (kṛt) or a secondary (taddhita) suffix, or which is a compound word (samāsa).”1 In Patañjali’s commentary on this passage, there is a famous discussion concerning the meaning of a nominal stem to which neither an affix nor a case ending has been added. Against an opponent who claims that a nominal stem, for example, /vṛksa-/, cannot be meaningful without an affix, Patañjali explains that since a śabda /vṛkṣaḥ/—which is the combination of

78

Akane Saito

the śabda /vṛkṣa-/ and the śabda /s/ (the ending of the nominative singular, changed into /ḥ/ by euphonic rules)—has the referent “tree” and the singular number, and since a śabda /vṛkṣau/ has the referent “tree” and the dual number, the nominal stem, namely the śabda /vṛkṣa-/, must have the referent “tree,” a neutral meaning without any qualification.2 Here, both the nominal stem and the affix are called śabda. Not only a word or sentence but also any kind of linguistic element, whether it is an affix, a case ending, or a stem, is in the domain of śabda. Therefore, śabda means a linguistic unit which has a particular referent.

2.2  Are Phonemes Meaningful? There is a discussion of whether phonemes are meaningful or not by Kātyāyana in his Vārttika on the fifth Maheśvarasūtra.3 First, it is said that phonemes are meaningful, but immediately after it is also stated that they are meaningless. In the first alternative, phonemes are regarded as meaningful, but only when a verbal root, a nominal stem, an affix, or a particle consists in a single phoneme. In other words, phonemes are meaningless except for in these special cases. In general, for Patañjali, phonemes are just constituents of a word or a sentence. Getting back to the discussion in 1.2.45, Patañjali claims that the entire word has one meaning or purpose, even though its components have no such meaning or purpose, just as a chariot possesses the capacity to move but its disassembled parts lack it (Mahābhāṣya, on 1.2.45, p. 220,22–24).4 Phonemes are the essential parts (aṅga) of the whole word or sentence, but when we observe a word focusing on what denotes the meaning, we cannot start from the component phonemes. Therefore, Patañjali concludes that phonemes cannot be meaningful. In the opening section of the Mahābhāṣya, Patañjali discusses the exact definition of śabda, concluding that, in the case of the word “cow,” “śabda is that which, when uttered, gives rise to the cognition of those entities which have a dewlap, a tail, a hump, hooves, and horns [i.e., cows].”5 This definition of śabda has been taken up by Maṇḍana and later Grammarians as the starting point of their discussion of sphoṭa. It is therefore easy to trace their sphoṭa theories back to the whole argument on the meaningful unit, namely śabda, in the Mahābhāṣya, but we have to keep in mind that Patañjali does not use the term sphoṭa anywhere in this specific discussion.6 He does use this term, but always in relation to sound.

The Theory of the Sphoṭa

79

2.3  Sphoṭa in Relation to Sound In the introductory part of the Mahābhāṣya, śabda is alternatively defined by Patañjali as “the word,” which makes the meaning understood, and as “physical sound.”7 The two aspects of being a word and a sound are not fully separable, since every word has its sound aspect. How does Patañjali argue this issue? In the Mahābhāṣya on the Aṣṭādhyāyī, 1.1.70, he divides śabda into sphoṭa and sound (dhvani-), using the example of a drummer. When he strikes his kettledrum, some sounds reach far away, and some do not; but there is a sphoṭa that has always the same extent. Thus śabda is a complex entity constituted of sounds, which are the external phenomenon, and of sphoṭa, the core of śabda to which the sound aspect is subservient.8 But what is this unchangeable entity, which has sound as its property? It is the nature of the word, or in other words, the word’s phonetic form (śabdasvarūpa), that is common to different individual variations of one word. The word agni, for instance, can be pronounced by different agents or in different ways, but it has its own essential form which is the basis of all the variations. The wordor sound-form is its most essential “object” (vācya). Each word possesses this unchangeable phonetic form as its core, and how it is heard is affected by the modalities of sound. And according to Patañjali, every word or even mere sound has sphoṭa at the core of its existence.

3  Bhartṛhari on the Sphoṭa The investigation of the word’s phonetic form is found in the Vākyapadīya of Bhartṛhari, one of the greatest Grammarians and philosopher, who had a huge influence on the later history of Indian philosophy. Bhartṛhari refined the few remarks on sphoṭa left by Patañjali into an elaborate philosophical theory. His sphoṭa has three characteristics: 1. Phonemes (varṇa), sounds (dhvani), and bodily resonance (nāda) are differentiated. 2. Sphoṭa is the indivisible sound-form manifested in both the utterance and the hearing perception of a word, which are both discussed by Bhartṛhari. 3. Various views on sound, the nature of sphoṭa, and the relation between them, are discussed by Bhartṛhari as alternatives.

80

Akane Saito

3.1  Utterance and Hearing of Sound In Bhartṛhari’s argument, sphoṭa is necessarily related to sound, which is transmitted through the ether from the speech organ to the auditory faculty, and he explains the transmission of sound using the words dhvani and nāda, both of which mean “sound” but conceivably have slightly different nuances. In the auto-commentary on the Vākyapadīya, I.47, on the one hand, it is told that there are subtle particles of sound (dhvani) that spread in space and are the fundamental cause of the manifestation of sphoṭa. These imperceptible particles are accumulated by the speech organ before the utterance and are transformed into gross sound (nāda).9 In the course of the manifestation of sphoṭa, there is the process of pronunciation in which the speech organ accumulates subtle external sounds and transforms them into the perceptible entities. Notably, it is just because of the sequentiality of the bodily resonances that we have the impression that the sphoṭa is sequential. In the Vākyapadīya, I.76, on the other hand, Bhartṛhari mentions the subclassification of dhvani on the basis of how we grasp them. Here dhvani is audible sound from the hearer’s viewpoint, and it is divided into primary sound (prākṛtadhvani) and secondary sound (vaikṛtadhvani), both of which are distinct from any imperceptible entity.10 The primary sound is the cause of the manifestation of sphoṭa, and it creates the specific form of a word delimited by time and size. Although we have the feeling that the time required to pronounce a monosyllabic word such as “cow” is shorter than that required for a polysyllabic one like “Isidore,” such a difference in size is not due the word’s inner form, and is rather caused by the primary sound. The secondary sound, by contrast, causes other inessential conditions such as intonation, pitch, accent, or tempo. In this way, in the course of the manifestation of sphoṭa, there is the process of hearing or cognition in which the sounds are limited by various imposed factors.11 Furthermore, Bhartṛhari’s explanation on pronunciation and hearing suggests that sphoṭa is not a “word” grounded in semantics, but rather a soundbased form, which is not necessarily meaningful.

3.2  How the Sphoṭa Is Manifested The indivisibility of sphoṭa is systematized by Bhartṛhari. The gist of his argument is that even though sounds are sequential, the sphoṭa manifested by them is without any sequence or parts. Here Bhartṛhari provided the most famous discussion on sphoṭa, describing how a unitary idea can manifest from sequential

The Theory of the Sphoṭa

81

sounds. He explains this as a non-analyzable (anupākhyeya) cognition, which gradually becomes clearer and clearer (Vākyapadīya, I.82–84).12 The nonanalyzable cognition is at first a vague image of the word. When the hearer perceives the series of physical sounds uttered by the speaker, at the moment of the initial sound the cognition is ambiguous, because it could become any kind of series of sound beginning with this first one. At that moment, such an initial sound just vaguely manifests the unitary word. Subsequently, each following sound leaves a seed in the hearer’s mind, which assists the perception of the immediately next sound and thus makes the initially vague image clearer and clearer. As this process is reiterated, the pronunciation of the final sound produces the cognition of the word-form in its full clarity. In this way, the sphoṭa is gradually made clearer by each sound’s instigation, until it is completely manifested. Here Bhartṛhari does not discuss how the hearer understands the meaning of a particular word. The process of manifestation of the sphoṭa explains how we perceive an entire word- or sound-form through the sequential process. If we only focused on this sequential process, we would easily conclude that a word is divisible into smaller components such as phonemes. In hearing, however, a word remains a word, and a sentence a sentence: the sequential perception does not affect the unitary nature. Bhartṛhari concedes that upon hearing a word, one inevitably feels like perceiving components such as phonemes. However, just as when we approach an object from a distant place and we gradually cognize it for what it really is, a hearer realizes in the end that a word or a sentence is intrinsically unitary (Vākyapadīya, I.89–91).13

3.3  Views on Sphoṭa and Sound Though in Bhartṛhari’s main argument the sphoṭa is the phonetic form of the word, revealed by individual sounds, he also offers other possibilities of interpretation. In the Vākyapadīya, I.104, a proponent claims that sounds (śabda) play the role of producing an additional and superficial resonance (nāda), which arises always together with the body of sound, namely, the sphoṭa.14 The relation between sound, superficial resonance, and body of sound is compared to that connecting a lamp (i.e., sound), a light (i.e., the resonance), and a fiery substance (i.e., the sphoṭa). Sphoṭa is considered to be the material cause (upādāna) of the resonance. Vṛṣabhadeva, a commentator on the Vākyapadīya, interprets this passage that the word sphoṭa is a sharp sound at the first moment of articulation, while nāda means resonance coming after the initial sound. In this case, sphoṭa is not a conceptual thing but audible sound.

82

Akane Saito

In the Vākyapadīya, the sphoṭa is still open for much discussion. In the above context, Bhartṛhari mentions that there are the holders of transient sphoṭa.15 The sphoṭa proponents generally rest on the premise that the sphoṭa is fixed and permanent (nitya), and there is no theory found on the basis of its impermanence in the later period. But we can see that it was not the case in the fifth century. Also in the Vākyapadīya, I.93, the view taking sphoṭa as the class (jāti) of the individual words is referred to as one of the alternative views.16 In the presentation of these different ideas, the concepts of sphoṭa and dhvani are sometimes not fully distinct, and there is some inaccuracy in the usage of nāda and dhvani.

3.4  Bhartṛhari’s Word’s Generic Form and Class In Bhartṛhari’s discussion on the generic form of words (śabdākṛti), we find another explanation of the word’s essential form (śabdasvarūpa). The generic form of a word is common to every variation of the word in pronunciation.17 It is not the actualized or superficial form of the word, but rather it is that which is perceived as its true nature, though concealed by sounds. Such a conceptual form arises gradually in the mind in the process of perceiving the word.18 This description of the generic form is strikingly similar to that of the sphoṭa. The class of a word (śabdajāti) is mentioned as an alternative in respects to the word’s generic form.19 It is not the class as the word, because otherwise all words would reduce to that concept, but rather the class of each individual word. To explain, the class of the word X is extracted from numerous variations of X, and it is equated with X’s own form (X-svarūpa). As such it is also equated with X’s generic form (X-ākṛti) and with the universal of X (X-sāmānya), expressed as X-ness (X-tva). It should now be clear how Bhartṛhari pursued the argument of the core or the nature of the word using not only sphoṭa but also other terms: the universal (sāmānya), generic form (ākṛti), class (jāti), and sphoṭa can all be taken as synonymous and have slightly differing aspects that need to be taken into consideration. As for sphoṭa, the concept is deeply connected with the discussion on sounds.

4  Maṇḍana Miśra’s Sphoṭa After Bhartṛhari, the theory of sphoṭa was exposed to various criticisms of anti-sphoṭa philosophers, who held phonemes (varṇa), the smallest units

The Theory of the Sphoṭa

83

of speech, to represent the fundamental aspect of language. Maṇḍana Miśra (eighth century CE) is the first sphoṭa proponent who answered their criticism. The purpose of his Sphoṭasiddhi was to tackle those criticisms and to lay the thoroughly theoretical foundations for the existence of the sphoṭa.20 These are the characteristics of his sphoṭa: 1. 2. 3. 4.

The main opponents are Mīmāṃsā and Buddhist authors. The existence of phonemes is rejected in the manifestation of sphoṭa. The focus is on the perception of the word. He refers to, but does not discuss, the sphoṭa at the level of the sentence.

Maṇḍana repeatedly uses the word “cow” (gauḥ) as an example of “word” (pada), and he never specifically discusses the sentence (vākya). Hence the sphoṭa in the Sphoṭasiddhi is restricted to the word-sphoṭa (padasphoṭa), in its premodern classification. When the opponent changes from the Mīmāṃsakas to the Buddhists in the latter half of the Sphoṭasiddhi, the term vākya is used frequently, but Maṇḍana never examines the sentence.

4.1  Maṇḍana’s Response to the Criticisms of the Mīmāṃsakas and the Buddhists 4.1.1  Against Kumārila’s Theory of Phonemes For proponents of phonemes such as Mīmāṃsā authors, the causal relationship between phonemes and the word’s referent is not direct. This is because the meaning is not grasped at the time of the pronunciation of each individual phoneme, but rather when the sequence of phonemes constituting a linguistic expression is completed. Therefore, Mīmāṃsā authors postulate a role of latent impressions (saṃskāra) between the perception of phonemes and the understanding of a meaning. Śabarasvāmin (fifth century CE), the great authority of the Mīmāṃsā school, claimed that what conveys the word’s meaning is the last phoneme accompanied by the latent impressions produced from the previous phonemes.21 After him, Kumārila Bhaṭṭa (sixth to seventh century CE) further developed this argument in the Sphoṭavāda-chapter of his Ślokavārttika. According to Kumārila the phonemes, which are grasped in a sequence through direct perception, bring about a single recollection, in the form of nonsequential phonemes with the help of latent impressions. He proposes three alternative causes to the understanding of the words’ meaning: either the phonemes, that is, śabda, or the recollection of the phonemes assisted by the perception of the final

84

Akane Saito

phoneme of the word, or just the recollection of all the phonemes, including the last one. After having perceived the final phoneme, there is a unitary idea of the word occurring in the hearer’s mind either in the form of mere recollection of all the phonemes or with the perception of the last phoneme. And since it is unitary, such recollection does not contain any sequence of phonemes inside, even though it exists at the time of perception. With this theory in mind, Kumārila claimed that sphoṭa is just a fabricated entity. For him the word’s own form is nothing but a construct of phonemes, and the cognition of the unitary word is attributed to the unifying recollection. Maṇḍana, however, argues that the latent impressions produced by the cognitions of the individual phonemes cannot be the cause of understanding the meaning of a word. This is because from the perception of a phoneme there will be a latent impression of that phoneme, and then a recollection of that phoneme alone; analogously, the perception of a word will cause the recollection of a word, and the same will apply to the sentence. For Maṇḍana, Kumārila’s latent impressions can only explain the recollection of phonemes, not the understanding of the word’s meaning.22 Against the alternative view that the final phoneme assisted by the recollection of previous phonemes is the cause of the understanding of the word’s meaning, Maṇḍana claims that when a person learns the relation between a word and its object, he requires the idea of a word and not of the final phoneme.23

4.1.2  Against Dharmakīrti’s Theory of Phonemes In the Pratyakṣapariccheda of his Pramāṇavārttika, at the sections dealing with “the refutation of eternity of phonemes and the sentence” as well as “the refutation of the opinion of Kumārila” (Pramāṇavārttika, 127–134, 160–161), Dharmakīrti (sixth to seventh century CE) criticized the idea that the unitary word is manifested by the sequential sounds on the basis of momentariness, the most basic tenet for the Buddhist logicians. In Dharmakīrti’s view, a single and indivisible entity cannot be produced from momentary phonemes, not even in the mind. He even criticized the Mīmāṃsā authors who accepted the concept of a unitary idea only in a limited sense through the unitary recollection. The uttered phonemes have a prior-posterior relation to each other, and this relation among phonemes is equivalent to the cause-effect relation of the mind-moments of the speaker and that of the hearer. The necessity of the prior-posterior relation is evident in light of the fact that even though it sounds the same, the /s/ of saraḥ and the /s/ of rasaḥ are quite distinct, because the latent impression of the respective /s/ is distinguished from the other through the cause-effect relation of

The Theory of the Sphoṭa

85

the mind-moments. The linguistic convention (saṅketa) serves the purpose of aligning the speaker’s and the hearer’s intentions, so that the hearer can infer the speaker’s intention, namely, a word or sentence meaning. Against such a stance, Maṇḍana brings up two prominent problems: (1) The issue of the universal: Buddhist logicians strictly deny the concept of the universal on the basis of momentariness. Against Dharmakīrti, however, Maṇḍana claims that the theory of momentariness cannot apply to phonemes. How could new, momentary phonemes become the cause of the understanding of a meaning shared by a speaker and a hearer? If we do not accept the universal of phonemes, we must accept the sphoṭa, akin to the universal of each individual word.24 (2) The issue of the linguistic convention: Dharmakīrti states that phonemes convey only the referent which has been conventionally established. A linguistic convention, in the form of “this is X,” represents the relation between a particular sequence of phonemes, which is taken as one conceptual unit, and an external object (viṣaya), and according to him it plays the role of bringing the sequential phonemes into the domain of the word. In Maṇḍana’s idea, the unitary image generated from what one hears is a prerequisite for the understanding of the relation between the word and its meaning. The unitary word-form is inevitable for learning its convention. For Maṇḍana the acquisition of language is not an objectbased relation, but a word-form-based relation. The real object at hand is not directly referred to.25 Maṇḍana also claims that the speaker is not a necessary condition that is needed by the hearer to learn the relation between the word and its meaning. To know the relation of cause and effect of the speaker’s mind-moments, one needs to identify the speaker, but if s/he is in a distant or hidden place, we would need to have the idea of hearing not merely a complex of sounds, but one word uttered to denote a particular meaning. In order to specify the speaker, the necessary condition is the cognition of the unitary word-form. Therefore, the causeeffect relationship of the speaker’s mind-moments is not indispensable for the understanding the meaning of the word.26

4.2  Maṇḍana’s Sphoṭa Theory Except for his refutation of the proponents of phonemes, Maṇḍana’s sphoṭa theory is based on Bhartṛhari’s arguments, and especially for the manifestation

86

Akane Saito

of sphoṭa he strictly follows Bhartṛhari by quoting his statements and emphasizing that there is no room for phonemes. The erroneous cognition (viparyāsa) of phonemes is examined by him slightly more in detail.27 This erroneous cognition is not a misunderstanding in a general sense, but rather a pre-understanding, preceding the identification of the object (word) as it is. An approximate example is that of the cognition of a rope in a dark place, which is at first mistaken for a snake but afterward manifests its own form. And this process of identification in the mind inevitably happens whenever we see an object and find (perceive) what it is.28 For the sake of the attainment of sphoṭa, this erroneous cognition is unavoidable, because only after the gradual analysis from phonemes to sentence one can realize that such a process of analysis is erroneous. Thus the erroneous cognition is said to be the cause of the correct understanding. Smaller components of language can be discarded only after bigger components are grasped, and the sequence of the analytical process is itself refuted in the end.29

4.3  Direct Perception and Understanding The sphoṭa is gradually grasped in its unitary form by the sequential phonemes. Such a graduality proves that the sphoṭa itself is in the domain of direct perception, which Maṇḍana maintains to function only progressively, from the mere or vague existence of something, to its specific character. For him the sphoṭa is not different from any other object of perception, and therefore its perception shows us the universal structure of direct perception.30 While the sphoṭa is the object of direct perception, the same does not apply to the referent (artha). Even though the proponents of phonemes are consistently addressing the relation between the word—or more precisely a sequence of phonemes—and its meaning, we should not assume that the proponents of the sphoṭa do the same, as is clear from Bhartṛhari’s stance. In Maṇḍana’s case, however, the starting point of the Sphoṭasiddhi is Patañjali’s statement that “the word is that which, when it has been uttered, gives rise to the cognition of those entities,” which unavoidably reminds us of the relation between the word and its meaning.31 This reference to Patañjali shows how Maṇḍana’s discussion of sphoṭa has a slight semantic angle—Bhartṛhari’s total distance from semantics has not been genuinely followed by him. It is certainly the result of his effort of refuting the anti-sphoṭa philosophers, but it is nonetheless crucial because his discussion somehow gave momentum to the later concept of sphoṭa as “the conveyer of the meaning.”32

The Theory of the Sphoṭa

87

4.4  Criticism of the Sphoṭa after Maṇḍana How did the philosophers contemporary to or immediately after Maṇḍana react to his sphoṭa theory? Śaṅkarācārya (eighth century?) is probably the first scholar who responded to Maṇḍana’s criticism of the phoneme theory. Although it is not an exact parallel, in his commentary on the Brahmasūtrabhāṣya, 1.3.28, Śaṅkara appears to summarize and criticize Maṇḍana’s arguments.33 He claims that the cognition of a word is adequately explained through the phonemes. Only phonemes exist, for instance the phoneme /g/, and no /g/-ness (gatva). One and the same phoneme /g/ is recognized in every perception of the words in which the phoneme /g/ occurs. Incidentally, this idea that there is only one phoneme, here, /g/, and that therefore there is no universal of /g/ had already been propounded by Kumārila. The minor difference of each phoneme is caused by momentary sound (dhvani). In other words, for him dhvani is not the cause of manifestation of the word or the sphoṭa, but just a property of phonemes. What manifests the word, instead, is the air or breath (vāyu). Even the unitary recollection of a word such as gauḥ is brought about by the phonemes /g/-/au/-/ḥ/, because we can assume that /g/ keeps having a causal efficiency even after its manifestation. Besides, Śaṅkara also underlines the importance of the sequence in the causal capacity of phonemes, in producing the cognition of a word. In this way, with the Mīmāṃsā authors, he concludes that there is no need to presuppose the sphoṭa. Other influential critics of the sphoṭa are Jayanta Bhaṭṭa and Vācaspati Miśra (ninth to tenth century). Jayanta presents two versions of the process of the manifestation of the sphoṭa, one through phonemes and another through sound (dhvani). Also, Jayanta does not stop at the word-sphoṭa and discusses further about the sentence-sphoṭa in detail, and even addresses the argument of the three levels of speech, which will be described below.34 Vācaspati discusses the sphoṭa both in his commentary Bhāmatī to Śaṅkara’s commentary on the Brahmasūtrabhāṣya, 1.3.28, and in his monograph on sentence meaning, the Tattvabindu, where he discussed in detail and criticizes Maṇḍana’s approach on the sphoṭa. Vācaspati’s description of the sphoṭa follows Maṇḍana’s argumentation and his main quibble is whether the sphoṭa can really be the object of direct perception. Vācaspati claims that what is perceived is not the sphoṭa or “a vague word-form,” but each uttered phoneme. His defense of the theory of the phonemes serves as support for the Bhāṭṭa Mīmāṃsā theory of sentence meaning (see Chapter 9 in this volume). What we notice from the arguments of the anti-sphoṭa philosophers is that, unlike in Bhartṛhari’s approach, utterance and hearing are not at the focus. They

88

Akane Saito

are rather investigating the relation between the parts and the whole, or the whole and the meaning. By doing so, they somehow oriented the following theory of the sphoṭa in a different direction.

5  Śaiva Philosophers’ Sphoṭa 5.1  Sphoṭa as the Highest Reality Brough (1951, 29, 32,30–33) pointed out that sphoṭa has been wrongly taken as a mystic entity in the early scholarly works.35 Already in premodern times, indeed, the theory of sphoṭa was merged with the philosophy of linguistic monism (śabdādvaita). The metaphysical aspect of the linguistic philosophy, however, has been “wrongly” attributed to the sphoṭa, and still we cannot ignore the influence of such an idea. But what are the early sources claiming that the sphoṭa is such a metaphysical, mysterious entity? Let us look at some Śaiva material. The Netratantra was composed in Kashmir between 700 and 850.36 In its twenty-first chapter, there is a reference to the sphoṭa in the context of the utterance of the sacred mantra oṃ: when the sphoṭa, whose nature is sound (dhvani), flows forth from the form of Śiva, filling the universe with sound, it is called nāda.37 Kṣemarāja (eleventh century), who commented on the Netratantra from the point of view of Śaiva nondualism,38 gives a semantic analysis of the word sphoṭa as “that from which the whole totality of speechunits is split open (sphuṭ-).”39 The premodern Grammarians gave two analyses of the word sphoṭa: one focusing on the perceptible word-form, “that which is split open by sounds” (yaḥ dhvanibhiḥ sphuṭyate), and the other focusing on its function as the bearer of meaning “that from which the meaning is split open” (sphuṭati artho yasmāt).40 Joshi (1967, 39–40), however, proposes that the latter analysis was not supported by Bhartṛhari,41 and Bronkhorst (2005, 2–3) positively accepts Joshi’s view, suggesting that the later analysis, or the investigation of “how the sphoṭa expresses the meaning” (not “how we grasp sphoṭa in the process of perception” as Bhartṛhari and Maṇḍana did), begins with the premodern Grammarians such as Bhaṭṭoji Dīkṣita (sixteenth to seventeenth century). Using the same analysis, already in the eighth century Śaṅkara was taking sphoṭa as the essential matter of the universe. It means that sphoṭa by his time has been taken or well known also as the entity from which the objects are

The Theory of the Sphoṭa

89

manifested. All signifiers and signified things come into existence from the sphoṭa, which in this sense is the origin of all the phenomenic differences.

5.2  Sphoṭa Discussed with Three Levels of Speech In the Vākyapadīya, Bhartṛhari introduces an ontological hierarchy of speech, dividing it into three levels.42 The first level is called “concrete speech” (vaikharī), which consists of audible speech and includes by extension any audible sound. The second one is “intermediate speech” (madhyamā), which is the internal murmur existing between the concrete speech and the “intuitive speech” (paśyantī), which is the ultimate form of speech, namely, the highest speech (parā vāc) that is self-luminous and beyond the order of phonemes or pronunciation. Already, Bhartṛhari introduced a soteriological dimension by stating that “a person who attains the truth of speech is freed from karmic duty.”43 The essence of paśyantī is the eternal and all-pervading means for liberation. This doctrine was adopted in the linguistic theory of the Śaiva philosophers, as we can see in the Śivadṛṣṭi of Somānanda (tenth century) and in its commentary by his pupil Utpaladeva, both of whom actually criticized Bhartṛhari’s view on the sphoṭa.44 In the second chapter of the Śivadṛṣṭi Somānanda starts criticizing the “Grammarians” (=Bhartṛhari), who claim the identity of the supreme brahman with the intuitive speech. Among various problems he points out, there is the one caused by the identification between the sphoṭa and the intuitive speech. If they are equivalent, the sphoṭa should be accorded the highest level, but how would it then be possible for it to be manifested by the unreal words or phonemes?45 And in the commentary, Utpaladeva identifies sphoṭa with the highest reality, with the highest level of eternality (kūṭasthanitya).46 When these different levels of speech are discussed by the Śaiva scholars, we find that the speaker’s aspect is assertively emphasized. When an intention to speak occurs, speech, whose essential nature is indeed paśyantī, arises in the mind in the intermediate form, adopting the sequence caused by expiration and inspiration; after that, the concrete speech manifests once speech reaches the mouth and gets released through the articulatory effort. It is a description of what happens in one’s body while uttering a word.

5.3  Nāgeśa’s Sphoṭa with the Process of Conceptualization Nāgeśa Bhaṭṭa (eighteenth century), who is one of the greatest Grammarians in premodern India, explains sphoṭa by combining it with the theory of paśyantī

90

Akane Saito

in his Paramalaghumañjūṣā.47 According to him, there is speech of four kinds. At the top of the hierarchy, there is the highest speech (parā vāc), the fourth level added by the Śaivas above Bhartṛhari’s triadic classification. This highest speech has the nature of being nonconceptual (nirvikalpa) and is equivalent to the “supreme and undivided speech” (śabdabrahman). By contrast, the second level, the intuitive speech (paśyantī), is defined as conceptual (savikalpa), clashing with the ideas of both Bhartṛhari and Somānanda, who stated that the intuitive speech is not conceptual. On the intuitive level, speech becomes the object of mental awareness. Both the highest and the intuitive speech are accessible to those who can enter into deep meditation on the speech form of brahman. When the speech comes to the intermediate level (madhyamā), it has the form of sphoṭa, which then expresses various meanings but cannot yet be perceived by the auditory faculty. Once such speech becomes manifest in the various points of articulation and is apprehended by the ears, it is called “concrete speech” (vaikharī). Here, the sphoṭa theory is completely overlapping the paśyantī theory. Furthermore, in contrast to Bhartṛhari or Maṇḍana, Nāgeśa clearly states that sphoṭa is subtle and imperceptible. By his time, one of the most distinguishing characters of the sphoṭa had been lost, and it had become the object of intellect (buddhi).

6  Premodern Grammarians’ Sphoṭa There are various great Grammarians in premodern time who shed the light on the sphoṭa again and tried to give comprehensive descriptions about it: Śeṣa Kṛṣṇa, Bhaṭṭoji Dīkṣṭia, Kauṇḍa Bhaṭṭa, and Nāgeśa Bhaṭṭa.48 They considered the sphoṭa to be the “meaning-bearing unit,” and classified it into different categories in accordance with the various segments of language, namely phoneme, word, and sentence.49 These are the specific characteristics of their sphoṭa: 1. The sphoṭa theory necessarily requires the concept of the phoneme (varṇa). 2. The sphoṭa is ultimately related to the highest reality. 3. All discussion of sphoṭa is focused on its role as the conveyor of the meaning (arthapratyāyaka).

The Theory of the Sphoṭa

91

The most important feature is that how sphoṭa, namely the word-form, is grasped is no more at the focus; instead, they discuss a lot about how it makes the meaning understood. Also most of their arguments below are easily traced back to what Bhartṛhari established in the Vākyapadīya on the word- or sentence meaning.50 Each type of sphoṭa, no matter if at the level of phoneme, word, or sentence, follows the principle of indivisibility. If we take into account a larger unit, then the smaller units are to be regarded as unreal entities. Let us see how Kauṇḍa Bhaṭṭa, a Grammarian of the seventeenth century who was a follower of Bhaṭṭoji Dīkṣita, has discussed the sphoṭa in his Sphoṭanirṇaya.

6.1  Phoneme-, Word-, and Sentence-Sphoṭa 6.1.1  Phoneme-Sphoṭa In the Sphoṭanirṇaya the phoneme-sphoṭa is defined not as each single phoneme but as the smallest and indivisible set of phonemes that constitute morphemes such as stems or suffixes, which means that the term varṇa is used in a new and specific acceptation, and not generically as “phoneme” anymore.51 The Grammarians then have to discuss what the meaningful unit is at the level of morpheme. In Pāṇinian grammar, all the word-forms are analyzed as being derived from an original stem or verbal root, and they pass through various substitutions of morphemes to reach their final form. For example, a word-form pacati (he cooks) consists of pac-a-ti, the final ti (or technically tip) of which is substituted for a prototype laṭ which signifies the present tense and the agent of an action. So which one of the two, laṭ or its substitute tip, is the denotative element? Against the Naiyāyikas who insist that the prototype is denotative, Kauṇḍa Bhaṭṭa claims that the substitute conveys the meaning. This is because even a person who has no knowledge of the prototype laṭ understands the meaning of pacati, and accordingly, laṭ is taken as a fictional element invented for the sake of grammatical analysis.

6.1.2  Word-Sphoṭa and Sentence-Sphoṭa However, it is certainly not the case that all meanings are explained by the phoneme-sphoṭa. Exploiting the intricacies of the rules of substitution of morphemes in the Pāṇinian grammar, Kauṇḍa Bhaṭṭa argues that words cannot unambiguously be divided into morphemes,52 and we cannot always divide a sentence into words.53 Furthermore, there are cases in which, even though we do not understand the particular referents of each word or parts of a word, we

92

Akane Saito

still understand the meaning of a sentence as a whole. Thus, Kauṇḍa Bhaṭṭa claims that besides the phoneme-sphoṭa, there must also be a word-sphoṭa and sentence-sphoṭa. In other words, the word is the denoter of the word meaning, and a sentence is that of the sentence meaning.54 Phonemes are indispensable as building blocks of words and sentences. Kauṇḍa Bhaṭṭa maintains that the word-sphoṭa consists in a series of phonemes that are uttered together in association with each other and that the sentence-sphoṭa is made up of such words. Phonemes, however, are always vulnerable to the criticism that they cannot coexist with each other, since they are momentary. Kauṇḍa Bhaṭṭa replies to this objection as follows: when two phonemes are uttered together, the second is perceived with the first that is present in the mind (upasthita) through a relation of uninterrupted succession (avyavahitottaratva); in the case of three phonemes, the third phoneme is perceived with the second phoneme specified with the first in the mind. In this way, the existence and the value of the individual phonemes are guaranteed.55 At the level of the sentence-sphoṭa, only the sentence is assumed to be the conveyor of the sentence meaning, but at the same time we still see that it is composed of words, stems, suffixes, and so on, each of which reveals its own referent in accordance with the structure of the sentence.56

6.2  Indivisible Word- and Sentence-Sphoṭa This concept of indivisible (akhaṇḍa) word-sphoṭa and sentence-sphoṭa moves the theory one step further, in comparison to the previous ideas word- and sentencesphoṭas. What is held as the conveyor of meaning, at this stage, is a word or a sentence that cannot be divided into subordinate grammatical elements,57 even though in its manifestation the listener grasps phonemes, syllables, and words. In other words, a bigger unit is no longer an aggregate of smaller units. Phonemes are not denotative, even though in the word gauḥ, for example, we do not perceive a sphoṭa that is different from the phonemes /g/-/au/-/ḥ/. This is because these phonemes are not different from the sphoṭa, just as threads are not different from a piece of cloth. Phonemes are just the manifesting agents of the sphoṭa, no matter if it is in the case of a word or of a sentence. In this theory, the role of the conveyor of meaning is taken by sphoṭa.

6.3  Universal-Sphoṭa The final level of meaningful unit is the universal-sphoṭa (jātisphoṭa), the generic form of the conveyor of the meaning.58 Each of the previously mentioned

The Theory of the Sphoṭa

93

types of sphoṭa has its own generic form, namely, a semantic universality, which is shared by all the homogeneous sphoṭas. For example, every particular instantiation of the sentence-sphoṭa “hare ’va” has a universal form (“hare ’va” -ness), which is the conveyor of meaning and signifies the same referent “O Hari, protect [me].” Even in the case of the universal-sphoṭa, Kauṇḍa Bhaṭṭa repeats that we have to accept the existence of phonemes, though we may accept only specific types of phonemes as conveyors of meaning, as mentioned above. This universal-sphoṭa is ultimately equated to the highest reality, in accordance with the elevation of the level of the universal.59 In other words, this sphoṭa becomes close to Bhartṛharian concept of brahman (śabdabrahman). From the perspective of this level, all the subordinate categories are denied.

Notes 1 arthavad adhātur apratyayaḥ prātipadikam // kṛttaddhitasamāsāś ca //. 2 See Mahābhāṣya, on 1.2.45, p. 219,20–27: iha vṛkṣa ity ukte kaś cic chabdaḥ śrūyate vṛkṣaśabdo ’kāntaḥ sakāraś ca pratyayaḥ / artho ’pi kaś cid gamyate mūlaskandhaphalapalāśavān ekatvaṃ ca / vṛkṣāv ity ukte kaś cic chabdo hīyate kaś cid upajāyate kaś cid anvayī—sakāro hīyate aukāra upajāyate vṛkṣaśabdo ’kārānto ’nvayī / artho ’pi kaś cid dhīyate kaś cid upajāyate kaś cid anvayī— ekatvaṃ hīyate dvitvam upajāyate mūlaskandhaphalapalāśavān anvayī / te manyāmahe—yaḥ śabdo hīyate tasyāsāv artho yo ’ rtho hīyate, yaḥ śabda upajāyate tasyāsāv artho yo ’rtha upajāyate, yaḥ śabdo ’nvayī tasyāsāv artho yo ’rtho ’nvayī // In the world, when /vṛkṣaḥ/ is said, a particular linguistic-unit (śabda) is heard, the stem /vṛkṣa/ ending with /a/ and the suffix /s/. A particular referent is also understood[, namely something] having root, trunk, fruits, and leaves, and the singular number. When /vṛkṣau/ is said, one linguistic-unit disappears, one appears and something remains: /s/ disappears, /au/ appears and the stem /vṛkṣa/ ending with /a/ remains. In the case of the meaning, too, something disappears, something appears and something remains: singularity disappears, duality appears and that which has the root, trunk, fruit, and leaves remains. So we (I) hold the following view — when a linguistic-unit disappears, its meaning disappears. When a linguistic-unit appears, its meaning appears. When a linguistic-unit remains, its meaning remains. See Cardona (1967) for a more detailed explanation on this passage. See also Scharf (1996, 40). 3 Vārttika, 9–15, pp. 30,2–31, 14: arthavanto varṇā dhātuprātipadika­ pratyayanipātānām ekavarṇānām arthadarśanāt, varṇavyatyaye cārthāntaragamanāt, varṇānupalabdhau cānarthagateḥ, saṃghātārthavattvāc ca / saṃghātasyaikārthyatvāt subabhāvo varṇāt / anarthakās tu prativarṇam arthānupalabdheḥ, varṇavyatyayāpāyopajanavikāreṣv arthadarśanāt // “Phonemes

94

4

5

6

7

8

Akane Saito are meaningful. This is because a meaning is seen in the verbal root or the nominal stem or the affix or the indeclinable which consists only in a single phoneme; and because another meaning is understood if phonemes are transposed; and because there is no understanding of the meaning if phonemes are not perceived; and because the complex (saṃghāta) [of phonemes] is meaningful. Since the complex [of phonemes] has one meaning, there is no case-ending [applied] after [each of those] phonemes. On the other hand, [phonemes] are meaningless because the meaning is not understood phoneme by phoneme. For, the meaning is [still] perceived when phonemes are transposed (vyatyaya), dropped (apāya), added (upajana) or substituted (vikāra).” yathā tarhi rathāṅgāni vihṛtāni pratyekaṃ vrajikriyāṃ praty asamarthāni bhavanti tatsamudāyaś ca rathaḥ samarthaḥ / evam eṣāṃ varṇānāṃ samudāyā arthavantaḥ, avayavā anarthakā iti // “As, then, the disassembled (vihṛta) parts (aṅga) of a chariot lack the capacity individually (pratyekam) to effect the action of locomotion (vrajikriyā) but a chariot, which is the composite of these [parts], possesses [this] capacity. In the same way, the combination of these phonemes is meaningful, while [individual] parts are meaningless.” Mahābhāṣya, Paspaśāhnika, p. 1,6–11: atha gaur ity atra kaḥ śabdaḥ / […] yenoccāritena sāsnālāṅgūlakakudakhuraviṣāṇināṃ saṃpratyayo bhavati sa śabdaḥ / See Joshi (1986, 12–23), which is the full annotated translation, for further information. Kaiyaṭa, a Grammarian in the eleventh century, explains in his commentary Pradīpa on the Paspaśāhnika of the Mahābhāṣya that this Patañjali’s statement on śabda teaches sphoṭa. See Joshi (1986, 9–16,23). Mahābhāṣya, Paspaśāhnika, p. 1,11–13: atha vā pratītapadārthako loke dhvaniḥ śabda ity ucyate / tad yathā / śabdaṃ kuru / mā śabdaṃ kārṣīḥ / śabdakāry ayaṃ māṇavaka iti / dhvaniṃ kurvann evam ucyate / tasmād dhvaniḥ śabdaḥ / “Alternatively, sound (dhvani), of which the meaning of the word is [well-]known in the world, is called śabda. For example: “Make a sound” (śabdaṃ kuru), “Do not make a sound” (mā śabdaṃ kārṣīḥ), “This boy is noisy” (śabdakāry ayaṃ māṇavakaḥ)—a person who is making sound is said in this way. Therefore, śabda means sound.” Mahābhāṣya, on Vārttika 5, 1.1.70, p. 181,19–24: evaṃ tarhi sphoṭaḥ śabdo dhvaniḥ śabdaguṇaḥ / katham / bheryāghātavat / tad yathā—bheryāghāto bherīm āhatya kaś cid viṃśatipadāni gacchati, kaś cit triṃśat, kaś cic catvāriṃśat / sphoṭaś ca tāvān eva bhavati / dhvanikṛtā vṛddhiḥ // dhvaniḥ sphoṭaś ca śabdānāṃ dhvanis tu khalu lakṣyate / alpo mahāṃś ca keṣāṃ cid, ubhayaṃ tatsvabhāvataḥ // “If it is so, then sphoṭa is śabda, and sound is the property of śabda. [Question:] How is it possible? [Response:] Like the striker of the kettle-drum. For example, after the striker of the kettle-drum strikes it, some [sound] goes twenty steps, another [sound] goes thirty [steps], and another [sound] goes forty [steps]. But

The Theory of the Sphoṭa

95

sphoṭa is of the exactly same extent [in each case]. The increase is caused by sound [alone]. Śabdas have sound and sphoṭa. As for sound, as is well-known, it is observed by some people as small and [by some people as] big. Both [sound and sphoṭa] are [established] by nature.” 9 Vākyapadīya, I.47: vitarkitaḥ purā buddhyā kva cid arthe niveśitaḥ / karaṇebhyo vivṛttena dhvaninā so ’nugṛhyate // “The [word itself], which has been ascertained (vitarkita) by the intellect before [the utterance] and which has been made to reside (niveśita) in a particular meaning (i.e., a word-form is assigned to a particular meaning), is seized (anu-grah-) through sound which has been transformed (vivṛtta) by the speech organs.” Auto-commentary on the Vākyapadīya, I.47 (pp. 105,6–106,3): karaṇebhyo vivṛttena iti / avikriyādharmakaṃ hi   śabdatattvaṃ dhvaniṃ vikriyādharmāṇam anu vikriyate / tac ca sūkṣme vyāpini dhvanau karaṇavyāpāreṇa pracīyamāne sthūlenābhrasaṃghātavad upalabhyena nādātmanā prāptavivartena tadvivartānukāreṇātyantam avivartamānaṃ vivartamānam iva gṛhyate // “Regarding ‘which has been transformed by the speech organs’ [in the verse]. Indeed, the reality of the word (śabdatattva) which is not characterized by the transformation (vikriyā) is transformed according to sound which is characterised by the transformation. Then when the subtle external sound, which pervades [the ether], is accumulated by the function of the [speech-] organ, by means of the gross bodily resonance (nāda) which has been transformed and which is the object of perception just like a cloud compacted [by the wind], the untransformable [word itself] is grasped as if it is transformed by imitating the [bodily resonance].” 10 Auto-commentary on the Vākyapadīya, I.76 (p. 142,1–3): iha dvividho dhvaniḥ prākṛto vaikṛtaś ca / tatra prākṛto nāma yena vinā sphoṭarūpam anabhivyaktaṃ na paricchidyate / vaikṛtas tu yenābhivyaktaṃ sphoṭarūpaṃ punaḥ punar avicchedena pracitataraṃ kālam upalabhyate / “In this world, there are two kinds of sound, namely the primary sound and the secondary sound. Among these two, the primary [sound] is, indeed, that without which the nature of sphoṭa which is unmanifested is not delineated (pari-chid-). On the other hand, the secondary [sound] is that by which sphoṭa, which is already manifested, is perceived for a longer time, again and again without being interruption.” See Brough (1951, 36–41). Note that in some of my previous papers, I even differentiated the word dhvani here from nāda, while not in this book to make the argument simple. 11 In Vākyapadīya, I.101, Bhartṛhari refers to the subclassification of nāda too, namely prākṛtanāda and vaikṛtanāda, both of which are explained in the same way as in the case of prākṛtadhvani and vaikṛtadhvani. It is possible to say that nāda can be the concept of sound related to the body or pronunciation at the speech organ, namely for the speaker’s side, while dhvani can be to the outside the body or perception of the external world, namely for the hearer’s side. But I confess that I

96

Akane Saito

cannot give clear distinction between prākṛtadhvani and prākṛtanāda, and between vaikṛtadhvani and vaikṛtanāda. We should also keep in mind that since Bhartṛhari introduces various alternative views, each term does not need to be consistent in its usage. 12 Vākyapadīya, I.82–84: yathānuvākaḥ śloko vā soḍhatvam upagacchati / āvṛttyā na tu sa granthaḥ pratyāvṛtti nirūpyate // pratyayair anupākhyeyair grahaṇānuguṇais tathā / dhvaniprakāśite śabde svarūpam avadhāryate // nādair āhitabījāyām antyena dhvaninā saha / āvṛttaparipākāyāṃ buddhau śabdo ’vadhāryate // “For example, a chapter (anuvāka) or a verse [of the Vedas] come to the state of endurable [for memory] (soḍhatva) through repetition. But the [whole] book (grantha) is not grasped in each repetition. In the same way (just as the memory of a vedic verse is strengthened by its repetition), through the unanalysable cognitions dependent on the grasping (perception), the [word’s] own form is ascertained when the word is manifested by the sounds (dhvani). In the intellect into which the seeds are imparted by the bodily resonances (nāda) and which has reached maturity through repetition, the word is ascertained together with the final sound.” 13 Vākyapadīya, I.89–91: yathaiva darśanaiḥ pūrvair dūrāt santamase ’pi vā /   anyathākṛtya viṣayam anyathaivādhyavasyati // vyajyamāne tathā vākye vākyābhivyaktihetubhiḥ / bhāgāvagraharūpeṇa pūrvaṃ buddhiḥ pravartate // yathānupūrvīniyamo vikāre kṣīrabījayoḥ / tathaiva pratipattṝṇāṃ niyato buddhiṣu kramaḥ // “Just as when [an object is] in a distant place or in a dark place, one identifies completely otherwise after taking the object differently by previous experiences. In the same way, when a sentence is being manifested by the causes that manifest the sentence, initially the [unreal] cognition occurs in a form that is separated into parts. Just as the prior-posterior relation is fixed in the products of milk or a seed [namely, yogurt or rice grains], the sequence [of erroneously perceiving the word] is fixed in every hearer’s intellect.” 14 Auto-commentary on the Vākyapadīya, I.104 (pp.170,6–171,4): iha ke cid ācāryā vyaktaṃ sphoṭaṃ sahajena dhvaninā sarvato dūravyāpinā prakāśasthānīyena gandhena yuktaṃ dravyaviśeṣam ivāvirbhāvakāla eva saṃbaddhaṃ manyante dhvaninā / yathaiva pradīpe ghaṭasaṃniviṣṭāvayavaṃ praty upādānaṃ tejodravyaṃ tadāśritaś ca tadvikriyānuparivartī prakāśaḥ, ghaṇṭāyāś cābhighātena vyaktatarau sphoṭanādau sarveṣāṃ varṇānām abhiniṣpattau dharma iti / “Here, some teachers think that the sphoṭa, manifested (vyakta) by the sound that arises along with it and spreading like light in all directions for some distance, just as certain substances arise simultaneously with their odour, is linked with it at the very moment of [the sphoṭa’s] manifestation. In case of the lamp, there is a fiery substance, which is the material cause (upādāna) [shining] on a [particular] part of the pot, and there is the light that depends on that, that changes (anuparivartin) by the transformation of the [material cause]. And, [in the same way], as a result of the striking of a bell,

The Theory of the Sphoṭa

15 16 17 18

19 20 21

22

97

both the sphoṭa and resonance [arise] in a very clear form, and [both] are a quality in the production of any phoneme.” Iyer (1937) investigates the holders of impermanent sphoṭa (anityasphoṭavādin) in Vākyapadīya, I.102–06 (Iyer 1937, vv. 103–05). See Bronkhorst (1991) for further information. Brough (1951, 44–45) explains this in relation to the premodern concepts of jātisphoṭa and vyaktisphoṭa. Scharf (1996, 23, 35–36) shows the Grammarians’ concept of ākṛti on the basis of the Mahābhāṣya. Auto-commentary on the Vākyapadīya, I.23 (p. 56,3–4): varṇāvayavāvagrahaprā­ ptasaṃskārābhiḥ kramotpannabhir buddhibhiḥ pūrvam agṛhītā, avyaktaṃ gṛhītā vā saṃskṛte ’ntaḥkaraṇe caramavijñānenākṛtiḥ paricchidyate / “The generic form, which is previously not grasped or just vaguely grasped by the cognitions which arise sequentially and possess the latent impressions that have been attained by the forms of the phonemes as parts, is discriminated by the final cognition in the predisposed internal organ (i.e., the mind).” In the Vākyapadīya, I.68–69, and in its auto-commentary, Bhartṛhari gives various views on jāti in relation to śabdasvarūpa. See Iyer (1966) and Biardeau (1958) for a full translation of the Sphoṭasiddhi. Vṛttikāragrantha, 1.1.3–5, p. 38,13: pūrvavarṇajanitasaṃskārasahito ’ntyo varṇo vācakaḥ (pratyāyakaḥ) / “The last phoneme, accompanied by the latent impression produced from the previous phonemes, conveys the meaning directly.” The following is a portion of Maṇḍana’s criticisms against the idea that after each perception of individual phonemes, there is a unitary apprehension that makes the sequence understood. Auto-commentary on the Sphoṭasiddhi, v. 8 (p. 66,4– p. 69,3): tathā hi svato varṇā nityatayā vibhutayā ca na deśanibandhanaṃ nāpi kālanibandhanaṃ parāparabhāvam anupatantīti prakhyānanibandhana eṣa samupāśrīyate / tac cedam advayam akramam, yadviparivartinas te ’rthapratyayahetavaḥ / na cedaṃ pūrvopalabdhisaṃbandhinīṃ parāparatāṃ gocarayati, varṇāvalambitayā tadupalabdhīnām aviṣayīkaraṇena; na cāsamīhitāvadhyavadhimadbhāvaṃ paurvaparyaṃ pratyetum arhati; na caikasmin nānāvarṇāvayavātmani pade bhinnam avadhyavadhimadbhāvam apekṣituṃ kṣamate; prativarṇopalambhanaprabhāvitāni ca bhāvanābījāni kāmaṃ saṃhatisamāsādanād ekam anekāvalambi smaraṇaṃ janayeyuḥ, tāvatā caritārthebhyo nāparo varṇātmasu viśeṣo labhyate / “To explain, phonemes, as being eternal and all-pervading by nature [according to you Mīmāṃsakas], do not follow the prior-posterior relation caused by either locations or time. So it is admitted that this [prior-posterior relation of phonemes] is caused by [a single] apprehension [that comes in the end]. But this [apprehension] is non-dual and nonsequential, those [phonemes] transforming (viparivartin) into which are the causes of the understanding of the referent. Also this [apprehension] does not have as its

98

Akane Saito

domain priority and posteriority of the preceding perceptions [of each phoneme]. For, as being resting upon [those] phonemes, it does not have the perceptions of the [phonemes] as its scope. Also [it] cannot recognize the prior-posterior relation without requiring the relation of a limit [in the form of time or space] and what is limited [by such a limit]. Nor can it require the relation of a limit and what is limited, that is distinct (split), in a single word [even though it is] consisting of multiple phonemes as its parts. It is acceptable that the seeds, i.e., the impressions produced from the perception of each phoneme, give rise to the single recollection (= apprehension) which is dependent on the multiple [phonemes], because they attain (samāsādana) combination (saṃhati); [but] the other distinguishing character[, namely the capacity to make the sequence understood,] is not attained in the phonemes themselves from those [impressions], which have [already] fulfilled their purpose in this much.” 23 Sphoṭasiddhi, v. 13: na cāntyavarṇamātrasya puraḥ saṃbandhavedanam / akṣavartmātivṛttatvāt saṃskārasya na tadvataḥ // “Neither can the final phoneme be known before [learning] the relation [between the word and its object]. Nor can be the latent impression together with the [final phoneme], because [the latent impression] is beyond the range of the sense organ (= it is not the object of direct perception).” Here the important premise for Maṇḍana is that the word, from which a particular referent is understood, must be the object of direct perception. See the following section for the matter of direct perception in the sphoṭa theory. 24 Sphoṭasiddhi, v. 30: utpattivādino varṇāḥ kāmaṃ santu prabhedinaḥ / na tv asādhāraṇas teṣāṃ bhedo ’rthajñānakāraṇam // “Let phonemes be differentiated [each other] for the one who insists on [phonemes] being produced. However, their unique (= always new) distinction cannot be the cause of the understanding of the meaning.” 25 Auto-commentary on the Sphoṭasiddhi, v. 32, pp. 250,3–251,2: tasmād eka eva śabdātmā artheṣu saṅketena niyujyate loke pratyayakāle cāśrīyate; anyathā vaktṛbhede durnivāraḥ pratyayaḥ syāt / “Thus, one and the same word-body (śabdātman = śabdasvarūpa = sphoṭa) is connected to different objects by the linguistic convention, and [that single word-body] is depended on at the time of understanding [the meaning] in the world. Otherwise, the understanding cannot be avoided even though the speakers are different.” The following verse, which appears in the criticism of Kumārila, shows Maṇḍana’s attitude toward the understanding of the word meaning: Sphoṭasiddhi, v. 26: arthasyādhigamo na rte padarūpāvadhāraṇāt / tad arthabodhād yadi ca vyaktam anyonyasaṃśrayaḥ // “The understanding of the meaning is not [possible] without ascertainment of the form of a word. And if the [ascertainment of the form of a word] is caused by the understanding of the meaning, there is clearly the fault of mutual dependence.”

The Theory of the Sphoṭa

99

26 Sphoṭasiddhi, v. 32: na hetuphalabhāvo ’ṅgaṃ samutthāpakacetasām / arthabuddher anāśritya saṅketajñānakālayoḥ // “Since [it is] not resorted at the time of [learning] the linguistic convention and of the cognition [of the word meaning], the cause-effect relationship of the [speaker’s] mind[-moments] which make arisen [phonemes] is not the essential element (aṅga) for the understanding of the meaning.” Auto-commentary on the Sphoṭasiddhi, v. 32, pp. 248,1–249,2: na khalu samutthāpakacittakāryakāraṇatā tatkṛto vā svabhāvabhedo ’rthapratyayāṅgam /   jñāpakatve hi svavijñānam apekṣeta / dṛśyate ca tirohitavyavahitaprayuktāc chabdād arthajñānam / na ca tatra samutthāpakacittakāryakāraṇatāṃ kaś cana niścetum arhati / vaktrekatve ca niścite sā niścīyeta / na cāntareṇa śabdajñānaṃ tanniścayaḥ / “As is well known, neither the causal relation of the mind[-moments of the speaker] which causes [phonemes] to arise nor the intrinsic distinction [of phonemes], which are made by the [causal relation], is the essential element for the understanding of the meaning. For, if they make known [the meaning of the word], there would be expectation of their own cognition. However, the cognition of the object (meaning) is experienced from the word which is used [by people who are] in a distance or hidden place. But in that case, no one can ascertain the causal relation of the mind[-moments of the speaker] which causes [phonemes] to arise. But it can be ascertained if the identity of the speaker would be ascertained. And it is not ascertained without the cognition of the word.” 27 The investigation of error occupies an important position in Maṇḍana’s philosophy. Another of his works, the Vibhramaviveka, is directly addressing the structure of erroneous cognitions, as its title suggests. 28 Auto-commentary on the Sphoṭasiddhi, v. 21, p. 150,3–p. 152,2: nimittam evedam īdṛśaṃ śabdatattvopalabdher yad viparyāsayad eva śabdam upalambhayatīti, niyatasārūpyatvāt; na hi śabdāntaraviṣayadhvanivilakṣaṇā dhvanayo ’nye tasya vyaktau naḥ santi, yenāviparyāso ’vasīyeta / ata eva ca tulyarūpaḥ sarvapratipattṝṇāṃ viparyāsaḥ, tannimittasya samānatvāt / “It is this trigger[, namely sounds (dhvani)], as described, of the perception of the reality of the word (śabdatattva), which creating error causes the word to be perceived. This is because the similarity [of sounds] is always there. For, the other sounds, [similar but] different from the sounds having “another word“[, namely a particular sphoṭa,] as their object, are not in my view for the purpose of the manifestation of that [reality of the word, namely that sphoṭa], so that non-erroneous cognition could be ascertained. And exactly because of this, all the perceivers have the erroneous cognition in the same form. This is because the cause of the [erroneous cognition] is the same.” Sound /a/ of the word aśvaḥ is similar but different from that of agniḥ or of the sentence aśvam ānaya. Such similarity causes the erroneous cognition that it is made of phonemes.

100

Akane Saito

29 Auto-commentary on the Sphoṭasiddhi, v. 22, p. 167,1–5: tathā hi— bhedenānvīyamānam (emended from bhedenānanvīyamānam) api maṇikṛpāṇadarpaṇādiṣu tattvasya bhedavirodhāt tattvapratyayena samutsāritanikhilabhedaṃ mukham avasīyate / tathā dīrghādibhedānugame ’pi varṇātmā pratyabhijñābalena vidhūtabheda eko ’vagamyate, tathā padam api svapratyayāvagamyamānaikasvabhāvaṃ kiṃcidbhedaparāmarśe ’py abhinnaṃ niścīyate / “To explain, the face, even though connected to different [appearances], is ascertained as that all of whose variety has been expelled (samutsārita) by the cognition of [its] reality, since the reality contradicts the varieties [reflected] on [the surface such as] jewel, sword, or mirror. In the same way, in spite of its being accompanied by such variety as [its being] long are understood, the single phoneme itself whose differences are abandoned (vidhūta) is understood by force of recognition. Likewise the word, whose single nature is understood through the cognition of its [own form], is also determined as being undivided, even though there is awareness of some kind of variety.” 30 Sphoṭasiddhi, v. 23: pratyakṣajñānaniyatā vyaktāvyaktāvabhāsitā / mānāntareṣu grahaṇam atha vā naiva hi grahaḥ // “To manifest both vague and clear [objects] is restricted to perceptual cognition. In the other means of knowledge [such as inference] the object is either grasped or not.” 31 Auto-commentary on the Sphoṭasiddhi, v. 3ab, pp. 10,1–11,1: kiṃ punar idaṃ padaṃ nāma? śabdaḥ / kaḥ punar iha śabdo ’bhipretaḥ? kiṃ varṇāḥ? nety āha /   api tu—[Sphoṭasiddhi, v. 3ab:] arthāvasāyaprasavanimittaṃ śabda iṣyate // yathoktaṃ “yenoccāritena […]” ityādi / “But what is the word (pada)? It is śabda. What kind of śabda is intended here then? Is it phonemes? No, śabda is rather regarded as the trigger of giving rise to the ascertainment of the object (artha)—as [Patañjali] taught [in the form] that ‘[Śabda is] that which, when it has been uttered[, gives rise to the cognition of those [entities] which have a dewlap, tail, hump, hooves, and horns].’” 32 As we saw in the Sphoṭasiddhi, v. 26, Maṇḍana clearly states that the understanding of the meaning is not possible without ascertaining the word-form (padarūpa), and his sphoṭa theory is how we get this word-form, namely sphoṭa, in the process of perception. 33 One should note, however, that Śaṅkara discusses the sphoṭa while commenting to the aphorism “The universe bursts forth from scripture,” ataḥ prabhavāt (Brahmasūtrabhāṣya, 1.3.28), and also the Sphoṭavādin there concludes that “the universe, which is the object of denotation, arises from the speech, whose nature is sphoṭa,” nityāc chabdāt sphoṭarūpād abhidhāyakāt […] jagad abhidheyabhūtaṃ prabhavati (ibid.), which reminds us of the Śaiva scholars’ way of understanding the sphoṭa. 34 For this particular discussion on the sentence-sphoṭa, I have not yet found other texts prior to him. It might be either Jayanta’s invention or it can be the case

The Theory of the Sphoṭa

101

that there is some scholar who discussed the sentence-sphoṭa between Maṇḍana and Jayanta. Note that Joshi (1967, 70–72) refers to the Nyāyamañjarī when he mentions the Naiyāyika opponents in the Sphoṭanirṇaya. When did the confusion of phonemes and sounds happen? It might be the case that Jayanta stands as the turning point. 35 This Brough’s statement is criticized by Bronkhorst (1991) and Pinchard (2011, p. 462, ll. 21–31). 36 Sanderson (2004, p. 242, l. 10–p. 243, l. 3; p. 293, l. 4–p. 294, l. 2). Brunner (1974) is the first scholarly work and summary of the Netratantra. 37 Netratantra, vol. 2, p. 287–88, vv. 62cd–63: dhvanirūpo yadā sphoṭas tv adṛṣṭāc chivavigrahāt // prasaraty ativegena dhvanināpūrayañ jagat / sa nādo devadeveśaḥ proktaś caiva sadāśivaḥ // “When sphoṭa, whose nature is sound, flows forth with great force from the form of Śiva, which is unseen, filling the universe with sound, it is called nāda, O Master of the gods, and Sadāśiva.” See Padoux (1990, 93, 96, and 97–98, fn. 33). 38 See Sanderson (2007, 398–401). 39 Uddyota on the Netratantra, vv. 62cd–63, vol. 2, p. 288, ll. 3–10: sphuṭati abhivyajyate asmāt viśvaḥ śabdagrāmaḥ iti sphoṭaḥ śabdabrahma, ata eva dhvanirūpaḥ śabdanasvabhāvaḥ, adṛṣṭād iti anākṛter draṣṭrekarūpāt paranādāmarśātmanaḥ prakāśānandaghanāt śivasvarūpād ativegena avyucchinnadrutanadīghoṣavat prasarati / kīdṛk / dhvaninā ghaṇṭānuraṇanarūpeṇa nādāntena jagat viśvam āpūrayan āmarśanena ātmasātkurvan / “Sphoṭa, from which the entire (viśva) totality of speech-units is split open (sphuṭati), [namely] becomes manifest, is śabdabrahman. For this very reason, [such sphoṭa], whose nature is sound (dhvani), [namely] verbalization (śabdana), flows forth with great force, [namely] just like uninterruptedly-rapid roaring of a river from the nature of Śiva that consists of nothing but (ghana) light and bliss, who is unseen (adṛṣṭa), [namely] formless, whose nature is one with the perceiver (draṣṭṛ) (so he cannot be seen), [in other words] whose nature is representation of the highest resonance (paranāda). [Question:] Of what kind [is the sphoṭa]? [Response:] [Sphoṭa, which is] filling the world or the universe, [namely] assimilating (internalizing) [the universe] in itself by means of sound (dhvani), [namely] by means of ‘the end of nāda’ (nādānta) in the form of reverberation (anuraṇana) of a bell.” 40 See Sphoṭanirṇaya, 57–60. See also Iyer (1947, 12–15, 134; 1966, 10, 21–25). Filliozat (1984, 139–140) introduces these two definitions, referring them to the Sarvadarśanasaṃgraha doxography of the Mādhava (fourteenth century). 41 In other words, according to Bhartṛhari, śabda surely has the nature of making an object understood (arthapratyāyakatva); but when he talks of sphoṭa, the focus is the manifestation of śabda.

102

Akane Saito

42 It is found first in the Vākyapadīya, I.134, pp. 213–21, and there is no trace of it before Bhartṛhari. 43 Vākyapadīya, I.134 Jha, Ña, Ṭa, pp. 219–20: seyam ākīryamāṇāpi nityam āgantukair malaiḥ / antyā kaleva somasya nātyantam abhibhūyate // tasyāṃ dṛṣṭasvarūpāyām adhikāro nivartate / puruṣe ṣoḍaśakale tām āhur amṛtāṃ kalām //   prāptoparāgarūpā sā viplavair anubandhibhiḥ / vaikharī sattvamātreva guṇair na vyavakīryate // This [paśyantī] is never conquered, even though it is always affected by adventitious impurities, just like the last digit of the moon. When its nature is perceived, one is freed from the [karmic] duty. [People] say that such [last digit] in the soul consisting in sixteen parts is the immortal part. It is in the form of being colored by adventitious impurities. But vaikharī is not affected by the attributes, as if it were the pure existence. 44 Torella (2009) teaches the difference of the attitudes between Somānanda and Utpaladeva. While Somānanda criticized Bhartṛhari’s metaphysics of linguistic monism, Utpaladeva is more in favor of Bhartṛhari’s view. Nonetheless, both are against his theory of sphoṭa (See fn.20 of Torella 2009). See Nemec (2011) and Torella (2014) for the studies of the Śivadṛṣṭi. For the previous scholarly works on sphoṭa and paśyantī, see Filliozat (1984); Padoux (1990, 166–222). 45 Śivadṛṣṭi, 2.60–61ab: sphoṭasyāsatyarūpair hi padādyair vyaṅgyatā katham / paśyantyāḥ satyarūpāyā asatyair vyaṅgyatā na ca // tādṛgvyañjanasāpekṣā sā na kiṃ cana jāyate / “How would it be possible for sphoṭa to be manifested by the word and so on whose nature is unreal? And paśyantī in the form of real is not manifested by unreal things. Such like she (paśyantī) which needs a means of manifestation (vyañjana) does not become anything.” 46 This is different from Helārāja’s stance in his commentary on the Vākyapadīya. Helārāja (ca. 980 CE) is a Kashmirian Grammarian who explains sphoṭa as well as the ontological hierarchy of speech. He follows Bhartṛhari in the understanding of the sphoṭa, and does not give it the status of the highest reality. Also the understanding of the intuitive speech equated with the highest speech (parā vāc) is unchanged from Bhartṛhari’s explanation. Indeed, the fact that Helārāja refers to the sphoṭa and paśyantī in the different chapters of the Vākyapadīya (sphoṭa in the Jātisamuddeśa and paśyantī in the Dravyasamuddeśa, both in the third book) shows us that he recognizes the difference between sphoṭa and paśyantī. In Kashmir in almost the same period, Jayanta, Somānanda, Utpaladeva and Helārāja had different views on the sphoṭa theory. Their common attitude is the reference to three levels of speech, which is, as far as I know, not discussed anymore outside Kashmir before premodern times. 47 Nāgeśa composed three versions, with different levels of abridgment or prolixity, of his Mañjūṣā: Paramalaghumañjūṣā, Laghusiddhāntamañjūṣā, and Mañjūṣā. See Coward and Kunjunni Raja (1990, 323).

The Theory of the Sphoṭa

103

48 There is another Sphoṭasiddhi composed by a sixteenth- to seventeenth-century philosopher Bharata Miśra. His approach is different from that of the premodern Grammarians. See Pinchard (2011) for further information. 49 According to Kauṇḍa Bhaṭṭa, there are eight categories of sphoṭa: [individual] phoneme-sphoṭa (varṇasphoṭa), [individual] word-sphoṭa (padasphoṭa), [individual] sentence-sphoṭa (vākyasphoṭa), [individual] indivisible wordsphoṭa (akhaṇḍapadasphoṭa), [individual] indivisible sentence-sphoṭa (akhaṇḍavākyasphoṭa), and also a universal-sphoṭa (jātisphoṭa) for each of phoneme, word, and sentence (i.e., varṇajātisphoṭa, padajātisphoṭa, vākyajātisphoṭa). See Seneviratne (2015) and Bronkhorst (2005) for Śeṣa Kṛṣṇa’s and Bhaṭṭoji Dīkṣīta’s classification of sphoṭa. 50 Those premodern Grammarians quote frequently Bhartṛhari’s Vākyapadīya as their ground for the argument, but they have almost no remark on Maṇḍana. Rather exceptionally, Seneviratne (2015, 5) points out that Śeṣa Kṛṣṇa has the identical half verse with the Sphoṭasiddhi, v. 23, though he does not directly refer to Maṇḍana. 51 See Cardona (1976, 303): “In the view of such later Pāṇinīyas the term varṇa does not mean ‘sound unit’ in this context; it denotes a unit lower than a word, namely a base or an affix.” See also Bronkhorst (2005, 5, fn.11). 52 Sphoṭanirṇaya, v. 63 (5): ghaṭenetyādiṣu na hi prakṛtyādibhidā sthitā/ vasnasādāv ivehāpi sampramoho hi dṛśyate// “Indeed, in [the words] such as ghaṭena, division into [the smaller units] such as a stem is not established. This is because confusion (sampramoha) [of how to divide it] is seen even in this [word ghaṭena], just as [there is confusion of how to divide] in [the words] such as vas or nas (the inflected forms of yuṣmad and asmad).” See Joshi (1967, 138–40). 53 Auto-commentary on the Sphoṭanirṇaya v. 64 (6): hare ’va viṣṇo ’vetyādau padayoḥ “eṅaḥ padāntād ati” ity ekādeśe sati na tadvibhāgaḥ sujñānaḥ / “When a single substitution replaces [the beginning and ending vowels of] two words in [the sentences] such as ‘O Hari, protect [me]’ or ‘O Viṣṇu, protect [me]’ on the basis of A 6.1.109, division of th[ose sentences into the words] is not easily known.” Aṣṭādhyāyī, 6.1.109: eṅaḥ padāntād ati “A single vowel /e/ or /o/ replaces /e/ or /o/ which occurs as the last vowel of a word, when a short vowel /a/ follows it.” See Joshi (1967, 150–51). 54 Auto-commentary on the Sphoṭanirṇaya v. 64 (6): vastutaḥ padaiḥ padārthabodhavad vākyena vākyārthabodha iti padārthaśaktiḥ padeṣv iva vākyārthaśaktir vākye ’bhyupeyeti padasphoṭavākyasphoṭau vyavasthitau / “In reality, just as the word meanings are understood by words, the sentence meaning is understood by a sentence. Therefore, just as the capacity [to denote] the wordmeanings is [admitted] in the words, the capacity [to denote] the sentence meaning is to be admitted in a sentence. In this way, a word-sphoṭa and sentence-sphoṭa have been established.”

104

Akane Saito

55 Auto-commentary on the Sphoṭanirṇaya v. 65 (7): uttaravarṇapratyakṣasamaye ’vyavahitottaratvasaṃbandhena upasthitapūrvavarṇavattvaṃ tathā taduttarapratyakṣakāla upasthitaviśiṣṭatadvarṇavattvaṃ tasmin sugraham iti tā dṛśānupūrvīghaṭitapadatvasyeva vākyatvasyāpi sugrahatvāt // “This is because we can easily understand that at the time of perceiving the subsequent phonemes, [the subsequent phoneme] is accompanied by the preceding phoneme which is brought [to the mind], through the relation of uninterrupted succession; in the same way, at the time of perceiving [another (third)] phoneme subsequent to that, the [third phoneme] is accompanied by that [second] phoneme specified [with the first phoneme] which have been brought [to the mind]; and accordingly, just as we can easily grasp that it is a word which consists of such a sequence [of phonemes], [it is the same] even for a sentence.” See Joshi (1967, 168–70). 56 Joshi (1967, 79): “The Vākyasphoṭa theory assumes that sentence alone is the conveyor of meaning, but it does not maintain categorically that the constituents of sentence have no meaning at all. According to it, it is possible to assign some meaning to the components on the basis of structural analysis. The main implication of the theory is that the meaning of word is always contextual, and no study of meaning apart from a complete sentence can be taken seriously. In other words, the words have meaning only when they form a part of sentence.” 57 Kauṇḍa Bhaṭṭa quotes Vākyapadīya, v. I.73 (Sphoṭanirṇaya, v. 66): pade na varṇā vidyante varṇeṣv avayavā na ca / vākyāt padānām atyantaṃ praviveko na kaś cana // “There are neither phonemes in words nor parts in phonemes. There is no absolute distinction of words from a sentence.” 58 See Sphoṭanirṇaya, v. 69, where the following verse of Bopadeva, according to Bhaṭṭoji Dīkṣita’s Śabdakaustubha, is quoted: śakyatva iva śaktatve jāter lāghavam īkṣyatām / aupādhiko vā bhedo ’stu varṇānāṃ tāramandavat // “It must be regarded that there is simplicity [in postulating] the universal of a denotative capacity (śaktatva), just as [we can assume the universal] of the nature of the denoted [objects]. Alternatively, let the distinction be caused by the imposed attributes, just like loudness or lowness of phonemes” 59 See the auto-commentary on the Sphoṭanirṇaya, v. 72.

Primary Sources Aṣṭādhyāyī

Sharma, Ram Nath, ed. 1987–2003. The Aṣṭādhyāyī of Pāṇini. Edited and translated, 6 vols. New Delhi: Munshiram Manoharlal.

Brahmasūtrabhāṣya

Narayan Ram Acharya, ed. 1948. The Brahmasūtrabhāṣya, Text with Foot-Notes and Variants. Third edition, with Notes, Various Readings etc. Bombay: Satyabhamabai Pandurang for the Nirnaya Sagar Press.

The Theory of the Sphoṭa

105

Mahābhāṣya

Kielhorn, Franz, ed. 1880–85. The Vyākaraṇa-Mahābhāṣya of Patañjali. Bombay: Government Central Book Depôt.

Netratantra

1939. The Netra Tantram, with Commentary by Kṣemarāja, Vol. 2. Ed. by Madhusudan Kaul Shastri. Kashmir Series of Texts and Studies 61. 2 vols. Bombay: Tattva-Vivechaka Press.

Paramalaghumañjūṣā

Shukla, Kalikaprasad, ed. 1961. Paramalaghumañjūṣā of Śri Nageśa Bhaṭṭa, with the Commentary Jyotsnā. Baroda: Baroda Sanskrit Mahavidyalaya, University of Baroda.

Pramāṇavārttika

Gnoli, Raniero, ed. 1960. The Pramāṇavārttikam of Dharmakīrti. The First Chapter with the Autocommentary. Serie Orientale Roma 23. Roma: Istituto Italiano per il Medio ed Estremo Oriente.

Śivadṛṛṣṭi

Shastri, Madhusudan Kaul, ed. 1934. The Śivadṛṣṭi of Śrī Somānandanātha with the Vṛtti by Utpaladeva. Kashmir Series of Texts and Studies 54. Srinagar: Research Dept.

Ślokavārttika

Rai, Ganga Sagar, ed. 1993. Ślokavārttika of Śrī Kumārila Bhaṭṭa with the Commentary Nyāyaratnākara of Śrī Pārthasārathi Miśra. 2nd edition. Varanasi: Ratna Publications.

Sphoṭanirṇaya

Joshi, S.D., ed. Sphoṭanirṇaya of Kauṇḍa Bhaṭṭa. In Joshi 1967.

Sphoṭasiddhi

Ramanatha Sastri, S.K., ed. 1931. The Sphoṭasiddhi of Ācārya Maṇḍanamiśra with the Gopālikā of Ṛṣiputra Parameśvara. Madras University Sanskrit Series 6. Madras: University of Madras.

Vākyapadīya

Subramania Iyer, K. A., ed. 1966–73. Vākyapadīya of Bhartṛhari with the Commentaries Vṛtti and Paddhati of Vṛṣabhadeva. Deccan College Monograph Series. 3 vols. Pune: Deccan College Postgraduate / Research Institute.

Vārttika

Kielhorn, Franz, ed. Vārttika commentary of Kātyayāyana on the Aṣṭādhyāyī. In Mahābhāṣya.

Vṛttikāragrantha

Frauwallner, Erich, ed. Vṛttikāragrantha. In Frauwallner 1968.

Secondary Sources Biardeau, Madeleine. 1958. Sphoṭa Siddhi (La démonstration du Sphoṭa) par Maṇḍana Miśra. Pondichery: Institut français d’indologie. Bronkhorst, Johannes. 1991. “Bhartṛhari on Sphoṭa and Universals.” Asiatische Studien/Études Asiatiques 45: 15–18.

106

Akane Saito

Bronkhorst, Johannes. 2005. “Bhaṭṭoji Dīkṣita on sphoṭa.” Journal of Indian Philosophy 33 (1): 3–41. Brough, John. 1951. “Theories of General Linguistics in the Sanskrit Grammarians.” Transactions of the Philological Society 50: 27–46. Brunner, Hélène. 1974. “Un Tantra du nord: le Netra Tantra.” Bulletin de l’École française d’Extrême-Orient 61: 125–97. Cardona, George. 1967. “Anvaya and Vyatireka in Indian Grammar.” The Adyar Library Bulletin 31: 313–52. Cardona, George. 1976. Pāṇini, A Survey of Research. The Hague, Paris: Mouton. Coward, Harold G., and K. Kunjunni Raja, eds. 1990. Encyclopedia of Indian Philosophies: The Philosophy of the Grammarians. General editor Karl H. Potter. Delhi: Varanasi: Patna: Motilal Banarsidass. Filliozat, Pierre. 1984. “On the Semantic Use of the Word Sphoṭa.” In Amṛtadhārā (Professor R.N. Dandekar Felicitation Volume), 133–40. Delhi: Ajanta Publications. Frauwallner, Erich. 1968. Materialien zur ältesten Erkenntnislehre der Karmamīmāṃsā. Veröffentlichungen der Kommission für Sprachen und Kulturen Süd- und Ostasiens 6. Wien: Österreichische Akademie der Wissenschaften. Iyer, K.A. Subramania. 1937. “Who Are the Anityasphoṭa-vādinaḥ?” Proceedings and Transactions of the All-India Oriental Conference Mysore, December 1935. Bangalore: Government Press, 258–63. Iyer, K.A. Subramania. 1947. “The Doctrine of Sphota.” Journal of the Ganganatha Jha Research Institute 5: 121–47. Iyer, K.A. Subramania. 1966. Sphoṭasiddhi of Maṇḍana Miśra. English translation. Poona: Deccan College Post-graduate and Research Institute. Joshi, S.D. 1967. The Sphoṭanirṇaya of Kauṇḍa Bhaṭṭa: With Introduction, Translation and Critical and Exegetical Notes. Poona: University of Poona. Joshi, S.D. 1986. Patañjali’s Vyākaraṇa-Mahābhāṣya Paspaśāhnika, Introduction, Text, Translation and Notes. Poona: University of Poona. Nemec, John. 2011. The Ubiquitous Śiva: Somānanda’s Śivadṛṣṭi and His Tantric Interlocutors. Oxford: Oxford University Press. Padoux, André. 1990. Vāc. The Concept of the Word in Selected Hindu Tantras. Albany: SUNY Press. Pinchard, Alexis. 2011. “The Argumentative Value of Āgamic Quotations in the Sphoṭasiddhi by Bharata Miśra.” Journal of Indian Philosophy 39 (4): 461–77. Sanderson, Alexis. 2004. “Religion and the State: Śaiva Officiants in the Territory of the King’s Brahmanical Chaplain.” Indo-Iranian Journal 47: 229–300. Sanderson, Alexis. 2007. “The Śaiva Exegesis of Kashmir.” In Mélanges tantriques à la mémoire d’Hélène Brunner/Tantric Studies in Memory of Hélène Brunner, ed. by Dominic Goodall and André Padoux, 231–442. Collection Indologie 106. Pondicherry: EFEO / IFP. Scharf, Peter M. 1996. The Denotation of Generic Terms in Ancient Indian Philosophy: Grammar, Nyāya, and Mīmāṃsā. Transactions of the American Philosophical

The Theory of the Sphoṭa

107

Society Held at Philadelphia for Promoting Useful Knowledge 86.3. Philadelphia: American Philosophical Society. Seneviratne, Rohana Pushpakumara. 2015. The Revival of Sphoṭa in Early Modern Benares: Śeṣakṛṣṇa’s Sphoṭatattvanirūpaṇa (A Thesis Submitted for the Degree of Doctor of Philosophy). Oxford: University of Oxford. Torella, Raffaele. 2009. “From an Adversary to the Main Ally: The Place of Bhartṛhari in the Kashmirian Śaivādvaita.” In Bhartṛhari: Language, Thought and Reality (Proceedings of the International Seminar on Bhartṛhari, December 12–14, 2003), 343–54. Delhi: Motilal Banarasidass. Torella, Raffaele. 2014. “Notes on the Śivadṛṣṭi by Somānanda and Its Commentary.” Journal of Indian Philosophy 42 (Issue 5): 551–601.

5

A Buddhist Refutation of Sphoṭa Sara McClintock

1 Introduction This contribution examines arguments against the sphoṭa theory of language1 made by the eighth-century Buddhist philosophers Śāntarakṣita and Kamalaśīla in their twin treatises, the verse composition Tattvasaṃgraha (TS) and its prose commentary, the Tattvasaṃgrahapañjikā (TSP; for convenience, we refer to the two texts jointly as TS/P).2 In keeping with their larger project to defend a metaphysics of emptiness (śūnyatā), which rejects the possibility of any ultimately real singleness or permanence, the primary arguments against sphoṭa are aimed mostly at its supposed indivisibility and eternality. Using mereological arguments similar to those they deploy against other such purportedly ultimately real entities as universals (sāmānya), natures (svabhāva), the Self (ātman), and God (īśvara), Śāntarakṣita and Kamalaśīla reject this theory that posits a unitary and permanent entity to account for the capacity of language to disclose its meanings. One line of their argument focuses on the problem of how sphoṭa, literally a bursting or disclosure, remains indivisible when language has parts— whether understood primarily in terms of phonemes (varṇa), words (pada), or sentences (vākya)—that must be arranged sequentially to express any meaning. In addition, these Buddhists also attack the idea that sphoṭa could be eternal, given that the cognition of meaning manifests only in the wake of particular linguistic expressions and not perpetually. One of the most surprising aspects of Śāntarakṣita and Kamalaśīla’s refutation of sphoṭa, at least initially, is their heavy reliance on arguments provided by the Mīmāṃsaka Kumārila, a figure who more typically appears in the TS/P as a target of attack. While disagreements remain, Śāntarakṣita and Kamalaśīla do not hesitate to make direct use of Kumārila’s rebuttals of sphoṭa as advanced in his Ślokavārttika (ŚV). Of course, Kumārila’s views on the

A Buddhist Refutation of Sphoṭa

109

eternality of language and the authority of the supposedly authorless Veda come in for much criticism throughout the TS/P.3 However, in this particular instance, the Buddhists find themselves aligned with their Mīmāṃsaka nemesis, and they make free use of his work.4 Śāntarakṣita and Kamalaśīla also draw heavily on previous Buddhist refutations of the sphoṭa theory, especially as articulated by Dharmakīrti in his Pramāṇavārttikasvavṛtti (PVSV),5 and further elaborated by Śākyabuddhi in his sub-commentary on the PVSV, the Pramāṇavārttikaṭīkā. Dharmakīrti’s arguments in turn rely on Vasubandhu’s earlier criticism of the little-studied Buddhist Vaibhāṣika theories resembling those of the Grammarians (Vaiyākaraṇas) on the topic of signification, but these controversies seem to have receded by the time of Śāntarakṣita and Kamalaśīla.6 We can note as well that the independently minded Maṇḍana Miśra in his famous Sphoṭasiddhi offered numerous rebuttals to both Buddhist and Mīmāṃsaka arguments against sphoṭa, and that these rebuttals were later addressed in turn by the Buddhist Karṇakagomin in his Pramāṇavārttikasvavṛttiṭīkā.7 Unfortunately, space does not allow us to examine all these materials here, and we must limit ourselves to the arguments as they appear in the TS/P.

2  Rejection of the Eternal Verbum We begin with the arguments found in the chapter known as the investigation of the eternal verbum (śabdabrahmaparīkṣā). According to the Grammarian Bhartṛhari, the entire world may be seen as the manifestation of an ultimate reality known as śabdabrahman or the eternal verbum—a kind of unchanging primordial essence (brahman) consisting in language or sound (śabda). The chapter takes up the problem of the eternal verbum with particular reference to Bhartṛhari’s magnum opus, the Vākyapadīya (VP). Even though the term sphoṭa does not appear in it, the chapter nevertheless sets out the baseline Buddhist objections to the idea of an eternal element of language that accounts for the capacity of words and sentences to convey meaning. As such, these arguments are relevant to the more explicit critique of sphoṭa found later in the TS/P. After all, the sphoṭa that explains the comprehension of language might easily be characterized as a particular form of the eternal verbum.8 Śāntarakṣita describes the foundational view of the eternal verbum in the first verse of this chapter as follows: This multitude of things is recognized to be the transmutation (pariṇāma) of that highest brahman, made of sound, which is untouched by destruction or production.9

110

Sara McClintock

As Kamalaśīla renders explicit through a direct quotation, this verse is a gloss on Bhartṛhari’s opening verse in the VP, here in B. K. Matilal’s translation: The essence of language has no beginning and no end. It is the imperishable Brahman, the ultimate consciousness, which is transformed in the form of meanings and which facilitates the functioning of the world.10

Although the precise nature of the equation of sphoṭa and the eternal verbum according to Bhartṛhari is a topic that exceeds the scope of this chapter, we can note that the two concepts share many features and are closely linked.11 Most pertinently, both are understood, at least by these Buddhist thinkers, to be eternal and single,12 and to stand in a relationship of serving as a basis or source of the various manifestations of language and, in the case of the eternal verbum, the entire world. For these Buddhists, therefore, both sphoṭa and the eternal verbum are subject to a similar set of arguments rooted in a mereological analysis which aim to demonstrate the impossibility of an eternal and single cause for that which is patently ephemeral and diverse in nature.

2.1  Eternality and Sequence Are Incompatible A fundamental axiom running throughout Śāntarakṣita and Kamalaśīla’s thought is that eternality is incompatible with sequence (krama) of any kind— whether of time, place, or manifestation. We find a version of this argument in the chapter on scripture, where the authors seek to refute that the Veda is eternal and authorless.13 Arguing that the Veda cannot be eternal—and that it must therefore have an author—since it manifests sequentially (i.e., in terms of its phonemes, words, sections, and so on), Śāntarakṣita states: And there is no sequential manifestation in what is eternal, since its manifestation is not reasonable.14

Kamalaśīla adds the following helpful commentary: That is to say: first of all, a temporal sequence—such as that of seed, seedling, and tendril—is not reasonable since all the linguistic elements of the Veda would be simultaneous due to being eternal. Nor is a spatial sequence—such as that of a line of ants and so on—reasonable, since all linguistic elements are situated in a single atmosphere. Nor is a sequence of manifestation reasonable, since there can be no manifestation of that which is eternal due to it not having supplementary qualities associated with it which could cause it to manifest at some times and not others.15

A Buddhist Refutation of Sphoṭa

111

For those Indian philosophers who wish to hold any element of language to be eternal, there is the problem of how to account for its sequential and temporally intermittent nature. One standard strategy is to propose that while language is eternal, its manifestation (vyakti) is not. For Buddhist thinkers, this is untenable. For the eternal language to manifest at some points in time and not others, there would have to be some special or supplemental quality (atiśaya) that could intermittently be associated with the eternal language to cause it to manifest at some times and not others.16 On the Buddhist analysis, however, this presents a contradiction since it means that the purportedly eternal, by which they also understand unchanging, language has undergone a change from a non-manifest to a manifest state, as well as from an independent to a dependent state. The very notion of sequence is at odds with the idea of something unchanging and autonomous—whether that is language itself or sphoṭa.17

2.2  Further Arguments on Eternality, Sequence, and Change We can now briefly examine two passages from the chapter on the eternal verbum that foreshadow the explicit critique of sphoṭa that occurs later in the text. The first passage deals with the supposed eternality of language, and the second with the problem of sequence. In both cases, the core issues have to do with the problem of the multiplicity and changeability of the manifestations and how to reconcile those with the singleness and eternality of either the eternal verbum or sphoṭa. We begin with two verses: And if for every entity there were asserted a single, undifferentiated nature of language, then everything would come to exist in a single location and all knowledge would have the same phenomenal content. But if the nature of language were different for each of its manifestations, then there would be the undesired consequence that brahman would become multiple, due to having the form of a nature that is distinct and multiple, in accord with the differences in the manifestations.18

Kamalaśīla’s commentary makes it clear that this twofold objection is offered in response to Bhartṛhari’s idea, seen in the quotation of VP, I.1 above, that the eternal verbum undergoes a transmutation (pariṇāma) in order to manifest as the various objects of the world. Here, the Buddhists resort to their standard mereological analysis to urge unwanted consequences by showing that the eternal verbum, as a single entity, can be neither the same as nor different from its purported transmutations, the manifold entities of the world.

112

Sara McClintock

This argument about the problem of the variegated manifestation of that which is single will return again in the specific context of the refutation of sphoṭa, examined below. Likewise, the following argument about the impossibility of sequence for an unchanging entity also foreshadows arguments specific to the refutation of sphoṭa. Śāntarakṣita lays out the idea in a verse: But even if the world consists in the eternal verbum in terms of being its effect, nevertheless, because the eternal verbum cannot be modified, there can be no sequential production from it at all.19

This reasoning is then cogently unpacked by Kamalaśīla in the commentary: Indeed, since language cannot be modified due to its being eternal, therefore, no sequential production can be obtained from it. Rather, everything would be produced simultaneously due to the cause being complete, unobstructed, and efficacious. For effects are delayed due to an incomplete cause; and if that eternal verbum is complete, then how could it be dependent on some other thing, such that its effects would not be simultaneous?20

Crucial to this line of thinking is the idea that an eternal entity must be changeless, and the very act of producing multiple entities entails some kind of change. For these Buddhists, it does not matter whether the eternal entity is conceptualized as producing effects or transmuting into a variety of things. The basic problem remains the same: sequences require change and variegation requires multiplicity. As we will see, these core arguments are critical to the extensive refutation of sphoṭa later in the TS/P.

2.3  Awareness Is Nonlinguistic Of course, Śāntarakṣita and Kamalaśīla have concerns that go beyond merely wanting to reject any theory that relies on an unchanging entity; they are also motivated to dismantle any claim that experience is necessarily bound up with language. This is in keeping with their general understanding of language and concepts as useful distortions that will ultimately be eliminated in the nonconceptual state of the full awakening of a Buddha. The notion that all forms of awareness are imbued with language also contradicts a fundamental tenet of their system, inherited from Dignāga and Dharmakīrti: namely, that perception (pratyakṣa) is always devoid of concepts (kalpanāpoḍha). Thus, it is not surprising that in the continuation of his commentary on TS 128, discussed above, Kamalaśīla cites another key verse from Bhartṛhari, here again with Matilal’s translation: There is no awareness in this world without its being intertwined with language. All cognitive awareness appears as if it is interpenetrated with language.21

A Buddhist Refutation of Sphoṭa

113

Later, Kamalaśīla returns to this verse to dispute its fundamental claim, explaining that “when the eye is viewing a form while the mind has traveled elsewhere, one experiences a cognition of blue and so on that is simply not burdened with language.”22 Kamalaśīla does not defend this claim here but indicates that it will be explained extensively later on in the text (it is treated in the chapter on perception, the pratyakṣalakṣaṇaparīkṣā, among other places). But the fact remains that one fundamental objection that Śāntarakṣita and Kamalaśīla have to the idea of the eternal verbum is that it introduces a linguistic element where they see none. Their objection to sphoṭa is similar in eschewing what they see as an unnecessary metaphysical element to explain linguistic signification.

3  Refutation of Sphoṭa The explicit refutation of sphoṭa in the TS/P occurs in the context of the investigation of scripture (śrutiparīkṣā). The term “scripture” here mainly indicates the Veda as conceived by the Mīmāṃsā tradition as understood by Kumārila. Thus, much of the chapter is concerned with refuting such ideas as the authorlessness of the Veda, the trustworthiness of scripture more generally, and the eternality and omnipresence of language in its various forms. With these last two qualities of language, we see similarities to the notion of the eternal verbum examined above. However, despite commonalities, the Mīmāṃsaka position differs from that of the Grammarians in several regards, including the crucial question of how language conveys meaning. Whereas the Grammarians assert a “trans-phonetic”23 entity, sphoṭa, as the imperishable and indivisible mechanism serving as the cause for the cognition of meaning, the Mīmāṃsaka understands phonemes arranged in a particular order to perform this function. In this, Kumārila is in close accord with Buddhists like Dharmakīrti, Śāntarakṣita, and Kamalaśīla, all of whom we can label varṇavādins, or “proponents of the doctrine that phonemes are the conveyor of meaning.”24 It is on the basis of this commonality that Śāntarakṣita directly deploys nine verses from Kumārila’s ŚV, where the opponent is understood to be the sphoṭavādin or “proponent of sphoṭa” despite the fact that the chapter on the whole is directed against Kumārila. We should not, however, conclude that Śāntarakṣita and Kamalaśīla are in perfect agreement with Kumārila when it comes to the question of how language conveys meaning. Significant differences remain. For example, for Kumārila, the phonemes that are arranged to convey meaning are themselves

114

Sara McClintock

eternal, as are the words into which the phonemes are arranged. For Śāntarakṣita and Kamalaśīla these claims are both impossible and absurd.

3.1  Words Are Ephemeral—Mīmāṃsakas and Buddhists Agree In the first part of this chapter, Kumārila is shown to be defending his notion that language is eternal and all-pervasive. This includes him responding to objections that center on contradictions that emerge when one considers that words, as they are uttered and deployed, are not eternal but are temporally structured in terms of the sequence of their components, the phonemes. An unnamed objector now urges an unwanted consequence on the Mīmāṃsaka in two verses: And without sequentiality, phonemes are not known to be productive of meaning. Therefore there is the undesired consequence for the Mīmāṃsaka that words must possess sequence. But the conceptualization of language as eternal would be meaningful only for those for whom the word is separate from the phonemes and is devoid of sequence.25

Here, in a somewhat roundabout fashion, the objector urges the Mīmāṃsaka to reconsider the doctrine that language expresses meaning through the articulation of words represented by phonemes and arranged in a particular order. The objector asks whether this can really be compatible with the doctrine that language is eternal (śabdanityatva). Implying that it cannot, he then suggests that the Grammarian’s doctrine of sphoṭa is better suited as the mechanism for the conveyance of meaning for one who, like the Mīmāṃsaka, holds language to be eternal.26 Since Kumārila roundly rejects the doctrine of sphoṭa as unnecessary, this is a galling challenge to the Mīmāṃsā position. The longer passage within which this objection is situated includes an extended Mīmāṃsaka argument responding to the more general objection that (1) because words require a temporal sequencing of phonemes they cannot be eternal even if the phonemes themselves are granted to be eternal; and (2) that since words, not phonemes, are what signify meaning for the Mīmāṃsaka, the signification of meaning therefore also cannot be eternal unless one introduces another entity like sphoṭa, as do the Grammarians. We do not have space to take up this extended argument in detail, especially as it involves a dispute between the Mīmāṃsakas and the Grammarians. Our task involves seeing how the Buddhists Śāntarakṣita and Kamalaśīla refute the sphoṭa doctrine. But this short foray into a challenge to the Mīmāṃsaka refutation of sphoṭa is nevertheless useful as it helps us to understand how and why these Buddhist authors rely on

A Buddhist Refutation of Sphoṭa

115

Kumārila to help with their own refutation at this point despite their significant differences with the Brahmanical author on questions of the eternality of letters, words, and language in general.

3.2  From Phonemes to Meanings—Dharmakīrti The section on the refutation of sphoṭa begins by recapping some key claims from Dharmakīrti about how meaning is conveyed through a process in which phonemes are spoken, heard, and remembered in sequence. Futoshi Ōmae nicely explains the theory as articulated by Dharmakīrti and restated by Śāntarakṣita and Kamalaśīla: In the continuity of the mind (saṃtāna) of the speaker there arise vivakṣās (desires to utter) of varṇas (e.g., “s” “a” “r” “a” “ḥ”) in due order. Among them the vivakṣā of each precedent varṇa (e.g., sa-kāra) is considered to be an immediate condition (samanantarapratyaya) of the vivakṣā of the succeeding varṇa (i.e., a-kāra). Therefore, the vivakṣā of the precedent varṇa (i.e., sa-kāra) brings about its result (i.e., sa-kāra) as well as the vivakṣā of the succeeding varṇa (i.e., a-kāra). Thus varṇas are uttered in due order. In the continuity of the mind of the hearer each uttered varṇa (e.g., a-kāra) directly produces its direct cognition (śruti) with the help of its predecessor (e.g., direct cognition of sa-kāra) and indirectly brings about its recollection (smṛti) with the help of its predecessor (i.e., recollection of sa-kāra). Thus varṇas are grasped and recollected in due order. Then how is the understanding of the meaning from varṇas possible? After hearing the varṇas in due order there arises a cognition of all varṇas in the mind of the hearer. Then he obtains the notion of pada (the word “saraḥ”) from this cognition.27

This succinct summary, which Ōmae helpfully further elucidates through a clever graphic illustration, presents the basic Buddhist theory, broadly shared by Kumārila, concerning how multiple phonemes work together to produce the cognition of a single word. As varṇavādins, both camps insist on the critical role of sequence in the presentation of phonemes for the production of meaning; otherwise, if sequence were not a factor, then two words with the same phonemes but in different orders (e.g., latā and tāla) would be indistinguishable. The problem of sequence, and its incompatibility with sphoṭa, will be tackled again shortly. But first, let us see how Śāntarakṣita and Kamalaśīla introduce their explicit refutation of sphoṭa. The opening section presents Dharmakīrti’s theory,28 briefly sketched above, for how phonemes work together to produce a cognition of a word:

116

Sara McClintock

But the presumption that all phonemes are eternal has been refuted. For the cause of the cognition of meaning has been established in relation to that which is impermanent. A phoneme which has been produced from a subsequent intention to speak which follows after an initial intention to speak is heard immediately following that initial phoneme. The knowledge that is produced from the knowledge of the previous phoneme, heard not too quickly, subsequently produces the memory of itself in dependence on the memory of that previous phoneme. This succession of the phonemes, located in persons, is both the producer and the cause respectively in relation to the awarenesses that give rise to and grasp it. Thus, it is utterly clear that the phonemes differ with each word according to the difference in their sequence, as for example with the words damaḥ, madaḥ, latā, and tāla. It is through this kind of sequence that these words are indicators of distinct meanings.29

It is helpful to notice here the emphasis on memory as a conditioning factor in the production of new knowledge. Later on, picking up the terminology used by Bhartṛhari in his VP, the authors speak of the conditioning (saṃskāra) of an awareness by a previous awareness. The notion that a single awareness can nevertheless be conditioned by previous awarenesses is one way that these authors attempt to overcome the problem of unity and multiplicity in linguistic cognitions. We will revisit this below. Having first briefly sketched their theory for how phonemes work together to produce a cognition of a word, Śāntarakṣita and Kamalaśīla now introduce their refutation of sphoṭa with a curt dismissal: And therefore the imagination of sphoṭa is also pointless here. For that sphoṭa is conceptualized by Grammarians to be for the sake of the comprehension of meaning. And the phonemes themselves have that capacity. Therefore the conceptualization of that sphoṭa is useless.30

Here, in an application of something like Occam’s razor, these Buddhists reject sphoṭa as unnecessary for the purpose of the production of a cognition of meaning. They now attack a key problem for the proponent of sphoṭa, namely, that sphoṭa is not perceived. And it is determined that it does not exist because it, being perceptible, is not perceived. But if it were held to be imperceptible, then it could not be an indicator (jñāpaka) in the manner of an inferential sign.31

The Grammarian’s idea that there is a trans-phonetic entity, sphoṭa, responsible for the production of meaning in the mind of the auditor begs the question of why we do not perceive or otherwise know of this entity. If the answer is that

A Buddhist Refutation of Sphoṭa

117

the entity is imperceptible, then the Grammarian faces the problem of how such an imperceptible entity can serve as an indicator (jñāpaka) of meaning. Just as in an inference evidence can be an indicator of a property to be proved only if that evidence is itself perceptible, sphoṭa similarly can be an indicator of meaning only if it is perceptible. And according to Śāntarakṣita and Kamalaśīla, following Dharmakīrti, if an entity is perceptible in theory but is not perceived in a particular place, then that entity is absent from that place.

3.3 An Eternal Partless Entity Cannot Manifest Occasionally or Dependently If sphoṭa is thought to function in a manner similar to a sense faculty such that it would be able to indicate meaning through its mere presence, then the Grammarian would face other conceptual problems, as Śāntarakṣita argues in the next two verses: But if it is claimed that sphoṭa is an indicator of meaning through its mere existence—like the eye—since it is established as being the cause of the awareness of that meaning, then it would always be [functioning as an indicator of that meaning]; even when there was ignorance of the convention or when the phonemes were not heard, there would be an awareness of the meaning brought about by that due to the presence of the cause which is capable (śakta) of indicating meaning.32

Śāntarakṣita and Kamalaśīla link this idea that sphoṭa functions through its mere presence to its purported eternality. With the following two verses, Śāntarakṣita brings us to the heart of the Buddhist critique: namely, that as an eternal entity, sphoṭa could not manifest intermittently or only occasionally, nor could it depend on another entity to provoke its occasional manifestation (as this would imply that it changes over time—a position incompatible with its eternality). Śāntarakṣita summarizes this argument succinctly: That is, this sphoṭa is eternal, and it can have no dependency whatsoever. And there can be no manifestation of it by sounds, conventions, or letters, since such is not observed. For it has been said that awareness itself is manifestation, and we do not find awareness of that sphoṭa. Therefore, this imagination of sphoṭa manifesting is totally pointless.33

The Buddhists here are attacking several ideas at once. First, they point out that if sphoṭa were real, we would observe it in our minds. Second, they argue that as a purportedly eternal entity, its occasional manifestation is simply impossible.

118

Sara McClintock

This is because as an eternal entity, it should not need any other contributing factor in order to come about. So either it should be manifest at all times or it should never manifest—in both cases, it would be useless. Śāntarakṣita now allows the voice of the Grammarian to come in through a direct quotation from Bhartṛhari’s VP: [Proponent of sphoṭa:] When the seeds (bīja) that have been planted in cognition by sounds (nāda) become ripened through the repetition of the other phonemes along with the final vocalization (dhvani), language (śabda) appears.34

Although at first glance this position looks similar to that of both the Mīmāṃsakas and the Buddhists, there are two important differences. The first is just that when Bhartṛhari uses the word “language” (śabda) here, he has in mind the special entity otherwise known as sphoṭa. The second is that Bhartṛhari does not seem to be advocating a theory in which the cognition of a meaning proceeds through a process whereby phonemes uttered in sequence leave traces in the mind of the hearer which collectively then appear in the form of a memory that aggregates the total to make the word. Rather, here, because sphoṭa is partless, the Grammarian holds that each phoneme possesses the full signifying power of sphoṭa such that hearing just one phoneme is theoretically sufficient to understand the meaning of the entire word (though in practice most people are not able to access sphoṭa so easily and they thus require the repetition of the sphoṭa through the pronunciation of the remaining phonemes). The first distinction is dismissed quickly by Śāntarakṣita who again maligns sphoṭa by calling it a “baseless imagining,” reminding us that “one does not observe any other linguistic element appearing there [in awareness].”35 That is, once again, the Buddhist feels there is no need to introduce a further entity beyond the phonemes to account for the production of meaning. Śāntarakṣita then moves to the second position distinguishing sphoṭa from the theories of the varṇavādins, namely, that it is partless and thus can have nothing to do with sequence of any kind. He points out: But whether it is produced by sequenced vocalizations or whether it is manifested by them, there is still a contradiction for those who maintain sequence in sphoṭa. Given that sphoṭa is partless, neither its sequential production nor its sequential manifestation makes sense, since they [i.e., the production or manifestation] would occur at all times, due to not existing apart from the single nature [i.e., the sphoṭa itself].36

The distinction in play here is between production (jāti) and manifestation (vyakti). For the Buddhists and the Mīmāṃsakas, meaning is produced in

A Buddhist Refutation of Sphoṭa

119

cognition through the sequential presentation of the phonemes. But for the sphoṭavādin, meaning is manifested. This is a crucial difference because supposedly manifestation does not require change. The sphoṭa, which is eternal and partless, remains the same. The only difference is that it is now manifested in the mind of the hearer. However, for Śāntarakṣita and Kamalaśīla, this distinction does not avoid the problem of sequence for the Grammarian.

3.4 Postulating Sphoṭa as Having Parts Is Futile Turning now to a hypothetical solution for the Grammarian, Śāntarakṣita considers whether the sphoṭa doctrine might be saved if it were granted to have parts after all. He says: If that sphoṭa had parts, then, like the phonemes, the sphoṭa parts would not be expressive of meaning due to their sequentiality. But why would the sphoṭa parts not be expressive of meaning? Why should invisible parts be imagined?37

Kamalaśīla supplies the implication of these rhetorical questions as follows: “If the sphoṭa parts are conceptualized to be signifiers, then it is better to allow the phoneme parts to be the signifiers.”38 In other words, the solution of introducing parts to sphoṭa is no solution as words and sentences already contain parts (the phonemes and letters) which already serve as signifiers according to the varṇavādins. Although the commentary to this verse appears in one place to focus more on the sphoṭa of an entire sentence (vākya), as opposed to that of a word (pada), the theoretical problems remain largely the same. In the case of the expression of a word, the proponent of sphoṭa must contend with the sequentially arranged parts of the word, that is, the phonemes (varṇa). In the case of the sentence, the proponent of sphoṭa must contend with sequentially arranged individual words. For the Buddhists seeking to refute sphoṭa, whether we are dealing with phonemes or words is less important than the question of how we can conceptualize the appearance of a single meaning in the wake of a linguistic event that has parts.

3.5  Postulating a Gradual Manifestation of Sphoṭa Is Futile Śāntarakṣita now allows the Grammarian to speak through two verses that closely mirror verses from Bhartṛhari’s VP: [Proponent of sphoṭa:] Even though the production or manifestation of the sphoṭa is accomplished all at once by a single vocalization, the other phonemes are

120

Sara McClintock

performed for the sake of a more explicit manifestation. Since the fundamental nature of that sphoṭa is difficult to determine when it is done like that [i.e., through a single phoneme], it is performed to a greater degree through more phonemes possessing a common capacity.39

The proponent of sphoṭa here seeks to avoid an unwanted consequence of the position that sphoṭa is partless: namely, if sphoṭa is partless, then it must be fully and equally present in all the components of speech that bring about its manifestation. If that is the case, then hearing just one phoneme of a word should be enough to provoke the full manifestation of sphoṭa in the mind of the listener. The Grammarian does not dispute this assessment, but nevertheless insists that most people are not capable of discerning the sphoṭa through just one phoneme or just one exposure to a verse. Predictably, this response does not impress the Buddhists. Śāntarakṣita replies by rhetorically wondering: Well then, why is it not the case that when one phoneme or another phoneme is repeated again and again the manifestation does not repeat, since there is no distinction in that way that you have described above?40

In other words, when the entire, partless sphoṭa is fully present in just one phoneme or any other single part of a linguistic expression, one should be able to provoke its manifestation simply by repeating that one phoneme or part again and again. The argument that the remaining parts have other causal capacities that are necessary for the manifestation of sphoṭa is likewise not tenable on the view that sphoṭa is partless.41

3.6  From Perception of Phonemes to Memory to Meanings At this point, Śāntarakṣita and Kamalaśīla shift to a heavy reliance on arguments advanced by Kumārila in his Ślokavārttika. Kamalaśīla introduces this new section by indicating that the previous arguments had all been made having provisionally accepted that the manifestation of sphoṭa was in theory possible. Yet this leads to the following problem: If something distinct, separate from the phonemes, which is the essence of language known as sphoṭa, were to appear, then its manifestation would be possible, since the nature of manifestation is to be perceived. Insofar as it does not appear, therefore, it is as previously presented [i.e., impossible].42

This verse serves as a transition to a new section dedicated to a more positive pursuit, namely, to an explanation of how words convey meanings even

A Buddhist Refutation of Sphoṭa

121

without the trans-phonetic special entity known as sphoṭa. We have already touched briefly on the Buddhist and Mīmāṃsaka theory of how this works. Śāntarakṣita now quotes six verses from the ŚV to more fully flesh out this theory: Therefore, it is the final cognition of the collection of phonemes which have been previously perceived in sequential awarenesses that is the cause of the awareness of meaning. For when there is the cognition of the final phoneme, a memory of all the phonemes which has been created by the conditioning of all [the previous phonemes] is simultaneously produced. And this mental cumulative awareness (samuccayajñāna) with regard to all things is asserted by all disputants (vādin), even if it is produced sequentially. And if one does not accept this, then there could not arise a cumulative vision with the form “one hundred” and so on even when [one hundred and so on items had been] seen sequentially. Therefore, even if the previous awareness of phonemes is sequential, in terms of its being aural or being mental, later on there would still be a simultaneous memory. Thus, since the phonemes taken up in that memory are not far from the understanding of a meaning, worldly beings proclaim the opinion that meaning comes from the word (śabda).43

As these verses make clear, Śāntarakṣita and Kamalaśīla share with Kumārila a basic understanding of how the pronunciation of words functions as a cause for a cognition of meaning through a process of sequential enunciation of phonemes. What happens, as we have briefly discussed, is that when each subsequent phoneme in a word (or word in a sentence) is articulated, the perception of the sound makes an impression in the mind of the listener. The impressions remain so that after the vocalization of the final phoneme is complete, there is the production of a cognition consisting in the memory of the previously sequentially heard phonemes. This memory then becomes the vehicle for understanding the meaning of the word (or sentence). Although this theory is discussed with an understanding of language as an aural phenomenon, the same principle could probably be applied to written words as the eye takes in the diverse letters in sequence. The cultural cache of the spoken word at the time of these compositions accounts for the emphasis on the oral and aural elements of linguistic expression.

122

Sara McClintock

3.7  Phonemes Are Also Impermanent After this long string of verses from the ŚV, Śāntarakṣita now inserts a few verses indicating some key differences between the Buddhist position and that of Kumārila. He first says: And all this makes sense only if cognition has phenomenal content (ākāra). For otherwise, how could those destroyed phonemes appear in memory?44

This verse raises a whole separate realm of discourse, one that concerns whether cognition must be endowed with a form, or what we can call “phenomenal content.”45 Although we do not have space to consider this further here, it is important to note that Śāntarakṣita and Kamalaśīla want to emphasize that their agreement with Kumārila is not without qualifications.46 Similarly, Śāntarakṣita now clarifies that the mechanism by which the previous phonemes condition the cumulative awareness that arises after the final phoneme is pronounced does not allow for eternal phonemes as the Mīmāṃsaka maintains. He says: But it is emphatically not the case that the previously known phonemes which are being remembered abide in the cumulative awareness with a concealed manifestation, because of the undesired consequence that they would appear clearly [if they were there]. Abiding has been previously rejected. If the phonemes did abide, then memory would occur exactly at the time of the awareness of the experience of the phonemes, since there would be a single cause [for both the phonemes and the memory].47

Again, Śāntarakṣita wants to combat any talk of manifestation, which is the mechanism by which the Grammarians try to reconcile the eternality and unchangeability of sphoṭa with its occasional and seemingly sequential appearance. But this statement is also addressed to the Mīmāṃsakas, who, unlike the Buddhists, hold that the phonemes are eternal and thus are always present, even if in hidden (tirobhuta) form in the final cumulative cognition. In a further nod to the Buddhist commitment to radical impermanence, Śāntarakṣita reminds his readers that any hypostasized abiding (sthiti) had already been disproven earlier.48

3.8  The Apparent Unity of Words With these differences from Kumārila’s position clarified, the final section of the explicit refutation of sphoṭa returns again to verses from the ŚV, this time

A Buddhist Refutation of Sphoṭa

123

focusing on the purported single nature of sphoṭa and of meanings. Śāntarakṣita first sets out a response to a question attributed to the Grammarians as to how in the absence of the single sphoṭa there could be a singular phenomenal content in cognition upon hearing a word.49 Śāntarakṣita follows Kumārila in responding: But we emphatically do not refute the idea that the word “cow” (gauḥ) is single. There would be the idea of singleness in regard to a word due to the singleness of its object and its meaning.50

In other words, no one is denying that there can be a single cow, at least conventionally speaking, and that this is also the single meaning of the word “cow” when uttered in the singular. But this is not in conflict with the idea that the awareness which knows this meaning is cumulative in nature, since the phonemes that collectively produce this cognition of the single meaning are multiple. While the word “cow” (in both English and Sanskrit) is quite short, an examination of longer words makes the phonetic multiplicity inherent in the word itself quite clear. Again, following Kumārila: That idea of singleness might pertain in the case of the word “cow” due to the speed of pronunciation and due to the shortness of the interval between the first and last phoneme. But the division in words like “Devadatta” is clearly cognized.51

In other words, while the phenomenal content of a “cow” may be of a single cow, and the meaning of the word “cow” may be single, the final cognition of the meaning is nevertheless a cumulative awareness, which means that multiple phonemes uttered in sequence have all contributed to the production of the final cognition of meaning. With this statement, Śāntarakṣita completes his presentation of the refutation of sphoṭa.

3.9 Epilogue Śāntarakṣita now summarizes the formal reasoning behind his arguments with two verses, the first one taken from the ŚV: And that cognition of a meaning arises from phonemes, since it arises subsequent to the awareness of them. For whatever is like that [i.e., whatever arises subsequent to the awareness of something] arises from that previous thing, just like the cognition of fire arises from the awareness of smoke. No cognition of meaning is found consisting in a subsequent awareness of an appearance of a

124

Sara McClintock

word apart from its phonemes. Therefore there is no other linguistic element that is expressive of meaning.52

Of course, this refutation leaves a host of unanswered questions regarding the nature of signification according to the Buddhist tradition. There are profound disagreements between these Buddhists and virtually all Brahmanical opponents. Since Śāntarakṣita and Kamalaśīla, like Dignāga and Dharmakīrti, reject universals, they are faced with difficult problems concerning how the same word (e.g., “cow”) can signify a wide range of individual cows. What is it that is same about the various cows such that a speaker can use a single word to refer to them all? The Grammarian tries to explain signification through a mysterious, partless, eternal entity called sphoṭa which could account for the sudden comprehension of meaning through the encounter with language. As Buddhists, any such entity would be immediately suspect. But a universal, such as was favored by Naiyāyikas and Mīmāṃsakas, would be equally problematic. As is well known, Buddhists developed the exclusion (apoha) theory of linguistic signification, a nominalist theory that Kumārila and others found great delight in ridiculing. Despite their temporary alliance with Kumārila, the battle over signification raged on in other sections of their respective works.

4 Conclusion In this chapter, we have considered Śāntarakṣita and Kamalaśīla as varṇavādins, or proponents of the doctrine that meaning is conveyed through phonemes and not through a mysterious trans-phonetic eternal entity like sphoṭa. But unlike Kumārila, who also holds phonemes to be the key to the process of linguistic signification, these Buddhists also do not ascribe any special metaphysical status to phonemes or any other linguistic element. Nor do they claim any special, eternal status for the relationship between words and their meanings, as these relationships are always a matter of transactional agreement (vyavahāra). The question of whether words or sentences are the primary units of meaning is also not of the same grave concern as it is for Mīmāṃsakas, Naiyāyikas, and Vaiyākaraṇas. That is, while these thinkers hold that words produce meaning through the sequential utterance of phonemes, they also maintain that meaning takes place at the level of sentences.53 Indeed, as Kamalaśīla attests in his introduction to the TSP, meaning can even be conveyed by whole treatises,

A Buddhist Refutation of Sphoṭa

125

which he labels “mega-sentences” (mahāvākya).54 Whatever the context, the principle is the same: meaning is conveyed through smaller linguistic units arranged sequentially, that is, phonemes arranged sequentially in words; words arranged sequentially in sentences; and sentences arranged sequentially in paragraphs and treatises. In every case, the smaller units produce awarenesses which condition subsequent awarenesses such that the later awarenesses contain the trace (vāsanā) of the previous ones in the form of a memory. An awareness of meaning can thus remain single, while nevertheless still being conditioned by a multitude of prior successive linguistic elements, all of which are understood as conditioning factors for the production of the final cognition of the meaning in the form of a single cumulative awareness (samuccayajñāna). In theory, these traces could be infinite, since Buddhists do not accept a beginning to time, but in no conceivable world can any element of language be partless and eternal. That final cumulative awareness that arises from the sequence of phonemes that we call a word is said by Śāntarakṣita to be an exclusion (apoha) in the form of a reflection (pratibimba); it is also equated with the meaning of a word (padārtha) known as a mental flash (pratibhā).55 Kamalaśīla adds the important specification that this reflection can equally be understood to be the meaning of a sentence.56 However, before one becomes too comfortable, thinking that we now have grasped the essence of linguistic signification in the form of this reflection, we must take heed of Śāntarakṣita’s warning, voiced elsewhere, that neither the signifier nor the signified is, in any case, ultimately real.57 (Interestingly, despite the sphoṭa doctrine, Bhartṛhari appears to hold a very similar position.)58 The arguments against the ultimate reality of not only language but even of all that language is said to signify lie at the heart of both the TS/P and Śāntarakṣita and Kamalaśīla’s explicitly Madhyamaka works. Like the arguments deployed against a single, partless, eternal sphoṭa standing in relation to diverse, sequential, transitory words, the more general arguments against language and its meanings—and indeed against all entities— rely on variations of mereological analysis to demonstrate the impossibility of any ultimate singleness or multiplicity in things. But as long as one does not probe too far, Śāntarakṣita and Kamalaśīla provisionally allow that words can produce meanings in the mind of a hearer through a cumulative mnemonic awareness conditioned by successive previous phonemes coming together in words, sentences, and even treatises. They thus uphold the possibility of the communication of meaning through language, all the while warning us not to imagine there to be anything fixed or permanent about any part of the process.

126

Sara McClintock

Notes 1 For an overview of the sphoṭa theory and its various stages of development, see Chapter 4 in this volume. Note that the bulk of the discussion in this contribution is negative, in the sense that we aim to reveal the arguments against the sphoṭa theory and not those that directly take up the Buddhist views on the nature of language and linguistic expression. For these topics, see Chapters 8 and 10 in this volume. 2 The arguments against sphoṭa are found mainly in two sections of the texts: the investigation of the eternal verbum, the śabdabrahmaparīkṣā (TS/P 128–52); and the refutation of the doctrine of sphoṭa, or the sphoṭavādakhaṇḍana, a section (TS/P 2698–732) of the investigation of scripture, the śrutiparīkṣā (TS/P 2084–809). Both the śabdabrahmaparīkṣā and the sphoṭavādakhaṇḍana have been critically edited by Paolo Giunta (2009), who also provides an Italian translation of the śabdabrahmaparīkṣā, and a thorough introduction clarifying the extant Sanskrit manuscripts and Tibetan translations. His work is extremely valuable given the poor state of the manuscripts and published editions for these sections. In this contribution, all readings from these sections follow Giunta’s edition. My deepest thanks to Paolo Giunta for providing me with a copy of this work. Note that for all verses after TS 525, the 1926 Krishnamacharya edition appears as one number higher than the 1963 Shastri edition. This was due to an error by Krishnamacharya, who believed a verse had been elided from the manuscripts at this point. However, none of the Tibetan translations attest to the missing verse, and the current standard numbering follows Shastri. Our numbers correspond to Shastri throughout. 3 Although the topic recurs throughout the TS/P, the most important location for the sustained critique of the Mīmāṃsaka position on the eternality of the Veda is the lengthy chapter on scripture, the śrutiparīkṣā. For Kumārila’s own views on language, see Chapters 9 and 15 in this volume. 4 Nine verses from the sphoṭavāda section of the ŚV appear directly in the sphoṭavādakhaṇḍana section of the śrutiparīkṣā with only slight modifications. The verses are ŚV sphoṭa 109, 112–16, 120–21, and 135; these correspond to TS 2719, 2720–24, 2728–29, and 2730. Thirteen other verses from the sphoṭavāda section of the ŚV are also included in an earlier section of the chapter at TS 2140–54; these verses correspond to ŚV sphoṭa 22, 32, 34, 28, 38–44, and 50–51. For an overview of the arguments and structure of the śrutiparīkṣā, see Verpoorten (1994). 5 See especially PVSV, 119,17–29; 126,24–129,21; and 134, 1–25. For analysis, see Eltschinger (2001, 2007). The latter contains a revised and condensed version of the former (Eltschinger 2007, 157–79) as well as a French translation of the relevant passages (Eltschinger 2007, 277–80; 319–33; and 353–57).

A Buddhist Refutation of Sphoṭa

127

6 Jaini (1959) details the earlier Buddhist arguments, noting that neither Vasubandhu nor his commentator Yaśomitra refer to sphoṭa despite clear parallels with the Vaibhāṣika theories they refute. In contrast, the author of the Abhidharmadīpa, not only “refutes the Sautrāntika position” but also “makes pointed reference to the theories of (verbal) sounds held by the Mīmāṃsaka and Vaiśeṣika, and briefly examines the sphoṭa theory of the Grammarians” (Jaini 1959, 103). Kamalaśīla in passing also connects the notion of sphoṭa with the Vaibhāṣika theories at TSP ad TS 2713–14: vaibhāṣikā hi kecit padakāryābhidhānena vākyasphoṭam anityatvāj janyaṃ pratipannāḥ /. See also Biardeau (1964, 390–400). 7 See Chapter 4 in this volume, Biardeau (1964) and Eltschinger (2001, 2007). 8 Matilal (1990, 95) makes clear the close relationship between sphoṭa and śabdabrahman for Bhartṛhari in particular: “Bhartṛhari’s philosophy of language is ultimately grounded in a monistic and idealistic metaphysical theory. He speaks of a transcendental word-essence (śabdatattva) as the first principle of the universe. His sphoṭa doctrine is finally aligned with the ultimate reality called śabdabrahman. A self-realized person attains unity with the word-principle—a man of perfect knowledge. There is no thought without language, no knowledge with word in it. Consciousness vibrates through words, and such vibrating consciousness or a particular cognitive mode motivates us to act and obtain results. Hence language offers the substratum upon which human activity is based. Language and meaning are not two separate realities such that one conveys the other. They are in essence the two sides of the same coin. The sphoṭa is this unitary principle where the symbol and what is signified are one. To understand each other’s speech and to communicate, we do separate the inseparable, the sound and its sense. This is only instrumental to our mutual understanding. At the ultimate level, they are one.” See also Bronkhorst (1991, 14–15). 9 TS, 128: nāśotpādāsamālīḍhaṃ brahma śabdamayaṃ param / yat tasya pariṇāmo ’yaṃ bhāvagrāmaḥ pratīyate //. 10 anādinidhanaṃ brahma śabdatattvaṃ yad akṣaram / vivartate ’rthabhāvena prakriyā jagato yataḥ // VP, I.1. Translation in Matilal (1990, 125). Cited in TSP ad TS 128. 11 In this volume Saito (Chapter 4, section 5), notes that while later thinkers, including Somānanda, equate sphoṭa with the highest level of language (paśyantī) and therefore also with brahman, the highest level of reality, neither Bhartṛhari nor Maṇḍana claim that sphoṭa or the word is unchangingly eternal. We should consider, however, that for Patañjali sphoṭa is understood to be the essence or “own form” (svarūpa) of language (śabda). Given that it is not a very far leap from śabdasvarūpa to the śabdatattva we see in Bhartṛhari’s VP, I.1, it is not unreasonable to consider the possibility of an equation between sphoṭa and śabdabrahman, at least in the works of Bhartṛhari or in the minds of his Buddhist

128

12

13 14 15

16 17 18

19 20

21

22 23 24

25

Sara McClintock and Mīmāṃsaka critics. See also Eltschinger (2001, 248), who notes that Kamalaśīla frequently equates śabdātman with sphoṭa in the TSP. TSP, ad TS 128: pūrvāparādidigvibhāgarahitam anutpannam avināśi yac cchabdamayaṃ brahma […]. Giunta (2009, 143) notes parallel passages in Vṛsabhadeva’s Paddhati commentary on VP. On the authorlessness (apauruṣeyatva) of the Veda, see Chapter 2 in this volume. TS, 2421cd: nābhivyaktikramaś cāsti nityatve vyaktyayogataḥ // TSP, ad TS 2421cd: tathā hi na tāvad bījāṅkuralatādivat kālakṛtaḥ kramo   yujyate nityatvena sarveṣāṃ samakālatvāt / nāpi pipīlikādipaṅktivad deśakṛtaḥ /   vyāpitvena sarveṣām ekanabhodeśāvasthānāt / nāpy abhivyaktikṛtaḥ / anādheyātiśayatvena nityasya vyakter ayogāt // Cf. also TSP, ad TS 2761–2765, which states basically the same thing. For an analysis of this argument as articulated by Dharmakīrti in PV and PVSV, see Eltschinger (2007, 189–96). For the problem of sequence in the sphoṭa theory, see Saitō (2012). TS, 136–137: pratibhāvaṃ ca yady ekaḥ śabdātmābhinna iṣyate / sarveṣām ekadeśatvam ekākārā ca vid bhavet // prativyakti tu bhede ’sya brahmānekaṃ prasajyate / vibhinnānekabhāvātmarūpatvād vyaktibhedavat // TS, 140: athāpi kāryarūpeṇa śabdabrahmamayaṃ jagat / tathā ’pi nirvikāratvāt tato naiva kramodayaḥ // TSP, ad TS 140: evam api śabdasya nityatvenāvikāritvāt tataḥ krameṇa kāryodayo na prāpnoti / sarveṣām avikalāpratibaddhasāmarthyakāraṇād yugapad evotpādaḥ syāt / kāraṇavaikalyād dhi kāryāṇi parilambante / tac ced avikalaṃ / tat kim aparam apekṣeran / yena yugapan na bhaveyuḥ // VP, I.131: na so ’sti pratyayo loke yaḥ śabdānugamād ṛte / anuviddham iva jñānaṃ sarvaṃ śabdena varttate //. There is a slight variation in Kamalaśīla’s citation of the verse as compared with the version as it comes down to us in the VP itself and on which Matilal’s translation is based. In the stanza of the VP we read the final pāda as: sarvaṃ śabdena bhāsate / Translation in Matilal (1990, 127–28). TSP, ad TS 147–48: anyatra gatamānaso ’pi cakṣuṣā rūpam īkṣamāṇo ’nāviṣṭābhilāpam eva nīlādipratyayam anubhavatīti /. Cf. also PV, 3.175. This term is adapted from Eltschinger (2001, 249) “le sphoṭa, le véhicule si j’ose dire ‘trans-phonétique’ de la signification.” Ōmae (1999) succinctly demonstrates how despite their rather different understandings of the nature of phonemes and language in general, both Kumārila and Dharmakīrti understand meaning to be produced through the utterance of phonemes in a particular sequence and thus both can be considered varṇavādins. The same applies to Śāntarakṣita and Kamalaśīla, as Ōmae (1999, 299) also notes. TS, 2281–82 = ŚV, śabdanityatā 281–82: na ca kramād vinā varṇā nirjñātāḥ pratipādakāḥ / kramasyaivaṃ padatvaṃ ca tasmād evaṃ prasajyate //

A Buddhist Refutation of Sphoṭa

26

27 28 29

30

31

32

33

34

129

padaṃ varṇātiriktaṃ tu yeṣāṃ syāt kramavarjitam / teṣām evārthavaty eṣā śabdanityatvakalpanā //. Note that there are minor variations between TS, 2282 and ŚV, śabdanityatā 282, which reads as follows: na ca kramād vinā varṇā vijñātāḥ pratipādakāḥ / kramasyaiva padatvaṃ vas tasmād evaṃ prasajyate // This interpretation of ŚV, śabdanityatā 282, reflected both in Jha’s translation and in Sucarita Miśra’s Kāśikā on the verse, is made clear by Kamalaśīla as well in TSP, ad TS 2283: yeṣāṃ punas tu vaiyākaraṇānāṃ varṇavyatiriktaṃ sphoṭākhyam anavayavaṃ padaṃ vācakam itīṣṭam / teṣām arthavaty eṣā śabdaniyatvakalpanā // Ōmae (1999, 298). See also Eltschinger (2007, 204–12). Cf. PV, I.302–03; French translation in Eltschinger (2007, 205–06), n. 55 and n. 56. TS/P 2698–703ab: nityatāyāṃ tu sarveṣām arthāpattir apākṛtā / arthapratītihetutvam anityeṣu hi sādhitam // yo yadvivakṣāsambhūtavi vakṣāntaratas tataḥ / varṇa utpadyate tasya śrutis tansamanantaram // pūrvavarṇavidudbhūtasaṃvin nātidrutaśrutiḥ / so ’pekṣya tatsmṛtiṃ paścāt kurute smṛtim ātmani // tatsamutthāpakagrāhijñānāni prati janyatā / hetutā vānupūrvīyaṃ varṇeṣu puruṣāśrayā // ataḥ pratipadaṃ bhinnā varṇā iti parisphuṭam / damo mado latā tāla ityādikramabhedataḥ // īdṛśena krameṇaite tv arthabhedopapādakaḥ /. Unfortunately, this section of the TSP manuscripts is quite corrupt. See Giunta (2009), 171–72 for extensive notes and a critical reconstruction based in part on the Tibetan translations as well as the reconstructions in the edition of Dvarikadasa Shastri (1968). TS, 2703cd–04: ata eva nirartheha sphoṭasyāpi prakalpanā // sa hy arthapratipattyarthaṃ śābdikaiḥ parikalpitaḥ / varṇā eva ca tacchaktā ity anarthāsya kalpanā // TS, 2705: dṛśyasyādṛṣṭitaś cāsya nāstitādhyavasīyate / adṛśyatve tu naivāyaṃ liṅavaj jñāpako bhavet // Giunta (2009, 176) notes parallel passages in the Pramāṇa­ vārttikaṭīkā and the Pramāṇavārttikasvavṛttiṭīkā. TS, 2706–07: sattāmātreṇa tajjñānahetubhāvavyavasthiteḥ / tasya jñāpakateṣṭā cen netravat sarvadā bhavet // saṅketānavabodhe ’pi varṇānām aśrutāv api / tadbhāvyartheṣu vijñānaṃ śaktakāraṇasannidheḥ // Giunta (2009, 176) notes parallel passages in the Pramāṇavārttikaṭīkā and the Pramāṇavārttikasvavṛttiṭīkā. TS, 2708–09: tathā hi nityasattvo ’yaṃ na cāpekṣāsya kācana / dhvanisaṃketavarṇaiś ca tadvyaktir nāpy adarśanāt // jñānaṃ hi vyaktir ity āhus tajjñānaṃ na ca vidyate / tato nirarthakaivāsya vyañjakasyāpi kalpanā // The citation of the phrase jñānaṃ hi vyaktir appears to be a reference to Dharmakīrti’s Pramāṇavārttika. See Giunta (2009, 177), who cites PV, 3.440c: jñānaṃ vyaktir. Giunta (2009, 177) also notes parallel passages in the Pramāṇavārttikasvavṛttiṭīkā. TS 2710: nādair āhitabījāyām antyena dhvaninā saha / āvṛttaparipākāyāṃ buddhau śabdo ’vabhāsate // This verse is nearly identical to VP, I.86, which reads

130

35 36

37

38 39

40 41

Sara McClintock nādenāhitabījāyām antyena dhvaninā saha / āvṛttaparipākāyāṃ buddhau śabdo ’vadhāryate //. Kamalaśīla also cites this verse at TSP, ad TS 2311. TS, 2711: ity etad api tenātra nirnimittaṃ prakalpitam / tasyām api na śabdo ’nyo bhāsamāno hi lakṣyate // TS, 2712–13: janyatāṃ vyajyatāṃ vāpi dhvanibhiḥ kramabhāvibhiḥ / ye ’pi sphoṭasya manyante kramas teṣāṃ virudhyate // na hi krameṇa yujyete vyaktijātī niraṃśake / ekarūpābahir bhāvāt te syātāṃ sarvathaiva hi //. Giunta (2009, 178) notes a parallel passage with PVSV, 128, 21–25, which appears in a French translation in Eltschinger (2007, 327–28). TS, 2714: sāṃśatve ’pi yathā varṇāḥ krameṇāpratipādakāḥ / sphoṭāṃśā api kiṃ naivaṃ kiṃ adṛṣṭāḥ prakalpitāḥ // Reading krameṇāpratipādakāḥ with Giunta (2009), 178. Cf. also PV, 1.248–249 and Eltschinger’s translation (Eltschinger 2007, 323–24). The TS/P verse and commentary both draw heavily from Dharmakīrti’s Pramāṇavārttika. TSP, ad TS 2714: sati ca kalpite vācakatve varaṃ varṇabhāgā eva santu vācakāḥ / TS, 2715–2716: jātau vyaktau kṛtāyāṃ ced ekena dhvaninā sakṛt / nitarāṃ vyaktisiddhyarthaṃ varṇānānyān prayuñjate // yato duravadhārāsya prakṛtiḥ sā tathā kṛtā / samānaśaktikair varṇair bhūyo ’pi vyajyate paraiḥ // Cf. VP, I.84–85: yathānuvākaḥ śloko vā soḍhatvam upagacchati / āvṛttyā na tu sa granthaḥ pratyāvṛtti nirūpyate // pratyayair anupākhyeyair grahaṇānuguṇais tathā / dhvaniprakāśite śabde svarūpam avadhāryate //. Biardeau (1958, 37) translates these two verses, together with VP, I.86 (cited by Śāntarakṣita at TS 2710 and translated above), as follows: “De même que l’on arrrive à posséder (par cœur) une section védique ou un śloka par répétition, mais que le texte n’est pas bien élucidé par chacune des répétitions, de même, c’est par des connaissance indéfinissables mais qui sont ordonnées à son appréhension, que la forme propre d’une entité verbale mise en lumière par les sons est déterminée. Dans l’intellect, où les résonances ont déposé des germes et qui s’est mûri à chaque répétition, l’entité verbale est déterminé au moment du dernier son.” TS, 2717: tasyaivānyasya vaikasya kiṃ nāvṛttau punaḥ punaḥ / vyaktirāvartate tasya nanv evam aviśeṣataḥ // TSP, ad TS 2717 spells this out: tathā hi tadvyaktyāvartanamātraphalāny uttarottaravarṇotccāraṇāni samānaśaktikatvāt sarveṣāṃ / tac cāvartanam ekenaiva punaḥ punar āvarttyamānena kartuṃ śakyata iti śeṣavarṇoccāraṇavaiyarthyam / nāpy uttarottaravarṇānāṃ bhinnaśaktikatvam abhyupagantavyam / niraṃśake viśeṣāntarasyādhātum aśakyatvād viśeṣāntarakaraṇāsambhave bhinnaśaktikalpanāvaiyarthyaprasaṅgāt //: “That is, the articulation of the subsequent phonemes has the effect of just repeating the manifestation of that sphoṭa, since all the phonemes have the same capacity. And the repetition of that manifestation can be accomplished by just one phoneme

A Buddhist Refutation of Sphoṭa

42

43

44 45

46

47

48

131

being repeated again and again. Therefore, the articulation of the remaining phonemes is pointless. Nor can you postulate that the subsequent phonemes have diverse capacities, since one cannot add other qualities to that which has no parts. Therefore, since there can be no cause in the form of another quality, you would have the undesired consequence that conceptualizing diverse capacities is also pointless.” TSP, ad 2718: yadi hi varṇavyatirekeṇāparaḥ sphoṭākhyaḥ śabdātmāvabhāseta tato ’syābhivyaktiḥ sambhaved vyakter upalabdhirūpatvāt / yāvatā nāvabhāsata iti pūrvam āveditam // TS, 2719–24: tasmāt pratyakṣataḥ pūrvaṃ kramajñāneṣu yat param / samastavarṇavijñānaṃ tadarthajñānakāraṇam // antyavarṇe hi vijñāte sarvasaṃskārakāritam / smaraṇaṃ yaugapadyena sarvavarṇeṣu jāyate // sarveṣu caitadartheṣu mānasaṃ sarvavādinām / iṣṭaṃ samuccayajñānaṃ kramajñāteṣu satsv api // na cet tadabhyupeyeta kramadṛṣṭeṣu naiva hi / śatādirūpaṃ jāyeta   tat samuccayadarśanam // tena śrotramanobhyāṃ syāt kramād varṇeṣu yady api /   pūrvajñānaṃ parastāt tu yugapat smaraṇaṃ bhavet // tadārūḍhās tato varṇā na dūrārthāvabodhanāt / śabdād atha matis tena laukikair abhidhīyate //. These verses correspond to ŚV, sphoṭa 109, 112–116. See the notes in Giunta (2009, 180–82) for minor variations between the verses as they appear in TS and in ŚV. TS, 2725: ākāravati vijñāne sarvam etac ca yujyate / anyathā hi vinaṣṭās te bhāseran smaraṇe katham // See McClintock (2014) for an argument that Śāntarakṣita and Kamalaśīla adopt various positions on the nature and status of phenomenal content depending on the rhetorical context of a particular argument. In the context of an argument with Mīmāṃsakas, where objects external to the mind are provisionally accepted, these Buddhists cleave to the sākāravāda (doctrine that awareness is endowed with phenomenal content) position while Kumārila holds awareness to be nirākāra (not endowed with phenomenal content). TSP, ad TS 2725: tataś ca yad etat / antyavarṇe hi vijñāte ityādinā kumārilena sphoṭavādinaṃ prati samuccayajñānaṃ varṇitam / tad asman mata eva yujyate /   na tu bhavatāṃ mīmāṃsakānāṃ nirākāravādināṃ mata ity uktaṃ bhavati //. “And therefore the cumulative awareness that has been urged by Kumārila against the sphoṭavādin in the verse that begins ‘For when there is the cognition of the final phoneme […]’ [TS, 2720 = ŚV, sphoṭa 112] makes sense in terms of just our opinion. But it does not [make sense] in terms of your Mīmāṃsaka opinion that holds that [cognition] is without phenomenal content. That is the idea here.” TS, 2726–27: atha varṇās tirobhūtavyaktayo viditāḥ purā / smaryante ’vasthitā eva na spaṣṭābhaprasaṅgataḥ // apāstā ca sthitiḥ pūrvaṃ tat sthitau smaraṇaṃ bhavet / varṇānubhavavijñānakāla evaikahetutaḥ // Kamalaśīla supplies the location as the “investigation of the three times” (traikālyaparīkṣā), the twenty-first chapter of the TS/P.

132

Sara McClintock

49 TSP, ad TS 2728: atra śābdikāś codayanti / yady eko nāsti sphoṭākhyaḥ śabdātmā tat kathaṃ gaurityekākārā gośabde buddhir bhavatīti / “Here the Grammarians object: But if there is no single essence of language known as sphoṭa, then how is it that cognition has the single form of ‘cow’ when the word ‘cow’ [is uttered]?” 50 TS, 2728 = ŚV, sphoṭa 120: gaurityekamatitvaṃ tu naivāsmābhir nivāryate / tadgrāhyaikārthatābhyāṃ ca śabde syād ekatāmatiḥ // 51 TS, 2729 = ŚV, sphoṭa 121: śaighryād alpāntaratvāc ca gośabde sā bhaved api / devadattādiśabdeṣu spaṣṭo bhedaḥ pratīyate // 52 TS, 2730–2731 (TS, 2730 = ŚV, sphoṭa 135: varṇotthā cārthadhīreṣā tajjñānānantarodbhavāt / yedṛśī sā tadutthā hi dhūmāder iva vahnidhīḥ // na varṇa­ bhinnaśabdābhajñānānantarabhāvinī / arthadhīr vedyate tena nānyaḥ śabdo ’sti vācakaḥ //) 53 For a study of earlier Mīmāṃsaka theories of sentence meaning, see Taber (1989). 54 See McClintock (2010, 47–48) for Kamalaśīla’s use of this term at TSP ad TS 1–6. For Śāntarakṣita and Kamalaśīla’s approach to word and sentence meaning, see Siderits (1985). 55 TS, 1027: pratibimbātmako ’pohaḥ padād apy upajāyate / pratibhākhyo jhaṭity eva padārtho ’py ayam eva naḥ //. See Hattori (1980, 64) for Dignāga’s debt to Bhartṛhari’s theory of pratibhā. Hattori (1980, 69) also notes that Śāntarakṣita’s revision of Dignāga’s theory of apoha to include the positive entity that is the reflection (pratibimba) in awareness functions to rebut Kumārila’s objections to the Buddhist apoha theory. 56 TSP, ad TS 1027: yasmāt padādapi pratibimbātmako ’poha utpadyata eva tenāsmākam ayam eva pratibimbātmako ’pohaḥ padārtho ’pi mato na kevalaṃ vākyārtha ity apiśabdaḥ //. Reading vākyārtha instead of bāhyārtha in accord with the Tibetan. 57 TS, 1089: na vācyaṃ vācakaṃ vāpi paramārthena kiñcana / kṣaṇabhaṅgiṣu bhāveṣu vyāpakatvaviyogataḥ // See also TS 1032–33. 58 See Chapter 4 in this volume.

Primary Sources PV

Pramāṇavārttika of Dharmakīrti. In PV-PVSV.

PV-PVSV Gnoli, Raniero, ed. 1960. The Pramāṇavārttikam of Dharmakīrti. The First Chapter with the Autocommentary. Serie Orientale Roma: 23. Roma: Istituto Italiano per il Medio ed Estremo Oriente. PVSV

Pramāṇavārttikasvavṛtti of Dharmakīrti. In PV-PVSV.

ŚV

Rai, Ganga Sagar, ed. 1993. Ślokavārttika of Śrī Kumārila Bhaṭta with the Commentary Nyāyaratnākara of Śrī Pārthasārathi Miśra. 2nd edition. Varanasi: Ratna Publications.

A Buddhist Refutation of Sphoṭa

133

TS

Tattvasaṃgraha of Ācārya Śāntarakṣita. In TS-TSP.

TS-TSP

Dvarikadasa Shastri, ed. 1968. Tattvasaṃgraha of Ācārya Śāntarakṣita with the Commentary Pañjikā of Śrī Kamalasīla. Bauddha Bharati Series, 1–2. Varanasi: Bauddha Bharati (N.B.: in this chapter the Sanskrit text found in TS and TSP was emended following Giunta 2009, listed below).

TSP

Tattvasaṃgrahapañjikā of Śrī Kamalasīla. In TS-TSP.

VP

Rau, Wilhelm, ed. 1977. Bhartṛharis Vākyapadīya, die mūlakārikās nach den Handschriften herausgegeben und mit einem pāda-Index versehen. Wiesbaden: Franz Steiner Verlag.

Secondary Sources Biardeau, Madeleine. 1958. Sphoṭa Siddhi (La démonstration du Sphoṭa) par Maṇḍana Miśra. Pondichéry: Institut français d’indologie. Biardeau, Madeleine. 1964. Théorie de la connaissance et philosophie de la parole dans le brahmanisme classique. Paris: Mouton. Bronkhorst, Johannes. 1991. “Bhartṛhari on Sphoṭa and Universals.” Asiatische Studien/Études Asiatiques 45: 15–18. Eltschinger, Vincent. 2001. “Dharmakīrti: Critique de la théorie du sphoṭa.” Asiatische Studien/Études Asiatiques 55: 243–89. Eltschinger, Vincent. 2007. Penser l’autorité des Écritures: La polémique de Dharmakīrti contra la notion brahmanique orthodoxe d’un Veda sans auteur. Wien: Verlag der Österreichischen Akademie der Wissenschaften. Giunta, Paolo. 2009. “Il buddhismo e la tradizione filosofico-grammaticale brahmanica. Il rapporto di Śāntarakṣita con Bhartṛhari alla luce delle citazioni del Vākyapadīya nel Tattvasaṅgraha e nel suo commento.” PhD thesis, “Sapienza” University of Rome. Hattori, Masaaki. 1980. “Apoha and Pratibhā.” In Sanskrit and Indian Studies: Essays in Honour of Daniel H.H. Ingalls, ed. by Masatoshi Nagatomi and et al., 61–73. Dordrecht/Boston: D. Reidel/Hingham, MA. Jaini, Padmanabh S. 1959. “The Vaibhāṣika Theory of Words and Meanings.” Bulletin of the School of Oriental and African Studies 22 (1): 95–107. Matilal, Bimal K. 1990. The Word and the World: India’s Contribution to the Study of Language. Delhi: Oxford University Press. McClintock, Sara. 2010. Omniscience and the Rhetoric of Reason: Śāntarakṣita and Kamalaśīla on Rationality, Argumentation, and Religious Authority. Boston: Wisdom Publications. McClintock, Sara. 2014. “Kamalaśīla on the Nature of Phenomenal Content (ākāra) in Cognition: A Close Reading of TSP ad TS 3626 and Related Passages.” Journal of Indian Philosophy 42 (2–3): 327–37.

134

Sara McClintock

Ōmae, Futoshi. 1999. “Dharmakīrti as a Varṇavādin.” In Dharmakīrti’s Thought and Its Impact on Indian and Tibetan Philosophy. Proceedings of the Third International Dharmakīrti Conference. Hiroshima, November 4–6, 1997, ed. by Shōryū Katsura, 295–300. Wien: Verlag der Österreichischen Akademie der Wissenschaften. Saitō, Akane. 2012. “The Notion of ‘Sequence’ as Conceived of in the Sphoṭa Theory.” Journal of Indian and Buddhist Studies (Indogaku Bukkyogaku Kenkyu) 60 (3): 1174–78. Siderits, Mark. 1985. “Word Meaning, Sentence Meaning, and Apoha.” Journal of Indian Philosophy 13 (2): 133–51. Taber, John A. 1989. “The Theory of the Sentence in Pūrva Mīmāṃsā and Western Philosophy.” Journal of Indian Philosophy 17 (4): 407–30. Verpoorten, Jean-Marie. 1994. “The 24th Chapter of the Tattvasaṃgraha: Refutation of the Mīmāṃsā Doctrine of Vedāpauruṣeyatva.” In Studies in Mīmāṃsā. Dr. Mandan Mishra Felicitation Volume, ed. by Ramchandra Dwivedi, 117–129. Delhi: Motilal Banarsidass.

6

The Place of Language in the Philosophy of the Recognition Marco Ferrante

1  The School of the Recognition The expression Pratyabhijñā (“Recognition”) refers to the doctrines of a number of scholars and spiritual teachers who lived and taught in Kashmir between the tenth and eleventh century CE. This philosophical tradition prospered in the religious framework of tantric Śaivism, and its primary purpose, already evident in its name, was to recognize the presence of Śiva at all levels of reality.1 The three main theoreticians of the school—Somānanda (900–950), Utpaladeva (925–975), and Abhinavagupta (975–1025)—established and refined the main tenets of the tradition in approximately one hundred years’ time.2 Somānanda, the founder of the school, emerged as a leading thinker and guru in the heterogeneous setting of tenth-century Kashmir, and his main work, the Śivadṛṣṭi, was probably addressed to an audience of tantric initiates.3 His pupil Utpaladeva broadened the perspective of the tradition and was responsible for much if not all its theoretical and argumentative elaboration. His major work, the Īśvarapratyabhijñākārikā (ĪPK),4 put the Pratyabhijñā in the wider arena of Indian philosophical debate, in active engagement with beliefs and arguments of other traditions. Finally, Abhinavagupta closed the most creative phase of the school by commenting on Utpaladeva’s work twice.5 He possibly was the most eclectic of the three authors, being able to combine the religious concern of Somānanda with the argumentative rigor of Utpaladeva. His final aim was the creation of a system of knowledge in which linguistic, theological, and aesthetic doctrines could coexist with the ritual practices described in tantric scriptures. The importance of language to the Pratyabhijñā’s theoretical construction is hard to overstate. In the ideology and rhetoric of tantric traditions sound is

136

Marco Ferrante

regarded as the foremost manifestation of the divine and as a means to access, control, and manipulate the energy of the deity. Still, unlike other tantric systems, Pratyabhijñā’s philosophical inclination required a more argumentative analysis of the problem. This resulted in the discussion of topics that were relevant to the broader inter-sectarian debate, such as the relationship between language and consciousness, the epistemological status of conceptual and nonconceptual cognitions, and the problem of sentence meaning. This chapter is precisely meant to explain how the Pratyabhijñā’s authors touched on these philosophical questions, without addressing the greater issue of the symbolic value they placed on language from religious, ideological, or ritual points of view.6 Nevertheless, readers should keep in mind that even the most argumentative passages of these philosophers reflect a worldview that is affected at all levels by what one may call “religious concerns”: in the end, as the name itself implies and Utpaladeva explicitly states,7 the Pratyabhijñā is first and foremost a path to salvation, although its promoters did their best to base its doctrines on firm rational grounds.

2  Pratyabhijñā’s Metaphysics and Epistemology 2.1 Metaphysics To appreciate the role language plays in the discussion of these thinkers, we must briefly outline their metaphysical and epistemological views. The Pratyabhijñā conceives the universe as a unitary manifestation of the god Śiva, thus arguing for a strict nondualistic ontology. The godhead informs and shapes all reality by means of his energy, which is identified with his female consort (śakti). The most innovative move of the tradition consisted in identifying Śiva with the individual self, thus espousing a version of nondualism in which consciousness is equated not only with the godhead but, crucially, also with the latter’s dynamic energy. The outcome is a one-of-a-kind nondualism that is rather different from the one advocated in the Upaniṣads or in the ensuing hermeneutics of the Advaita Vedānta. The Śaivas did not posit a distinction between ordinary and absolute reality, but rather regarded all things as manifestations of a unitary consciousness, which is free to incessantly contract and expand. In terms of salvific path, the practitioner aims to reach the highest level of consciousnessexpansion, where she will finally recognize both the presence of the divine in all things and the identity between the godhead and herself.

The Place of Language in the Philosophy of the Recognition

137

Putting such a stress on consciousness posed a serious theoretical challenge to the Pratyabhijñā, a question that became central to Utpaladeva and which set the tone for both his ĪPK and Abhinavagupta’s commentaries thereon. The issue at stake, simply put, is whether a consciousness, a self, is existing at all. The emphasis on the question is significant for it reveals how deep the impact of Buddhist notions was—especially those advanced by the so-called logicalepistemological school of Dignāga and Dharmakīrti (sixth to seventh century CE)—on the theoretical development of the Pratyabhijñā. Crucial among these ideas was the rejection of the idea of self (nairātmyavāda). The Pratyabhijñā’s thinkers directed their most strenuous efforts to counter this position, but in the process they also carefully considered and eventually dismissed several other beliefs the Buddhists had about knowledge, agency, language, and soteriology. Interestingly enough, Buddhist conceptions were not always flatly rejected. On occasion, the Pratyabhijñā’s authors integrated some in their system, with the caveat that they may be correct only if one is also assuming the real existence of the notion of personhood.

2.2 Epistemology Much of the epistemological doctrine of the Pratyabhijñā is advanced in the first chapter of Utpaladeva’s ĪPK. As anticipated above, the Śaivas need to establish beyond any reasonable doubt the existence of a self, more specifically the existence of a permanent component of the psychophysical makeup that is able to cognize and act. The testing ground for this thesis is an analysis of memory; the reason is easy to see: memory being a cognitive event in which a present cognition is directed at a past one, it seems reasonable to explain it by resorting to a permanent knower who is the subject of both the original cognition and the later recollection. Utpaladeva examines the question by adopting the usual pūrvapakṣa/siddhānta scheme, by first presenting the Buddhist account of memory, then showing its flaws, and finally putting forward his own definitive position. In short, the Buddhists conceive memory as a series of momentary cognitions and take a stand against the idea of a permanent knower. They claim that knowledge is a singular event that is produced, exists, and ceases instantaneously. Still, crucially, a cognition is also able to produce a trace of itself (saṃskāra) that is transferred to the successive ones up to the point in time in which recollection occurs. This means that for the Buddhists, memory is nothing more than a reification of a chain of instantaneous knowledge

138

Marco Ferrante

events that gain temporal extension only due to the mental traces they are able to generate: instantaneous cognitions and mental traces are therefore sufficient to bridge the gap between the original cognition and the later recollection. Utpaladeva, for his part, refuses this picture by appealing to one of the most typical arguments of his school: conceiving memory as a series of cognitions glued together by mental traces would imply that cognitions are the object of one another.8 But then one would treat a cognition precisely as a content, thus negating its most defining quality: knowledge, the Śaivas argue, can never be objectified, otherwise it would simply cease to be what it is. The conclusion is that the only way to explain memory is to recognize the role of a knowing self that is permanent and unchanging. In this view, recollection is explained as the capacity of the subject to restage in the present an earlier knowledge-event as it happened in the past, and the original perception is therefore conceived as being, at least partially, identical to its recollection. As Abhinavagupta explains, all cognitions have two aspects. There is a “content-aspect” (arthāṃśa), which is self-confined, always restricted to the past, and absolutely inaccessible to any subsequent cognition. Once this aspect has occurred and gone, it is irretrievable. If cognitions were made up of this aspect only, memory would be impossible. But there is another aspect, the “consciousness-one” (svātmāṃśa), that is not affected by time and is perpetual; this is the one that guarantees the continuity and, using a crucial term of the school, the dynamic unification (anusandhāna or pratisandhāna) of all cognitions: it corresponds to the reflexive awareness of the knowing subject (vimarśa, pratyavamarśa) and is, in the end, the hallmark of self and consciousness. One must be careful to note the difference from the Buddhist position: vimarśa does not have the former cognition as its object. It is the former cognition, precisely its svātma part. Therefore, any cognition can be regarded as an extroversion of the individual consciousness, which cognizes contents that are only partially different from itself. This epistemological position has obvious ontological consequences: if cognitions and contents share the same nature, the former are not different from the latter. This opens an avenue for a strong ontological nondualism where everything is seen as the manifestation of a unitary principle that, depending on the context, is alternatively called vimarśa, consciousness or self. On a religious level, the individual consciousness is then identified with the supreme form of the godhead, Śiva Maheśvara.

The Place of Language in the Philosophy of the Recognition

139

3  Language and Consciousness In addition to all this, the Pratyabhijñā authors stress another quality of reflexive awareness (vimarśa), indeed its most defining characteristic. They argue that consciousness is unconceivable without language. The main arguments in support of the claim are contained in two specific stanzas. In 1.5.13, Utpaladeva sets the tone of the argument by stating: The essential nature of consciousness is reflexive awareness, it is the supreme word that arises freely. It is primarily freedom, it is the sovereignty of the supreme self.9

In the Vimarśinī (ĪPV), Abhinavagupta comments the stanza by observing that the ultimate nature of consciousness consists in an act of self-savoring of the self. Crucially, what makes an individual conscious is the fact that she has a subjective feeling of herself as the owner of a certain cognition, that is, of a certain mental state. The word citi, derived from cetayati, “to make conscious,” indicates the activity of consciousness, whose essential nature, its essence, is a reflexive awareness characterized by self-savoring. To explain: a pot does not have savoring with respect to itself; it does not have a reflective awareness as a self, it does not cognize with respect to itself, nor it shines as having an uninterrupted nature. This is why it is said to be unconscious. On the other hand, a person named Caitra has savoring with respect to himself, for he has the power to produce an effort towards a raised state, that is the “I”; he has reflexive awareness as a self and cognizes precisely with respect to himself.10

Abhinavagupta then adds that consciousness’s ultimate nature consists in the activity of an internal speech, which is different from the public one and is also independent from linguistic conventions. He calls it śabdana, a peculiar term I translate as “languageing,” and by which I seek to convey the idea that in order to have a coherent organization of cognitions one necessarily needs a syntactical arrangement of mental states: Furthermore, reflexive awareness’s essential nature is “languageing,” consisting in inner speech. This “languageing” is independent from any linguistic convention and is an uninterrupted spontaneous savoring. It is like an internal nodding of the head; it is the life of the conventional sounds existing at the Māyā level, such as “a” etc., because it is the internal support of a reflexive awareness

140

Marco Ferrante

such as “this is blue” or “I am Caitra” and so on. Since speech denotes, conveys, all things through reflexive awareness, consciousness is said to be the “word.”11

Still, what we have said so far does not provide substantial evidence in support of the claim that consciousness involves language. To find a more solid argument we must refer to a second stanza, ĪPK, 1.5.19, where Utpaladeva and Abhinavagupta go straight to the heart of the question, by arguing that language informs all kinds of cognitions, including perceptual awareness.12 The thesis is put forward by Utpaladeva as follows: There is self-awareness even in the case of immediate cognitions, otherwise how can actions like running etc. be accomplished if they lack dynamic unification?13

The main argument in support of this affirmation is hinted at in the rhetorical question Utpaladeva formulates in the second part of the verse. Without reflexive awareness, which always implies the presence of concepts and language, it would be impossible to carry out any action, for actions always necessitate a synthesis of cognitions, their dynamic unification. Abhinavagupta expands on the stanza by backing the concept with two arguments. The first concerns language acquisition. Without a cognitive synthesis, he claims, a child would never be able to put together the nonconceptual cognitions he experiences in a first moment. To begin with, in this world, the essential characteristic of consciousness is the association with language, in a way that is comparable to the indication of something with a finger etc. If it were not so, a child would not learn a linguistic usage when he sees a human exchange for the first time. In fact, a child first hears the word in the form of a series of non-conceptual cognitions, then she sees the external referent of the word before his eyes, and thereafter a surface in which the external referent is no longer present. On hearing linguistic expressions such as “bring the pot” or “carry the pot,” how can the idea “this is the meaning of that,” namely this is the meaning of the word “pot,” this is the meaning of the word “bring” and of the word “carry” appear in the heart of the child ? The answer is that to say “this object is the meaning of that word” is essentially an act of unification, and unification is a conceptual activity.14

The second justification, more explicitly connected to Utpaladeva’s verse, concerns the impossibility to carry out quick actions without a cognitive synthesis, that is, without vimarśa. One can explain the performance of these

The Place of Language in the Philosophy of the Recognition

141

fast actions only by positing the presence of a subtle verbal potentiality that is present in perceptual awareness too: Even admitting that an immediate cognition is by nature momentary, in that too there is reflexive awareness. This is necessary. How could it be otherwise? In fact, if it were not so [i.e., if there was no reflexive awareness in the case of immediate cognitions], then a person proceeding rapidly with a specific purpose, or reciting letters quickly, or reading fast a book of mantras, would not reach, pronounce or recite what she has in mind. To explain: how can a person reach the place she has in mind without a reflexive awareness, whose essential aspect are unification and separation, and consists of the knowledge of the place, the will to proceed, the actual proceeding, the awareness of the movement, the unification of another purpose, the desire to leave, the knowledge of a different place, the desire to proceed towards it and so forth? The same applies to reading quickly, speaking etc. where, specifically, there is a union between places and organs of articulation etc. Since here one does not experience a gross conceptual cognition as following the immediate one, there is quickness. Therefore, there must be a subtle reflexive-awareness consisting of verbal potentiality in a contracted form, because one gets a gross conceptual cognition when a verbal potentiality becomes manifest by expansion.15

Although presented in a non-systematic way, Pratyabhijñā’s arguments on the necessary relationship that obtains between knowledge and language can be recast more clearly as follows: 1. The first thing to notice is the stress on action. What emerges from the quoted passages is that in order to be reliable, a cognition must have a causal efficiency (arthakriyā).16 In other words, a piece of knowledge is veridical only if one is able to initiate an action on its basis. 2. In order to be able to initiate an action, a cognition must be conceptualized. This means that direct perceptions, that is, first-order, world-directed mental states, must be the content of higher order mental states, that is, metacognitions. Metacognitions are conceptual.17 The problem with this account is that it is apparently at odds with the key principle we discussed in Section 2, the one whereby cognitions, that is, first-order mental states, are self-revealing and never the object of another. A solution to this conundrum is to entertain the possibility that first-order cognitions have two sides, a perceptual and a conceptual one, occurring at the same time. This entails that no cognition is actually the content of another.18

142

Marco Ferrante

3. High-order, conceptual cognitions are language based. There is no specific evidence in support of this affirmation in the texts but one can relate this stance to an assumption shared by most Indian philosophers according to which conceptualization always involves language. 4. The conclusion is that knowledge and consciousness imply language.

4  The Influence of Bhartṛhari and Pratyabhijñā’s Innovations As I have shown in detail elsewhere, the scheme just sketched is drawn from the work of the philosopher and grammarian Bhartṛhari (fifth c. CE), without any substantial modification.19 Bhartṛhari’s discussion of the issue takes place around a well-known stanza of his Vākyapadīya (VP), the one in which the author claims that language pervades all knowledge.20 In regard to this, Bhartṛhari holds that the criterion of veridical cognition is causal efficiency.21 He maintains that there is no knowledge without a cognitive synthesis that binds together nonconscious cognitions, and that this higher order cognitive synthesis is conceptual and linguistic.22 Utpaladeva and Abhinavagupta knew the VP well, and almost all the stanzas that corroborate this basic scheme are quoted in the Pratyabhijñā literature.23 There is nonetheless an aspect that is absent in Bhartṛhari’s formulation and that probably represents Pratyabhijñā’s most original contribution to the problem of consciousness. The innovation revolves around the notion that we have encountered above whereby “to be conscious” implies a cognitive state these authors indicate with the term camatkāra, literally “wonder” or “savoring.” It is not my intention to embark upon a detailed discussion of a complex and surely polysemic term.24 I will then limit myself to suggest that in talking about camatkāra, Utpaladeva and Abhinavagupta are probably referring to the subjective, private dimension of consciousness. They have in mind what contemporary philosophers of mind call qualia, that is, the “what it is like to be” feeling one has when is in a given mental state.25 According to some contemporary theorists, this aspect constitutes the hard problem of consciousness, that which resists the attempts of a materialistic explanation.26 If that is the case, it is not surprising that the Pratyabhijñā authors resorted to metaphorical usages to pinpoint an elusive dimension of consciousness that is almost by definition incommunicable to others.

The Place of Language in the Philosophy of the Recognition

143

5  Phonemic Consciousness The general outlook we have investigated so far has a bearing on the widely debated question of the relationship between a linguistic expression and its meaning. Pratyabhijñā’s ideas on the issue are especially conspicuous in Abhinavagupta’s works. Differently from one may expect, the Śaivas do not follow the sphoṭa theory promoted by Bhartṛhari and endorsed by the ensuing Vyākaraṇa tradition. The position of the school is actually closer to the Mīmāṃsā view that the meaning of a word is due to the phonemes (varṇas). As Abhinavagupta puts it in the Parātriṃśikāvivaraṇa, “the expressive power of a linguistic expression, which is one with its content, belongs to the phonemes only.”27 Nevertheless, one must be aware that the Śaivas’ understanding of phonemes is quite different from that of the Mīmāṃsā. Abhinavagupta sees the varṇas as the acoustic manifestation of consciousness and the building blocks of all sounds. In accordance with the general position of the school, he regards them not as mere contents of knowledge, rather as consciousness itself unfolding. As we saw in a previous passage of the Vimarśinī, the relationship between a word and a meaning is explained in terms of convention. The point is that this convention must be established by something or someone in order to avoid infinite regress. If the Nyāya’s solution is to postulate the existence of a god who established the convention, the Pratyabhijñā’s thinkers believe that the common ground of all conventional sound is a “phonemic consciousness” (varṇasaṃvit) that is inherently linguistic. Although all these considerations are clearly influenced by the theological elaborations on sound so frequent in tantric literature, the picture that one eventually gets is no doubt philosophically original. The Pratyabhijñā is in fact distancing itself from all other traditions. In its view, the relationship between a linguistic expression and its meaning is not nitya, permanent, as the Mīmāṃsā and the Vyākaraṇa would argue; it is neither purely conventional as the Nyāya claims, nor does the school maintain that larger linguistic elements render meaningless the smaller ones, as the sphoṭavādins would claim. Although Pratyabhijñā’s “phonemic consciousness” is highly indebted to Bhartṛhari’s conceptions,28 Abhinavagupta remains loyal to the basic premise of his school, according to which consciousness embraces and molds all reality without gaps. Accordingly, Bhartṛhari’s belief that phonemes and words are unreal means to reach the only real meaning, that of the sentence, is eventually to be discarded.

144

Marco Ferrante

Notes 1 The school is perhaps the most sophisticated outcome of that composite movement of systematization and elaboration of tantric material that emerged at the end of the first millennium CE and which is sometimes labeled “post-scriptural exegesis.” Pratyabhijñā’s staunch nondualism is rooted in scriptural sources that were held as authoritative in contemporary Kashmir and which foregrounded the dynamic and unifying role of the feminine power of the god, the śakti, like the Siddhayogeśvarī matatantra, the Mālinīvijayottaratantra, and the Vijñānabhairava. 2 For the dates of these authors see Sanderson (2007, 441). 3 On Somānanda’s cultural milieu see Nemec (2011, 20). 4 Utpaladeva himself wrote two commentaries on this work, the short Vṛtti and the longer Vivṛti, the latter largely lost. 5 A first time in a work dedicated to the stanzas (Īśvarapratyabhijñāvimarśinī), then in another devoted to the Vivṛti (Īśvarapratyabhijñāvivṛtivimarśinī). 6 For an overview of these problems and a broader discussion of the role of language in tantric traditions, see Padoux (1990). 7 See ĪPK, 4.16. 8 For a discussion of this point and Bhartṛhari’s influence on the Pratyabhijñā’s elaboration see Ferrante (2017). 9 citiḥ pratyavamarśātmā parā vāk svarasoditā / svātantryam etan mukhyaṃ tadaiśvaryaṃ paramātmanaḥ // ĪPK, 1.5.13. 10 cetayati ity atra yā citiḥ citikriyā tasyāḥ pratyavamarśaḥ svātmacamatkāralakṣaṇa ātmā svabhāvaḥ. tathā hi ghaṭena svātmani na camatkriyate, svātmā na parāmṛśyate, nā svātmani tena prakāśyate, na aparicchinnatayā bhāsyate, tato na cetyata iti ucyate. caitreṇa tu svātmani aham iti saṃrambhodyogollāsavibhūtiyogāt camatkriyate, svātmā parāmṛśyate, svātmany eva prakāśyate. 11 pratyavamarśaś ca āntarābhilāpātmakaśabdanasvabhāvaḥ, tac ca śabdanaṃ saṃketanirapekṣam eva avicchinnacamatkārātmakam antarmukhaśironirdeśaprakhyam akārādimāyīyasāṃketikaśabdajīvitabhūtam. nīlam idaṃ caitro ’haṃ ityādipratyavamarśāntarabhittibhūtatvāt, pūrṇatvāt parā, vakti viśvam abhilapati pratyavamarśena iti ca vāk. 12 Indian thinkers normally believe that language is involved in conceptual types of knowledge, such as inference or verbal testimony. The main difficulty for the Pratyabhijñā authors is to prove that conceptualization is at work also in perceptual knowledge. Abhinavagupta raises the question very straightforwardly: nanu pratyavamarśātmatvaṃ citiśakteḥ saṃkalpasmaraṇādiśaktiṣu savikalpātmikāsu bhavatu. yā tu nirvikalparūpā sākṣātkaraṇalakṣanā anubhavaśaktiḥ, tatra katham. pratyavamarśo hi abhilāpaviśeṣayojanāmayaḥ, abhilāpaviśeṣayojanā ca saṃketasmaraṇam apekṣate. tat ca saṃskāraprabodham. so ’pi tādrśadṛśam, ity

The Place of Language in the Philosophy of the Recognition

13 14

15

16 17 18

145

evaṃ prathamasamaye katham abhilāpayogaḥ. “One might object: let us admit that consciousness intrinsically consists in self-awareness in the case of those capacities of the self such as will, memory etc., which are conceptual. But how can self-awareness be present in the capacity that we call perception, which is defined by immediacy and has a non-conceptual nature? In fact, reflexive awareness consists in the usage of a specific linguistic expression that requires the recollection of a linguistic convention. This recollection is based on the reawakening of mental traces, and these, in turn, on similar previous cognitions. So, how can there be linguistic usage at the first moment of cognition?” Īśvarapratyabhijñāvimarśinī, on ĪPK 1.5.18, vol. 1, pp. 283–84. The objection probably refers to Dharmakīrti’s Pramāṇavārttika, 3.124–25 and 3.174–91ab. See Torella (2001, 868). sāksātkārakṣaṇe ’py asti vimarśaḥ katham anyathā / dhāvanādy upapadyeta pratisaṃdhānavarjitam // ĪPK, 1.5.19. iha tāvat caitanasya ātmabhūto ’ṅgulinirdeśādiprakhyo ‘bhilāpayogaḥ, anyathā bālasya prathamaṃ vyavahāre dṛśyamāne vyutpattir eva na syāt. nirvikalpakajñānaparaṃparayā hi taṃ śabdaṃ śṛṇoti, tataḥ tam arthaṃ puraḥ paśyati, punas tadviviktaṃ bhūtalaṃ paśyati iti ghaṭam ānaya naya iti vyavahārāt katham asya ayam artha iti hṛdi parisphureta, idam ghaṭa iti, idam ānaya iti, idaṃ naya iti, itiyojanāprāṇo hi ayam arthaḥ yojanā ca vikalpavyāpāraḥ. ĪPK, 1.5.19, vol. 1, pp. 284–86. bhavatu vā kṣaṇamātrasvabhāva sākṣātkāraḥ, tatrāpi asti vimarśaḥ. avaśyam caitat. ‘katham anyathā’ iti. yadi sa na syāt tat ekābhisaṃdhānena javāt gacchan, tvaritaṃ varṇān paṭhan, drutaṃ ca mantrapustakaṃ vācayan, na abhimatam eva gacchet, uccārayet vā, vācayet vā. tathā hi tasmin deśe jñāna​m-āci​krami​ṣā-āk​ramaṇ​am-āk​rānta​tājñā​nampr​ayoja​nānta​rānus​ aṃdhā​nam-t​ityak​ṣā-de​śānta​rānus​aṃdhi​ḥ, tatrāpi ācikramiṣā ity ādinā yojanāviyojanarūpeṇa pratyavamarśena vinābhimatadeśāvāptiḥ kathaṃ bhavet. evaṃ tvaritodgrahaṇavācanādau mantavyam. tatra viśeṣataḥ sthānakaraṇākramaṇādiyogaḥ. atra ca yataḥ paścāt bhāvisthūlavikalpakalpanā na saṃvedyate, tata eva tvaritatvaṃ iti sūkṣmeṇa pratyavamarśena saṃvartitaśabdabhāvanāmayena bhāvyam eva. saṃvartitā hi śabdabhāvanā prasāraṇena vivartyamānā sthūlo vikalpaḥ. Īśvarapratyabhijñāvimarśinī, on ĪPK 1.5.19, vol. 1, pp. 290–93. The concept of “causal efficiency” is attested in the Pratyabhijñā’s literature. See, for instance, ĪPK, 2.3.12. For an overview of what contemporary philosophers of mind call “Higher-Order Theories of Consciousness” see Carruthers (2016). The picture is quite similar to Dharmakīrti’s position, according to which perceptions, in his view the only type of veridical knowledge, generate a

146

19

20

21

22

23

Marco Ferrante judgmental, conceptual counterpart (niścaya, ekapratyavamarśa) at the very moment in which they occur. See Ferrante 2019. Affinities between Bhartṛhari and the Kashmiri school are also discussed, with different degree of emphasis, in Rastogi (2009), Ratié (2011), Torella (2008) and Vergiani (2016). VP, 1.131: na so ’sti pratyayo loke yaḥ śabdānugamād ṛte / anuviddham iva jñānaṃ sarvaṃ śabdena bhāsate //. “In the world there is no cognition without the pervasion of language. All knowledge shines as if pierced by language.” VP, 1.129: iti kartavyatā loke sarvā śabdavyapāśrayā / yāṃ pūrvāhitasaṃskāro bālo ’pi pratipadyate //. “In the world all that is required to be done is dependent on language. A child too understands it on the basis of the traces of previous births.” See Vākyapadīya-Vṛtti on 1.131: yathāsya saṃhṛtarūpā śabdabhāvanā tathā jñeyeṣv artheṣūtpannenāpy avikalpena jñānena kāryaṃ na kriyate. “Just like one has a verbal potentiality in a contracted form, similarly no effect is produced by indeterminate knowledge, even if it arises in relation with objects that are [already] known.” And finally VP, 1.135: arthakriyāsu vāk sarvān samīhayati dehinaḥ / tadutkrāntau visaṃjño ’yaṃ dṛśyate kāṣṭhakuḍyavat //. “Language urges all creatures towards successful activities. If it were to disappear, all this world would appear as unconscious as a piece of wood.” Vṛtti on VP, 1.132: vāgrūpatāyāṃ cāsatyām utpanno ’pi prakāśaḥ pararūpam anaṅgīkurvan prakāśanakriyāsādhanarūpatāyāṃ na vyavatiṣṭhate. bhinnarūpāṇāṃ cānupakāriṇām ātmāntarānātmanām itaretarasya vastumātrājñāne pratyavabhāsamāne yad uttarakālam anusaṃdhānaṃ pratyavamarśa ekārthakāritvam avibhāgena śaktisaṃsargayogopagrahaḥ tad vāgrūpatāyāṃ baddham. “If knowledge were not by nature linguistic, that very vague cognition that has been produced, being unable to contribute to another cognition, would not function as a means to accomplish the action of knowing. In fact, once there is a cognition of a bare object, whose different parts are not functional to each other and are mutually independent, immediately after comes a unification, a synthesis, the production of a unitary content, that is, the grasping of conjoined capacities as undifferentiated; all this is connected to the linguistic nature of knowledge.” Somānanda was very critical of Bhartṛhari, especially of his well-known differentiation of language into three levels. Still, Utpaladeva and Abhinavagupta realized the importance of this thinker for their own theory and deliberately adopted his most innovative ideas and often quoted the VP. VP, 1.131 is partially quoted by Somānanda in Śivadṛṣṭi, 2.9 and by Abhinavagupta in both the Īśvarapratyabhijñāvimarśinī (Vol. 2, p. 265) and the Īśvarapratyabhijñāvivṛtivi­ marśinī (Vol. 2, p. 206). VP, 1.132 is quoted by Utpaladeva in his commentary on Śivadṛṣṭi, 2.2 and by Abhinavagupta in both the Īśvarapratyabhijñāvimarśinī (Vol. 2, p. 265) and the Īśvarapratyabhijñāvivṛtivimarśinī (Vol. 2, p. 206). VP,

The Place of Language in the Philosophy of the Recognition

24 25 26 27 28

147

1.134 is quoted by Abhinavagupta in both the Īśvarapratyabhijñāvimarśinī (Vol. 2, p. 265) and the Īśvarapratyabhijñāvivṛtivimarśinī (Vol. 3, p. 380). VP, 1.135 is quoted by Abhinavagupta in the Īśvarapratyabhijñāvivṛtivimarśinī (Vol. 1, p. 100). VP, 1.159, which states the tripartition of language, is not quoted verbatim but the levels of speech are frequently mentioned and debated by all three main thinkers of the Pratyabhijñā. On the Pratyabhijñā’s change of attitude toward Bhartṛhari, see Torella (2008). On the point, see Torella (1999, 37) and Torella (1994, 118–19, fn. 23). See Ferrante (forthcoming). Most notably Chalmers (1996). vārṇānām eva ca paramārthato ’rthatādātmyalakṣanaṃ vācakatvam. See Parātriṃ śikāvivaraṇa, p. 251, l. 9. Torella speaks of “Mīmāṃsā elements cooked in Bhartṛharian sauce.” See Torella (2004, 183).

Primary Sources ĪPV

Subramanya Iyer, K.A., and K.C. Pandey, eds. 1986a. Īśvarapratyabhijñāvimarśinī of Abhinavagupta. Vols. I–II. Delhi: Motilal Banarsidass.

Īśvarapratyabhijñākārikā

Īśvarapratyabhijñākārikā of Utpaladeva. 1994. In The Īśvarapratyabhijñākārikā of Utpaladeva with the Author’s Vṛtti.

Īśvarapratyabhijñāvimarśinī

Subramanya Iyer, K.A., and K.C. Pandey, eds. 1986b. Īśvarapratyabhijñāvimarśinī of Abhinavagupta. Vols. I–II. Delhi: Motilal Banarsidass.

Īśvarapratyabhijñāvivṛtivimarśinī

Shastri, Madhusudan Kaul, ed. 1938–43. Īśvara pratyabhijñāvivṛtivimarśinī [of Abhinavagupta]. Kashmir Series of Texts and Studies, 60, 62, 65. Vols. I–III. Bombay: Research Dept. Jammu / Kashmir State.

Parātriṃśikāvivaraṇa

Gnoli, Raniero. Parātriṃśikāvivaraṇa of Abhinavagupta. In Gnoli 1985.

Pramāṇavārttika

Sāṅkṛtyāyana, Rāhula. 1938–40. “Dharmakīrti’s Pramāṇavārttika with a Commentary by Manorathanandin.” Journal of the Bihar and Orissa Research Society: New Series 24–26: 349–84.

148

Marco Ferrante

Śivadṛṣṭi

Shastri, Madhusudan Kaul, ed. 1934. The Śivadṛṣṭi of Śrī Somānandanātha with the Vṛtti by Utpaladeva. Kashmir Series of Texts and Studies 54. Srinagar: Research Dept.

Vākyapadīya

Subramania Iyer, K.A., ed. 1966a–73. Vākyapadīya of Bhartṛhari with the Commentaries Vṛtti and Paddhati of Vṛṣabhadeva. Deccan College Monograph Series. 3 vols. Pune: Deccan College Postgraduate / Research Institute.

VP

Subramania Iyer, K.A., ed. 1966b–73. Vākyapadīya of Bhartṛhari with the Commentaries Vṛtti and Paddhati of Vṛṣabhadeva. Deccan College Monograph Series. 3 vols. Pune: Deccan College Postgraduate / Research Institute.

Secondary Sources Carruthers, Peter. 2016. “Higher-Order Theories of Consciousness.” In The Stanford Encyclopedia of Philosophy, ed. by Edward N. Zalta. Fall 2016 Edition, https:// plato.stanford. edu/a​rchiv​es/fa​ll201​6/ent​ries/​consc​iousn​ess-h​igher​/. Chalmers, David J. 1996. The Conscious Mind: In Search of a Fundamental Theory. New York: Oxford University Press. Ferrante, Marco. 2017. “Studies on Bhartṛhari and the Pratyabhijñā: The Case of Svasaṃvedana.” Religions 8 (8): 145. doi:https://doi.org/10.3390/rel8080145. Ferrante, Marco. 2019. “Studies on Bhartṛhari and the Pratyabhijñā: Language, Knowledge and Consciousness”. Journal of Indian Philosophy, https://doi. org/10.10007/s10781-019-09412-2. Ferrante, Marco. Forthcoming. “The Pratyabhijñā on Consciousness and SelfConsciousness: A Comparative Perspective.” Gnoli, Raniero. 1985. Il commento di Abhinavagupta alla Parātriṃśikā (Parātriṃśikātattvavivaraṇa): traduzione e testo. Roma: Istituto Italiano per il Medio ed Estremo Oriente. Nemec, John. 2011. The Ubiquitous Śiva: Somānanda’s Śivadṛṣṭi and His Tantric Interlocutors. Oxford: Oxford University Press. Padoux, André. 1990. Vāc. The Concept of the Word in Selected Hindu Tantras. Albany: SUNY Press. Rastogi, Navjivan. 2009. “Vāk as Pratyavamarśa: Bhartṛhari from Abhinavan Perspective.” In Bhartṛhari: Language, Thought and Reality. Proceedings of the

The Place of Language in the Philosophy of the Recognition

149

International Seminar, Delhi, December 12–14, 2003, ed. by Mithilesh Chaturvedi, 301–41. Delhi: Motilal Banarsidass. Ratié, Isabelle. 2011. Le soi et l’autre. Identité, différence et altérité dans la philosophie de la Pratyabhijñā. Leiden/Boston: Brill. Sanderson, Alexis. 2007. “The Śaiva Exegesis of Kashmir.” In Mélanges tantriques à la mémoire d’Hélène Brunner/Tantric Studies in Memory of Hélène Brunner, ed. by Dominic Goodall and André Padoux, 231–442. Collection Indologie 106. Pondicherry: EFEO / IFP. Torella, Raffaele, ed. 1994. The Īśvarapratyabhijñākārikā of Utpaladeva with the Author’s Vṛtti. Serie Orientale Roma 71. Roma: Istituto Italiano per il Medio ed Estremo Oriente. Torella, Raffaele, ed. 1999. Vasugupta. Gli aforismi di Śiva con il commento di Kṣemarāja ((Śivasutravimarśinī). Milano: Mimesis. Torella, Raffaele. 2001. “The Word in Abhinavagupta’s Bṛhad-vimarśinī.” In Le parole e i marmi: studi in onore di Raniero Gnoli nel suo 70 compleanno, ed. by Raffaele Torella et al., 853–74. Serie Orientale Roma 92. Roma: Istituto Italiano per l’Africa e l’Oriente. Torella, Raffaele. 2004. “How Is Verbal Signification Possible: Understanding Abhinavagupta’s Reply.” Journal of Indian Philosophy 32. Torella, Raffaele. 2008. “From an Adversary to the Main Ally: The Place of Bhartṛhari in the Kashmirian Śaiva Advaita.” In Linguistic Traditions of Kashmir. Essays in Memory of Paṇḍit Dinanath Yaksha, ed. by Mrinal Kaul and Ashok Aklujkar, 508–24. New Delhi/Jammu: D.K. Printworld/The Harabhatta Shastri Indological Research Institute. Vergiani, Vincenzo. 2016. “Helārāja on Omniscience, Āgama, and the Origin of Language.” In Around Abhinavagupta: Aspects of the Intellectual History of Kashmir from the 9th to the 11th Centuries, ed. by Eli Franco and Isabelle Ratié, 531–608. Leipziger Studien zu Kultur und Geschichte Süd- und Zentralasiens 6. Berlin: LIT Verlag.

150

Part Two

Of Word Meanings

152

Introduction

In Part One we discussed the most influential views on what speech units are, culminating in the exposition of the arguments in favor and against the theory of the sphoṭa. The most extreme sphoṭa theorists, true “semantic monists,” even denied the most basic signifier-signified duality. Now, after examining theses on the nature of the signifier, it is time to discuss the signified. Early Nyāya authors (Chapter 7) argued that words convey propertiesbearing individuals (tadvat). The word “cow” denotes an individual being possessing the universal “cowness.” Therefore, even individual words convey complex meanings that are not radically distinct from sentence meanings. A further implication of this theory is that words refer to external entities. Such realists postulate a hardwired relation between a word and its referent, and base their concept of denotation on such a relation; the Buddhist Dignāga, instead, proposes a negative relation, and accordingly a weaker denotation which consists of “non-exclusion” (Chapter 8). This method of non-exclusion may be consistently applied to word meanings. While examining sentence meanings, however, Mīmāṃsā authors had nuanced arguments in defense of their basic tenet that words denote universals, but only sentences can convey specific meanings (Chapter 9). These authors, from Nyāya, Buddhist, and Mīmāṃsā schools, claimed that what is signified by a word (śabdārtha) is respectively a property-bearer, the counter-positive of an exclusion, and a universal. But how to render this phrase, śabdārtha, in English language? Is this artha akin to our philosophical use of “sense” or “referent”? The issue, again from a Buddhist viewpoint, is investigated in Chapter 10. As presented in Chapter 7, earlier Nyāya authors focused on the signification of individual words, arguing in favor of the universal-possessing individual as their referent. Their tradition, however, lacked a theory of meaning at the level of the sentence. Jayanta, while furthering the theory of the property-bearer, filled this gap by developing his theory of a second causal power, besides their denotative power, which he called the “intentionality” (tātparya) of words (Chapter 11).

154

Introduction

Such reflections triggered questions about the nature of the relation among words and their referents. Is this a fixed, natural relation or an arbitrary, artificial one? Mīmāṃsā authors maintained that words and meanings cannot be conceived outside their relation, so such a relation cannot be artificially construed by any person at any given moment in time. The relation is by definition fixed and permanent. Nyāya and the Buddhist epistemological school, by contrast, maintained that this word-meaning relation is conventional, not fixed or permanent in any way, because if the very relata are not permanent their relation must also be impermanent. In Chapter 12 we find the take of some Navyanyāya authors on this question. According to this author, the word-meaning relation is defined by God at the beginning of creation, so to learn the meaning of a word is also to know God’s will. While God has infinite knowledge and eternal wish, however, the human cognitive horizon is limited. How can our knowledge extend to the infinite number of cows and the infinite other referents in the past, present, and future? Navyanyāya authors use their epistemic notion of sāmānyalakṣaṇā to support their theory of meaning.

7

Early Nyāya on the Meaning of Common Nouns Matthew R. Dasti

When the Naiyāyikas develop positions within the philosophy of language, they often nest such discussions under other categories that are more fundamental to their system. For example, they argue for the conventional basis of denotation while defending testimony’s status as an irreducible source of knowledge (NSū, 2.1.49–56 in Gautamīyanyāyadarśana), and they explain secondary reference while analyzing equivocation within debate (NSū, 1.2.10–14). These two discussions are under the headings pramāṇa and chala, respectively, the first and fourteenth of classical Nyāya’s sixteen core topics.1 The concern of this chapter, the meaning of common nouns, is considered under the broad heading of pramāṇa, in NSū, book 2, section 2. It is the culmination of a line of inquiry that arises abruptly after consideration of the exact number of irreducible pramāṇa types. From NSū, 2.2.13 until the end of the book there are three major debates: over the eternality of sound (NSū, 2.2.13– 39), whether letters are transformed or replaced when, for example, i becomes y under sandhi rules (NSū, 2.2.40–57), and finally over the meaning of common nouns (NSū, 2.2.58–69), the focus of this chapter. The entire discussion centers on what we may call the metaphysics of meaningful utterances, and it seems to be at some remove from the core concern with pramāṇas, but in his concluding verse for the entire second chapter, Uddyotakara ties them all together: Here we have discussed: reflection on and determination of doubt and the sources of knowledge, the true nature of sound and its status as a source of knowledge, and the meaning of words (NV, ad NS 2.2.69, p. 318,14–15).2

Uddyotakara’s summary reiterates the way in which Nyāya treats the identifiable knowledge source testimony as an instance of the pervasive and fundamental power of words to generate veridical understanding.3 Therefore, it is fitting

156

Matthew R. Dasti

for us to think of the object of denotation as that which we cognize when we understand aptly used words.

1  The Alternatives: Individual, Universal, and Form When trying to settle the meaning of words in general, the sūtras themselves restrict the question to the meaning of common nouns; “cow” is employed as a stock example. Three competing alternatives are put forth as the primary meaning of such words: an individual, a universal, and a form or shape. In its characteristic style, the sūtras introduce doubt in order to stimulate inquiry (NSū, 2.2.59). There is doubt owing to usage that is directed to the individual, form, and universal, which are inseparably connected.4

Let us examine the rationale given for each.

1.1  The Individual (vyakti) Vātsyāyana remarks that “individual” (vyakti) is here synonymous with “substance” (dravya). The case for the individual as the primary referent of a word is that it is the only way to account for usages that require granular specification. It’s the individual, since the usage is for an individual in the case of the word “which” as well as with words for group, offering, possession, number, growth, reduction, and color, and in compound words—offspring, too. (NSū, 2.2.60)5

Vātsyāyana: “‘The cow which is standing,’ ‘The cow which is sitting’— in phrases like these, the word ‘cow’ does not denote a universal, because a universal is undifferentiated in its instances. […]”6 In each of the situations mentioned, only individuals can account for usage. Universals are singular and undivided and thus cannot serve as the meaning of usages that require plurality or specification. Uddyotakara adds: “A universal is singular. Qualification of it is meaningless.”7

1.2  The Universal (jāti) The case for the universal is that in the absence of a universal, a “bare individual” (vyaktimātra) could not play any role within language.

Early Nyāya on the Meaning of Common Nouns

157

No (it is not the individual), since there would be nothing to ground usage (NSū, 2.2.61).8

Vātsyāyana: “In the case of the word ‘which’ and the other items mentioned, that which is distinguished or qualified is the meaning of the word ‘cow.’ In phrases like ‘The cow which is standing,’ ‘The cow which is sitting,’ it is not the unqualified, bare individual substance that is denoted, without the universal (cowhood) […]. Therefore, the individual is not by itself the meaning of the word.” Uddyotakara adds: “The word ‘cow’ does not denote a bare individual alone. If it did, then it would generate awareness of any and all individuals.”9 Here, Uddyotakara appeals to an important early Nyāya causal principle. Understanding that a is a cause does not merely account for the fact that a generates b, but also for why it does not generate c, d, e, f, and so on. Causal influences are restrictive (cf. NSū, 3.1.23). Universals likewise allow for the regularities of language use in ways that mere individuals cannot. This is a compelling case. But how then to respond to the problem of finegrained usage advanced in NSū, 2.2.60? The sūtras reply, There is common usage of certain words (like “stick”), with reference to (things like) brāḥmaṇas […] because of conditions like association, even though these are not what the words primarily refer to (NSū, 2.2.62).10

Here, the sūtras and commentators provide a list of standard examples of secondary usage, where nonstandard meanings are affixed to terms by the force of certain contextual conditions. “In (e.g.) the sentence ‘Feed the stick,’ a brāḥmaṇa associated with a walking stick is denoted” (NBh, p. 130,13).11 Vātsyāyana connects this to the problem at hand, “Out of these, a word denoting a universal is directed toward the individual through association or connection” (NBh, p. 131,1–2).12 On this account, universals are the proper referents of common nouns. But through secondary usage, individuals may be spoken of, just as a brāḥmaṇa may be spoken of through the word “stick.”

1.3  The Form (ākṛti) At this point of the discussion, the third and final option is put forth, that the primary denotation of a word is neither an individual nor a universal, but a form or shape. It is the form that is the primary meaning, because determination of what exists depends upon it. (NSū, 2.2.63)13

158

Matthew R. Dasti

The third entrant is cleverly motivated by taking seriously the rationale provided for the universal. The primary referent for common nouns must be something that makes it possible to distinguish between types of things when we communicate. While universals provide a metaphysical basis for such distinctions, this interlocutor contends that they don’t provide an epistemological basis, because our knowledge of objects as belonging to certain classes itself is dependent on our identification of repeatable forms or shapes. Such form or shape amounts to a fixed arrangement of something’s parts along with its sub-parts. When shape is grasped, what exists is determined, and we say “That’s a cow,” “That’s a horse.” Without this being grasped, such instances of knowledge would not occur. The rule is: the thing that is properly denoted by a word is that which, through its being grasped, results in what exists being determined. That is its meaning (NBh, p. 131,6–8).14

Word use depends upon distinctions and hence, universals, but to know something as possessing a universal itself usually requires the recognition of a particular form. Therefore, it is argued, such forms or shapes are the primary referent of common nouns. Anticipating what will be said by Uddyotakara in his commentary on 2.2.66, below, we may also add that taking shape to be the referent helps make sense of references to cow-shaped pastries and other artifacts which are spoken of as “cow” but lack the corresponding universal. A father says to his daughter “I like that cow you are holding” while she plays with her animal figurines. What problems are faced by this third entrant? The core problem advanced is that form is metaphysically unimportant compared to the universal and individual. A word denotes something that has a connection with a universal; something qualified by the universal. For example, “That’s a cow.” And it’s not the arrangement of parts which is connected to the universal. “What is connected to the universal then?” A thing, an individual substance that has a fixed arrangement of parts. Therefore it is not the form itself that is the meaning of the word (NBh, p. 131,11–13).15

Uddyotakara sharpens this point by arguing that the form or shape fares no better than the bare individual. “Here too, what was said earlier applies. No, since there would be nothing to ground it” (NV, p. 302,3).16 What is needed is a metaphysical basis for use of common nouns, and form is ultimately an individual feature of an individual substance. The next sūtra argues that a cow shape can exist in an individual substance without the corresponding universal, and thus it cannot play the role that is needed.

Early Nyāya on the Meaning of Common Nouns

159

The primary referent is the universal, because a clay cow, though endowed with individuality and shape, cannot be the subject of the consecrating ceremonies of washing, and so on. (NSū, 2.2.64)17

Vātsyāyana makes the case plain: “‘Wash the cow,’ ‘Bring the cow,’ ‘Donate a cow’—these would not be used for a cow made of clay. Why? Because a clay cow would not belong to the proper class; it lacks the appropriate universal” (NBh, 2.2.64, p. 132,1–2).18 The final salvo in this debate is fired on behalf of both the individual and the shape, against the universal. In effect, the interlocutor stamps his foot to the ground while reiterating that it is epistemic priority that matters here, not metaphysical priority. Wrong, because the manifestation of the universal depends upon the form and the individual. (NSū, 2.2.65)19

Vātsyāyana: “The manifestation of the universal depends on the form and the individual. It is not the case that when the form and the individual are not grasped, the pure universal is grasped by itself. Therefore the universal is not the meaning of the word” (NBh, p. 132,6–7).20 Let us, then, take stock of reasons for and against each of the candidates (Table 7.1). Here, the Nyāya solution to this apparent trilemma is to propose that all three serve together, as the complex object(s) of reference. Nevertheless, the individual, the form, and the universal constitute the word’s meaning (NSū, 2.2.66).21

Vātsyāyana: “There is no rule about which of the three would be the meaning element that is predominant or subordinate in any given case of a word’s usage. For, when one wants to talk about a distinct thing and in the manner of specificity, then it is the individual that is the predominant meaning and the universal and shape subordinate. But when one does not want to talk about distinctness or difference, but is speaking in the mode of generalization, then it is the universal that is the predominant meaning and the individual and shape subordinate. All of this is richly illustrated in common usage. The predominance of shape may be likewise understood” (NBh, p. 132,11–15).22 Uddyotakara provides an example of the latter: “Make rice-flour cows” (NV, p. 307,3).23 On the surface, this triadic, unified account takes the object of common nouns to be a conjunction: individual and universal and form, while allowing that certain conjuncts are predominant or subordinate according to context and

160

Matthew R. Dasti

Table 7.1  Arguments in Favor Of and Against Each Candidate Candidate

For

Against

Individual (vyakti)

Provides grounds for fine-grained, granular reference; is epistemically fundamental, even when grasping universals.

Is incapable of providing grounds for regularities of reference.

Universal (jāti)

Provides metaphysical basis for regularities of reference; accounts for usage which distinguish between real members of a class and objects that merely have their shape (a cow vs. a clay cow).

Is epistemically secondary, as it is only grasped after grasping an individual and its shape.

Shape (ākṛti)

Accounts for what is epistemically immediate when identifying class membership; accounts for usage pertaining to similarly shaped artifacts (a clay cow is spoken of with “cow”).

Is as much an individual as the particular substance which it qualifies. As such, is also incapable of providing grounds for regularities of reference; things with “cow” shapes, but lacking the universal cowness are sometimes outside of the scope of the word “cow.”

the desire of the speaker. To say “that cow is drinking water” is to prioritize the individual. Still, the universal and form are included in the act, since the individual is spoken of as a cow. To say “cows are not aquatic animals” is to prioritize the universal; but of course, we are speaking of the totality of individual cows, each of which exhibits a form that allows one to recognize the universal “cowhood.” Finally, to say “whatever has a dewlap is not a horse” is to prioritize form, while the universal and individual are required as the metaphysical background conditions for this sort of reference. This surface reading works well for paradigm cases, like “cow,” that include a property-bearing substance that instantiates both a universal which demarcates the sort of substances denoted by the term and a form which acts as an indicator of the universal in question. But it becomes strained by instances where one of the three is obviously not present within the direct object of reference. Most problematic for this account are instances of reference to form where the universal is absent. For statements like “make rice-flour cows,” there is direct reference of form and individuals, but not the universal “cowhood” (as explicitly stated in (NSū, 2.2.64) quoted above).24 In this case, there are three available options: (i) to defend that common nouns directly denote a conjunction of the three elements, (ii) to introduce a hybrid account where they denote some of the three while indirectly referring

Early Nyāya on the Meaning of Common Nouns

161

to others, or (iii) to allow for a disjunctive account, where some of the disjuncts need not be present at all. In support of (i), an argument could be made that even in the case of cow pastries, the term “cow” refers to all three elements, because the use of “cow” even for cow-shaped pastries, depends on a semantic connection to actual cows. Such parasitism arguments are a staple of Nyāya realism, but there seems to be a difference between requiring knowledge of the universal cowhood as a background condition to think and refer to objects associated with cowhood, and claiming that every use of the term “cow,” even when referring to artifacts, somehow contains reference to the universal directly. This sort of creative exegesis requires us to extrapolate far beyond what is said by the early thinkers themselves.25 Option (ii) would allow a hybrid account which combines reference and indication or secondary meaning. Here, the word “cow” directly refers to the individual and the universal, and through something like indication or secondary meaning (lakṣaṇā) refers to the form. This would allow cow pastries to be conveyed by the term “cow” but not through direct denotation. One problem with this attempt is that Naiyāyikas think that the form of a pastry cow is indeed the same sort of thing as the form of a real cow. For early Naiyāyikas, form is simply a basic shape that usually indicates a distinct universal. Another problem is that our Naiyāyikas resist such hybrid explanations generally. Vācaspati, in particular, devotes his introductory remarks on sūtra 2.2.66 to a refutation of hybrid accounts (in particular, Mīmāṃsā views that take the primary object of denotation to be the universal, with the individual only known after, through secondary reference or indication.) He stresses the unified nature of the object of denotation as experienced upon hearing a word (ekopalambhagocaratva (NVT, 2.2.66)). Option (iii) would be to take sūtra 2.2.66 as a disjunction: the individual or the universal or the form are the meaning, with the or being inclusive. In paradigm cases, all three disjuncts are present even as some are primary and some are subordinate. But in non-paradigm cases, fewer than three are directly denoted. This is probably the best reading in that—in a context slightly different from that of a cow pastry—Vātsyāyana explicitly allows for it: “When a universal is not made known by means of a form, in cases like clay, gold, and silver, form is excluded. The condition of being the object of the term is removed” (NBh, 2.2.68). Clay, gold, and silver are malleable; their observable form is not what indicates their nature. Therefore, the common noun “gold” refers only to an individual piece of gold and the universal goldness. This is an explicit admission that in nonparadigm cases, the object of reference need not include all three disjuncts, even as they are all typically present in paradigm cases.26

162

Matthew R. Dasti

2  Holism Reaffirmed With its irenic solution, Nyāya takes recourse to a strategy seen elsewhere in the sūtras: competing, independent options are found insufficient to account for an explanandum, but they are found to serve collectively as an appropriate answer.27 This is itself an illustration of two principles that lie deep within Nyāya’s siddhāntic structure: first, when proposing explanations, a propensity to countenance a cluster of necessary conditions over a single sufficient condition, and second, that real objects are typically multifaceted complexes with a variety of properties. The metaphysical correlate to an explanatory stress on necessary conditions is the acceptance of stable properties as real features of objects. These principles are typically highlighted in the context of Nyāya’s holism (cf. NSū, 2.1.33–6) and denial of momentariness (cf. NSū, 4.1.14–18). The former is an endorsement of metaphysically stable, spatially coherent clusters of properties unified in an object at any given time; the latter is the endorsement of the existence of such unified objects and properties through time. Nyāya’s view of the object of denotation is consistent with these more famous cases. For example, when we refer to a cow, we typically speak of a substance (dravya), possessing a distinct shape (ākṛti), and a universal (jāti) appropriate to all cows. Even when some properties of an object are secondary, latent, or merely dispositional, they are still metaphysically integrated with the object. From the epistemic perspective, some properties may be at the forefront of a cognitive event, without consigning others to nonexistence.

Notes 1 Citations of the Gautama Akṣapāda’s Nyāyasūtra, Vātsyāyana’s Nyāyabhāṣya, Uddyotakara’s Nyāyavārttika, Vācaspatimiśra’s Nyāyavārttikatatpāryaṭīkā are, respectively, from NBh, pp. 128–134 in Gautamīyanyāyadarśana, NV, pp. 298–318, and NVT, pp. 426–451. Some translations are taken or modified from Dasti and Phillips (2017). I would like to thank David Buchta for many helpful comments on a penultimate draft of this chapter. 2 saṃśayasya pramāṇānām vicāras tad-vyavasthitiḥ / śabdasya tattvaṃ prāmāṇyaṃ padārthaś ceha kīrtitāḥ //. Vācaspati notes that having concluded the discussion about the impermanence of sounds, our discussion “investigates words, which facilitate the status of sounds as sources of knowledge” (śabdaprāmāṇyopayogi padaṃ nirūpayati, NVT, ad NS 2.2.68, p. 426,6–7).

Early Nyāya on the Meaning of Common Nouns

163

3 Cf. NSū, 2.1.52, which argues that knowledge from words in general depends on āpta upadeśa, the assertions of an authority, echoing the definition of testimony in NSū, 1.1.7. 4 vyakty-ākṛti-jāti-sannidhāu upacārāt saṃśayaḥ. 5 yā-śa​bda-s​amūha​-tyāg​a-par​igrah​a-saṃ​khyā-​vṛddh​y-apa​caya-​varṇa​-samā​sa-anu­ band​hānāṃ​ vyaktāu upacārād vyaktiḥ. 6 yā gaus tiṣṭhati yā gaur niṣaṇṇeti, nedaṃ vākyaṃ jāter abhidhāyakam abhedāt (NBh, 2.2.60, p. 129,6). 7 ekā jātiḥ. tasyā viśeṣaṇam anarthakam (NV, 2.2.60, p. 300,8–9). 8 na, tad-anavasthānāt. 9 na anena go-śabdena vyakti-mātraṃ śuddham ucyate. yady ayaṃ vyaktimātrābhidhāyako ’bhaviṣyat, tena yasyāṃ kasyāñcid vyaktau pratyayo ’bhaviṣyat (NV, p. 300,20–21). 10 sahacaraṇa-[…]-sādhanādhipatyebhyo brāhmaṇa-[…]-puruṣeṣv atadbhāve ’pi tadupacāraḥ. 11 yaṣṭikā-sahacarito brahmaṇo ’bhidhīyata iti. 12 tatra ayaṃ sahacaraṇād yogād vā jāti-śabdo vyaktau prayujyata iti. 13 ākṛtis tad-apekṣatvāt sattva-vyavasthāna-siddheḥ. 14 sattva-avayavānāṃ tad-avayavānāṃ ca niyato vyūha ākṛtiḥ. tasyāṃ gṛhyamāṇāyāṃ sattva-vyavasthānaṃ sidhyati, ayaṃ gaur ayam aśva iti na agṛhyamāṇāyām. yasya grahaṇāt sattva-vyavasthānaṃ sidhyati taṃ śabdo ’bhidhātum arhati, so ’sya artha iti. 15 yasya jātyā yogas tad atra jāti-viśiṣṭam abhidhīyate gaur iti. na ca avayavavyūhasya jātyā yogaḥ. kasya tarhi niyata-avayava-vyūhasya dravyasya. tasmān na ākṛtiḥ padārthaḥ. 16 atrāpi tad eva upasthitam, na tad anavasthānād iti. 17 vyakty-ākṛti-yukte ’py aprasaṅgāt prokṣaṇādīnāṃ mṛd-gavake jātiḥ. 18 gāṃ prokṣaya, gām ānaya, gāṃ dehi iti na etāni mṛd-gavake prayujyante. kasmāt jāter abhāvāt. 19 na ākṛti-vyakti-apekṣatvāj jāty abhivyakteḥ. 20 jāter abhivyaktir ākṛti-vyaktī apekṣate. na agṛhyamāṇāyām ākṛtau vyaktau ca jātimātraṃ śuddhaṃ gṛhyate. tasmān na jātiḥ padārtha iti. 21 vyakti-ākṛti-jātayas tu padārthaḥ. 22 pradhāna-aṅga-bhāvasya aniyamena padārthatvam iti. yadā hi bheda-vivakṣā viśeṣa-gatiś ca tadā vyaktiḥ pradhānam, aṅgaṃ tu jāty-ākṛtī. yadā tu bhedo ’vivakṣitaḥ, sāmānya-gatiś ca, tadā jātiḥ pradhānam, aṅgaṃ tu vyakty-ākṛti. tad etad bahulaṃ prayogeṣu. ākṛtes tu pradhānabhāva utprekṣitavyaḥ. 23 piṣtakamayyo gāvaḥ kriyantām iti. 24 For an astute identification of other distinctively metaphysical problems (that range beyond the concerns of this chapter) stemming from Nyāya’s view that a clay cow

164

Matthew R. Dasti

has the form of a cow without instantiating the universal “cowhood,” see Scharf (1996, 159–63). 25 Also see Doctor (2012, 59–60) for an argument that Nyāya’s view of universals undergirding denotation is at odds with its account of secondary meaning. 26 The discussion in the sūtras closes with definitions of each of the primary referents of a word. These are not directly relevant to our analysis, so I include them here in the notes. “The individual is that manifest object that is the locus of a distinct property (2.2.67).” “‘Form’ is the name of the indicator of a universal (2.2.68).” “The universal generates the conception of commonality (2.2.69).” 27 Famously, NSū, 4.1.19–21, which argues that neither human action nor God’s power independently account for the results of action; rather, they work in tandem.

Primary Sources Gautamīyanyāyadarśana

Thakur, Anantalal, ed. 1997a. Gautamīyanyāyadarśana with Bhāṣya of Vātsyāyana. Nyāyacaturgranthikā 1. New Delhi: Indian Council of Philosophical Research.

NBh

Nyāyabhāṣya of Vātsyāyana. In Gautamīyanyāyadarśana.

NSū

Nyāyasūtra of Gautama. In Gautamīyanyāyadarśana.

NV

Thakur, Anantalal, ed. 1997b. Nyāyabhāṣyavārttika of Bhāradvāja Uddyotakara. Nyāyacaturgranthikā 2. New Delhi: Indian Council of Philosophical Research.

NVT

Thakur, Anantalal, ed. 1996. Nyāyavārttikatātparyaṭīkā of Vācaspatimiśra. Nyāyacaturgranthikā 3. New Delhi: Indian Council of Philosophical Research.

Secondary Sources Dasti, Matthew, and Stephen Phillips. 2017. The Nyāya-sūtra: Selections with Early Commentaries. Indianapolis: Hackett Publishing. Doctor, Payal. 2012. “Meaning and Metaphor in the Early Nyāya School.” Journal of Indian Philosophy and Religion 17: 38–67. Scharf, Peter M. 1996. The Denotation of Generic Terms in Ancient Indian Philosophy: Grammar, Nyāya, and Mīmāṃsā. Transactions of the American Philosophical Society Held at Philadelphia for Promoting Useful Knowledge 86.3. Philadelphia: American Philosophical Society.

8

Dignāga on Relation Kei Kataoka

1  The Scope of This Chapter Buddhist epistemologists are well known to hold a special semantic theory characterized by exclusion (apoha). Dignāga (ca. AD 470–530) is the founder of this negative semantics in the Buddhist tradition. In this chapter, the present author tries to elucidate Dignāga’s theory of semantics in general and his notion of the relation between a word and its referent in particular, as observed in the fifth chapter of his Pramāṇasamuccaya and its Vṛtti.1

2  The Three Levels of Existence It is necessary to distinguish between three levels of existence when reading texts of Yogācāra authors: 1. The ultimately existent (paramārtha/pariniṣpanna) 2. The causally existent (dravyasat/paratantra) 3. The conceptually constructed existent (prajñaptisat/parikalpita) For Buddhist scholars, as clear from Nāgārjuna’s severe criticism of the realistic view of language, words operate only in the sphere of the third level, that is, they communicate only conceptually constructed things called “general characteristics” (sāmānyalakṣaṇa). They do not communicate the “particular characteristics” (svalakṣaṇa) present at the second level, nor do they operate on the first level.2 Language, however, functions successfully on the mundane level and is to be distinguished from an erroneous cognition such as a dreamingcognition. Dignāga, who is strongly influenced by the view of the grammarian Bhartṛhari, tries to elucidate the operation of language within the sphere of the third level, without focusing too strongly on its erroneousness with regard to

166

Kei Kataoka

the second level. Dignāga tries to explain how meaning is conveyed by a word, although both relata are in fact characterized by the negation of other things, and, ontologically speaking, absences. Dignāga’s attitude may be contrasted with that of Dharmakīrti (ca. AD 600– 660), who is forced to give greater attention to the boundary or gap between the second and the third levels in order to respond to Kumārila’s critique of Dignāga’s theory of apoha. Dharmakīrti bridges the gap between the two levels by explaining that an inferential cognition, although essentially erroneous, does not belie (avisaṃvādin) one’s expectation. His introduction of the notion of “judgment” (adhyavasāya) is the key to linking the two levels. In making this connection, his semantic theory, which is included in his theory of conceptual cognition in general, changes drastically from Dignāga’s. Whereas Dharmakīrti’s semantic theory is specifically based on the Buddhist ontology, Dignāga’s is not necessarily confined to the Buddhist system, as is also suggested by the fact that Dignāga’s main idea of “exclusion” semantics comes directly from the grammarian tradition.3 When Dharmakīrti emphasizes the fallacious aspect of language, he shows the arbitrariness of the relationship made up and agreed upon by human beings. Language has nothing to do with external reality. By contrast, for Dignāga, the semantic relationship is assumed to be relatively stable in a given society once established and conventionalized, although essentially transient (anitya) inasmuch as it is an artificial relation (saṃbandha). The communicative power of a word is not innate. Rather, it is based on the artificial connection established through the negative method of co-absence (vyatireka) that is based on nonperception (adarśana).4 Although mundane usage conventionalized as such (lokarūḍha) is not taken to refer to reality, it is observed (anu-gam) according to the dictates of the society (lokavat), as stated by Dignāga: “[What is] current in the world is not adhered to” and “therefore we too do not adhere to expressions current in the world whether they have a cause of application or are technical designations as having real referents, but observe them exactly as the world does. And in the world the word ‘color’ is only acknowledged to denote [the color] blue, and so on, but not to denote taste, etc.”5 In other words, the language of a society is followed, though not “believed in.”

3  Language as Inference The proper use of language by a reliable person (āpta) is regarded as a type of genuine inference on the part of a listener, as indicated by the main theme of the

Dignāga on Relation

167

Apoha chapter of Dignāga’s Pramāṇasamuccaya. There, Dignāga claims that verbal cognition (śābda) is not an independent, separate type of means of valid cognition, but that it is rather included in and reducible to inference.6 He presupposes a parallel structure between inference and verbal cognition.7 At the concluding part of the Apoha chapter, he states: “For, the relation (between word and referent) is mentally constructed after having perceived the word ‘jack-fruit’ and its referent by other means of valid cognition [namely perception] with the thought ‘This word refers to that,’ as with the relation between an inferential reason and its target.8 Therefore, verbal cognition is not a separate means of valid cognition.”9 For Dharmakīrti, on the other hand, although verbal cognition is accepted as a kind of inference with respect to a speaker’s intention (vivakṣā), it is regarded as an inferior type of inference and thus it is to be strictly distinguished from a reality-based inference (vastubalapravṛttānumāna) such as an inference of fire from smoke. In his semantic theory, he emphasizes that a word merely conveys a meaning that is arbitrarily established by human beings and that it essentially has nothing to do with an external entity, whereas Dignāga assumes a word to be a reliable conveyor of a meaning to the extent that the meaning is agreed upon in a society. By rejecting other theories and establishing his own, Dignāga intends to offer, within the third level, a better theory of denotation based on his nominalist view of language as “not touching reality” (arthāsaṃsparśitā).

4  Dignāga’s Basic Scheme From Dignāga’s perspective, when a speaker utters the word “cow” a listener receives it as equivalent to “cow only,” that is, “only a cow and not a noncow such as a horse, an elephant, a lion, etc.” The word “cow” excludes noncows as a whole (PS, 5.43b: sāmānyena nirākṛteḥ) and not one by one, just as the inferential reason “horned-ness” (viṣāṇitva) excludes horses and so on as a whole and not one by one.10 The hidden particle “only” (eva) restricts the semantic domain to that which is excluded from non-cows.11 Thus, a word conveys exclusion of other things.12 As Dignāga states in the opening verse of the Apoha chapter of the Pramāṇasamuccaya, a word expresses its own referent by excluding other things.13 Here, Dignāga presupposes four items in his semantic theory of exclusion: 1. A word such as “cow,” which denotes its own referent. 2. Exclusion (apoha, vyāvṛtti) or negation (nivṛtti), which can ontologically be identified as absence or nonexistence (abhāva).

168

Kei Kataoka

3. Non-cows, which are excluded (apohya). 4. Those things or aspects qualified by exclusion (apohaviśiṣṭa), which we can regard as the complement, that is, that which remains after the exclusion of other things. To put it in Dignāga’s own words, “A word expresses things that are always qualified by the negation of other referents (arthāntaranivṛtti).”14 In Dignāga’s semantics, the referent of a word is “qualified by the exclusion of other referents” (Figure 8.1).15 Thus, according to Dignāga, the word “cow,” an example of a generic term (jātiśabda), can express, in a general form, anything that has not been experienced as a non-cow.16 In other words, the domain of the denotation extends to the entire complement that has not been experienced as non-cows. Those things left over, which have not been excluded as being non-cows, are the referents of “cow.”17 That is to say, the nonspecific domain that remains, not having been excluded, is the general characteristic (sāmānyalakṣaṇa) made manifest by a word.

5  The Background behind Dignāga’s Theory and His Motivation Dignāga explicitly states in the opening verse of the Apoha chapter that verbal cognition functions in the same way as inference: Verbal cognition is not a valid source of knowledge separate from inference, because [a word] expresses its own referent by excluding other referents, just as [an inferential reason] such as “being produced” does.18

To explain, there is a parallel between a word (e.g., “cow” or “horse”) and a logical reason (e.g., being produced or smoke). His claim implies the following: Just as smoke enables one to cognize fire only when one has learned the relationship (i.e., the invariable concomitance) between smoke and fire, in the same way the word “cow” enables one to cognize its referent only when one has Figure 8.1  Four Items in Dignāga’s Semantic Theory 1

cow’

2

exclusion 4

things

3

non-cows

Dignāga on Relation

169

learned the relationship between the two. The application of both inference and language requires the prior learning of a particular relationship. Dignāga states: “And [generally speaking] a word cannot denote a referent if a relation has not been constructed, because [in that case] only its own [phonic] form [would] be understood.”19 Two methods are proposed, one positive and one negative, for establishing the invariable concomitance between smoke and fire (and similarly for establishing the relation between “cow” and a cow):20 1. Establishment of concurrent existence (anvaya) of fire and smoke, that is, establishing the existence of smoke and fire in a similar case.21 2. Establishment of concurrent absence (vyatireka), that is, establishing the nonexistence of smoke in the absence of fire,22 in other words, knowing that where there is no fire, there is no smoke.23 Dignāga considers only the latter, the negative approach, to be a possible method for establishing an invariable concomitance.24 This is because the positive method cannot cover infinite instances,25 whereas the negative method can.26 He states: “Inference [is possible] only through co-absence.”27 Invariable concomitance can also be formulated as follows by adding “only” to the concurrent existence (sapakṣa eva sadbhāvaḥ): A. Only where there is fire, there is smoke. B. Only when there is a cow, there is cow. (Cow is applied only to a cow.) People have never experienced the word “cow” legitimately applied to noncows, just as people have never seen smoke without fire. Thus, there is a concurrent absence between a word and its referent. The device that enables Dignāga to claim an invariable concomitance (vyāpti) or the necessity of the relationship is “non-perception” (adarśana, or anupalabdhi).28 He presupposes that the non-perception of x guarantees the absence of x, as many scholars before Dharmakīrti generally did. Later, Dharmakīrti severely criticizes this idea by labeling it as “mere non-perception” (adarśanamātra). According to Dharmakīrti, “mere non-perception” of x does not necessarily guarantee its absence. Only the non-perception of a perceivable object (dṛśyānupalabdhi) necessarily guarantees its absence.29 Dignāga’s claim of parallelism between inference and verbal cognition might seem strange to us because he does not distinguish between a natural relationship based on causation and a relationship that is socially agreed upon. For Dignāga, however, both relationships similarly stand in the mundane level

170

Kei Kataoka

and both are conceptual in nature inasmuch as their related items are general characteristics (sāmānyalakṣaṇa), that is, essentially, the exclusion of other things (anyāpoha).30 Although they deal with general characteristics in this way, both inferential and verbal cognition function successfully in the world. Thus, they can be distinguished from invalid cognitions such as error.

6  Tadvat and apohavat One may raise the following question regarding the structure of exclusion (apoha) presupposed by Dignāga: What is the crucial difference between Dignāga’s theory and his opponent’s theory of tadvat? Dignāga presupposes four opposing theories of reference in his Pramāṇasamuccaya. 1. 2. 3. 4.

“Cow” denotes only particular, individual cows (bheda). “Cow” denotes only the universal, that is, cowness (jātimātra). “Cow” denotes the relation with the universal (jātiyoga). “Cow” denotes anything qualified by the universal, that is, cowness (tadvat or jātimanmātra).

The first view has a problem with the existence of an infinite number of referents (PS, 5.2b: ānantya). This infinitude makes it impossible to connect a word and its referent. One cannot learn all instances of a thing, that is, one cannot connect all individual instances of a referent with a word, because the referents are infinite. The first view also has the problem of deviation because there are many particulars that could be connected as the referent of the word (PS, 5.2b: vyabhicāra). Thus, after hearing the word, there may be the doubt “which one?” instead of a definite denotation (abhidhāna). The second and third views, although solving the above problems, cannot explain co-referential expressions such as “an existent substance” (sad dravyam), “an existing pot” (san ghaṭaḥ), and so on, because the universal, existence (sattā), or its relation is a property, whereas the substance is a substratum of a property. They cannot be equated on the same level. Thus, the second and third views cannot explain that the two words have a common referent (PS, 5.2cd: sāmānādhikaraṇya). The fourth theory presents the greatest obstacle for Dignāga because “anything that has the universal in question” (tadvat or jātimanmātra) and “anything that has the exclusion of other things” (apohavat) have the same

Dignāga on Relation

171

structure. What is then the main feature that distinguishes between the two competing theories? It is the ontological difference between the two qualifiers: cowness is real, whereas exclusion is unreal (adravya). The latter is a mere absence and it has no ontological existence. Therefore, in Dignāga’s theory, the word “cow” can denote its own referent without incurring into the fault of independence (svātantrya). Thus, the word “cow” can directly refer to those things that are qualified by the exclusion of non-cows (PS, 36c: sākṣād vṛtteḥ). In other words, it does not have the fault of being secondary.31 According to the tadvat theory, on the other hand, the word “cow” has to rely on universal cowness to express the things qualified by cowness, that is, it has the fault of dependency.32 Thus, it can denote its referent not directly (na sākṣāt) but only indirectly through the intervention of the universal. Therefore, the expression of the referents would be secondary.33 Because of this dependence, the theory meets with another problem. It cannot explain the linguistic fact that the word “existent” (sat) can imply or include (ā-kṣip, vy-āp) things lower in the hierarchy such as a substance (dravya) or a pot (ghaṭa), as demonstrated in the co-referential expression “an existent substance” (sad dravyam) or “an existing pot” (san ghaṭaḥ). If one follows the tadvat theory of the realists, the word “cow” includes only up to things that have universal cowness, but it does not include other things beyond them because they are not included as its referent. Dignāga states: “Without inclusion there is no sāmānādhikaraṇya, i.e., there is no scope for the two words to have the same referent.”34 In the case of an existent substance, the word “existent” reaches only up to existent things and not beyond them, because the word has a connection only with the existent. The word “existent” does not refer to a substance (Figure 8.2).35 Dignāga gives as an example the word “white” (PSV ad 5.4a). Since “white” denotes only white things (i.e., substances qualified by a white color), it should not be able to include or imply sweet things and so on.36 Sweet things Figure 8.2  Non-Inclusion of Substance Due to Dependence 1

‘existent’ 2

existence

existence-possessor ?

‘substance ’

3

substance

172

Kei Kataoka

Figure 8.3  Inclusion of Substance Due to Non-Independence ‘existent’

0

exclusion of non-existent 1

‘substance ’

2

existent

substance

are beyond the range of the word “white”; the word “white” does not refer to sweet things. Similarly, because of the dependence on the universal, the word “existent” cannot include or imply a substance. Thus, the theory of tadvat cannot explain the co-referential expression “an existent substance,” in which the word “existent” should cover a substance. Thus, there is the fault of “not implying its lower referent due to dependence.”37 Similarly, the tadvat theory cannot explain the fact that “blue” covers a blue water lily (PSV ad 5.18a). One may still argue that the word “existent” can cover a pot, because it is implied indirectly through an existent thing. Anticipating this objection, Dignāga points out the fault of uncertainty.38 There is no ascertainment of, for example, a pot upon hearing “existent.” Thus, because of ambiguity, a pot is not implied indirectly.39 Dignāga has no difficulty in explaining the co-referential expression “an existent substance,” because the word “existent” independently denotes an existent thing without the intervention of exclusion and can imply a substance. This is because exclusion is absence in nature (Figure 8.3).40 To explain further why the word “existent” cannot include/cover/imply a substance according to the realist theory, whereas it can in Dignāga’s theory, we need to look more closely at the nature of the relationship and the denotation presupposed by the two camps.

7  Relationship and Denotation: Strong versus Weak An important factor in Dignāga’s rejection of the tadvat theory is its positive character with respect to the learning of the relation through concurrent existence (anvaya). Through the positive method of concurrent existence, one learns the hard relationship between two definite relata. We can call this the

Dignāga on Relation

173

Figure 8.4  The Strong Relation and Denotation ‘cow’

anything possessing cowness

denotation: ‘cow’

anything possessing cowness

relation:

“strong relation” in comparison to the “weak relation” presupposed by Dignāga. For the realists, a word positively and definitely denotes (abhi-dhā) certain things through the means of the learned relationship. We can call this “strong denotation” (Figure 8.4). Following this theory, the word “existent” does not cover substances, because one has learned it only in relation to existent things and not substances. Its connection is learned only with a thing that has existence (sattā). To put it another way, “existent” does not include substances as its referent and therefore cannot be juxtaposed with “substance” in reference to the same object, as seen in the expression “an existent substance.” By contrast, the relation accepted by Dignāga can be regarded as a weak relation in that it is not directly learned like the relation between two definite relata in the tadvat theory. Rather, it is learned as a relationship between two remainders (i.e., śabdāpoha and arthāpoha), that is, the words excluded from the words for non-cows and the referents excluded from non-cows. That is to say, the complement that is excluded from non-cows is the target sphere of the word “cow.” The words also have the nature of being excluded, that is, they are distinct from other words such as “horse.” The relationship learned through the method of concurrent absence (vyatireka) is negative in nature. Accordingly, the process of denotation is also negative for Dignāga. In other words, non-exclusion (anapohana or apratikṣepa) is the real nature of the process of denotation (abhidhāna). Following this theory, “existent” does not exclude a substance or a pot, because in learning its referents they were not excluded and therefore are included in its referents. Similarly, the word “blue,” according to the apoha theory, denotes blue substances by excluding non-blue things. Here a blue water lily is included (antar-bhū) in these blue substances, because it has not been excluded (PS, 5.18b: anapohanāt). Inclusion is based on non-exclusion. This is possible only in the theory of weak denotation based on weak connection.

174

Kei Kataoka

Dignāga himself is well aware of the difference between the strong denotation (abhidhāna) presupposed by the opponent and his weak denotation which has the nature of non-exclusion (anapohana). He states: [Q] Then, [why was it] stated that lower and particular [referents] (viśeṣa) are included in the higher and general [referents] (sāmānya)? [A] This was stated not because particulars are denoted (abhidheyatvāt), but rather because they are not excluded (anapohanāt).41

Here Dignāga contrasts strong denotation (abhidhāna) with weak denotation, that is, non-exclusion (anapohana). If one were to follow the opposing theory of strong denotation, one would have to conclude that blue water lilies are not included in blue things, because they are not denoted by the word “blue.” Following Dignāga’s theory of weak denotation, conversely, the referent of “blue” is the remainder, that is, anything not non-blue, which is left over after the exclusion of non-blue things. Therefore, the semantic domain of “blue” includes its own referent, blue things, as well as higher and general referents (e.g., color or quality) and lower and particular referents (e.g., a blue water lily), because they are not excluded. Thus, the ability of the words “a blue water lily” (nīlam utpalam) to convey a blue water lily is explainable only in Dignāga’s theory of non-exclusion. This distinction between strong and weak denotation is probably one of the reasons why Dignāga uses the expression “makes manifest” (dyotayati) instead of “denotes” (abhidhatte) in the opening part of the Pramāṇasamuccayavṛtti.42 He states: “A word … makes manifest (dyotayati) that [part] (i.e., the referent) through the exclusion of other referents, just as [an inferential reason such as] ‘being produced’ [manifests the object of inference].”43 Dignāga probably refers to this weak denotation with the word dyotayati, because for him a wordreferent is something made manifest indirectly, through the exclusion of other things, just as an object’s aspect “being transient” (anityatva) is made manifest by its other aspect, “being produced” (kṛtakatva) through a similar process of exclusion.

8  Higher and Lower Referents Only a “weak” theory of semantics can explain the linguistic phenomena of general terms (sāmānyaśabda) and specific terms (bhedaśabda). For example, the word “tree” can definitely imply/cover higher/general referents such as

175

Dignāga on Relation Figure 8.5  The Higher and the Lower Referents substance earthen ‘tree’

non-tree

tree pine

non-earthen

cherry

an earthen thing (pārthiva), a substance (dravya), an existent (sat), and so on, because they are not excluded inasmuch as the word “tree” has not been seen to apply to non-earthen things and so on.44 The word “tree” also implies lower/specific referents such as pine trees or cherry trees, due to the same reason, that is, non-exclusion. With regard to the higher referents, the word produces a certainty (PS, 5.35: niścaya), that is, a tree is certainly an earthen thing, a substance, and so on, whereas with respect to the lower referents it produces a doubt (saṃdeha), for example, regarding whether the tree is a pine tree or a cherry tree. In other words, the word “tree,” according to the apoha theory, neither adopts (PS, 5.26: nopāttaḥ) cherry nor abandons it (PS, 5.26: nojjhitaḥ); rather it is indifferent to both possibilities (PSV, ad 5.25cd: upekṣate) (Figure 8.5).45 These linguistic phenomena cannot be explained by the tadvat theory of strong relation and denotation. Why not? As already seen in the case of the word “existent,” the word “tree” would have a relation only with something that has treeness and not with anything beyond that. It would cover neither the higher referents nor the lower referents due to the limitation of the strong relation. Therefore, the proponents of the theory cannot explain why one has a doubt such as “cherry or pine?” when hearing “tree.”46 If they insist that the word “tree” produces a doubt with regard to lower referents, then it should similarly produce a doubt with regard to higher referents.47 With regard to higher referents, however, one has certainty.

9 Conclusion Dignāga regards the linguistic relation between a word and its referent as equal to the inferential relation between an inferential reason and its object, such as being produced (kṛtakatva) and being transient (anityatva). In other words,

176

Kei Kataoka

the acquisition of a semantic relation is considered parallel to the learning of invariable concomitance in inference. This comes from Dignāga’s claim that verbal cognition is not a valid source of knowledge separate from inference. Therefore, verbal cognition can be reduced to inference. The establishment of the invariable concomitance in inference and the learning of a linguistic relation are based solely on a concurrent absence (vyatireka) that is determined by non-perception (adarśana/anupalabdhi). Just as one has never seen smoke without fire, one has never seen the word “cow” legitimately applied to non-cows. Concurrent absence is the foundation for establishing the universal relation applicable to all instances of a word and its referents. Concurrent existence (anvaya), by contrast, cannot cover all instances and thus cannot be the basis for establishing an undeviating, universal relation. For the reasons that follow, only the apoha theory can properly explain various linguistic phenomena: ●●

●●

●●

●●

●●

●●

It is free from the impossibility of connection that arises due to an infinite number of instances. It is free from the uncertainty caused by deviation. It is free from the fault of dependence, such as the necessity of intervention by a real universal. It is free from the fault of being a secondary expression. It can explain co-referential expressions such as “an existent substance,” “an existing pot,” “a blue water lily,” and so on, in which two words refer to the same referent. It can explain the fact that a listener has certainty with regard to higher and general referents and an uncertainty with regard to lower and specific referents. For example, when hearing “tree,” one is sure that it is an earthen thing, a substance, an existent thing, and so on, and one is not sure about whether it is a pine tree or a cherry tree.

The realists presuppose a positive strong relation between a word and its referent and accordingly a strong denotation, whereas Dignāga presupposes a negative weak relation and accordingly a weak denotation that is of the nature of non-exclusion. A word covers/includes/implies a referent because the referent in question is not excluded. In other words, non-exclusion (anapohana/apratikṣepa) is the real nature of denotation for Dignāga. This is possible because the referent x, that is, things (bhāvāḥ) qualified by the exclusion of other things (arthāntaraviśiṣṭāḥ), is a complement excluded from all non-x-s.

Dignāga on Relation

177

Notes 1 Based on a newly discovered manuscript of Jinendrabuddhi’s commentary on Dignāga’s PSV, two Tibetan translations of the PSV, and parallel passages found in later works, Pind (2015) (originally a doctoral dissertation submitted to Universität Wien in 2009) reconstructs the original Sanskrit text of the PSV as far as possible and translates it with detailed notes. Pind’s study is the main source for my study of the fifth chapter of Dignāga’s PSV. 2 The Mahāvibhāṣā (Taishō, 1545, 27, 865c6–7) emphasizes that all names and concepts are arbitrarily established and therefore necessarily untrue. The Yogācārabhūmi (Taishō, 1579, 30, 362c19–20) refers to linguistic referents as tentatively established by mundane convention. It is also to be noted that the Yogācāra notion of prajñaptisat is parallel to Bhartṛhari’s notion of upacārasat. 3 Ogawa 2010, 420, n. 20: “It is to be noted that the word called viśeṣa-śabda denotes an entity excluded from others (vyāvṛttârthâbhidhāyin), referring to the exclusion (vyāvṛtti), see Vākyapadīya, 3.5.4cd: viśeṣa-śabdair ucyante vyāvṛttârthâbhidhāyibhiḥ /” 4 Cf. PSV, ad 5.35: na ca saṃbandhadvāraṃ muktvā śabdasya liṅgasya vā svārthakhyāpanaśaktir asti. Pind (2015, II 127): “And without the means of connection the word or the inferential indicator is incapable of indicating its own referent.” 5 PSV, 5.37b: lokarūḍho na mṛśyate. Pind (2015, II 142, PSV, ad 5.37b): tasmād asmābhir api lokavyavahārā naimittikā vā pāribhāṣikā vā bhūtārthatvena na mṛśyante, lokavad evānugamyante. siddhaś ca rūpaśabdo loke nīlādiṣv eva, na rasādiṣu (Pind 2015, II 142–143). 6 Dignāga accepts only two types of pramāṇas, that is, perception and inference, just like the Vaiśeṣikas. Verbal testimony is included in inference. Cf. ŚV, śabda 15ab: tatrānumānam evedaṃ bauddhair vaiśeṣikaiḥ śritam / 7 The parallel structure between a word and an inferential mark is criticized by Kumārila in ŚV, śabda 68cd–98, in particular regarding the three conditions of an inferential reason, when applied to a word: pakṣadharmatva (68cd–84), anvaya (85–95), and vyatireka (96). 8 Kumārila argues (ŚV, apoha 78–79) that neither perception nor inference can be the means for grasping the two relata of verbal cognition. 9 PSV, ad 5.50c: saṃbandho hi panasaśabdārthau pramāṇāntareṇopalabhya “asyāyam” iti manasā kalpyate, anumānānumeyasaṃbandhavat. tato na śābdaṃ pramāṇāntaram. Cf. the translation by Pind (2015, II 188). 10 PSV, ad 43b: na tu karkādīn pratyekam apohate (Cf. Kataoka 2016, 876–79). 11 Cf. Pind (2015, II 147–148, n. 499). 12 PS, 5.11d: anyāpohakṛc chrutiḥ.

178

Kei Kataoka

13 PS, 5.1: tathā hi saḥ … svārtham anyāpohena bhāṣate; PSV, ad 5.1: śabdo … taṃ … arthāntaravyavacchedena dyotayati. 14 PSV, ad 5.36d: śabdo ’rthāntaranivṛttiviśiṣṭān eva bhāvān āha. 15 PS, ad 5.38d: arthāntarāpohaviśiṣṭe ’rthe. 16 For Mādhava’s criticism of Dignāga’s theory of apoha, see, for example, his criticism regarding the first cognition (ādyapratyaya). Dignāga accepts this fault as non-problematic on the ground of the assumption that the mundane world has no beginning (PS, 5.43a: anāditvāt) (See Pind 2015, II 157–58). 17 Kumārila points out the fault of mutual reliance between a cow and a non-cow in ŚV, apoha 83–85ab. 18 PS, 5.1: na pramāṇāntaraṃ śābdam anumānāt. tathā hi saḥ / kṛtakatvādivat svārtham anyāpohena bhāṣate // 19 PSV, ad 5.2b: na cākṛtasaṃbandhe śabde ’rthābhidhānaṃ yuktam, svarūpamātrapratīteḥ. 20 Concurrent existence and concurrent absence can be the means by which a word denotes a referent. See PSV, ad 5.34: anvayavyatirekau hi śabdasyārthābhidhāne dvāram. 21 sapakṣe sadbhāvaḥ; tulye vṛttiḥ. 22 sādhyābhāve hetor nāstitā; atulye ’vṛttiḥ. 23 Both cases, that is, tulye vṛttiḥ and atulya ’vṛttiḥ are also mentioned in the Apoha chapter. PS, ad 5.34: tau [= anvayavyatirekau] ca tulyātulyayor vṛttyavṛttī. 24 However, at least one instance of concurrent existence (anvaya) is considered necessary, as stated in PS, 5.34ab: svārthasyāṃśe ’pi darśanāt. But this is because Dignāga has to exclude exceptional cases, classified as asādhāraṇānaikāntika, such as śṛāvaṇatva, a type of pseudo-reason that fulfills vyatireka but not anvaya. It is clearly not the case that a single instance of anvaya contributes to establishing invariable concomitance (vyāpti). For details, see Kataoka (2015). Furthermore, when Dignāga states that at least a single instance of concurrent existence is necessary, he does not mean concurrent existence between “cow” and a cow, but between two anyāpohaviśiṣṭas, that is, agośabdāpohaviśiṣṭa and ago ’pohaviśiṣṭa. For Dignāga, who does not accept the existence of real universals, anvaya and vyatireka are not substantially different from each other. Cf. PSV, ad 5.38d (Pind 2015, I 49, 7–8): jātivyatirekeṇa tu … arthāntarāpohaviśiṣṭe ’rthe śabdasyānvayavyatirekau na bhinnārthau. Pind (2015, II 150): “But if the referent is qualified by the exclusion of other referents without the general property, the word’s joint presence and joint absence do not have different referents.” 25 PSV, ad 5.34: tatra tu tulye nāvaśyaṃ sarvatra vṛttir ākhyeyā, kvacit, ānantye ’rthasyākhyānāsaṃbhavāt. “Among the two, however, occurrence in every similar case cannot necessarily be stated, rather [it can be stated] only in some cases, because it is impossible to state [it in every case] as the referent is infinite.”

Dignāga on Relation

179

26 PSV, ad 5.34: atulye tu saty apy ānantye śakyam adarśanamātreṇāvṛtter ākhyānam. “Absence in a dissimilar case, however, can be stated merely by nonperception despite the infinitude of cases.” 27 PSV, ad 5.35: vyatirekamukhenaivānumānam. 28 Kumārila criticizes Dignāga’s adarśanamātra in ŚV, apoha 75. 29 Cf. Kataoka, forthcoming. 30 For the parallel structure between the inferential reason “horned-ness” (viṣāṇitva) and a word, see Kataoka (2016). 31 PSV, ad 36c: bhāktadoṣo ’pi nāsti. 32 PS, 5.4a: asvatantratva, that is, pāratantrya. 33 PS, 5.4b: upacāra. 34 PS, ad 5.4a: na hy asatyāṃ vyāptau sāmānādhikaraṇyabhāvaḥ. 35 Dignāga does not take into consideration the distinction between substance-ness and substance-ness-possessor probably to avoid complication. 36 PSV, ad 5.4a: svābh​idhey​aguṇa​mātra​viśiṣ​ṭadra​vyābh​idhān​āt saty api dravye madhurādīn nākṣipati. 37 PSV, ad 5.36c: pāratantryeṇa svabhedānākṣepadoṣaḥ. 38 PS, 5.11c: arthākṣepe ’py anekāntaḥ. 39 PSV, ad 11c: iha punaḥ sad ity ukte na ghaṭādiṣu niścaya iti saṃśaye sati nāsty arthākṣepaḥ. In Dignāga’s usage, ā-kṣip can be paraphrased with vy-āp as attested in PSV, ad 5.4a. 40 Dignāga omits introducing the distinction between the exclusion of non-substance and its possessor probably for the sake of clarity. 41 PSV, ad 5.18b: yat tarhīdam uktam “antarbhūtaviśeṣaṃ sāmānyam” iti. naitad uktam abhidheyatvāt, kiṃ tarhi—“anapohanāt” (PS, 5.18b). 42 Jinendrabuddhi’s explanation (Pind 2015, II 2, n. 7) regarding Dignāga’s choice of dyotayati to gloss bhāṣate in the verse (PS 5.1) is different. Jinendrabuddhi says nothing about the nuance of dyotayati. 43 PSV, ad 5.1: śabdo … taṃ kṛtakatvādivad arthāntaravyavacchedena dyotayati. 44 PSV, ad 5.35: vṛkṣaśabdo ’pārthivādiṣu na dṛṣṭaḥ. 45 For the hierarchy of universals, see Katsura (1979). 46 PSV, ad 5.35: anvayadvāreṇa cānumāne vṛkṣaśabdād ekasmin vastuni śiṃśapādyābhāsaḥ saṃśayo na syāt. “Furthermore, if one inferred through [the positive method of] coexistence, from [hearing] the word ‘tree’ one would not have a doubt in which a North Indian Rosewood and so on appear with respect to one and the same entity.” 47 PSV, ad 5.35: tatsaṃśayavat pārthivatvadravyatvādyābhāso ’pi saṃśayaḥ syāt. “One should also have a doubt in which earthen-ness, substance-ness, and so on appear, just like a doubt [in which, for example, a North Indian Rosewood appears].”

180

Kei Kataoka

Primary Sources PS

Pind, Ole Hoten. Dignāga’s Philosophy of Language: Pramāṇasamuccayavṛtti V on Anyāpoha. Part 1 and Part 2. In Pind 2015.

PSV

Pind, Ole Hoten. Dignāga’s Philosophy of Language: Pramāṇasamuccayavṛtti V on Anyāpoha. Part 1 and Part 2. In Pind 2015.

ŚV

Dvarikadasa Shastri, ed. 1978. Ślokavārttika of Śrī Kumārila Bhaṭṭa with the Commentary Nyāyaratnākara of Śrī Pārthasārathi Miśra. Prachyabharati Series 10. Varanasi: Tara Publications.

Taishō

“Taishō Shinshū Daizōkyō or The Taishō-era Revised Tripiṭaka, 1924–1934.” 1924–34. Available online in the SAT Taishō Shinshū Daizōkyō Text Database. http:​//21d​zk.l.​u-tok​yo.ac​.jp/S​ AT/in​dex_e​n.htm​l.

Vākyapadīya

Rau, Wilhelm, ed. 1977. Bhartṛharis Vākyapadīya, die mūlakārikās nach den Handschriften herausgegeben und mit einem pāda-Index versehen. Wiesbaden: Franz Steiner Verlag.

Secondary Sources Kataoka, Kei. 2015. “Apoha no Henjū Haaku—Dignāga to Kumārila.” Journal of Indian and Buddhist Studies (Indogaku Bukkyogaku Kenkyu) 64 (1): 450(75)– 444(81). Kataoka, Kei. 2016. “Horns in Dignāga’s Theory of Apoha.” Journal of Indian Philosophy 44 (5): 867–82. Kataoka, Kei. Forthcoming. “Adarśanamātra and Utprekṣā: A Study of a Bṛhaṭṭīkā Fragment.” In Transmission and Tradition: Proceedings of the Matsumoto Conference on the Meaning and the Role of “Fragments” in Indian Philosophy, Shinshu University, August 20–24, 2012, ed. by Hiroshi Marui and Ernst Prets. Katsura, Shōryū. 1979. “The Apoha Theory of Dignāga.” Journal of Indian and Buddhist Studies (Indogaku Bukkyogaku Kenkyu) 28 (1): 493(16)–89(20). Ogawa, Hideyo. 2010. “Bhartṛhari on Unnameable Things.” In Logic and Belief in Indian Philosophy, ed. by Piotr Balcerowicz, 403–18. Warsaw Indological Studies 3. Delhi: Motilal Banarsidass. Pind, Ole Hoten. 2015. Dignāga’s Philosophy of Language: Pramāṇasamuccayavṛtti V on Anyāpoha. Part 1 and Part 2. Beiträge zur Kultur-und Geistesgeschichte Asiens 92. Wien: Verlag der Österreichischen Akademie der Wissenschaften.

9

Meanings of Words and Sentences in Mīmāṃsā1 Elisa Freschi

1  General Problem: What Is Meaning? What Expresses It? What is meaning? What is the carrier of meaning? In other words, how can one identify the linguistically relevant elements on the side of meaning and on the side of the expressive factors? The typical candidates for the latter role are phonemes, individual morphemes or words, whole sentences, or the indivisible sphoṭa. As for meaning, different schools identify it as corresponding to the exclusion of everything else, to an entity in the external world, to a complex state of affairs (so the Bhāṭṭa interpretation of worldly language and the Vedānta one, see below), to a duty (so the Prābhākaras). One view influences the other, since if one identifies the meaning conveyed in linguistic communication as a duty, one needs to take into account the perspective of a whole sentence, whereas other theories may favor a more atomistic approach. In this contribution I will analyze the Mīmāṃsā answer to these questions, and I will further discuss the sentence meaning, specifically as expressing something to be done, in Chapter 15 of this volume. A short terminological premiss is also needed: Indian authors developed sophisticated theories of language, which need at times to be understood in their own terms, since there is not always a corresponding term in English (as well as in any other European language). One such case is the opposition between signifier and signified, usually expressed by the pair śabda and artha. The former cannot be translated as “word,” since part of the focus of Indian theories is indeed on the identification of what is the signifier element: it can be identified with varṇa (phoneme),2 pada (word), vākya (sentence), etc. Accordingly,

182

Elisa Freschi

here I will translate śabda as “linguistic expression” or “linguistic unit,” by that indicating a linguistic element capable of (contributing to) conveying a meaning. A similar discussion could be applied to the term artha, which I will translate as “meaning,” in order to designate what is meant by a linguistic expression, leaving it open for the various schools to identify it, for instance, as an external referent or an internal idea.3 It remains firmly established in each case that reflections on language are linked to reflections on epistemology and that, therefore, most schools see the linguistic artha as directly connected to the object of valid cognition (some schools, such as the Buddhist logicoepistemological one, disagree with this perspective, but the way they disagree with it presupposes the existence of this view as the mainstream one). In the debate reproduced in Section 2.1 below, for instance, Mīmāṃsā authors will take it for granted that they can use the same word, namely artha, for both the contents conveyed by linguistic communication and the ones conveyed by sense perception and inference.

1.1  Premisses Shared by All Mīmāṃsakas The Mīmāṃsā school recognizes as its basic text Jaimini’s Pūrvamīmāṃsāsūtra (henceforth PMS), possibly composed in the last centuries BCE. The PMS has been commented upon several times, but Śabara’s Śābarabhāṣya (henceforth ŚBh), possibly composed in the fourth or fifth century CE, became the standard commentary. In the sixth or seventh century two authors, Kumārila Bhaṭṭa and Prabhākara Miśra, offered divergent sub-commentaries on the ŚBh, thus creating two sub-schools related to the one and the other interpretation and known as the Bhāṭṭa and the Prābhākara one, respectively. Kumārila dedicated to the first part of the ŚBh an extensive commentary in verses, the Ślokavārttika, to the intermediate part of the ŚBh a commentary in prose with intermixed verses, the Tantravārttika, and to the last part a shorter commentary in prose, the Ṭupṭīkā. A further treatise on the same topics of the Ślokavārttika, the Bṛhaṭṭīkā, is preserved only through quotations. The Ślokavārttika has been further commented upon by Uṃveka Bhaṭṭa (eighth century), Sucarita Miśra (tenth century), and Pārthasārathi Miśra (eleventh century). Prabhākara’s only extant work is his Bṛhatī, a commentary on the first part of the ŚBh. Prabhākara’s ideas have, therefore, often be reconstructed through his commentator Śālikanātha (eighth c.?), who is, however, a great thinker on his own right, so that one needs to carefully distinguish his innovations from what are or might have been Prabhākara’s central theses.4

Meanings of Words and Sentences in Mīmāṃsā

183

The initial concern of the Mīmāṃsā school is the exegesis of Vedic texts, and the Vedas remain therefore the paradigmatic case of language for many Mīmāṃsā authors. Nonetheless, they always consider worldly (laukika) usages of language along the Vedic ones. Mīmāṃsakas (i.e., Mīmāṃsā authors) of both the Bhāṭṭa and the Prābhākara sub-schools refute the idea of a sphoṭa carrying the meaning and being different from what we experience, namely phonemes and words, since this contradicts both the principle of parsimony and our common experience. Why should one refute her common experience in favor of a not-assessable sphoṭa? And why postulate an additional entity, the sphoṭa, when one already experiences phonemes and words and knows through experience and concomitant presence and absence that they positively and negatively correlate with the apprehension of a meaning? Accordingly, Mīmāṃsā authors claim that phonemes really exist and that they together constitute words. Mīmāṃsā authors were known as varṇavādins (upholders of phonemes) and this label is due to their insistence in recognizing the role of phonemes as opposed to larger linguistic units (such as words or sentences). What is then the ontological status of words? The answer may partly differ in the various Mīmāṃsā sub-schools. The grammarian Patañjali is possibly the main initiator of this reflection since he asks at the beginning of his Mahābhāṣya (see Candotti, Chapter 1 in this volume) “What is the linguistic unit (śabda) when we utter ‘cow’ (in Sanskrit gauḥ)?” The Mīmāṃsaka Śabara extensively reuses Patañjali (though without naming him) and seems to emphasize the role of phonemes only when he rephrases his well-known question (ŚBh ad PMS 1.1.5): Now, when we utter gauḥ, what is the linguistic unit? The revered Upavarṣa says:5 “The phonemes g, au and ḥ.” In fact, the expression “linguistic unit” is well-known as expressing that which is grasped by one’s hearing organ, and the phonemes are grasped by the hearing organ. [Objection:] If it were so, there could be no communication of the meaning, since no meaning is understood when one cognises the single phonemes, nor is there any collection apart from the phonemes which would communicate the meaning. In fact, when there is the g there is no au nor ḥ and when there are au and ḥ there is no g. Therefore, there is no linguistic expression “gauḥ” apart from the phonemes g etc., through which the meaning could be conveyed. […] [Reply:] The last phoneme, together with the traces of the previous ones communicates the meaning, therefore there is no flaw in our theory.6

184

Elisa Freschi

Note that Śabara does not deny that phonemes are the real linguistic units and that there is no additional collection on top of them. His definition of what we could call “word” is pretty minimalistic, namely “the last phoneme together with the traces (saṃskāras) left by the previous ones.” Does this mean that words do not exist at all and that one could reach the end of a sentence without the need of stopping at each word, with the last phoneme of the sentence—with the traces of the previous ones—designating the whole sentence meaning? This does not seem to have been Śabara’s theory, although it is discussed by other authors. In his commentary on PMS, 1.1.25 Śabara is quite clear about the fact that there are intermediate steps in the understanding of the meaning of a sentence, namely the word-meanings. This implies that one needs to stop at each word, which is therefore an independent reality, distinguishable from the flux of phonemes making a sentence: “In fact, there is no [communicative] power of the last phoneme of a sentence, independent of the word-meanings.”7 Moreover, Śabara refers to “word” (pada) several times in the ŚBh, in passages which are crucial to his linguistic theory and in which therefore the mention of “word” cannot be considered just a concession to worldly usage (ŚBh, ad PMS 1.1.24): “The word means the universal” (sāmānye padam) and “words have exhausted their function [in conveying word-meanings]” (padāni nivṛttavyāpārāṇi). Mīmāṃsā authors thus subscribe to the idea that words really exist. They also insist that words are intrinsically linked to their respective word-meanings, and thus refute the Bhartṛharian holism (for which see Saito, Chapter 4 in this volume), again because this idea is more parsimonious and because it is confirmed by common experience and it is a basic thesis of Mīmāṃsā that common experience should be trusted unless there is a valid reason not to. In fact, human beings commonly experience that one needs to understand the words composing a sentence in order to understand its meaning. Moreover, human beings also agree about the fact that words (and not complex texts only) are related to a distinct meaning. This experience cannot be denied in favor of a view focusing on the text as a whole and rejecting without compelling reasons our prima facie experience of words as meaningful units. Given that one can thus establish that words are meaningful, what exactly do they convey? And how do they convey it? As for the second question, Mīmāṃsā authors agree that the relation between a word as meaningful unit and its meaning is fixed (nitya), as it is proved by our common experience of language. As for the first question, mainstream Mīmāṃsā authors side with Śabara’s claim

Meanings of Words and Sentences in Mīmāṃsā

185

and hold—against Nyāya ones—that words convey universals (see again ŚBh, ad PMS 1.1.24, “the word conveys the universal”). This is, again, confirmed, by our common experience, in which words figure again and again, designating the same element recurring in several particular items, namely their underlying universal aspect. For instance, the word “cow” designates in every sentence in which it occurs the universal cowness, which is shared by all individual cows. The opposite thesis, namely, that words designated individuals, would lead to the paradox that no communication would be possible, since one would spend one’s entire life learning how each single individual entity is called or learning one by one all the individuals which can be referred to by a certain word. However, the thesis that words convey universals seems at first sight to imply that words would never be able to convey a complex state of affairs on their own accord, and would therefore be almost useless. Human language would be constituted only by extremely general statements about universals and, even more important for Mīmāṃsakas, no specific actions could be enjoined. In fact, each order presuppose a specification (one cannot bring the universal cowness, but only a particular cow; one cannot sacrifice the universal goatness, but only a particular goat). In order to solve this difficulty, Mīmāṃsakas claim that a complex state of affairs (viśiṣṭārtha in the Mīmāṃsā jargon) is conveyed by a sentence: “The sentence conveys the specific” (viśeṣe vākyam, ŚBh, ad PMS 1.1.24). This means that the sentence meaning is more than the sheer sum of word-meanings, insofar as at the level of sentence meaning one moves from one level (that of universals) to the other (that of specific meanings). This solution, however, leads to a further question, namely: How are these two different levels reached? Do the same words lead to the one and then to the next?

2  From Words to Sentence Meaning: Bhāṭṭas versus Prābhākaras 2.1  The Bhāṭṭa View: From Words, to Word-Meanings, to the Sentence Meaning The process of sentence-signification, leading from words to the sentence meaning, is distinctly explained by the two main Mīmāṃsā sub-schools, Bhāṭṭa Mīmāṃsā and Prābhākara Mīmāṃsā. Both sub-schools agree on the basic tenets seen so far, but they differ on the path leading from the words signifying universals to the sentence signifying a particular state of affairs. According to

186

Elisa Freschi

Bhāṭṭa Mīmāṃsā authors, words conclude their function in designating their own universal meanings. They ground this view in a statement by Śabara, describing words as “having concluded their function” (nivṛttavyāpārāṇi), after having conveyed their meaning (ŚBh, ad PMS 1.1.25). Thus, it is the word-meanings, conveyed by words, which convey the sentence meaning once connected together. In favor of this view Bhāṭṭa authors mention the fact that there is an invariable concomitance between understanding the meaning of a sentence and having understood the meanings of the words composing it. It is, in other words, not enough to hear the words composing a sentence; in order to understand the sentence meaning, one also needs to understand the individual word-meanings, which therefore need to have been designated by words before one goes on understanding the whole sentence meaning. Some Prābhākaras will reply, as will be discussed below, that this is true, but that the word-meanings do not need to be designated and can be only remembered. One might object that in the case of the Bhāṭṭa view the sentence meaning is no longer conveyed directly by words, but rather by their meanings and that it is therefore no longer, strictly speaking, linguistic. Prābhākaras claim that, in this way, the sentence meaning would no longer be conveyed through linguistic communication as a distinct instrument of knowledge. This might look like a sheer terminological problem, but for Mīmāṃsā authors it has a much deeper relevance. In fact, Mīmāṃsā authors explain that only the Vedas can convey knowledge of dharma (for more details, see Freschi, Chapter 15 in this volume). This means that any knowledge of dharma obtained through another source is invariably unreliable. Therefore, if the sentence meaning were not linguistic, then even the sentence meanings about dharma would no longer be directly conveyed by Vedic sentences, and would therefore end up being unreliable. Bhāṭṭa authors reply that the sentence meaning is even in their account still a function of words, although via their meanings. Bhāṭṭas, therefore, distinguish a direct designation (abhidhā) of words, through which universals (i.e., wordmeanings) are designated, and a secondary signification (lakṣaṇā), through which complex sentence meanings are conveyed. Secondary signification works once the word-meanings have been designated by words. The directly designated universals/word-meanings are blocked (bādha) and the words, through secondary signification, let word-meanings designate the sentence meaning (see McCrea, Chapter 18 in this volume). Although the sentence meaning is achieved only via the word-meanings, still it can be considered as the result of words. Citing a simile found in Kumārila’s Ślokavārttika, chapter

Meanings of Words and Sentences in Mīmāṃsā

187

on sentences, the process works just like cooking, which is the result of wood logs, although only via their burning. Prābhākara authors object in three ways: (1) They claim that secondary signification should be possible only once the direct designation is impossible, for instance, in the case of “the village on the Ganges,” one comes to understand that the village is on the Ganges’s bank because the primary meaning would be impossible. But what exactly is incongruous in the connected sentence meaning? And how could one dare claiming that one needs to resort to secondary signification for each single sentence? (2) How do word-meanings connect to each other? If they do because the words bestow into them the capacity to connect to each other, then it is more economical to just postulate that the words themselves convey the sentence meaning, without the intermediate step of the word-meanings. In fact, the Bhāṭṭa theory implies the need to postulate word-meanings (an additional entity) and to postulate that words give them the ability to connect (an additional potency). Thus, it would be more economical to postulate just one additional potency of words, without postulating any additional entity. (3) If word-meanings can automatically connect among themselves, then why don’t they do it unless once in a sentence? (In this connection it is important to remember that “artha” means both a linguistically conveyed meaning and a cognitively acquired one (see above, Section 1).) Why is it not the case, for instance, that the arthas we acquire through senseperception automatically connect into a sentence meaning? A plausible answer to (1) would point to the fact that the connection of various universals leads, in fact, to an impossibility, since, for instance, in the case of “Bring a cow!” one cannot bring the universal cowness. One might also suggest that lakṣaṇā in the Bhāṭṭa account acquires a technical meaning, different from the one it assumes in accounts of, for example, implicature, so that it does not need to be as exceptional as secondary signification in cases of implicature. As for (3) and (2), Kumārila Bhaṭṭa answers that word-meanings do, in fact, connect automatically and that this does actually occur even outside of sentences. The example Kumārila mentions will be discussed by generations of authors and will remain the only one discussed in this connection: A person sees an indistinct white shape, hears a neighing, and perceives the sound of hooves. These three unconnected arthas automatically connect into the complex

188

Elisa Freschi

meaning “A white horse is running.” The important thing to note here is that the three arthas (whiteness, sound of hooves, and neighing) connect to each other automatically, without the need of additional efforts on the part of the knower. Therefore, claim the Bhāṭṭas, there is no need of an additional activity on the part of the knower to move from the level of the word-meanings to that of the sentence meaning. This argument is used by the Bhāṭṭas against the Prābhākaras to claim that there is no need to postulate a specific potency of words toward the sentence meaning, since the single potency leading to the word-meaning is enough.8 By contrast, Prābhākara authors will discuss the white horse example by saying that there the knower infers a horse and connects it with the other data he derives from sense perception. Thus, the Prābhākara concludes that there is no automatic connection of the word-meanings just by themselves. In contemporary terms, one might distinguish with François Recanati (Recanati 2004) a lexical meaning (the one found in dictionaries), a prototypical meaning (the default one), and the contextual one (the one the word conveys in the context of a sentence). Mīmāṃsā authors don’t necessarily distinguish between the first two, possibly because the role of dictionaries in precolonial India was different than the one they covered in Europe (dictionaries were meant to expand on semantic connections among words, not to fix each word’s meaning and giving “the last word on words” (“History of the Oxford English Dictionary,” see also Lynch 2006, 48 and Tomkins 1984, 249, 258)). Moreover, they add to the scheme the idea of words signifying universals, a thesis linguists like Recanati don’t even take into account. Accordingly, the Mīmāṃsaka idea of a sheer word-meaning (kevalapadārtha) is closer to the first two categories of Recanati, whereas the sentence meaning is the result of the combination of the contextual word-meanings achieved through fitness and syntactical expectancy (see Section 3). In other words, according to Kumārila words convey first (through direct designation, abhidhā) their own meanings and then secondarily delimit them to a specified sentence meaning. Recanati speaks in this context of the restriction to one contextual meaning among the many possible lexical meanings; Kumārila speaks also of a restriction leading from universals to a particularized sentence meaning. Word-meanings, once connected, designate a particularized sentence meaning which cannot be identical to the sum of their universal meanings. In this volume, McCrea (Chapter 18 in this volume) rightly highlights in this connection that the universal meaning of the unconnected word-meanings needs to be abandoned in order to move to the specific sentence meaning.

Meanings of Words and Sentences in Mīmāṃsā

189

2.2  The Prābhākara View: From Words to the Sentence Meaning By contrast, Prābhākara Mīmāṃsā authors, and especially Prabhākara’s main commentator, Śālikanātha, state that words first get connected and then designate the specified sentence meaning only once connected. This assures that the sentence meaning can be said to be linguistically conveyed, since there is not the intermediary step of word-meanings, a conclusion which is very important for the Mīmāṃsā epistemology, regarding linguistic communication as a distinct instrument of knowledge (see also Freschi, Chapter 15 in this volume). However, this explanation altogether skips the role of word-meanings. Thus, Prābhākara authors have to explain the fact that the own meanings of single words appear to do have a role to play in the process, since there is an invariable concomitance between knowing the words’ individual meanings and knowing the sentence’s one. At the same time, Prābhākara authors don’t want to avoid the Scylla of atomic word-meanings only to end up in the Charybdis of sentenceholism à la Bhartṛhari. This tension between the opposing risks of atomism and holism is dealt with differently by various authors of the Prābhākara school. Prabhākara himself seems to present the most basic version of the theory, where individual word-meanings just don’t play a role in the apprehension of the sentence meaning. Words do contribute individually to the sentence meaning, but cannot designate their own meanings before the sentence meaning or independently of it. The theory seems in this sense to be a form of radical contextualism. Some forms of contextualism are so much shared by all users of language that they are not even highlighted, for instance, in the sentence “I am going to bring your car to the park,” it is impossible to determine whether the word “your” means belonging to the single person the utterer is talking to or belonging to the people the utterer is talking to, not to speak of the meaning of units such as “to” or “the.” Prabhākara seems to have gone one step further, claiming that in fact we don’t even know what “car” means until the whole sentence meaning has been grasped. And it would indeed be easy to think of other sentences in which “car” as well as any other word appears to convey a different meaning, for instance, “Each car in the wheel of the local amusementpark was painted in green.” Prabhākara does not say that in such cases the same word conveys two different meanings, but rather that it contributes to two different sentence meanings. This also means that the problem of “car” is not one of polysemy (it would be relatively easy to learn by heart the multiple meanings of a restricted set of words). Rather, the problem is that although “car” does have a meaning (the opposite theory would lead to the sphoṭa holism), it

190

Elisa Freschi

does not convey it unless in the specific context of a particularized sentence utterance. The Nyāya philosopher and historian of philosophy Bhaṭṭa Jayanta (ninth century) reconstructs the Prābhākara view as entailing that words would be able to designate their own meanings if used separately, but they never are. Moreover, even when they are used together to convey the sentence meaning, Jayanta claims, one can still recognize their contribution to this general purpose, so that the Prābhākara thesis is clearly distinct from the one of the upholder of the sphoṭa. Jayanta compares the contribution of single words once connected to that of the parts of a wagon to the single purpose of carrying goods: one by one, they would not be able to achieve anything, but put together they can co-operate to the general purpose by delivering their own individual contribution. By contrast, Śālikanātha and the Prābhakara opponent embedded in the commentary on Kumārila’s Ślokavārttika by Sucarita (who deeply depends on Śālikanātha, but inserts a few innovations) start discussing the role that the memory of the individual word-meaning plays in the process. Words would accordingly cause one to remember their own meanings, then get related to one another, and then designate the complex sentence meaning. The individual word-meanings would therefore be recollected, but not designated by words. What do these individual, disconnected word-meanings amount to? According to the pan-Mīmāṃsā view of word-meanings, one remembers a universal. This is different than the role words will have in the sentence, since the sentence meaning is always specified and therefore particular. Words, once connected, therefore designate a particularized sentence meaning which cannot be identical to the sum of their universal meanings. The Prābhākara view as reformulated by Śālikanātha comes clearly closer to the Bhāṭṭa one, insofar as it also recognizes that word-meanings have a role in the process of linguistic communication. The main difference between the two views at that point seems to be that words cause one to recollect their individual word-meanings, whereas they all together designate the sentence meaning. This might seem like too much a concession to the Bhāṭṭa view, but Śālikanātha might have considered it needed, because of the following reasons (more on each one below): (1) Without the intermediate step of word-meanings, one would be forced to admit that the process of language learning (vyutpatti) would link sentences and sentence meanings, which is not acceptable since it would be noneconomic and would be falsified by the case of one’s understanding of new sentences.

Meanings of Words and Sentences in Mīmāṃsā

191

(2) Bhāṭṭa authors objected that the Prābhākara refusal to acknowledge the intermediate step of the designation of word-meanings led to either circularity or redundancy. (3) It would be difficult to speak of semantic fitness (see Section 3) as leading the process from words to the sentence meaning unless one has some preliminary understanding of the word-meanings. (4) The Mīmāṃsā textual hermeneutics is based on the application of rules (nyāyas), which need to be applied to word-meanings. Most importantly, (4) concerns a specific hermeneutic device of Mīmāṃsā, namely, its hermeneutic rules (nyāyas).9 Through such rules one can identify, for instance, what is the topic and what is the comment of a given sentence, what is meant and what is said without meaning it, what is principal and what is subordinate. Before the application of these rules, the specific sentence meaning cannot be ascertained. For instance, a sentence such as “He should take a pill” could mean that a specific male person should take a pill or that a generic person should do that. It could also mean that one should take a pill (and not a liquid), or that one should take one pill (and not many). Thus, nyāyas are absolutely needed to understand a sentence, and not just in the Veda, but even in worldly communication. However, the rules need to be applied to some preliminary understanding of the word-meanings, since the sheer words would not be enough. It is also noteworthy that Prābhākara authors implicitly reject the idea of first understanding a given sentence meaning and then applying to it the nyāyas, and change the first meaning to the definitive one. The sentence meaning one understands needs to be the definitive one, and not an occasion for trials and errors.10 As for (2), Sucarita objects to the Prābhākara reconstruction that out of a heap of words one could never get a sentence meaning. If the first word depended on the second to express its meaning and vice versa, there would be mutual dependency. If, by contrast, the sentence meaning were already completely present in the first word, which would already express its meaning as connected with all other words, then these would be redundant. The Prābhākara replies by pointing out that one does not start randomly from the first word of a sentence. Rather, one starts with the word expressing a duty (kārya), which is identified by typical exhortative suffixes such as the optative (liṇ), the imperative (loṭ), and the gerundive (kṛtya) suffixes; one’s analysis is then led by the criteria of proximity, fitness, and syntactical expectancy (see Section 3). The application of semantic fitness, however, leads to the problem listed above (3).

192

Elisa Freschi

Concerning (1), Śālikanātha has also the merit of devising an argument which will have a long fortune in the history of philosophy of language in India, namely, the one about language learning. Śālikanātha explains that the way people learn language is an evidence in favor of the Prābhākara view. Language learning occurs, he explains, by immersion in the interactions among expert speakers. The scenario he imagines is that of a child looking at an older person giving orders to a younger one, who is older than herself. For instance, we might imagine a child looking at her grandfather giving orders to her father, or a child looking at her father giving orders to her older sibling. The child does not learn the meanings of the single words. Rather, she learns that to the sequence “Bringthecow!” corresponds to the action of bringing a given animal, and she infers that the sequence must impart a command having the bringing of the cow as content when she sees its repeated effect. Only through repeated instances of commands using the same words in different combinations (e.g., “Tie the cow!,” “Bring the horse!,” or “Milk the cow!”) the child can isolate the meaning of each single word.11 The process through which the child extracts the meaning of single words out of sentences she has understood is called extraction (niṣkarṣa). Niṣkarṣa is said to occur through concomitant absence and presence of a given word and its designated meaning in various sentences. For instance, a child hears sentences such as “Bring the cow!” and “Bring the horse!” and observes that whenever the sequence “cow” is present, the associated meaning is present and whenever the sequence “cow” is absent, its meaning is also absent. In summary, word-meanings need to have a role and they need to be remembered before the designation of the sentence meaning. However, the introduction of memory might lead to problems, since memory is generally believed not to be an instrument of knowledge (pramāṇa) by Indian philosophers. Śālikanātha stresses, therefore, that also in Kumārila’s theory memory is not avoidable, since one needs to remember the single word-meanings in order to combine them into the sentence meaning. Apart from the role of memory in remembering word-meanings, Śālikanātha stresses the role of word-meanings over the one attributed to them by Prabhākara through the introduction of the idea of a ground or reason (nimitta). In fact, Śālikanātha explains that through extraction (niṣkarṣa) one understands that a cow is the ground of the use of the word “cow.” One does not learn that a cow is the designated meaning, since the designation process occurs only within a sentence and isolated words cannot designate their own meanings. This idea of ground distinguished from the designated entity might seem preposterous, and

Meanings of Words and Sentences in Mīmāṃsā

193

Śālikanātha feels therefore the need to offer a noncontroversial example of a case of ground. This is the case of the word daṇḍin, which is formed out of the word daṇḍa (rod) and the possessive suffix “-in,” and literally means rod bearer, and consequently also king or ascetic. The word’s nimitta, that is, the reason for the usage of “rod bearer,” is a rod. However, the rod is not its meaning. The Bhāṭṭa objector voiced by Sucarita objects that this distinction between the ground and the designated meaning only suits the case of complex words, such as “rod bearer,” but would not work in the case of simple ones like “cow.” One might push the Bhāṭṭa objection even further and observe that the case of daṇḍin is actually one of a micro-sentence, since daṇḍin, in fact, designates a complex meaning, that is, “person endowed with a stick,” in a way which can be embedded in a single word only by the morphology of languages like Sanskrit.

2.3  The Problem of Metaphorical Expressions What happens in the case of secondary signification? Kumārila and Śālikanātha distinguish in this connection between lakṣaṇā and gauṇavṛtti.12 The first expresses a secondary signification in which a word does not signify its usual meaning, but something connected with it, for example, “cradle” for baby or “Ganges” for bank of the Ganges. The second expresses a secondary signification in which a word does not signify its usual meaning, but rather something else sharing the same qualities, for example, “lion” for brave or “fire” for lively. How can this process make sense within the two competing Mīmāṃsā theories of compositionality, the Bhāṭṭa theory of abhihitānvaya and the Prābhākara’s anvitābhidhāna? In the case of Kumārila’s version of the Bhāṭṭa theory, there would be a double lakṣaṇā. Let me take the example of “There is a hamlet on the Ganges.” According to Kumārila, first, each word designates its own meaning, then these meanings get connected after a blockage (bādha) of their own meanings—since their own meanings are universals and would not make sense once connected— and through the first secondary signification. By means of this secondary signification, one switches from the designation of the universal Ganges and the universal hamlet to the particularized sentence meaning, which concerns one specific hamlet on the Ganges. At this point the listener realizes that a hamlet cannot be on the Ganges, a second blockage occurs, and she thus moves from the meaning on the Ganges to the secondarily signified meaning on the Ganges’s bank. The same applies to the case of gauṇavṛtti.

194

Elisa Freschi

Śālikanātha clearly states that only words expressing their own meaning, which he calls “meaning directly expressed” (śrautārtha) convey the connected sentence meaning. Secondarily used words cannot do that. Rather, they merely cause one to recollect their own meanings (as described above, Section 2.2), which, in turn, evoke a suitable connected meaning. Accordingly, in the sentence “the spears are approaching” the word “spears” just causes one to remember its isolated word-meaning, which then evokes something connected with spears—namely, people bearing spears—and suitable to be connected with “are approaching.” The phrase “are approaching,” by contrast, conveys a meaning which is connected with the meaning evoked by the word “spears” via its sheer meaning spears, as something connected with it (see Vākyārthamātṛkā, first pariccheda, ad verse 12, p. 405). This entails the obvious consequence that no sentence with just words having a secondary signification could be understood. One needs to start somewhere to make sense of a sentence and one cannot, maintains Śālikanātha, but start from words directly designating their meaning. Within this volume, Hugo David (Chapter 16) discusses the case of the Advaita Vedānta author Prakāśātman who, by contrast, modifies this theory in order to accommodate for sentences whose words are all to be metaphorically understood. Why this difference? Because Mīmāṃsā authors use common experience as their usual parameter, whereas Advaita Vedānta ones think of sentences speaking of the brahman, which by definition cannot be directly designated by words.

3  Proximity, Fitness, and Syntactical Expectancy According to the Bhāṭṭa authors, words get connected into a complex sentence meaning through proximity, semantic fitness, and syntactic expectancy. These three criteria correspond to the requirement of being uttered one after the other with no intervening time (unlike in the case of the words “a cow” and “runs,” pronounced on two different days), being semantically fit to connect (unlike the words “watering” and “with fire”) and being linkable through syntactic expectancy (as in the case of a verb and its arguments). It is in this connection noteworthy that the examples of expectancy always refer to syntax rather than semantics and typically have a verb expecting a complement or vice versa. The Prābhākaras also adopt the same criteria in order to rule the understanding of the sentence meaning out of the connected words and avoid the objections (mentioned above, in Section 2.2) about the fact that out of a

Meanings of Words and Sentences in Mīmāṃsā

195

random heap of words one would not know how to start to get to the sentence meaning. But what exactly do proximity, fitness, and syntactical expectancy refer to? The question arises because they seem to conflate different levels, insofar as fitness is necessarily semantic and therefore appears to refer to word-meanings, expectancy is syntactic and proximity seems to refer only to the proximity of the uttered words. However, in Sucarita’s discussion proximity is also attributed to the mental proximity of words or meanings. This can make it possible for the Prābhākara opponent voiced by Sucarita to get to the complex sentence meaning “The door is to be closed!” out of “The door!” only, since the phrase “is to be closed” is mentally proximate. This is needed also in order to explain how one-word sentences can designate a meaning, although, according to Prābhākaras, words designate a meaning only once connected. In fact, explains the Prābhākara opponent embedded in Sucarita’s text, even one-word sentences such as “pacati” ([s/he] cooks) designate a complex sentence meaning together with other words which are proximate in one’s mind, for example, “rice” or “pulses.” At this point, Sucarita has a powerful objection: how does one choose between the many possible counterparts which are evoked in one’s mind and the ones which are actually present in the sentence, given that they are all proximate? In other words, suppose I utter the sentence “He cooks rice.” If the word “cooks” evokes in the listener’s mind pulses and vegetables too, should they also be combined in a single sentence meaning? The answer is that they should not. One resorts to mental proximity only if there are no uttered words nearby. That is, mental proximity is only a second choice, and one first connects what is actually heard (or, a contemporary author might say, read). This preference for what is actually heard also explains away the problem in cases such as the following: 1. Bring the cow! 2. Bring the cow, the white one! Since both are correct sentences, syntactical expectancy must be fulfilled after “cow” in the first case, but not in the second. How is this possible? Just because “white” is not uttered in the first case, explains Śālikanātha in the Vākyārthamātṛkā. The mental aspect of proximity discussed above makes it possible for the Bhāṭṭas to interpret all three criteria as referring to meanings, whereas Prābhākaras would still need to understand at least fitness as referring to meanings, not words.

196

Elisa Freschi

3.1  Supplying Words and Supplying Word-Meanings Mental proximity might come close to the idea of the role of the context as highlighted in pragmatics and textual linguistics. Bhāṭṭas can accordingly add to the sentence meaning also word-meanings one obtains through contextual information. The typical examples delivered in this connection regard ambiguous or polysemic words, whose precise meaning is ascertained on the basis of the text’s contextual meaning. This is called tātparya (see Graheli 2016), a derivative from tatpara (aiming at). The words of a given text all aim at something, that is, the sentence meaning. The situation for Prābhākaras may be more complicated, since they should speak only of mental proximity of words, not of word-meanings, since wordmeanings are not designated before the sentence meaning. This position can work in the case of stereotyped sentences such as “Faster!.” A young girl going to school with her father will automatically add to it “walk,” since she has heard “Walk faster!” many times in the past. But what about more complex sentences, in which one does not know which exact word is needed, but is still able to supply the relevant word-meaning(s)? Mīmāṃsā authors discuss this topic in connection with the instrument of knowledge called arthāpatti. This is the instrument leading one from a given state of affairs, which seems to entail a contradiction, to its resolution, which was not previously known. The standard example is as follows. Devadatta (the Sanskrit version of John Smith), who is known to be alive and is usually always home, is not there. After an initial moment of puzzlement, one concludes that he must be outside. Bhāṭṭa authors claim that there is a variety of arthāpatti called śrutārthāpatti in which one supplies words, not meanings. The example they mention evokes Devadatta again and says: “The fat Devadatta does not eat during the day.” One does not just conclude with the thought that he must eat at night (Indian authors did not consider the possibility of metabolic diseases), but rather supplies the actual sentence “He eats at night.” At that point, the postulated sentence conveys its meaning. Prābhākara authors oppose this view and say that the knowledge that Devadatta eats at night is not obtained through language, but rather through the normal arthāpatti. In other words, one comes to know that Devadatta eats at night, but without the need to postulate the intermediate step of the corresponding sentence. As an evidence for that, they mention the fact that one could also postulate a sentence such as “He consumes food after sunset” and achieve the same result.

Meanings of Words and Sentences in Mīmāṃsā

197

Noteworthy here is the consensus about the fact that linguistic communication only covers what is conveyed by language (uttered or mentally supplied). Extralinguistic elements, such as one’s knowledge about the way metabolism works, are not included in the process of linguistic communication (but they may play a role in arthāpatti or in other instruments of knowledge). Linguistic communication (or śābda) is therefore restricted to language-based communication and cannot include such other elements. Therefore, unless there is a distinct śrutārthāpatti, we know that Devadatta eats at night through arthāpatti, not through linguistic communication.

4  Can We Adjudicate the Debate? The Bhāṭṭa and the Prābhākara views both appear to capture some significant insights about the way language works and it may be that they could both work as explanations of linguistic understanding in different situations. Basically, the idea of an understanding guided by the sentence itself (as in the Prābhākara view), without the intermediate step of the word-meanings, seems to reflect the linguistic apprehension of meanings out of short sentences, especially of usual or stereotypical ones. By contrast, readers of a second-language or dealing with a difficult and technical text might need to reflect about the meaning of each individual word before understanding the sentence meaning, as predicted by the Bhāṭṭa theory. The first case is the one which is clearly more present to Mīmāṃsā authors (and likely to Sanskrit authors in general). Śālikanātha (Vākyārthamātṛkā, first pariccheda, ad verse 12, p. 404) speaks in this case of sentences which have been repeated so often that they have become stereotyped (atyantābhyāsta). He opposes them to sentences about something unknown and, in a different context, to new sentences created by a poet. Śālikanātha also uses as an evidence for his theory the process of language learning (see above, Section 2.2), in which it is clear that the understanding of a connected sentence meaning precedes the knowledge of the individual word-meanings. Last, the adjustments of the Prābhākara theory proposed by Śālikanātha— the idea of a ground (nimitta) and the crucial role of memory (smṛti)—seem to weaken its intuitive core strength. Would this have been avoidable, given the powerful objections from the Bhāṭṭa side? Possibly not, and indeed also the Bhāṭṭa theory increased its degree of sophistication with time, although Kumārila was so systematic a thinker that he took care of the most part of it.

198

Elisa Freschi

It is in this connection also interesting to note that Recanati (2017) reproduces a discussion among the meaning of words and of sentences which evokes the Bhāṭṭa versus Prābhākara controversy, thus showing that the oscillation is due not just to the accidental creation of two sub-schools within Mīmāṃsā, but most probably also to the different nature of various events of linguistic communication.

4.1  A Further Extension of Prābhākara’s Theory How can Prābhākaras still claim that the three criteria of proximity, fitness, and expectancy lead one to get to the sentence meaning without the intermediate step of the word-meanings? In other words, if word-meanings do not play any role, how can fitness play a role? A possible way out is Śālikanātha’s suggestion that word-meanings, though not designated by words, are remembered by them. In this case, one might speak of fitness among the remembered word-meanings as leading one’s understanding of the designated sentence meaning. Alternatively, a contemporary Prābhākara might suggest that some preliminary understanding of word-meaning is immediately designated by each word, but that each new word adjusts the meaning of the previous one through the above mentioned criteria in a hermeneutic circle. This solution is not explicitly discussed, at least in the texts I am aware of, possibly because it implies a preliminary (and therefore epistemologically unsound) step within linguistic communication and could have therefore jeopardized the role of linguistic communication as an instrument of knowledge. The reason for upholding this view, by contrast, is that the distinction between the words causing one to remember (smār-) the individual word-meanings, and their designating (abhidhā-) the sentence meaning seems to be based only on a lexical technicality. Since the effect is the same, namely, that one becomes aware of a meaning, the reason for calling the former remembered and the latter designated seems to be only due to the anxiety of keeping the word “designation” for the sentence meaning, which appears something like an ad hoc explanation. It is more relevant, I think, to say that at the first stage the understanding of the individual word-meanings is not yet determined. For example, suppose one hears a child speaking of her horse. Until the end of the sentence one might be uncertain about whether she means a real horse or a toy, so that the word at stake, namely “aśva” designates its individual word-meaning (kevalapadārtha) only in a preliminary way. In contrast to that, by the time the whole text has been uttered, one can determine that an animal horse is running or that a toy horse is standing on a child’s bed.

Meanings of Words and Sentences in Mīmāṃsā

199

Notes 1 Research for this chapter has been supported by the WWTF project MA 16_028. I could read and discuss the texts upon which this chapter is based during several workshops organized at the IKGA (see https​://mi​mamsa​langu​age.w​ordpr​ess.c​om/ pa​ge/).​ I am grateful to Marco Lauri for improving the English. 2 There have been important discussions (e.g., in Deshpande 1997 and more recently in Dharurkar 2018) on the possibility to translate varṇa as phoneme. The main problem in this translation lies in the fact that the concept of phoneme has been historically developed in connection with the idea of a minimal pair, whereas this idea, even if present, is not determinant for the concept of varṇa. However, the translation of varṇa as speech sound seems to me more misleading, since Mīmāṃsā authors are extremely careful in distinguishing varṇas (unchangeable) and their phonic representations through sounds (variable according to the speaker). 3 For a thorough discussion on the meanings of artha, see Freschi and Keidan (2017). 4 For an overview on all the above dates, see Kataoka (2011, Introduction). 5 Upavarṣa is an early Mīmāṃsā author, whose works are lost, who is held in high esteem by Śabara. 6 ŚBh, ad PMS 1.1.5: atha gaur ity atra kaḥ śabdaḥ? gakāraukāravisarjanīyā iti bhagavān upavarṣaḥ. śrotragrahaṇe hy arthe loke śabdaśabdaḥ prasiddhaḥ. te ca śrotragrahaṇāḥ. yady evam, arthapratyayo nopapadyate. katham? ekaikākṣaravijñāne hy artho nopalabhyate. na cākṣaravyatirikto ’nyaḥ kaścid asti samudāyo nāma, yato ’rthapratipattiḥ syāt. yadā hi gakāraḥ, na tadaukārādivisarjanīyau. yadaukāravisarjanīyau, na tadā gakāraḥ. ato gakārādivyatirikto ’nyo gośabdo ’sti, yato ’rthapratipattiḥ syāt. antarhite ’pi śabde smaraṇād arthapratyaya iti cet, na. smṛter api kṣaṇikatvād akṣarais tulyatā. pūrvavarṇajanitasaṃskārasahito ’ntyo varṇaḥ pratyāyaka ity adoṣaḥ. 7 ŚBh, ad PMS 1.1.25: na hy anapekṣitapadārthasya vākyāntyavarṇasya […] śaktir asti. 8 Bhāṭṭa and Prābhākara authors will keep on counting the number of additional potency they are postulating for several generations (see, for example, Rāmānujācārya’s Tantrarahasya, chapter III). 9 On nyāyas, see Freschi (2018). 10 The topic is elaborated by Śālikanātha in Vākyārthamātṛkā, first pariccheda, ad verse 12, p. 404. 11 Further aspects of Śālikanātha’s argument, for example, its impact on the idea of the prescriptive nature of language, are discussed in this volume by Freschi, Chapter 15, and Kataoka, Chapter 14. 12 See Tantravārttika, ad 1.4.11–17, discussed in Keating (2017) and Vākyārthamātṛkā, first pariccheda, p. 405.

200

Elisa Freschi

Primary Sources Mīmāṃsādarśana

Abhyankar, Kashinath Vasudev, and Ganesh Shastri Ambadas Joshi, eds. 1970a–1974. Mīmāṃsādarśana. Anandasramasamskrtagranthavali 97. Pune: Anandasrama.

PMS

Pūrva Mīmāṃsā Sūtra of Jaimini. In Mīmāṃsādarśana.

Prakaraṇa Pañcikā

Subrahmanya Sastri, A. ed. 1961. Prakaraṇa Pañcikā of Śālikanātha Miśra with the Nyāya-Siddhi of Jaipuri Nārāyaṇa Bhaṭṭa. Benares: Benares Hindu University.

ŚBh

Śābarabhāṣya of Śabara. In Mīmāṃsādarśana.

Ślokavārttika

Dvarikadasa Shastri, ed. 1978. Ślokavārttika of Śrī Kumārila Bhaṭṭa with the Commentary Nyāyaratnākara of Śrī Pārthasārathi Miśra. Prachyabharati Series 10. Varanasi: Tara Publications.

Tantrarahasya

Rāmānujācārya. 1956. Tantrarahasya: A Primer of Prābhākara Mīmāṃsā. Critically edited with Introduction and Appendices, ed. by K.S. Rāmaswami Śāstri. Baroda: Oriental Institute.

Tantravārttika

Tantravārttika of Kumārila Bhaṭṭa. In Mīmāṃsādarśana.

Vākyārthamātṛkā

Vākyārthamātṛkā of Śālikanātha. In Prakaraṇa Pañcikā of Śālikanātha Miśra with the Nyāya-Siddhi of Jaipuri Nārāyaṇa Bhaṭṭa.

Secondary Sources Deshpande, Madhav Murlidhar. 1997. “Building Blocks or Useful Fictions: Changing View of Morphology in Ancient Indian Thought.” In India and Beyond: Aspects of Literature, Meaning, Ritual and Thought: Essays in Honour of Frits Staal, ed. by Dick van der Meij, 71–127. London: Kegan Paul International. Dharurkar, Chinmay Vijay. 2018. “Comparing Phoneme, Morpheme and Zero in Pāṇini and Modern Linguistics.” PhD thesis, Department of Humanities and Social Sciences. Indian Institute of Technology. Freschi, Elisa. 2018. “The Role of Paribhāṣās in Mīmāṃsā: Rational Rules of Textual Exegesis.” Ed. by Gianni Pellegrini, Asiatische Studien/Études Asiatiques 72 (2): 567–95. Freschi, Elisa, and Artemij Keidan. 2017. “Understanding a Philosophical Text: A Multi-Disciplinary Approach to the Problem of ‘Meaning’ in Jayanta’s Nyayamañjarī, book 5.” In Reading Bhaṭṭa Jayanta on Buddhist Nominalism, ed.

Meanings of Words and Sentences in Mīmāṃsā

201

by Patrick McAllister, 251–90. Wien: Verlag der Österreichischen Akademie der Wissenschaften. Graheli, Alessandro. 2016. “The Force of Tātparya: Bhaṭṭa Jayanta and Abhinavagupta.” In Around Abhinavagupta: Aspects of the Intellectual History of Kashmir from the 9th to the 11th Centuries, ed. by Eli Franco and Isabelle Ratié. Leipziger Studien zu Kultur und Geschichte Süd- und Zentralasiens 6. Proceedings of the Abhinavagupta conference, Leipzig 2013. Berlin: LIT Verlag. “History of the Oxford English Dictionary.” Visited on 09/03/2018. http:​//pub​lic.o​ed.co​ m/his​tory-​of-th​e-oed​/. Kataoka, Kei. 2011. Kumārila on Truth, Omniscience, and Killing. Part 2. An Annotated Translation of Mīmāṃsā-Ślokavārttika ad 1.1.2 (Codanāsūtra). Beiträge zur Kultur- und Geistesgeschichte Asiens 68. Wien: Verlag der Österreichischen Akademie der Wissenschaften. Keating, Malcolm. 2017. “Metonymy and Metaphor as Verbal Postulation: The Epistemic Status of Non-Literal Speech in Indian Philosophy.” Journal of World Philosophies 2 (1): 67–80. Lynch, Jack. 2006. “Disgraced by Miscarriage: Four and a Half Centuries of Lexicographical Belligerence.” The Journal of the Rutgers University Libraries 62: 35–50. Recanati, François. 2004. Literal Meaning. Cambridge: Cambridge University Press. Recanati, François. 2017. “Contextualism and Polysemy.” Dialectica 71 (3): 379–97. Tomkins, Sandra. 1984. “Noah Webster’s Conservative Radicalism. The Myth of the American Spelling Book.” McGill Journal of Education 19 (3): 249–61.

10

Śabdārtha as Sense or Reference: Dharmakīrti on Synonymy Patrick McAllister

The term “śabdārtha” is a very common compound in classical Indian theories of language. It is usually translated as “meaning,” “reference,” or “object” (artha) of a word (śabda). Certainly none of these translations is wrong. But what could puzzle a modern reader of these translations, especially if she has an acquaintance with work in language philosophy done after Frege, is how exactly these translations should be understood: Is śabdārtha what a word refers to or is it what a word means? And what might puzzle a translator of such Indian material into a modern context is how to even answer her question. In this chapter such a translator tries to answer that question in a way that makes sense within the theoretical framework of a particular historical context, namely, the Buddhist theory of what words mean (the apoha theory). First, however, a few words have to be said both about the method employed in the answer and about earlier answers.

1  Asking about Sense and Reference The question whether Frege’s famous distinction between sense (Sinn) and reference (Bedeutung) applies to theories of language developed in premodern India has been asked before.

1.1  Previous Answers The answers that have been given about the meaning of śabdārtha in the context of Indian language theories exhaust all possibilities with respect to the

Śabdārtha as Sense or Reference: Dharmakīrti on Synonymy

203

distinction of sense and reference: modern scholars have maintained that the term śabdārtha was used (a) for what a modern philosopher of language would call the reference of a word, (b) indiscriminately for both the reference and the sense of a word, or (c) for the sense of a word. These answers mainly depended on the ontological positions that the various Indian systems maintained about universals: realists would tend toward reference theories and nominalists toward sense theories, with some play for intermediate positions. This is, of course, a simplification, and the positions maintained by various scholars are often much more nuanced. A short overview of some characteristic positions is as follows: ●●

●●

●●

●●

●●

Staal (1969, 516): “The nominalistic tendency in Indian thought, which was clearly expressed in Buddhism, reacted against the realism that characterizes much of Hindu thought and against the reification that arises in Sanskrit even more easily than in many other languages.” He concludes that the Buddhists developed “a theory of meaning […] from which direct reference was systematically excluded” (Staal 1969, 516 sq.). Siderits (1991, 65): “It is generally accepted that Indian philosophers of language do not posit sense as a component of the meaning of an expression in addition to its reference. […] I shall nonetheless argue that, predominant though the reference theory might be in their tradition, certain Indian philosophers of language were still forced to recognize something akin to sense as a distinct element in meaning.” According to Siderits, one of these philosophers is Dharmakīrti (ca. mid-sixth century).1 Scharf (1996, 1, fn. 3) does not distinguish between sense and reference (connotation and denotation). Apparently, he thinks that that distinction would not help much for his discussion of how generic terms denote. Bronkhorst (1999a,b) explores the “correspondence principle” that many Indian thinkers of all denominations share, according to which there is a direct, one-to-one correspondence between every word in a statement and the matter of fact that the statement is about. Ganeri (1999) states that Naiyāyikas maintain, due to their position “as the primary proponents of realism in the Indian tradition,” “a ‘direct reference’ account of meaning, according to which the meanings of words are identified with real, external entities of some sort” (Ganeri 1999, 2 sq.).2 The Buddhist theory of apoha, on the other hand, is characterized by Ganeri (1999, 118), as “a non-referential theory of meaning for such expressions [i.e., nominals—PMA].” Ganeri (1999, 118, fn. 15) explains that the theory involves a “clear treatment of general terms as predicates.”

204 ●●

●●

Patrick McAllister

Patil (2003, 2009) speaks of “semantic value,” saying that this term “is supposed to be neutral to whatever it is that a semantic theory associates with the expressions of the language it interprets” (Patil 2003, 244, n. 3). Patil thus seeks to avoid a decision about the ontological status of a word’s meaning. He does, however, consider that compatible theories could posit an expression’s meaning to be either an external object or a mindinternal object (“a facet of awareness (buddhy-ākāra),” so Patil 2003, 230). Both such theories would be reference theories, and the question of an expression’s sense would remain unanswered. But since this interpretation is not primarily developed with regard to Dharmakīrti’s writings and its purpose is to not answer the question asked here, its possible applicability will not be further explored. Tillemans (2011), who notes that Apohavādins, like “many other Indian philosophers” (Tillemans 2011, 52), do not have to differentiate between the reference and the meaning of a term.

It is clear from this necessarily short, but hopefully not entirely unfair, overview that there is quite a variety of positions about what the reference of a word is according to the various theories of language developed in India, and even with regard to the apoha theory. It thus seems that the distinction of reference and sense would nicely correspond to whether a theory maintains realist or nominalist positions about universals. Many of the scholars who have given more or less explicit answers to our question were speaking about their understanding of the term śabdārtha in comparison to a very general idea of what Frege’s distinction is. Others, in particular Siderits and Ganeri, had a very clear concept of Frege’s distinction in mind when they gave their respective answers. Ganeri’s opinions are formed on the basis of Nyāya and Navya-Nyāya literature, and his treatment of the apoha theory is not extensively argued. It will therefore not be useful for our purposes to enter into a detailed discussion of the positions he takes. For our inquiry, the most immediately relevant discussion of śabdārtha is provided in the work of Mark Siderits. Siderits (1986, 1991) answered a very similar question to ours with an understanding of śabdārtha derived from Dharmakīrti’s “Pramāṇavārttikasvavṛtti” and Śāntarakṣita’s (ca. 725– 788) “Tattvasaṅgraha.” The difference between sense and reference used by Siderits for interpreting these texts is a generalized version of the distinction made by Frege: for Siderits, “sense” is the propositional element in a referring expression.3 Siderits came to the conclusion that Dharmakīrti introduced an element akin to Frege’s sense into “a tradition dominated by the reference

Śabdārtha as Sense or Reference: Dharmakīrti on Synonymy

205

theory” (Siderits 1991, 99). The core of Siderits’s argument for this assessment is his analysis of Dharmakīrti’s solution to the informativeness puzzle. Siderits considers the problem that Dharmakīrti solves to be analogous to the problem that “the morning star is the morning star” and “the morning star is the evening star” have different information value but are about the same things.4 Explaining the role of the two properties “being impermanent” and “being a product” in an inference that allows a person to know a thing’s impermanence from its being a product, Siderits (1991, 99 sq.) says: The problematic inference shows that the meaning of such a term [i.e., a predicate term—PMA] in a sentence is not just its demonstratum. In order for this inference to be informative, the sentences, “This is a product,” and “This is impermanent,” must somehow differ in propositional content when asserted of the same particular. The difference lies in the manner in which the particular is presented in the two sentences: in the first, by way of not belonging to the exclusion class formed by the paradigm image for “product”; in the second, by way of not belonging to the exclusion class for “impermanent.” […] To say of a term that part of its meaning is its sense is to say that the term presents its referent in a particular way, that understanding a use of the term involves thinking of the referent in a manner that involves propositional structure. This is just what Dharmakīrti is claiming. To understand the statement of identity “This is a product” is to grasp the particular denoted by “this” twice: once as just this unique particular that is different from every other particular; but also as a particular that is different from those particulars in the exclusion class formed by the paradigm for “product.” The first grasping is direct and nonpropositional; the second is a grasping of that particular by way of the sense of “product.”

What a Fregean sense, the way in which a referent is shown,5 would actually be on this interpretation is an other-exclusion.6 The way other-exclusion is understood by Siderits here is heavily dependent on Śāntarakṣita’s explanations (cf. Siderits 1991, 93, fn. 14). Since this way of understanding Dharmakīrti’s theory on this point is debatable, and since Siderits himself now considers this approach somewhat problematic (2011, 295 sq.), it will perhaps suffice to point out that this sense is interpreted both to have a propositional structure and to be a mode of showing a particular. This analysis by Siderits of sense as involving propositional structure can be understood in at least two ways: first, as equating a general term with a proposition, perhaps by considering that the general term is, for the Buddhists, essentially a predicate (“cow” as “is not not a cow,” a double negation or

206

Patrick McAllister

other-exclusion). But since this would be a difficult position to maintain, it will be better to focus on the second way of taking his explanations: a general term has a sense in the context of a proposition. This means that the word “cow” would not, outside of a sentence, have a sense. In other words, the sense of a general term like “cow” is something that becomes manifest, at least at first, in the sentences in which it occurs, for example, “This is a cow.” I will, in the following, take this to be the intention of Siderits’s proposal. To avoid terminological confusion, we should note here also that the assertion, in the last part of the long quotation above, that “this” allows us to grasp a referent directly and nonpropositionally, is problematic if taken as a historical position: for Dharmakīrti, it is evidently not the case that any kind of linguistic activity makes a particular, in the strict ontological sense, known. One strong reason for this is that if that could ever happen, all other means of valid cognition would be useless.7 But if “this” cannot refer to a particular, then what does it refer to? And if, as is to be expected, it refers to anyāpoha like all other concepts, then do we suddenly have two terms in a sentence playing different epistemological roles: a direct reference, on the one hand, and a reference in a representational mode, on the other? And do they both refer to the same thing? Due to these difficult questions, the analysis of a sentence “This is a cow.” will here be that it consist of two general terms: “cow,” explained as “not not a cow,” and “this,” explained as “not not this.” In the latter case, the entity referred to is not what Dharmakīrti would call a particular (svalakṣaṇa), but rather a chain of causes and effects which one would classify as constituting the moments that the life of the indicated cow consists of. “This” and “cow” in the sample sentence would thus both have to be taken, on Siderits’s interpretation, as referring to a thing and as presenting that thing in a certain mode. His theory with regard to the issue of the sense of a general is thus not modified.

1.2  Proposition as a Working Definition of Sense The notion of meaning that I would like to make the focus of attention here is that of Quine, as he discusses it in Word and Object. Among the reasons for choosing this presentation is, on the one hand, that it presents a generalized criticism of meaning as some sort of entity linking words and their objects and, on the other, that this criticism allows us to narrow down our line of inquiry quite a lot, namely, to a question about synonymy. A further reason for using Quine’s model here is that it scrutinizes a theory of sense that is structurally not too distant from the one proposed by Siderits, insofar as the sense of a general

Śabdārtha as Sense or Reference: Dharmakīrti on Synonymy

207

term is found in propositions where this general term is used. The difference between the two is, however, large: Quine, as is commonly known, does not accept the existence (or usefulness) of a general term as such. For Siderits, on the other hand, the Buddhist theory needs general terms to have sense, even if only in propositions, to work. Moreover, it is not necessary to think that Quine’s criticism of meaning is correct. It is enough to understand his notion and criticism of meaning to such a degree that it can be employed in reading a group of statements by Dharmakīrti and that it allows a judgment of how it fares in that context. Quine (1960, §§ 40 and 42) makes the point that a proposition can be seen as what is designated by an eternal sentence that is treated as a name. Problems arise as soon as one tries to explain how two such eternal sentences are judged to express the same proposition. Quine’s sample sentence here is “The door is open.” This can be made into an eternal sentence by adding specifiers, for example: “At 1:57 pm on the 23rd of July 2012, the door to office 304 in Apostelgasse 23, 1030 Vienna, Austria, is open.” This sentence is true (the door to my office was actually open then), and it will be true forever. Different eternal sentences could have been made by any number of syntactical or semantical changes, switching the position of address and time around, for example, or by giving the same time in a different format. All such sentences would normally be counted as expressing the same proposition. But why? “A usual answer is that the sentences are to be synonymous. One who gives this answer can as well go on to say that the proposition is the meaning of the sentence” (Quine 1960, 200). In this argument, we can see that synonymy8 is a good measure for understanding what role a proposition plays in a theory of language: if synonymy is appealed to in order to explain why two sentences express the same proposition, then the proposition can here be equated to “meaning.” Quine gives two characteristics of a proposition in this sense: it is, obviously, not a physical object (cf. Quine 1960, 201), and it must be identically the same in all of its instances (cf. Quine 1960, 203). The former trait makes it, and hence meaning or sense, an abstract object.9 The second demands that synonymy be a clear notion, precise enough to allow a person to be completely certain about the synonymity of two sentences.10 Another, purely practical, advantage of choosing Quine’s interpretation here is that it is, at least at the first look, close to the notion that Siderits was

208

Patrick McAllister

working with when studying the informativeness problem in Dharmakīrti. We might therefore expect to be able to find out where the exact difference in interpretation lies between Siderits’s understanding and the one developed here. Siderits does not touch on the problem of synonymy as such, but does enter into a short discussion of a priori truths. Noting the interpretational difficulties involved in the question, Siderits (1991, 101 sq.) raises the question whether, for Dharmakīrti, it can be considered an a priori cognition that an elm is a tree. He thinks not, since he takes Dharmakīrti as maintaining that the extensions of “elms” and of “trees” are empirically founded (Siderits 1991, 202). But nevertheless, the exclusion classes—senses on this interpretation of Dharmakīrti’s language theory—decide for Siderits whether there are a priori truths or not. So it can be assumed that synonymy too would be explained by an appeal to exclusion classes. We can see this from an argument like this: Imagine that two people learn to use the words “elm” and “tree” in two different valleys. On meeting in a third valley, they discover that they say “elm” about the same trees, none of which either of them has ever seen. They both agree, furthermore, that everything they call “elm” is what they would be prepared to call “tree.” On Siderits’s explanation, they would have to be agreeing on the exclusion classes in order to be able to agree in such a way. For as they have both learned to use the word “elm” with wholly different sets of trees, and hence the extension of “not not a tree” and “not not an elm” is, up to the current situation, to different particulars for each of them, their agreement must be about the exclusion class. And if the exclusion class is the sense, and not the referent, then synonymy—in this case, the usage of the same word, “elm,” by different speakers—is explained by appeal to sense.

2  Dharmakīrti on Synonymy As already noted by Frauwallner (1932, 248), Dharmakīrti first introduces the apoha theory, the Buddhist theory about the object of words and concepts, in response to an objection against a particular kind of logical reason: the svabhāvahetu, the reason consisting in the nature of the thing that an inference is about. The classical example, given above in the terms of elm and tree, is “This is a tree because it is an elm.”11

Śabdārtha as Sense or Reference: Dharmakīrti on Synonymy

209

The background to the objection is that Dharmakīrti has just claimed that it is possible to infer an essential property of a thing from another essential property, PV I, 39cd and the commentary: If this state that is to be proven did not exist, the state which proves it12 would not exist itself, since they are not different. Only that state which, following on the mere existence of another state, is called “nature of a thing,” is really in and of itself that other state. And how should it exist having abandoned itself?13

From this it results that two properties, the property to be proven (like tree-ness) and the property that proves it (like elm-ness), are, in some unspecified sense, not different. The opponent picks up on this point and criticizes Dharmakīrti by saying that there is a logical problem: the fault of pratijñārthaikadeśa—that the reason is part of the object of the thesis. This objection is as follows: Then, exactly that which is produced is impermanent, because there is no difference between these two. So the reason consisting in the nature of a thing, i.e., being produced, would be a part of the object of the thesis, i.e., of what is expressed by saying “This is impermanent.”14

Referring to a foundation of the proof of universal momentariness,15 this statement pulls into doubt that a thing’s being “produced” is a valid reason for inferring its being “impermanent.” For our inquiry about synonymy, the most important question about this passage regards the “non-difference” of what it is to be produced (kṛtaka) and what it is to be impermanent (anitya), the first of which can prove the latter: how are these two properties not different?16 Dharmakīrti’s first defense against this objection is given in PV I, 40–42. As Steinkellner (1971, 198) has noted, these verses give, in a nutshell, Dharmakīrti’s theory of what conceptual cognition is really about: classes of things, analyzed as common differences among particulars. Steinkellner (1971) has discussed these verses very closely, and there is nothing I can add to that discussion at the moment. A point to consider here is the somewhat peculiar usage of the term “dharma” in PV I, 42: Therefore, a distinction of a thing, which is cognized through some property, cannot be cognized through a property other than that one. Hence, there is a differentiated determination of two properties of the same thing.17

In the two preceding verses, Dharmakīrti said that a particular is completely (spatially and temporally) unique. With this statement, he has committed

210

Patrick McAllister

himself to the claim that a particular has as many differences as there are other things. Based on whatever these differences are differences from, things can be imagined to belong to certain classes: All entities have, through their own nature, a part in the differentiation from both the same and other entities because they subsist in their very own nature. For this reason, different genera are mentally constructed, which are based on whatever the referents are differentiated from, and which fathom the particulars’ specific differences.”18

Steinkellner (1971, 198), pointing out Dharmakīrti’s own paraphrase by “name” (nāman), translates dharma here as Bestimmung, a term that perfectly transports the difficulty into German: Bestimmung can mean attribute in the sense of “actual property of a thing,” as well as in the sense of something only ascribed to a thing.19 Corresponding English terms would be “predicate” or “attribute.”20 Dharmakīrti explains the verse himself: Therefore, that distinction—that is, difference—, which is cognized through some property—that is, through a name21—, even though there is no difference in the thing’s nature, that cannot be cognized through a property/name other than that one. So not all words have the same object. Therefore, this reason, consisting in the thing’s nature, is not a part of the object of the thesis.22

This concludes Dharmakīrti’s first refutation of the objection that a thing’s being produced should be an invalid reason for inferring its impermanence because these two properties are not separated in the thing and are hence identical, leading to the problem that the reason is part of the thesis. At first sight, it seems ontological and epistemological matters are being mixed together: properties, differences among things, and their individual natures and states, on the one hand, and imagined classes and names, on the other. Dharmakīrti equates property (dharma) and name (nāman) in this consideration. He says of the different classes that they are imagined or constructed (pari√kalp), and that this has its objective basis in the things’ various differences, which result from their natures. Even though we have only taken a cursory look at this passage, we can clearly see that the synonymy of two terms involves some conceptual activity: the world consists only of unrepeatable point-instants, and any notion of a “class” or commonness of things is only imagined. This lands Dharmakīrti’s discussion of synonymy squarely in the subjective realm: that two terms are synonymous depends, at least in part, on cognitive events. We have not yet seen whether this means that the intension of the terms is appealed to for their being

Śabdārtha as Sense or Reference: Dharmakīrti on Synonymy

211

synonymous (or, closer to the text, for two predicates of the same thing to not be synonymous). So far, we have only noted that Dharmakīrti’s account clearly involves a subjective component: the conceptual construction, or imagination, of different genera of things. This imagined class is, furthermore, only cognizable through a certain property (or name). That this is so, we shall see, is only due to convention. From this argument of Dharmakīrti’s we can conclude that a necessary condition for synonymy is that two terms lead to the cognition of the same class that a thing belongs to. In other words, the same differentiation, this sameness understood as the sameness of what is excluded, is what makes two words synonymous. So the next question must be: Is this differentiation in some way a proposition? For this, we can focus on its role in language: Is this differentiation, for Dharmakīrti, identically the same for two words? And is it something “internal” to cognition, as opposed to an objective fact such that two terms are synonymous because they refer to the same thing?23 There are difficulties in understanding identity of difference. Are the differences of two full glasses from an empty glass the same difference? How to judge the identity of these relations? Let us focus on Dharmakīrti’s approach to the matter, and investigate what he has to say about this differentiation as regards its identity in multiple instances. The relation “difference” is, as we saw in the verses above, bound to or based on the nature of particulars, but is nevertheless imagined. We can try to understand this as stating that, from the mass of differences that a thing has, a person can arbitrarily pick one out and classify things according to it. This way of understanding has the benefit of easing the tension between Dharmakīrti’s two claims that a thing has a differentiation but that the different classes based on this thing’s differences are merely imagined. In the following passage, Dharmakīrti describes how a linguistic convention is taught. A linguistic convention is, for Dharmakīrti, an act in which a term is agreed upon for the purpose of naming a difference. It is useful to look at this passage because, if words ever express differences that are elements identically subsisting throughout instances of the word’s usage, then certainly the minimum expectation would be that in teaching a word the pupil would be said to understand it correctly only if she were to understand the same as the teacher wishes her to understand. For, if Dharmakīrti does not hold that the pupil understands the same as the teacher even when the former successfully learns a word from the latter, then surely his theory of linguistic communication

212

Patrick McAllister

does not require that two people ever have to understand the same meaning so that they can communicate. The passage, itemized for clarity, is as follows: It is due to the very nature24 of these things that the causes for such25 a conceptual cognition—which, in turn, is due to an experience of these things—are continuous.26 Because of this continuity in the causes of a conceptual cognition, (6) a learner27 should cognize the reasons for this conceptual cognition as different from others—those reasons which (1) occur in the mind of someone seeing this thing producing the same effect, (3) are determined as being a reason for this cognition and as excluding what is other than this reason for this cognition, even though (2) they are not really like this, and (4) have a difference of external and internal form that is not distinguished; the learner does this (5a) in accordance with his cognition that “These are trees,” as he is (5b) based upon a conceptual cognition (5c) after having been shown those things appearing as the same in his own and the other’s conceptual cognition. Therefore the learner binds an expression to the difference of these things from those things which are not reasons for that, i.e., which are not a reason for the conceptual cognition of tree.28

This is how Dharmakīrti analyzes why two people can use the same word for something. There are two candidates for meaning involved here: the difference that a word is bound too and that which has the same appearance (ekapratibhāsa) in each person’s awareness.

3  Form (ākāra) as Sense? Let us first take a closer look at that which has this same appearance. For, if this is something that remains the same throughout the different occurrences of a word in the spoken utterances of different people and is not what is referred to, this would be a good candidate for a “meaning” entity. Karṇakagomin makes it strikingly clear that this is not, strictly speaking, the same, and gives the following example about two persons suffering from the eye disease timira, due to which they both see two moons where there is only one, PVSVṬ, 241.20–25: This is said: Like one Taimirika, upon seeing a double moon, pointing it out to another Taimirika, points out only the double moon seen by himself, not the one seen by the other, because that one is not perceivable by him. And now such a thought occurs to him: “I have made him cognize just this.” And the other, perceiving the form of a double moon that exists only in his own mental

Śabdārtha as Sense or Reference: Dharmakīrti on Synonymy

213

continuum, thinks: “I have cognized precisely that which this person, who is letting me cognize this, has made me cognize.” In this way a convention is made between the cognizer and the one making him cognize, because they determine as identical the form of awareness whose form is determined as external, even though it is different for both of them; and there is a cognition of just that same form at the time of every-day activity, because this form is determined as identical with the form at the time the convention is made.29

Karṇakagomin here maintains that there is no real identity between what appears to the minds of the two persons. He also clarifies the situation in the ensuing everyday usage of a term: the person who has been introduced to the convention, when using the word at a later time, determines that second usage as identical with the previous one. In other words, even the same person does not, in understanding or using the same word twice, have the exact same cognition in each of these instances without determining the one as the other. But let us see if this is indeed how Dharmakīrti intends this passage to be understood. A good example of how strictly the word “same” (eka) should be interpreted is given in PV I, 109: A cognition is not different because it is the reason for the same judgement. The particular things too are undifferentiated, since they are reasons for the same cognition.30

As Dunne (2004, 121 sq.) has pointed out in his interpretation of this verse, the criterion for the “sameness” of things is thus the same judgment, or, in other words, a determination of sameness. It is this kind of sameness that is, in my opinion, also meant when the person learning a word understands that what the teacher is showing her is actually “the same appearance” as in the teacher’s awareness (cf. Section 2). We can now follow Karṇakagomin’s elaboration and see that this sameness even applies when a person uses (or understands) the word “tree” repeatedly: she thinks that the cognition’s form (or content) is the same as in the previous cases of hearing or saying this word. Can any element involved here be something that remains the same throughout different instances of a word’s employment? Not really, one has to answer, because, even though taken as the same, nothing—except this (erroneous) judgment that two things are the same—is actually repeated in one instance and the other. This is a question that is raised in PV I, 107ab and the commentary, PVSV, 54.18–55.6:

214

Patrick McAllister

If this is thought by an opponent: “There is no cognition with an undifferentiated appearance for different things.” That is: “We opponents do not say that what is not the same (aneka) does not create the same (eka) effect. Rather, we say that an awareness, insofar as the form of different objects is laid into it, cannot have an undifferentiated appearance for those different objects.” Then we Buddhists answer: Indeed there is no appearance of particulars, which are not the same, in awareness events grasping a commonness, first, because31 these awareness events exist even if those particulars do not exist; second, because of the appearance in their own cognition, i.e., perceptual cognition, with another form, i.e., other than the form of the commonness; and third, because there is an overreaching consequence32 since a single thing is not consistent with multiple forms. Therefore this awareness of a commonness does not appear as grasping different objects, as undifferentiated for different particulars, or as arising from these different objects. Even though it has no appearance of this particular, conceptual awareness causes the world to engage in everyday activity because of the confusion resulting from determination. But this form, appearing in this conceptual awareness, does not exist in the objects, other (anyatra) than as a difference that has no difference.33 But this common form is unreal. In this way this awareness grasping only this unreal common form goes astray. This has already been explained.34

The central argument of Dharmakīrti here is that the form appearing in a verbal or conceptual cognition is not something that positively or substantially exists in, among, or as a part of real things. This form is a result of the common difference among things, which, in turn, is an insubstantial and relational property of the things. This, in Dharmakīrti’s analysis, is not a real thing, without any externally existing nature or form (note the use of arūpa here). Hence, the cognition showing that form is classified as an error. So according to Dharmakīrti, a form that appears in verbal cognition is “the same”—or undifferentiated—in that it appears there in virtue of a difference among the things. As we have seen above, this “in virtue of” is to be understood in terms of a causal relation. The term “ekapratibhāsa” as understood in the passage about a word convention must therefore mean an appearance that is the same because it is caused by the same kind of thing (i.e., things of the same difference). A person having being introduced to the usage of “tree” in winter could well be confused in summer when everyone is talking of the shade found under trees. That person would have simply picked the wrong difference for distinguishing trees.

Śabdārtha as Sense or Reference: Dharmakīrti on Synonymy

215

Since for Dharmakīrti a thing’s existence, involvement in a causal relation, and momentariness are essentially dependent on each other,35 we can easily conclude that the form appearing in conceptual awareness must be a particular, and hence unique in time and space. It is, therefore, impossible as a candidate for an entity that should persist as the meaning of words.

4  Difference as Sense? The other element we have noticed, the differences to which words are bound (cf. p. 212), is the last candidate for a meaning entity. This, however, must be discarded on the basis of the passages we have seen so far: it is not an identical “thing” in any meaningful way. Differences are constituted by things’ natures.36 Differences between real things will then be as transient as the real things themselves. When one of the involved things vanishes, this particular relation will vanish with it. Since real things are momentary for Dharmakīrti, the differences will be momentary as well. Accordingly, to pick out a certain difference (say between trees and non-trees) will have practical use as long as both sides of this relation exist. But this difference is the same only in the above sense of causing a judgment of sameness, which in turn has nothing persistently the same about it.37 In addition, this difference among the things is external, so that one would speak rather of a referent, and it is not something that should be likened to a proposition. Even if it is an “imagined or constructed difference,” it is still something that is referred to as an object. This is a form of nominalism which is not realism simply because the universals don’t really exist. Most other elements are still in place.38 But how can even the same word, used twice, be synonymous according to Dharmakīrti? Surely, if everything involved in linguistic practice is momentary and unique—the differences the things have to each other, which a word is based on, and the form of cognition each conceptual cognition involves—there cannot be any kind of continuity? But according to Dharmakīrti, the error of seeing this continuity, and imposing it on our cognitions and on the things which these cognitions are in turn based on, is persistent. On identifying the tree in front of the house with the tree there yesterday, we are actively construing an identity that is not really there. If anything, what persists is the repeated mistake in conceptual cognition. It is conditioned by habits, preferences, expectations, and so on,39 and is, if “correct,” advantageous for interaction with causally efficient particulars in that it can prompt activity in approximately the right direction. But it does not show the world as it really is.

216

Patrick McAllister

5 Conclusion Which elements of Quine’s analysis of meaning sketched above (Section 1.2) are to be found in Dharmakīrti’s theory? Neither a shareable “sense” nor, as we have seen, anything representing a proposition is used to explain why two statements (“These are trees,” said at the time of learning a convention and later when applying it) mean the same. The explanation of this fact goes along causal lines. Ultimately, two instances of a sentence are judged by us to be the same only because of a certain lack of knowledge about reality. What is understood by and meant with “These are trees” are objects external to cognition. Even though there is a form of cognition that belongs to each instance of this sentence for every speaker, no appeal to the uniformity of this form for every speaker or even for the same speaker at different occasions is made. What the speaker intends or means with that sentence are the external objects, and their relation (that “treeness” qualifies that which is “this”). That they do not exist as they are intended does not really matter, their status as externally posited things, objects one can interact with, remains untouched. In answer to the leading question, I would therefore say that I don’t think Dharmakīrti was talking about “sense” that would correspond either to Frege’s notion or to the emended version where the sense of an expression is a proposition. A translation of śabdārtha as “word object” or “referent” would therefore cause least confusion.

Notes 1 For Dharmakīrti, follow the dating given by Krasser (2012). In the following, the dates of ancient Indian Buddhists are given in accordance to Frauwallner (1961) unless noted otherwise. 2 Hinted at in Ganeri (1996, 17), and more fully explicated in Ganeri (1999), is the observation that later (post-Gaṅgeśa) Naiyāyikas recognized a need for either introducing a meaning element into their language theory or extending a pure reference theory in a way to “do duty for a full theory of meaning” (Ganeri 1996, 17). 3 Cf. Siderits (1991, 75); 99 sq. Siderits (1991, 73) makes “the following emendation to the orthodox Fregean view: the reference of a sentence is not a truth value, but rather a proposition, understood in accordance with the naive theory.” This “naive theory,” in accordance to Salmon (1982), states that a proposition consists in the information values of its components (such as names, predicates, etc.), and that

Śabdārtha as Sense or Reference: Dharmakīrti on Synonymy

4

5 6

7

8 9

10

11 12 13

217

that information value is the thing referred to (cf. Siderits 1991, 67). For other candidates of this strand of sense-reference distinction, see Siderits (1991, 74 sqq.). The error that Dharmakīrti is accused of making is that the reason is “part of the object of the thesis” of the inference (pratijñārthaikadeśa). See the explanations at p. 209. For example, Frege (1892): “Sinn des Zeichens […], worin die Art des Gegebenseins enthalten ist.” The theory of other-exclusion (anyāpoha) states that what things have in common is nothing substantial (like a property, genus, or similar), but only a difference to other things. For a closer analysis of this theory, see Chapter 8 in this volume. For the general argument, see PV I, 43 and commentary. A translation and summary is given in Kellner 2004, 4 sqq. Cf. McAllister 2011, fns. 87, 88 for some pointers to the role of demonstratives in the apoha theory. Synonymy here means broad or general synonymy, the classical notion of interchangeability without variance in truth value (cf. Quine 1960, 61 sq.). Quine, of course, sees no big problem in this, saying that meaning could be seen as “the very class of all those mutually synonymous sentences that are said to have it” (Quine 1960, 201). A different solution is, perhaps ironically, considered in Quine 1980, 9 “one might quite plausibly explain meanings as ideas in the mind, supposing we can make clear sense in turn of the idea of ideas in the mind.” This is impossible for Quine, and is a reason that identity of propositions cannot be defined through a notion of synonymy: the general one is not possible, and Quine’s “graded notion of synonymy” (Quine 1960, 203) is inadequate for judging identity of propositions. “Elm” is a substitute śiṃśapā, which Böhtlingk and Roth 1875, 7.173.1 says is the name for Dalbergia Sisu, the Indian redwood. PVSVṬ, 107.27: bhāvasya hetutvenopāttasyābhāvaḥ syāt, “the state, i.e., that state which is taken (upā-dā) to be the reason, would not exist.” PV I, 39cd and PVSV, 24.14–15: tadabhāve svayam bhāvasyābhāvaḥ syād abhedataḥ // ya eva bhāvo bhāvamātrānurodhī svabhāva ity ucyate, sa eva svayaṃ vastuto bhāvaḥ. sa cātmānaṃ parityajya kathaṃ bhavet? This verse, with the surrounding passage, was translated into German by Steinkellner (1971, 206, fn. 97) (with some corrections made in Steinkellner 1996, 258). That translation reflects the problems connected with the terms svabhāva and bhāva far better than the one given here. A proper estimation of these terms is hard, and not wholly relevant for our inquiry. I have chosen to translate “bhāva” as state here because it seems to me that this way most of the term’s aspects can be expressed: it can mean both “thing” (that which is) as also what (or even how) something is. Steinkellner (1971, 206, fn. 97) translates this term as “Beschaffenheit,” and Iwata (2003, 62, 67) translates it as “property” in the context of PV I, 2c’d and “thing”

218

14 15

16

17 18

19

20 21 22

Patrick McAllister in this context. It seems to me that “state” does justice to both without sounding too awkward. For a general discussion of Dharmakīrti’s statements about the svabhāvahetu, see Steinkellner 1996. PVSV, 24.16–17 ya eva tarhi kṛtakaḥ sa evānityo bhedābhāvāt. pratijñārthaikadeśo hetuḥ syāt. There are several variations of this proof and its supplementary theories, and even five hundred years later it was still a major topic in the Indian intellectual landscape (see, for example, Kyuma 2005; see Sakai 2010, xxii sqq. for a study of its development up to Dharmottara). The gist of at least one strand of the argument, hinted at in this remark of the opponent, is that produced things perish at every moment as they do not need a cause for their destruction. This is possible, however, only when everything that is produced is shown to be impermanent, that is, temporally finite but not necessarily momentary. Linked to this question is the further question of what type of error (doṣa, acc. to PVSV, 24.17) pratijñārthaikadeśa is. Toshikazu Watanabe has worked on this question, also taking PVin III into account. If I understand correctly, Watanabe (2012) concludes that a reason of this type would be of no practical use for an inference, since it would merely repeat the thesis, and would therefore be considered asiddha. PV I, 42: tasmād yo yena dharmeṇa viśeṣaḥ saṃpratīyate / na sa śakyas tato ’nyena tena bhinnā vyavasthitiḥ // See PV I, 40–41: sarve bhāvāḥ svabhāvena svasvabhāvavyavasthiteḥ / svabhāvapa­ rabhāvābhyāṃ yasmād vyāvṛttibhāginaḥ // tasmād yato yato arthānāṃ vyāvṛttis tannibandhanāḥ / jātibhedāḥ prakalpyante tadviśeṣāvagāhinaḥ // Vetter (1964, 48 sq.), paraphrasing these verses, takes dharma as “Eigenschaft,” property. Frauwallner (1932, 368), translating from the Tibetan, does likewise. He apparently also found this a bit odd, since he quotes the explanation in PVṬt, and points out that the V(ṛtti) reads “ming gi chos” (more fully, PVSVD, 275a3 and PVSVP, 420b7 read “ming gi chos gang gis,” which one might consider emending to “ming gis chos gang gis” according to “yena yena dharmeṇa namnā” (PVSV, 25.24)). The terms “predicate” and “subject” are used by Dunne (2004, 153 sqq.) when he explains the svabhāva as property. Given that “produced” is an example for a name as used here, one could also translate “term.” tasmāt svabhāvābhede ’pi yena yena dharmeṇa nāmnā yo viśeṣo bhedaḥ pratīyate na sa śakyo ’nyena pratyāyayitum iti naikārthāḥ sarvaśabdāḥ / tan na pratijñārthaikadeśo hetur iti. Śākyabuddhi (ca. 660–720) and Karṇakagomin (ca. 770–830) explain the usage of dharma here (and in PV I, 42): (1) PVṬt, 56b4 (explaining yena dharmeṇa in PV I, 42): mi rtag pa zhes bya ba’i sgra can gyi

Śabdārtha as Sense or Reference: Dharmakīrti on Synonymy

219

chos gang gis “through which property that has a word that is associated with it, like ‘impermanent’” (2) PVSVṬ, 114.25–26: yena dharmeṇa yena śabdena, yathā ’nityaśabdena. śabdo ’pi dharmavācakatvād dharma ucyate. “Through which property, i.e., through which word, like through the word ‘impermanent.’ A word is also called a property because it names a property” (3). PVṬt, 59b5: chos gang gis zhes bya ba ’di’i rnam par ’grel pa ni, ming zhes bya ba ’di yin no. sgra zhes bya ba’i don te brjod par ’dod pa’i dbang gis chos zhes bya ba’i sgra ni sgra la bya’o. “Dharmakīrti’s explanation of this expression ‘through which property’ is this expression, ‘name.’ It is due to the wish of explaining that its meaning is ‘word’ that he uses the word ‘property’ in the sense of word” (4). PVSVṬ, 120.16–17: dharmeṇety* asya vivaraṇaṃ nāmneti śabdenety arthaḥ. (*Read PVSVṬms, 46b4 dharmeṇety instead of the printed dharmiṇety. Trl.: “Dharmakīrti explains this expression, ‘through which property,’ with the word nāmna, ‘through the name.’ That is the intention of this passage.” 23 Realism about universals is very common in Indian writings of the Brahmanical schools. See section 1.1. Asking about the “internalness” of the conditions for synonymy is an attempt to clarify Dharmakīrti’s position in this respect. 24 This passage is also translated and discussed in McCrea and Patil 2006, 312 sqq. The translation presented here builds heavily on Karṇakagomin’s analysis of the passage (PVSVṬ, 240.10–241.25), and is (sometimes for that reason) different from the one by McCrea and Patil (2006) in various details. The main difference lies in the interpretation of the atathābhūtatathādhyavasita clause. McCrea and Patil (2006, 312) translate: “These things, which are not really so [i.e., do not really have the same appearance in his own and the other person’s conceptual awareness] are determined to be so, […] in virtue of their being the causes of that conceptual awareness, and in virtue of the exclusion of what is other than them [i.e., what does not cause that conceptual awareness].” They take the two instrumental phrases to be reasons for the determination, and determination to be the identification of what appears in their own and the other person’s conceptual cognition. While this is possible, and I do think that this identification is actually a form of determination, Karṇakagomin (PVSVṬ, 240.23–28) explains the determination mentioned by Dharmakīrti as follows: tajjñānahetutayā tasya vikalpajñānasya hetutayā tadanyavyāvṛttyā cety ekākārapratyabhijñānahetubhyo ye ’nye tathābhū­ tavikalpāhetavaḥ, tebhyo vyāvṛttyā ca, atathābhūtān api. na hi te vikalpārūḍhās taddhetavaḥ, bahiravidyamānatvāt. ata evāheturūpavikalatvam* apy asat, teṣām avastusattvāt. tathādhyavasitān tajjñānahetutayā tadanyavyāvṛttyā cāropitān. anena bhāto hetutayā dhiyaḥ. aheturūpavikalān iveti vyākhyātam. (*Emendation against PVSVṬms, 88b5 aheturūpavikalpatvam acc. to PVṬt, 135a4: rgyu ma yin pa’i ngo bo dang bral ba nyid.)” Karṇakagomin is here saying that this passage

220

25 26

27 28

29

30 31

32

Patrick McAllister explains PV I, 120bc (bhāto hetutayā dhiyaḥ. aheturūpavikalān ivā). But I believe his interpretation of the two instrumental clauses should not be taken causally here: in the paraphrase of tathādhyavasitān by tajjñānahetutayā tadanyavyāvṛttyā cāropitān, “being the reason of this cognition” and “excluding (or as exclusion of) that different from this” are most naturally taken as explicating the “tathā.” That is, the conceptual cognition of a thing as the same as another, “called the same judgement” (ekapratyavamarśākhya PV I, 119a). “Continuity” here translates the Sanskrit anvaya. Though a technical term, the notion generally includes a consistency or continuity between two factors, such as cause and effect: similar causes will produce similar effects, and similar effects are signs of similar causes. In the present case, the causes for repeated conceptual cognitions of the form “This is the same as/similar to that” are therefore continuous in that they regularly or consistently produce what is judged to be the same effect. PVSVṬ, 241.7: yasmai saṃketaḥ kriyate, sa pratipattā. “She, for whose sake a convention is made, is a learner.” PVSV, 60.23–61.1 teṣāṃ prakṛtyaiva pratyayavaśāt tathābhūtavikalpakāraṇānām anvayāt (1) taddraṣṭur buddhau viparivartamānān (2) tajjñānahetutayā tadanyavyāvṛttyā cātathābhūtān api (3) tathādhyavasitān (4) avibhaktabāhyādhyātmikabhedān (5) pratipattā (5a) pratipattim anusṛtyaite vṛkṣā iti (5b) svaparavikalpeṣv ekapratibhāsān ādarśya (5c) vikalpavijñāne vyavasthitas (6) tadvijñānahetūn bhedena pratipadyetety uktim ataddhetubhyo bhede niyuṅkte.” PVSVṬ, 241.20–25: etad uktam bhavati—yathaikas taimiriko dvicandraṃ dṛṣṭvānyataimirikāyopadiśan svadṛṣṭam evopadiśati na paradṛṣṭam, apratyakṣatvāt. atha ca tasyaivam bhavati—ayam eva mayā parasmai pratipādita iti. paro ’pi ca svasantānabhāvinam eva dvicandrākāraṃ pratiyan——ya eva pratipādakena mama pratipāditas sa eva mayā pratipanna iti manyate. tadvad pratipādyapratipādakayor buddhyākārasyādhyavasitabāhyarūpasya bhede ’py ekatvādhyavasāyāt saṅketakaraṇaṃ vyavahārakāle ca tasyaiva pratītir ekatvādhyavasāyāt. PV I, 109: ekapratyavamarśasya hetutvād dhīr abhedinī / ekadhīhetubhāvena vyaktīnām apy abhinnatā // The three reasons given now are, according to PVSVṬ, 221.25–222.10, aimed against three different opinions about the relation between something that appears and particulars or universals: first, the form in which a particular appears to perception is the same as the form in which it appears to conceptual cognition. Second, the form appearing in the awareness of a universal is the form of the particulars. Third, the same individual has a twofold form: with one it appears to perception, with the other to conceptual cognition. PVSVṬ, 222.8–10: tṛtīyam pakṣaṃ nirākartum āha—anekākārāyogād iti. ekasyānekatvam ayuktam ekānekatvayor virodhāt. atiprasaṅgāc cety

Śabdārtha as Sense or Reference: Dharmakīrti on Synonymy

221

ekasyānekatvakalpanāyāṃ na kvacid ekatvaṃ syād ity arthaḥ. “In order to refute the third position (cf. section 31), Dharmakīrti said: ‘Because it is not consistent with multiple forms.’ Being many is not consistent for one, because being one and being many are contradictory. And because there is the overreaching consequence, i.e., there would not be oneness for anything when there is the option that one thing is many things. This is the meaning.” 33 For this way of construing anyatra, cf. Böhtlingk and Roth 1855, anyatra, item 7. A difference that has no difference, is nothing but other-exclusion. See also PV I, 56.21: sa tv eṣām abhinno bheda ity ucyate. Cf. PVSVṬ, 222.22–25: kathaṃ tarhi vyaktiṣv abhinnākārapratibhāsa ity āha—anyatra bhedād abhedina iti. bhedo ’nyāpohaḥ sa eva prativyaktyabhedī. tathā hi yathaikā govyaktir agovyāvṛttā tathānyāpi. tad anena prakāreṇa svalakṣaṇāny eva vijātīyavyāvṛttāny abhedīni bheda ity ucyante. anyatraśabdaś cāyaṃ vibhaktyantapratirūpako nipātaḥ. anyaśabdasamānārthaḥ. na tv ayaṃ tralpratyayāntaḥ saptamyarthasyāvivakṣitatvāt. tenāyam arthaḥ—yathoktena prakāreṇa svalakṣaṇātmakād bhedād abhedino ’nyaḥ pratibhāsamāna ākāro ’rtheṣu nāsti. kiṃ tu svalakṣaṇātmaka eva bhedo vijātīyavyāvṛtter abhedī sarvatra vidyate ’bhedādhyavasāyāt. abhedādhyavasāyasya ca sa eva bhedaḥ pāramparyeṇa nimittam. “Because of the question: ‘How then is there an appearance of a non-different form in particulars?,’ Dharmakīrti said: other than a difference that has no difference. This difference, i.e., other-exclusion alone has no difference for multiple particulars. For it is so: As one cow-particular is differentiated from non-cow, so also another cow-particular is differentiated from non-cow. Thus, by this method, the particulars alone, which are differentiated from that of another genus, i.e., non-different, are called ‘difference.’ And this word ‘anyatra’ is an indeclinable that accords to a case ending. It has the same meaning as ‘other.’ But this is not the tral ending, i.e., tra, because the sense of the seventh, locative, case is not meant. Thus this is the meaning: An appearing form, which is different from the non-different difference—which has the nature of a particular in the way explained—does not exist in the objects. Rather, a difference, which only has the nature of particulars, and which is without difference due to a differentiation from things of a different kind, is seen in all particulars of the same class because non-difference is determined. And for a determination of non-difference precisely this difference is the indirect cause.” 34 PVSV, 54.18–55.6 abhinnapratibhāsā dhīr na bhinneṣv iti cen matam / na brūmo ’nekam ekakāryakṛn na bhavatīti. kiṃ tarhi. na bhinneṣv artheṣv arpitatadākārā buddhir abhinnapratibhāsinī syāt. na vai sāmānyagrāhiṇīṣu svalakṣaṇapratibhā­ saḥ, tadabhāve ’pi tāsāṃ bhāvāt, ākārāntareṇa ca svajñāne pratibhāsanāt, anekākārāyogād ekasya ’tiprasaṅgāc ca. tasmān neyaṃ bhinnārthagrāhiṇy abhinnā pratibhāti tadudbhavā. atatpratibhāsiny apy adhyavasāyavibhramād vyavahārayati lokam. sa tu tasyāṃ pratibhāsamāna ākāro na artheṣv asti. anyatra bhedād abhedinaḥ. sa cārūpaḥ. tam evaiṣā gṛhṇatī tathā viplavata ity uktaṃ prāk.

222

35 36 37

38

39

Patrick McAllister As pointed out by Gnoli (1960, 190), acc. to PVSVṬ, 223.9–10 the reference in the last sentence is to PVSV, 50.16–17. Cf. Dunne 2004, 91 sqq. See PV I, 40–41 (trl. section 5). A second, and perhaps stronger, argument for differences or exclusions is found in PV I, 169ab and commentary, translated by Dunne (2004, 127) as follows: “A negation has no nature; hence, one cannot conceive of it as having ‘perdurance’ or ‘non-perdurance.’ […] That is, there is no such thing at all as an ‘other-exclusion.’ And concepts of that exclusion’s perduring or ceasing by its nature, which would follow from it having a nature, do not make sense (na kalpante).” And this is also where, I think, it becomes clear that Dharmakīrti’s theory fits quite neatly into the realism that many scholars (cf. 1.1) judge dominant in Indian philosophy. See Kellner 2004 for a careful analysis of all “secondary” elements involved in conceptual cognitions.

Primary Sources PV I

Dharmakīrti. 1960a. “Pramāṇavārttike Prathamaḥ Paricchedaḥ.” In The Pramāṇavārttikam of Dharmakīrti: The First Chapter with the Autocommentary; Text and Critical Notes, ed. by Raniero Gnoli, 1–176. Serie Orientale Roma 23. The chapters of the Pramāṇavārttika are counted as follows: (1) svārthānumāna, (2) pramāṇasiddhi, (3) pratyakṣa, and (4) parārthānumāna. Rome: Istituto Italiano per il Medio ed Estremo Oriente.

PVin III

Dharmakīrti. 2011. “Pramāṇaviniścaya 3.” In Dharmakīrti’s Pramāṇaviniścaya: Chapter 3, ed. by Pascale Hugon and Toru Tomabechi. Sanskrit Texts from the Tibetan Autonomous Region 8. Beijing, Vienna: China Tibetology Research Center, Austrian Academy of Sciences.

PVSV

Dharmakīrti. 1960b. “Pramāṇavārttikasvavṛtti.” In The Pramāṇavārttikam of Dharmakīrti: The First Chapter with the Autocommentary; Text and Critical Notes, ed. by Raniero Gnoli, 1–176. Serie Orientale Roma 23. Rome: Istituto Italiano per il Medio ed Estremo Oriente.

PVSVD

Dharmakīrti. 1981. “Tshad ma rnam ḥgrel gyi ḥgrel pa.” In Tibetan Tripiṭaka sde dge Edition: Bstan ḥgyur; Preserved at the Faculty of Letters, University of Tokyo, ed. by Kyōshō Hayashima et al., vol. Tshad ma 1, ce 261b1–365a7 (No. 4216). Tokyo: Sekai Seiten Kanko Kyokai Co., Ltd., for the Faculty of Letters, University of Tokyo.

Śabdārtha as Sense or Reference: Dharmakīrti on Synonymy

223

PVSVP

Dharmakīrti. 1957. “Tshad ma rnam ’grel gyi ’grel pa.” In The Tibetan Tripitaka: Peking Edition: Kept in the Library of the Otani University, Kyoto; Reprinted under the Supervision of the Otani University, Kyoto, ed. by Daisetz T. Suzuki, vol. 130, ce 404b3–535a4 (No. 5717(a)). Tokyo, Kyoto: Tibetan Tripitaka Research Institute.

PVSVṬ

Karṇakagomin. 1943. “Pramāṇavārttikavṛttiṭīkā.” In Ācārya-Dharmakīrteḥ Pramāṇavārttikam (svārthānumānaparicchedaḥ) svopajñavṛttyā, Karṇakagomiviracitayā taṭṭīkayā ca sahitam, ed. by Rāhula Sāṅkṛtyāyana, 1–619. Ilāhābād: Kitāba Mahala.

PVSVṬms Karṇakagomin. 1998. “Karṇakagomin’s Pramāṇavārttika(sva) vṛttiṭīkā: Manuscript A.” In Sanskrit Manuscripts of Karṇakagomin’s Pramāṇavārttika(sva)vṛttiṭīkā, ed. by Shoren Ihara, 1–56. The Sanskrit Commentaries on the Pramāṇavārttikam from the Rāhula Sāṅkṛtyāyana’s Collection of Negatives 2. Patna, Narita: Bihar Research Society/Naritasan Institute for Buddhist Studies. PVṬt

Śākyabuddhi. 1982. “Tshad ma rnam ’grel gyi ’grel bśad.” In Tibetan Tripiṭaka sde dge Edition: Bstan ḥgyur; Preserved at the Faculty of Letters, University of Tokyo, ed. by Kyōshō Hayashima et al., vols. Tshad ma 3–4, je, 1b1–nye 282a7 (No. 4220). Tokyo: Sekai Seiten Kanko Kyokai Co., Ltd., for the Faculty of Letters, University of Tokyo.

TS2

Śāntarakṣita. 1981. “Tattvasaṅgraha.” In Tattvasaṅgraha of Ācārya Shāntarakṣita. With the Commentary “Pañjikā” of Shrī Kamalashīla: With the Commentary “Pañjikā” of Shrī Kamalashīla, 2nd ed., ed. by Svāmī Dvārikādāsa Śāstrī. 2 vols. Bauddha Bharati Series, 1–2. Varanasi: Bauddha Bhāratī.

Secondary Sources Böhtlingk, Otto, and Rudolph Roth. 1855. Die Vocale. Vol. 1 of Sanskrit Wörterbuch. St. Petersburg: Kaiserliche Akademie der Wissenschaften. Böhtlingk, Otto, and Rudolph Roth. 1875. śa–ha: Nebst den Verbesserungen und Nachträgen zum ganzen Werke. Vol. 7 of Sanskrit Wörterbuch. St. Petersburg: Kaiserliche Akademie der Wissenschaften. Bronkhorst, Johannes. 1999a. Langage et Réalité: sur un épisode de la pensée indienne. Bibliotheque de l’École des Hautes Études Section des Sciences Religieuses 105. Turnhout: Brepols Publishers. Bronkhorst, Johannes. 1999b. “Nāgārjuna and Apoha.” In Proceedings of the Third International Dharmakīrti Conference Hiroshima, November 4–6, 1997, ed. by

224

Patrick McAllister

Shōryū Katsura, 17–23. Beiträge zur Kultur- und Geistesgeschichte Asiens 32. Vienna: Verlag der Österreichischen Akademie der Wissenschaften. Dunne, John D. 2004. Foundations of Dharmakīrti’s Philosophy. Studies in Indian and Tibetan Buddhism. Boston: Wisdom Publications. Frauwallner, Erich. 1932. “Beiträge zur Apohalehre: I. Dharmakīrti; Übersetzung.” Wiener Zeitschrift für die Kunde des Morgenlandes 39: 247–85. Frauwallner, Erich. 1961. “Landmarks in the History of Indian Logic.” Wiener Zeitschrift für die Kunde Süd- und Ostasiens und Archiv für indische Philosophie 5: 125–48. Frege, Gottlob. 1892. “Über Sinn und Bedeutung.” In Funktion, Begriff, Bedeutung: Fünf logische Studien, ed. by Günther Patzig, 23–46. First published in: Zeitschrift für Philosophie und philosophische Kritik, NF 100, 1892, pp. 25–50. Page numbers given in text refer to the first publication. Göttingen: Vandenhoeck & Ruprecht. Ganeri, Jonardon. 1996. “Meaning and Reference in Classical India.” Journal of Indian Philosophy 24 (1): 1–19. Ganeri, Jonardon. 1999. Semantic Powers: Meaning and the Means of Knowing in Classical Indian Philosophy. Oxford: Clarendon Press. Gnoli, Raniero, ed. 1960. Pramāṇavārttike Prathamaḥ Paricchedaḥ. Serie Orientale Roma 23. Rome: Istituto Italiano per il Medio ed Estremo Oriente. Iwata, Takashi. 2003. “An Interpretation of Dharmakīrti’s svabhāva-hetu.” Journal of Indian Philosophy 31 (1–3): 61–87. Kellner, Birgit. 2004. “Why Infer and Not Just Look? Dharmakīrti on the Psychology of Inferential Processes.” In The Role of the Example (dṛṣṭānta) in Classical Indian Logic, ed. by Ernst Steinkellner and Shōryū Katsura, 1–51. Wiener Studien zur Tibetologie und Buddhismuskunde 58. Vienna: Arbeitskreis für Tibetische und Buddhistische Studien Universität Wien. Krasser, Helmut. 2012. “Bhāviveka, Dharmakīrti and Kumārila.” In Devadattīyam: Johannes Bronkhorst Felicitation Volume, ed. by François Voegeli et al., 535–94. Worlds of South and Inner Asia 5. Bern, etc.: Peter Lang. Kyuma, Taiken. 2005. Sein und Wirklichkeit in der Augenblicklichkeitslehre Jñānaśrīmitras: Kṣaṇabhaṅgādhyāya I, Pakṣadharmatādhikāra. Wiener Studien zur Tibetologie und Buddhismuskunde 62. Vienna: Arbeitskreis für Tibetische und Buddhistische Studien Universität Wien. McAllister, Patrick. 2011. “Ratnakīrti’s Apohasiddhi: A Critical Edition, Annotated Translation, and Study.” PhD thesis, Institute for South Asian, Tibetan, and Buddhist Studies, University of Vienna. McCrea, Lawrence J., and Parimal G. Patil. 2006. “Traditionalism and Innovation: Philosophy, Exegesis, and Intellectual History in Jñānaśrīmitra’s Apohaprakaraṇa.” Journal of Indian Philosophy 34 (4): 303–66. Patil, Parimal G. 2003. “On What It Is That Buddhists Think About: apoha in the Ratnakīrti-Nibandhāvali.” Journal of Indian Philosophy 31 (1–3): 229–56.

Śabdārtha as Sense or Reference: Dharmakīrti on Synonymy

225

Patil, Parimal G. 2009. Against a Hindu God: Buddhist Philosophy of Religion in India. New York: Columbia University Press. Quine, Willard Van Orman. 1960. Word and Object. Cambridge: MIT Press. Quine, Willard Van Orman. 1980. “On What There Is: Nine Logico-Philosophical Essays,” 2nd ed., 1–19. Cambridge, MA and London: Harvard University Press. Sakai, Masamichi. 2010. “Dharmottaras Erklärung von Dharmakīrtis kṣaṇikatvānumāna: Pramāṇaviniścayaṭīkā zu Pramāṇaviniścaya 2 vv. 53-55 mit Prosa.” PhD thesis, Institute for South Asian, Tibetan, and Buddhist Studies, University of Vienna. http://othes.univie.ac.at/9623/. Salmon, Nathan U. 1982. Reference and Essence. Princeton: Princeton University Press. Scharf, Peter M. 1996. The Denotation of Generic Terms in Ancient Indian Philosophy: Grammar, Nyāya, and Mīmāṃsā. Transactions of the American Philosophical Society Held at Philadelphia for Promoting Useful Knowledge 86.3. Philadelphia: American Philosophical Society. Siderits, Mark. 1986. “The Sense-Reference Distinction in Indian Philosophy of Language.” Synthese 69 (1): 81–106. Siderits, Mark. 1991. Indian Philosophy of Language: Studies in Selected Issues. Studies in Linguistics and Philosophy 46. Dordrecht etc.: Kluwer Academic Publishers. Siderits, Mark. 2011. “Śrughna by Dusk.” In Apoha: Buddhist Nominalism and Human Cognition, ed. by Mark Siderits, Tom J.F. Tillemans, and Arindam Chakrabarti, 283–304. New York: Columbia University Press. Staal, J. F. 1969. “Sanskrit Philosophy of Language.” In Linguistics in South Asia, ed. by Thomas A. Sebeok, 499–531. Current Trends in Linguistics 5. The Hague: Mouton. Steinkellner, Ernst. 1971. “Wirklichkeit und Begriff bei Dharmakīrti.” Wiener Zeitschrift für die Kunde Südasiens und Archiv für indische Philosophie 15: 179–211. Steinkellner, Ernst. 1996. “An Explanation of Dharmakīrti’s Svabhāvahetu Definitions.” In Festschrift Dieter Schlingloff zur Vollendung des 65. Lebensjahres dargebracht von Schülern, Freunden und Kollegen, ed. by Friedrich Wilhelm, 257–68. Reinbek: Verlag für Orientalische Fachpublikationen. Tillemans, Tom J.F. 2011. “How to Talk about Ineffable Things: Dignāga and Dharmakīrti on Apoha.” In Apoha: Buddhist Nominalism and Human Cognition, ed. by Mark Siderits, Tom J.F. Tillemans, and Arindam Chakrabarti, 50–63. New York: Columbia University Press. Vetter, Tilmann. 1964. Erkenntnisprobleme bei Dharmakīrti. Veröffentlichungen der Kommission für Sprachen und Kulturen Süd- und Ostasiens 1. Vienna: Kommissionsverlag der Österreichischen Akademie der Wissenschaften. Watanabe, Toshikazu. 2012. “Dharmakīrti on pratijñārthaikadeśa.” Presentation at the 15th World Sanskrit Conference in New Delhi. To be published in the proceedings thereof.

11

Semantic Relations and Causation of Verbal Knowledge Alessandro Graheli

1 Introduction The Nyāya appraisal of what constitutes a word is based on its theories of causation, which are a central feature in this system of thought. The Kashmirian author Jayanta (ninth century CE) concludes that the necessary and proximate cause of such knowledge is constituted by phonemes, the minimal segments of speech which, when uttered in certain sequences, become what we call “words.” The cognitive process behind their apprehension involves acts of perception (anubhava), subconscious dispositions (saṃskāra), and recollections (smaraṇa), as discussed in Chapter 3 in this volume.1 In the following pages, accordingly, by “word” one should actually understand “a string of phonemic segments.” Such words convey word-meanings by dint of their causal force (śakti), that is, their potentiality to cause knowledge of their referent (padārtha) under certain conditions, though the ontological status of such “forces” needs some clarification. According to Jayanta, words possess two such capacities, “referentiality” (abhidhā) and “intentionality” (tātparya, see Chapter 19 in this volume). The referentiality of words explains the individual words’ causation of knowledge of their referents, which interests us here. What is the ontological nature of such referents? Are they universals or particulars? I will here try to decode Jayanta’s peculiar theory of tadvat, the “that-possessor.” But first of all, what exactly is a śakti according to him?

2  Jayanta on Causation As an exponent of Nyāya, adept to theories of causation, Jayanta certainly uses the term śakti with purpose and defines it as “a concomitance of assisting factors

Semantic Relations and Causation of Verbal Knowledge

227

qualified by the capability to bring forth an effect,” in short, a causal force.2 Jayanta frequently resorts to this term when discussing causal relations. But what exactly is a causal force, in the metaphysical map of Nyāya-Vaiśeṣika? It is certainly not counted as a substance, while it is discussed as a property of substances. For instance, Jayanta mentions śakti along with other properties as distinguishing factors that prove the plurality of entities (dharmin) (NM, II.42,15–43,2),3 while his commentator Cakradhara includes śakti among the “delimiting properties” (dharma2).4 The exact collocation of śakti in the NyāyaVaiśeṣika paradigm would require a deeper discussion, but for the present purpose, it suffices to say that referentiality and intentionality are understood by Jayanta as the two types of causal forces belonging to units of meaning, that is, words. In his general discussion on causation, Jayanta distinguishes two types of causal forces: (1) fixed or inherent (avasthitā) and (2) situational or accidental (āgantukī)—in the causation of a earth vessel, for instance, the material causes, the atoms of earth, have an inherent causality, while the combination of the potter’s stick and wheel has an accidental causality.5 Now, in the linguistic context that Jayanta takes into consideration, the two causal forces discussed are referentiality and intentionality. Referentiality is the causal link between words and their referents, and Jayanta, as every other Nyāya author, clearly considers it as a conventional link and therefore a contingent force, in disagreement with the Mīmāṃsā tenets of permanent sound and of a fixed word-meaning relation. As for intentionality, which is conditioned by the situational presence of other factors, it must necessarily be contingent as well. Jayanta tries to find a synthesis of the semantic views of Nyāya and Mīmāṃsā also in their description of language acquisition (vyutpatti). Yet he cannot help but remark that the two schools clash in their understanding of the temporal status of the relation between words and their referents, which begins at the time of creation according to Nyāya, but it is beginningless according to Mīmāṃsā. This theoretical quibble, however, may not impact the observation of presentday learning of language. Mīmāṃsā authors, in any case, have an additional problem to solve, namely that a fixed word-referent relation implies a natural or inherent referentiality of words, which leads them to complicate the process of language acquisition with needless postulations that one can spare in the Nyāya model (NM, I 600,13–6).6 Referentiality must be a conventional force because it is based on the wordreferent relation established by God at the dawn of creation (NM, II 251,5–7).7

228

Alessandro Graheli

Even when a child learns a language by observing the senior speakers’ usage, what he learns is the relation of given words to given referents. Such wordreferent relations cannot be permanent or fixed, which means that referentiality is also not a fixed or permanent power, but rather the causal force depending upon God’s stipulation (NM, I 600,6–10).8 Jayanta, consequently, argues that this word-referent causal relation can be known by means of just two epistemic instruments—namely, perception and inference—and not with the burden of a third one, as the Bhāṭṭas believe (ŚV, sambandha 141). This is because one does not need to resort to postulation (arthāpatti) in order to know the role of the invisible denotative force (NM, I 600,11–16).9 In conclusion, the word-referent relation is not permanent or fixed, and the denotative force is not natural (naisargikā), but is rather contingent, since it is based on the convention established by God (NM, I 601,17–602,21).10 Applying this distinction to the two causal forces of words, it appears that both referentiality and intentionality are situational forces. The conventional nature of the signifier-signified relation is a necessary assumption in the Nyāya epistemology of testimony, which is founded on the trustworthiness of the authority and, consequently, on the authoriality of the Veda.

3  Ontology of Referents In Mīmāṃsā, the meaning of a word is distinct from that of a sentence, according to the principle formulated by Śabara as “the word conveys the universal (sāmānya), the sentence the particular (viśeṣa).”11 This is Jayanta’s own frame of reference when he argues that the referent of a word is neither a universal nor a particular, but rather a complex entity that he calls “thatpossessor” (tadvat), an ambiguous expression that accounts for both aspects of a referent, universal and particular (see Chapter 7 in this volume for an appraisal of tadvat before Jayanta’s time).12 This difference between the Mīmāṃsā and the Nyāya versions of direct reference is an important factor in the development and in the assessment of the two parallel theories. One must keep in mind that both mainstream Nyāya and Mīmāṃsā schools pride themselves with a realist and empiricist approach to ontology. Jayanta, accordingly, assumes an external reality of objects of knowledge, independently of our thinking.13 This certainly applies even to objects known through words. A denoted object, according to Nyāyasūtra, 2.2.67

Semantic Relations and Causation of Verbal Knowledge

229

(vyaktyākṛtijātayas tu padārthaḥ), can be a particular, a form, or a universal. While referring to this aphorism, Jayanta asks (NM, II 47,5–6): Having thus established that the referent is external […] now the question is raised: is [a word] the conveyor of a form, of an individual, or of a universal?14

Even before Jayanta, this passage was the topic of debate among Buddhists, Mīmāṃsakas, and Naiyāyikas. Dignāga rejected the theory that common nouns refer to universals and saw, in particular, co-referential usages (sāmānādhikaraṇya) as a problem in this theory: “If the word ‘existent’ referred the universal ‘existence,’ then it would not be co-referential with the words denoting particulars, such as ‘substance,’ etc., and there would not be such expressions as ‘existent substance,’ ‘existent quality’ and the like.”15 Dignāga also hinted, with disapproval, at the Nyāya notion that the referent of words is a that-possessor: “A class noun is not the referent of a particular possessing a universal, because it is not independent in referring to such a universalpossessing particular.”16 Jayanta’s position on this that-possessor has been articulated in Ganeri 1996 and Ganeri 1999, § 4.1, 4.2, so here I will address some aspects not discussed by Ganeri, as well as some nuances of his interpretation.17 Jayanta introduces the concept of “possessor of that” in defense of the Nyāya theory of reference, to the two main objections raised by Buddhist and Mīmāṃsā authors on the need to account for co-referential usages and for indirect or figurative speech, respectively: In other usages, such as [the injunction] “donate a cow!,” they said that the padārtha of a word is the that-possessor (tadvat), as proven by its practical efficacy (arthakriyā). The word directly expresses (abhijalpati) the artha, i.e., the that-possessor.18 And there is no impeded flow of thought, nor is [the word] overburdened [by the two tasks of direct and indirect signification]. Since even the co-referential usage (sāmānādhikaraṇya) can be justified by direct reference, it does not need to be explained in another way [by figurative signification, etc.]. Therefore the referent of the word is the that-possessor.19

Jayanta continues with the definition of tadvat: [Objection:] What exactly is this “that-possessor”? [Reply:] The very particular (viśeṣa) that possesses the universal (sāmānya) is called “that-possessor,” i.e., “this possesses that.” If the particular were referred to by a word, endlessness and ambiguity would ensue. The universal, however, while not expressed by a word, is not a contingent characteristic [of a particular,

230

Alessandro Graheli

i.e., it must be its inherent characteristic]. And if both the particular and the universal are to be referred to by one same word, there is an excessive burden on that word.20

Here Jayanta refers to the two great puzzles that have engaged all Sanskrit authors involved in theories of reference: endlessness and ambiguity. If common nouns such as “cow” are taken to refer to particulars, two major problems arise. By “endlessness” (ānantya) it is meant that since there are extensionally endless referents of a common noun, it is absurd to claim that it refers to an individual thing, and by “ambiguity” (vyabhicāra) that a common noun may then refer to heterogeneous entities such as substances, qualities, and so on, as in the case of the word sat (“existent,” “real thing”), which may refer to a quality as well as to a substance (Hattori 1996, 337; 2000, 141). In Ganeri’s rendition (Ganeri 1999, 103, translating NMShu, 296), the solution of the “that-possessor” is further articulated as follows: [Jayanta replies]: What is meant is this. The “property-possessor” (tadvān) is not a particular individual, such as Śābaleya, which is indicated by the word “this” [in “this has that”], and it is not the collection of all the individual [cows, say] in the world. It is the substratum of a universal. The aforementioned particular Śābaleya is said to be the “tadvān” because it is the substratum of the universal [cowhood], and so neither infinity nor discrepancy are relevant [objections]. Nor do we admit that a word denotates the qualificant [i.e., the particular] without denotating the qualifier [i.e., the property]. Since [someone who understands the word] knows a relation [between it] and a property-substratum, [the word] just means a tadvān. So where is the word’s excessive [semantic] burden?

An issue with this interpretation of Jayanta’s words is Ganeri’s use of the term “property,” where Jayanta’s speaks instead of sāmānya. This results in a paradigm shift, away from Jayanta’s realism, because in Jayanta’s passage sāmānya is intended as a universal, existing out there in the world. The following is my rendition of the same passage: [Jayanta’s synthesis:] We say that the that-possessor is not a particular, such as Śābaleya, enumerated by individuation (idantā). Nor is such a that-possessor the collection of all the individuals that pervades the world. Rather, any unspecified particular, such as the cow named “Śābaleya,” is a that-possessor. Since it is the substratum of a universal, there is no scope for the charges of endlessness and ambiguity. And we do not assume that śabda refers to a qualifier and then conveys a qualified particular, which would cause a fallacy of excessive burden on a single word. Since the conventional relation is already learned in relation to

Semantic Relations and Causation of Verbal Knowledge

231

this substratum of the universal, what would be the excessive burden of a word that expresses this from the very beginning? In this way, objections such as Dignāga’s “[the referent] cannot be the that-possessor, because of heteronomy” (tadvato nāsvatantratvād) stand refuted.21

This that-possessor, in Jayanta’s system, is an external reality and not a mental construct. The particular (e.g., a particular cow) is the substratum of the respective universal (e.g., cowness). This particular qualified by its universal (e.g., cow qualified by cowness), or that-possessor, is the referent of a common noun (e.g., “cow”). Only in this way the objections of endlessness and ambiguity, caused by particularism, can be neutralized. Since the very referent is a particular qualified by a qualifier, the operation occurs at once, by direct reference, and not in two separate moments. And so the charge of a semantic burden is defused. It is also worthwhile mentioning that there is some degree of flexibility in the application of the tadvat concept, according to context, in terms of the predominance of the particular, of the universal and the form. Although this flexibility may just be an apologetic attempt to respect the letter of Nyāyasūtra, 2.2.65, if taken seriously it means that for Jayanta the that-possessor is not just a universal-possessor, as often glossed and as most frequently used, but also a form-possessor, or even a particular-possessor. Of the two relata, the possessor is the predominant relatum, and the possessed aspect is the secondary one. In Jayanta’s words, Being the referent of a word is not restricted [to any of those three], due to a primary-secondary relation. [To explain:] Once established that what is expressed is the that-possessor, (1) in same cases the universal is primary and the individual secondary, as in the injunction “a cow should not be touched with the feet.” (2) In other cases the individual is primary and the universal is secondary, as in “release this cow,” “tie this cow” […]. (3) In other cases the form is primary and the individual is secondary, while the universal is not there at all, as in “cows made of flour should be modeled” […]. It is thus established that the referent of words such as “cow,” “horse,” etc., is the thatpossessor.22

4  Epistemology of Referents If the “that-possessor” is the referent of individual words, surely the content of inferential knowledge must be something different from that, since it concerns

232

Alessandro Graheli

complex states of affairs, that is, full-fledged sentences and not individual words. Jayanta writes:23 As far as their epistemic content, a word and an inferential mark are dissimilar. It will be established how the referent of a word is a “that-possessor.” Inference, instead, has the meaning of a sentence as its object, because from an inference a knowledge in the form of “here there is fire,” “there is a fire-possessing hill” derives. And it was already stated24 that the probandum of an inference is a property-possessor (dharmin) qualified by a property.25

Said otherwise, the content of an inference is a property possessed by an apprehended locus (pakṣa or dharmin) qualified by an apprehended property.26 The content of knowledge is thus the already perceived locus qualified by a previously unknown property (e.g., the hill qualified by fire); this is the meaning of a sentence, not of a single word. One may argue that this is not a sentence, but rather a complex word: why do we need to say that “the fire-qualified hill” is a sentence? The answer is that a necessary condition of a sentence is the satisfaction of expectancy (ākāṅkṣā):27 [Objection] There are also words with the function of the sentence meaning, such as “owner of cattle, descendant of Upagu, maker of pots.” [Counter-objection] True, but in those there is still expectancy, because without other words the fulfillment of expectancy is not achieved; the question “which owner of cattle?” is not satisfied.28

A further difference is that in inferences the qualifier of a qualified, that is, the property of a locus, is the object of discovery, while in words-derived knowledge it is the qualified, that is, the that-possessor: Moreover, from an inferential sign, knowledge of a qualifier such as “fire,” arises, based upon [perceptual] knowledge of the qualified, such as “hill.” From a word, instead, knowledge of the qualified arises, based on knowledge of the qualifier.29 Thus there is a different epistemic content.30

In inferential knowledge the previously unknown entity, that is, the probandum, is the qualifier (e.g., fire), which is known on the basis of the perception of a qualified entity (e.g., the hill). On the contrary, in word-derived knowledge the unknown entity that is revealed by the word “cow” is an individual “cow” qualified by a generic “cowness.” Even in the case of a complex word such as “cow-possessor,” the artha would still be an individual “cow-possessor” qualified by a generic “cowpossessorness.”

Semantic Relations and Causation of Verbal Knowledge

233

5 Conclusion Ganeri 1999, 104–105, argues that Jayanta’s tadvat theory is flawed for two reasons: (1) Jayanta “clearly cannot take ‘property-possessor,’ the direct object in his meaning specification, as standing for a certain particular”; (2) “if this phrase is mentioned,” it becomes tautological, because “the meaning clause becomes a mere restatement of the fact that ‘A-hood-possessor’ (A-tvavān) and ‘A’ are synonyms, which is a consequence of the fact that the abstraction and possession affixes are inverses of one another.” Yet Jayanta endorses an external existence of particulars qualified by universals, and something possessing a universal is by definition a particular substance. Moreover, in a system in which universals are real, external entities, to say that “cowness-possessor” and “cow” are synonyms does not seem right. This because a universal “cowness” is not a mere abstraction derived from the word “cow,” but rather an entity as real as the particular “cow.” The distinction between a that-possessor and a dharmin, discussed in Section 4 above, shows by derivation a distinction among universal and property, if by “property” we intend dharma: while the former is a universal in an ontological sense, the latter does not need to be so, because a property, in Nyāya inference, can well be any of the Nyāya-Vaiśeṣika categories. Lastly, it is important to stress how this “that-possessor” is specifically intended as the referent of words, as not the meaning of sentences, for which Jayanta developed the quite peculiar theory involving a second power of speech, intentionality.

Notes 1 The etiology of verbal cognition is studied in depth in NM, II.143–184, in the context of Jayanta’s refutation of the sphoṭa, and in NM, II.185–208, in the context of the signification of the sentence. 2 For example, NM, I.182,11: svarūpasahakārisvabhāvaiva; NM, I.403,8: yogya​ tāvac​chinn​asvar​ūpasa​hakār​isann​idhān​am eva. See also NK, s.v. kāraṇaniṣṭhaḥ kāryotpādanayogyo dharmaviśeṣaḥ. For this and other quoted passages of the NM, I will supply substantive variants from the relevant manuscript materials and occasionally edit the text. A list of available manuscript sources of the NM can be found in Graheli 2012, and an explanation of the criteria used in the choice of the best sources in Graheli. An in-depth study of the history of the transmission of the NM can be found in Graheli 2015.

234

Alessandro Graheli

3 api cāsmanmate bhinnaiḥ dharmair yuktasya dharmiṇaḥ / dharmo ’sya kenacit kaścit pratyayena grahīṣyate // vicitrasahakāryādiśaktibhedaś ca dharmiṇaḥ / nān opādhyupakārāṅgaśaktyabhinnātmatā kutaḥ // 4 avacchedā dharmāḥ, cf. Granthibhaṅga, 136,11–17. 5 NM, II.403,8–10: saiveyaṃ dvividhā śaktir ucyate avasthitā āgantukī ca. mṛttvādyavacchinnnaṃ svarūpaṃ avasthitā śaktiḥ. āgantukī ca daṇḍacakrādisaṃgarūpā. 6 eṣa eva cāvayor viśeṣaḥ yad eṣa śabdārthasambandhavyavahāraḥ, tavānādiḥ mama tu jagatsargāt prabhṛti pravṛtta iti / adyatve tu śabdārthasambandhavyu­ tpattau tulya evāvayoḥ panthāḥ / tatrāpi tv ayaṃ viśeṣaḥ yat tava śaktiparyantā vyutpattiḥ mama tu tadvarjam iti / 7 nanu naiyāyikānāṃ vā na samayaḥ pratipattyupāyaḥ / satyam / sa tv īśvarapraṇītaḥ prathamasargāt prabhṛti pravṛttaḥ mīmāṃsakābhyupagata naisargikaśaktisodarya eva na mādṛśaracitaparimitaviṣayasamayasamānaḥ / 8 yatrāpi ca vṛddhebhyaḥ vyavaharamāṇebhyaḥ vyutpadyate tatrāpi iyad evāsau jānāti, ayam arthaḥ amutaḥ śabdāt anena pratipanna iti / na tv anyā asya kācic chaktir astīti / iyatyaiva ca vyutpattyā śabdād arthapratyayopapatteḥ asyāś cāparihāryatvāt / adhikakalpanābījābhāvāc ca na nityaḥ śabdārthasambandhaḥ / 9 ata eva ca sambandhaḥ tripramāṇakaḥ iti yat tvayocyate tad asmābhir na mṛṣyate / śabdavṛddhābhidheyāṃś ca pratyakṣeṇātra paśyati iti satyam / anyathānupapattyā tu vetti śaktiṃ dvayātmikām ity etat tu na satyam / śrotuś ca pratipannatvaṃ anumānena ceṣṭayā ity etad api satyam / anyathāpy upapatter ity uktatvāt / tasmāt dvipramāṇakaḥ sambandhaniścayaḥ na tripramāṇakaḥ / 10 tad evaṃ śabdasya naisargikaśaktyātmakasambandhābhāvāt īśvaraviracitasamayanibandhanaḥ śabdārthavyavahārah nānādiḥ / […] tasmāt īśvar​avira​citas​amban​dhādh​igamo​pāyab​hūtav​ṛddha​vyava​hāral​abdha​tadvy​utpatti­ sāp​ekṣaḥ​ śabdaḥ artham avagamayatīti siddham / 11 ŚBh, 1.1.24, p. 112,1: sāmānye hi padaṃ pravartate viśeṣe vākyam. 12 The tadvat theory was already discussed before Jayanta’s time, by Uddyotakara in response to the Buddhist Dignāga. See Nyāyavārttika, 1.1.1, p. 4,16; 1.1.3, p. 28,3; 1.1.29, p. 100,3; 2.2.66, pp. 306–312, and Chapter 8 in this volume. 13 Cf. NM, II.540,16, where Jayanta distinguishes the teleological from the ontological use of the word (arthaḥ arthyamānaḥ ucyate, na vasturūpa eva, abhāvasyāpi prayojanatvasaṃbhavāt) 14 evaṃ siddhe bāhye ’rthe […] adhunā vivicyate gośabdaḥ kim ākṛter vācakaḥ uta vyakteḥ atha jāter iti. 15 See Hattori 1996, 387 and Kataoka, Chapter 8, in this volume. 16 Here I adapted the translation from Hattori 2000, 142. The Sanskrit version of the Pramāṇasamuccaya passage has been reconstructed in Dvādaśāram Nayacakram, 607 as tadvato nāsvatantratvād upacārād asambhavāt / bhinnatvād buddhirūpasya

Semantic Relations and Causation of Verbal Knowledge

17

18 19

20

21

22

235

rājñi bhṛtyopacāravat. It is also quoted in Granthibhaṅga, 137–138, albeit with vṛttirūpasya bhinnatvād in the third foot of the verse. Ganeri evaluated Jayanta’s views on tadvat on the basis of the following passage of the NM, using a less reliable edition of the NM (NMShu). In this chapter I’m using the more reliable text of NM and, wherever necessary, emended it on the basis of the two best manuscripts, BORI 390 and MDUC 2606 (NM, II 59,13–16). My translation can be compared with Ganeri’s own provided in note. In the BORI 390 reading, “they unhesitatingly assert that the artha is exactly the that-possessor, i.e., of the word.” anyeṣu tu prayogeṣu gāṃ dehīty(dehīty] dogdhīty MDUC 2606) evamādiṣu / tadvato ’rthakriyāyogāt tasyaivāhuḥ padārthatām // padaṃ tadvantam (padaṃ tadvantam] padatadvantam BORI 390) evārtham āñjasyenābhijalpati (-bhijalpati] -bhijalpanti BORI 390) / na ca vyavahitā buddhir na ca bhārasya gauravam // sāmānādhikaraṇyādivyavahāraś ca (ca] pi NM) mukhyayā / vṛttyopapadyamānaḥ (vṛttyopapadyamānaḥ] vṛttyopāpādyamānaḥ NM) san nānyathā yojayiṣyate // tasmāt tadvānn eva padārthaḥ (padārthaḥ] śabdārthā MDUC 2606) // nanu ko ’yaṃ tadvān nāma. tad asyāstīti tadvān iti viśeṣa eva sāmānyavān ucyate. viśeṣavācyatve cānantyavyabhicārau tadavasthau. sāmāṇyaṃ tu śabdenānucyamānaṃ nopalakṣaṇaṃ (nopalakṣaṇaṃ] nopalakṣyamāṇaṃ NMShu NM) bhavati. ubhayābhidhāne ca śabdasyātibhāra ity uktam. Ganeri (1999, 103), quoting NMShu, 295, translates: “[An objection is raised to this]: What is this thing called a ‘tadvān’ (‘that-possessor’)? ‘Tadvān’ literally means ‘this has that’ (tad asyāsti), so what is meant is that a particular is the owner of a property. But if it is the particular which is the denotatum, then the infinity and discrepancy faults recur, [especially] since the property is not [considered by you to be] an undenotated indicator (upalakṣaṇa). And if both [particular and property] are denotated, then the word has an excessive [semantic] burden.” NM, II 63,14–64,6: ucyate. nedantānirdiśyamānaḥ śābaleyādiviśeṣas tadvān, na ca sarvas trailokyavartī vyaktivrātas tadvān, kintu sāmānyāśrayaḥ (sāmānyāśrayaḥ […] pīḍayema] om. MDUC 2606) kaścid anullikhitaśābaleyādiviśeṣaḥ tadvān ity ucyate. sāṃāṇyāśrayatvān (sāṃāṇyāśrayatvān] sāṃāṇyāśrayatvāc ca NMShu NM) nānantyavyabhicārayos tatrāvasaraḥ. na ca viśeṣaṇam abhidhāya (abhidhāya] anabhidhāya NM) viśeṣyam abhivadati) (abhivadati] abhidadhāti NMShu NM) śabda ity upagacchāmaḥ, (upagacchāmaḥ] abhyupagacchāmaḥ NMShu NM) yenainam atibhāreṇa pīḍayema. (pīḍayema] pīḍayemahi NMShu NM) sāmānyāśrayamātre saṅketagrahaṇāt tāvanmātraṃ vadataḥ śabdasya ko (ko] kataro NMShu) ’tibhāraḥ. evaṃ ca (ca] om. NMShu NM) tadvato (tadvato] tadvator NMShu) nāsvatantratvād ityādidūṣaṇaṃ parihṛtaṃ bhavati. guṇapradhānabhāvasyāniyamena śabdārthatvam. sthite ’pi tadvato vācyatve kvacit prayoge jāteḥ prādhānyaṃ vyakter aṅgabhāvaḥ yathā gaur na padā spraṣṭavyā iti

236

23

24 25

26

27 28

29 30

Alessandro Graheli sarvagavīṣu pratiṣedho’vagamyate. kvacid vyakteḥ prādhānyaṃ jāter aṅgabhāvaḥ yathā gāṃ muṃca gāṃ badhāna iti niyatāṃ kāṃcid vyaktim uddiśya prayujyate. kvacid ākṛteḥ prādhānyaṃ vyakter aṅgabhāvaḥ jātis tu (tu] om. NM) nāsty eva yathā piṣṭamayyo gāvaḥ kriyantām iti tat (tat] om. NM) sanniveśacikīrṣayā prayoga iti. sarvasarvagatatve ’pi jāter (pi jāter] tyajāter MDUC 2606) na mṛdgavakādau (mṛdgavakādau] mṛdgavakādo MDUC 2606; mṛdgavādau NM) viśiṣṭā (viśiṣṭā] iṣṭā BORI 390) vṛttir ity uktam. tad evaṃ gavāśvādi (gavāśvādi] gavādi MDUC 2606) śabdānāṃ tāvat tadvān artha iti siddham. Cf. ŚVRa, śabda, 56–57ab: dharmī dharmaviśiṣṭaś ca liṅgīty etac ca sādhitam / na tāvad anumānaṃ hi yāvat tadviṣayaṃ na tat // sāmānyād atiriktaṃ tu śābde vākyasya gocaraḥ. In ŚVRa, anumāna 47cd, tasmād dharmaviśiṣṭasya dharmiṇaḥ syāt prameyatā; or in the NM section on inference, see NM, I 309,10. NM, I 404,10–12: viṣayas tāvad visadṛśa (visadṛśa] dhisadṛśa MDUC 2606) eva padaliṅgayoḥ. tadvanmātraṃ padasyārtha iti ca (ca] hi MDUC 2606) sthāpayiṣyate. anumānaṃ tu vākyārthaviṣayam, atrāgniḥ, agnimān parvata iti tataḥ (tataḥ] om. MDUC 2606) pratipatteḥ. uktaṃ ca tatra dharmaviśiṣṭo dharmī sādhya iti. In NBh, ad 1.1.35 the probandum of an inference is said to be either the property qualified by the property-possessor or the property-possessor qualified by the property (sādhyaṃ ca dvividhaṃ dharmiviśiṣṭo vā dharmaḥ śabdasyānityatvaṃ dharmaviśiṣṭo vā dharmy anityaḥ śabda iti). Elsewhere, in NM, I 310,5, however, Jayanta seems to dismiss the latter option. Cf. also ŚVRa, śabda, 59cd: vākyārthe ’pi padaṃ yatra gomadādi prayujyate. NM, I 405,1–5: nanu padāny api vākyārthavṛttīni saṃsanti (saṃsanti] śaṃsati MDUC 2606; santi NM) gomān aupagavaḥ kumbhakāra iti. satyam, kintu teṣv api sākāṅkṣatāsty eva, padāntaram antareṇa nirākāṅkṣapratyayānutpādāt. gomān ka ity ākāṅkṣāyā anivṛtteḥ (ākāṅkṣāyā anivṛtteḥ] ākāṅkṣān ativṛtteḥ MDUC 2606). Granthibhaṅga, 72: “because from the word ‘cow’ comes knowledge of an individual object qualified by cowness” (gośabdād gotvaviśiṣṭapiṇḍāvagateḥ). NM, I 405,6–7: api ca parvatādiviśeṣyapratipattipūrvikā pāvakādiviśeṣaṇāvagatir liṅgād udeti. padāt tu viśeṣaṇāvagatipūrvikā viśeṣyāvagatir iti viṣayabhedaḥ.

Primary Sources BORI 390

Pune, Bhandarkar Oriental Research Institute, Manuscript no. 390/1875–76. 432 foll. birch-bark, bound, Śāradā script.

Nyāyadarśana

Thakur, Anantalal, ed. 1997a. Gautamīyanyāyadarśana with Bhāṣya of Vātsyāyana. Nyāyacaturgranthikā 1. New Delhi: Indian Council of Philosophical Research.

Semantic Relations and Causation of Verbal Knowledge

237

MDUC 2606

Thenjipalam (Malappuram District), Malayalam Department of the University of Calicut, Manuscript no. 2606. 188 foll. palmleaf, Grantha-Malāyālam script.

Mīmāṃsādarśana

Abhyankar, Kashinath Vasudev, and Ganesh Shastri Ambadas Joshi, eds.1970a–1974. Mīmāṃsādarśana. Anandasramasamskrtagranthavali 97. Pune: Anandasrama.

NBh

Nyāyabhāṣya of Vātsyāyana. In Nyāyadarśana.

NM

Shukla, Suryanarayana, ed. 1936. Nyāyamañjarī: Gautamasūtratātpāryavivṛti. 2 vols. Kashi Sanskrit Series 106. Benares: Jaya Krishna Das Haridas Gupta, Vidya Vilas Press.

Shu

Nyāyasūtra

Nyāyasūtra of Gautama. In Nyāyadarśana.

Nyāyavārttika

Thakur, Anantalal, ed. 1997b. Nyāyabhāṣyavārttika of Bhāradvāja Uddyotakara. Nyāyacaturgranthikā 2. New Delhi: Indian Council of Philosophical Research.

ŚBh

Śābarabhāṣya of Śabara. In Mīmāṃsādarśana.

ŚV

Dvārikādāsa Śāstrī, ed. 1978. Ślokavārttika of Śrī Kumārila Bhaṭṭa with the Commentary Nyāyaratnākara of Śrī Pārthasārathi Miśra. Prāchyabhārati Series 10. Varanasi: Tara Publications.

Secondary Sources Ganeri, Jonardon. 1996. “Meaning and Reference in Classical India.” Journal of Indian Philosophy 24 (1): 1–19. Ganeri, Jonardon. 1999. Semantic Powers. New York: Oxford University Press. Graheli, Alessandro. “The Choice of the Best Reading in Jayanta Bhaṭṭa’s Nyāyamañjarī.” In The Study of Asia between Antiquity and Modernity, ed. by Elisa Freschi et al., 107–22. Special issue of RSO (Rivista degli Studi Orientali), no. 84. Rome: University of Rome “Sapienza.” Graheli, Alessandro. 2012. “A Preliminary List and Description of the Nyāyamañjarī Manuscripts.” Journal of Indian Philosophy 40: 317–37. doi:10.1007/s10781-0129155-2. Graheli, Alessandro. 2015. History and Transmission of the Nyāyamañjarī. Critical Edition of the Section on the Sphoṭa. Wien: Verlag der Österreichischen Akademie der Wissenschaften. Hattori, Masaaki. 1996. “Discussions on Jātimat as the Meaning of a Word.” In Śrījñānāmṛtam. A Memorial Volume in Honour of Prof. Shri Niwas Shastri, ed. by Vijaya Rani. Delhi: Parimal Publications.

238

Alessandro Graheli

Hattori, Masaaki. 2000. “Dignāga’s Theory of Meaning. An Annotated Translation of the Pramāṇasamuccayavṛtti. Chapter V: Anyāpoha-parīkṣā (I).” In Wisdom, Compassion, and the Search for Understanding. The Buddhist Studies Legacy of Gadjin M. Nagao, ed. by J.A. Silk, 137–46. Studies in the Buddhist Traditions 3. Honolulu: University of Hawai’i Press. Jhalakīkara, Bhīmācārya. 1996. Nyāyakośa or Dictionary of Technical Terms of Indian Philosophy. Bombay Sanskrit Series 49. Or. ed. 1875. Pune: Bhandarkar Oriental Research Institute.

12

Human Intellect and God’s Will in Navya Nyāya Semantics Yoichi Iwasaki

1  Fake Liberty? A close study of the debate between Navya Nyāya and Mīmāṃsā on the meaning of general nouns (jātiśabda) indicates that the Mīmāṃsā hermeneutics implies the concept of “meaning” (artha) of a word as what ought to be understood by a word, whereas the Nyāya epistemology regards the meaning as what is actually understood by a word. This difference is perhaps a consequence of the difference in their scholarly projects—Mīmāṃsā is concerned with rule interpretation, while Nyāya mainly studies human intellectual faculties and epistemology. The Nyāya semantics, which is built upon the actual use of language, grants our intellect the right to define the meaning of words, thereby contrasting with Mīmāṃsā’s normative semantics. However, when we compound this perspective with another Navya Nyāya semantic thesis—that the meaning of a word is defined by God—it is revealed that we have no cognitive liberty, and we can know only the things God wants us to know. This chapter exposes this hidden bondage of our intellect. It should be noted, first, that there are different levels of meaning in Nyāya semantics. A direct meaning (śakya) is the object directly related with the word. A secondary or figurative meaning (lakṣya) is related with the word via the direct meaning. The direct relationship between a word and its direct meaning is called śakti, which I translate as “semantic relationship.”1 There are more levels of meaning recognized by Indian philosophers, but the present chapter is only concerned with the direct meaning. The direct meaning has the feature that it is understood purely through one’s linguistic faculty, while the secondary meaning requires one to be aware of the context and to employ inference as well.2 In this

240

Yoichi Iwasaki

framework, I will examine two questions in the following order: “What is the (direct) meaning of a word?” and “What is the semantic relationship?”

2  Meaning of “Meaning”3 The perennial debate on the meaning of general nouns—Is it the universal (jāti), the particular (vyakti), or both?—still goes on in the period of Navya Nyāya.4 Gaṅgeśa (fourteenth century) of the Navya Nyāya tradition discusses this issue in the section of general semantics (śaktivāda) of his Tattvacintāmaṇi. According to it, all parties that join the debate—the Bhāṭṭa school of Mīmāṃsā, a person called “Maṇḍana,”5 the standard Prābhākara school of Mīmāṃsā, a group of the Prābhākara school called “supporters of the curbed-denotation theory” (kubjaśaktivādin), and Gaṅgeśa himself—agree that we grasp both the universal and the particular when we hear a general noun. For example, when we hear the expression, “Bring a cow,” both the feature shared by all cows—let us call it “cowness”—and a particular cow specimen appear in our mind. All Mīmāṃsā philosophers conclude that only the universal, for example, cowness, is the meaning of the word, and that the particular is something else. How do we grasp the particular, if it is not meant by the word? They have different answers to this question. The Bhāṭṭa school claims that we grasp it by force of logical or ontological implication (ākṣepa),6 while the person called “Maṇḍana” is said to hold the view that the particular is brought into our cognition by the force of figurative signification (lakṣaṇā).7 Both of them assume that the universal and the particular are grasped by different acts of cognition, but the Prābhākara school does not agree with this assumption. This school holds the rule that the universal and the particular have to be grasped by single cognition (eka-vittivedyatva-nyāya or the rule of single cognition). Yet, the meaning of the word is still considered to be only the universal. The school gives various explanations of how we grasp the particular, all of which are based on the assumption that we grasp it by the innate capacity of our cognitive (perhaps nonlinguistic) faculty, without the word’s direct reference to it.8 Gaṅgeśa also accepts the rule of single cognition, but he supports his school’s legacy view that the particular is also a part of the word’s meaning. Since the universal and the particular are always present together in our cognition, the particular must also be meant by the word. Therefore, although the Prābhākara school and Gaṅgeśa share most of the points, they nevertheless differ in conclusion: the Prābhākara insists that the

Human Intellect and God’s Will in Navya Nyāya Semantics

241

particular must not be called the meaning, while Gaṅgeśa considers it a part of the meaning. These respective conclusions have in fact been established since the time of the Mīmāṃsāsūtra and Nyāyasūtra. The argument in the Tattvacintāmaṇi does not add anything substantially new, yet it gives us new insight into philosophical semantics, because it juxtaposes the two philosophies’ differences in their conception of “meaning.” I hypothesize as follows: Mīmāṃsā philosophers generally seem to share the understanding that the meaning of a word is what ought to be understood by the word. The word “cow” in a Vedic text requires us to understand cowness, but it does not specify which particular cow should be understood. The understanding of the particular is the interpreter’s responsibility. Gaṅgeśa’s argument, by contrast, shows his understanding that the meaning of a word is what is actually understood from the word. If something is understood from a word, and nothing but the word is affecting its understanding, it must be meant by the word.

3  God’s Will as Semantic Relationship There is another semantic argument of Nyāya philosophers that should be considered here. An aphorism of Nyāyasūtra states that the understanding of word meaning is dependent on a samaya,9 which literally means “coming together.” It is, according to the commentator, “the rule as to the fixed correspondence between a word and the meaning.”10 A philological study suggests that we should understand it as “a convention which relates things so far unrelated” rather than the “established custom” (Houben 1992). A convention may be created by somebody or some people. But who created this semantic convention? Vācaspati (tenth century) in his Tātparyaṭīkā said that it was God,11 after which it has commonly been accepted by Navya Nyāya philosophers that the semantic relationship is ontologically equal to God’s will, as God creates the relationship at will. For instance, since He desired that a pot is understood by the word “pot,” the pot is the meaning of the word “pot.” No listener can understand the meaning of a word without knowing God’s will as the semantic relationship. At first, she has to learn the relationship, although she is not aware that it is God’s will, and in the next encounter with the word, the word will remind her of its meaning—which according to Nyāya is both the universal and the particular, or more accurately, the particular qualified by the universal.12

242

Yoichi Iwasaki

4  Cognition Governed by God’s Will I am deriving seemingly conflicting conclusions of Navya Nyāya philosophers from my investigations into the two questions, “What is the meaning of a word?” and “What is the semantic relationship?” 1. The meaning is what is actually understood by the word. 2. The meaning is what ought to be understood by the word. We must be aware that the two arguments pertain to different domains of semantics: the first argument is concerned with the semantic function or mechanism of signification of words, and the second argument is about the lexical semantics or vocabulary. To be more accurate, Navya Nyāya semantics is nonnormative as to the word’s function, but normative as to the lexical semantics. There is no contradiction as long as these two domains are not confused. We can ask, however, whether or not the second conclusion is applicable to the argument of the word’s function, and the answer is, I think, yes. To ask what is meant by a word is to ask what is related to a word by the semantic relationship. The Nyāya’s answer is that both the universal and the particular are related, because we are grasping both of them at once. This is what the first conclusion implies. Now, the second argument entails that the relationship between a word and the meaning is identified with God’s will, and God wants both the universal and the particular to be understood through a word. From this, it follows that we are grasping what ought to be grasped. The two conclusions are not contradictory, and so we can accept both. It is rather necessary to assume that we are grasping what God wants us to grasp. In the process of verbal understanding, for the listener to remember the word meaning, the semantic relationship has to be learned first. Therefore the chronological order should be as follows: (1) God desires the potness and the pots to be grasped by the word “pot,” thereby establishing the semantic relationship. (2) The listener learns the relationship by various methods of lexical acquisition. (3) Someday the same person hears the word “pot” being uttered. (4) She remembers the semantic relationship. (5) She cognizes the potness and a pot or pots. This suggests that it is not our intellect that defined the particular to be a meaning, but that God defined it. If He has no will that connects particulars

Human Intellect and God’s Will in Navya Nyāya Semantics

243

and the word, we would not be able to cognize particulars upon hearing words. Gaṅgeśa’s argument is actually not intended to define what is meant by a word, but to discover what God has defined as meaning. Let us ask why God wanted the particular to be grasped. To generalize the issue so that even atheists can join, we can also ask why a contemporary namer—for example, developers of Apple Computer—wanted not only iPhoneness or the essence of all iPhones, but also a particular iPhone to be grasped by the word “iPhone,” in the manner of those Navya Nyāya philosophers who acknowledge even human will as the semantic relationship, in the case of newly created words. Apple Computer included the particular into the range of meaning, because nobody in the company can think of iPhone-ness in isolation; the rule of single cognition teaches that the universal and the particular have to be grasped together. Due to this limitation of human intellect, Apple Computer cannot name only iPhone-ness as “iPhone.” If this limitation applies to God’s intellect, it is possible that God did not intentionally include the particulars in the meaning, but he was rather forced to do so due to this limitation. This would allow us to claim that the cognitive faculty, whether human or divine, defines what is meant by a word, and thus to keep the nonnormative semantics. It is also important to ask whether it is actually God who defined the lexical convention, because if the convention is human made we can ignore the issue of God’s transcendence of the limitation of human intellect. The lexical convention was originally attributed to naming or social agreement. It is not clear why later Naiyāyikas converted it to God’s convention. What was wrong for them with having a convention of human origin? This should be further investigated.

Notes 1 Nyāyasiddhāntamuktāvalī, 265: śaktiś ca padena saha padārthasya sambandhaḥ; Tattvacintāmaṇi, vol. 4 (2), p. 690: śakyasambandha eva lakṣaṇā. 2 There is a debate on whether or not the understanding of the secondary meaning is linguistic. See Das 2011 and Guha 2012. I take the position that the understanding of sentence meaning that contains secondary meanings is linguistic (śābda) in the sense that it is word generated, yet it depends on inference. 3 For a more detailed study on this topic see Iwasaki 2017. 4 For an earlier history of this debate see Deshpande 2007, Scharf 1996, and Dasti (Chapter 7) in this volume. 5 Later Navya Nyāya authors usually identify this person with Maṇḍana Miśra who wrote Sphoṭasiddhi and so on, but I have not found textual evidences for this identification.

244

Yoichi Iwasaki

6 Tattvacintāmaṇi, vol. 4 (2), p. 578: bhaṭṭamate tu jātir eva śakyā, lāghavāt. vyaktis tv ākṣepalabhyā. 7 Tattvacintāmaṇi, vol. 4 (2), p. 587: ata eva na vyakter ākṣepaḥ, kin tu lakṣaṇayā gopadād gaur iti vyaktidhīr iti maṇḍanaḥ. 8 Tattvacintāmaṇi, vol. 4 (2), pp. 562–64, which seems a reproduction of Śalikanātha’s argument that is presented in Ollett’s Chapter 13 of this volume. 9 Nyāyasūtra, 2.1.55: na, sāmayikatvāc chabdārthasampratyayasya. 10 Nyāyabhāṣya, on Nyāyasūtra 2.1.55, p. 89: kaḥ punar ayaṃ samahaḥ? asya śabdasyedam arthajātam abhidheyam iti abhidhānābhidheyaniyamaniyogaḥ. 11 Tātparyaṭīkā, on Nyāyasūtra 2.1.55, p. 370: […] tasmin niyogo boddhavya iti bhagavataḥ parameśvarasya sargādau, so ’yaṃ samaya ity arthaḥ. 12 Tattvacintāmaṇi, vol. 4 (2), pp. 588–89: tasmād ekavittivedyatvaniyamāt jātiviśiṣṭaṃ śakyam.

Primary Sources Nyāyadarśana

Thakur, Anantalal, ed. 1997. Gautamīyanyāyadarśana with Bhāṣya of Vātsyāyana. Nyāyacaturgranthikā 1. New Delhi: Indian Council of Philosophical Research.

Nyāyabhāṣya

Nyāyabhāṣya of Vātsyāyana. In Nyāyadarśana.

Nyāyasiddhāntamuktāvalī

Śukla, Harirāma, ed. 1997. Nyāyasiddhāntamuktāvalī of Viśvanātha Pañchānan Bhaṭṭāchārya. Kashi Sanskrit Series 6. Fifth edition. Varanasi: Chaukhambha Sanskrit Sansthan.

Nyāyasūtra

Nyāyasūtra of Gautama. In Nyāyadarśana.

Tātparyaṭīkā

Thakur, Anantalal, ed. 1996. Nyāyavārttikatātparyaṭīkā of Vācaspatimiśra. Nyāyacaturgranthikā 3. New Delhi: Indian Council of Philosophical Research.

Tattvacintāmaṇi

Kamakhyanath Sharma, Tarkavagish, ed. 1990. The Tattvacintāmaṇi of Gaṅgeśa Upādhyāya. 4 vols. Delhi: Chaukhamba Sanskrit Pratishthan. Original edition Calcutta: Asiatic Society, 1888–1901.

Secondary Sources Das, Nilanjan. 2011. “Lakṣaṇā as Inference.” Journal of Indian Philosophy 39: 353–66. Deshpande, Madhav M. 2007. The Meaning of Nouns: Semantic Theory in Classical and Medieval India. New Delhi: D.K. Printworld. Original edition Dordrecht: Kluwer Academic Publishers, 1992.

Human Intellect and God’s Will in Navya Nyāya Semantics

245

Guha, Nirmalya. 2012. “Lakṣaṇā as a Creative Function of Language.” Journal of Indian Philosophy 40: 489–509. Houben, J.E.M. 1992. “Bhartṛhari’s Samaya / Helārāja’s Saṃketa: A Contribution to the Reconstruction of the Grammarians’ Discussion with the Vaiśeṣikas on the Relation between ‘śabda’ and ‘artha.’” Journal of Indian Philosophy 20: 219–42. Iwasaki, Yoichi. 2017. “Meaning of ‘Meaning’: A Debate between Navya-Nyāya and Prābhākara Mīmāṃsā.” Indogaku Bukkyōgaku kenkyū (Journal of Indian and Buddhist Studies) 65 (3): 1082–88. Scharf, Peter M. 1996. The Denotation of Generic Terms in Ancient Indian Philosophy: Grammar, Nyāya, and Mīmāṃsā. Transactions of the American Philosophical Society Held at Philadelphia for Promoting Useful Knowledge 86.3. Philadelphia: American Philosophical Society.

246

Part Three

Of Sentence Meanings

248

Introduction

Even admitting that there are segments of speech below the sentence level, everyone agrees that effective speech, in actual practice, functions by sentences, not by isolated words and even less so by isolated phonemes. In the second part of the volume we have already seen (Chapter 9) how Mīmāṃsā authors were interested in sentences and their meanings, rather than in the semantics of individual words. Chapters 13 and 14 investigate the two sections of Śālikanātha’s well-reputed treatise on sentence meaning (Vākyārthamātṛkā), respectively. In the first section, Śālikanātha, an exponent of the Prābhākara sub-school of Mīmāṃsā, facing the criticism of the problematic epistemological value of speech at the level of the sentence, defended and improved the Mīmāṃsā stance by devising a new model of compositionality that leads from words, through their universal referents, to the sentence meaning, without resorting to the Bhāṭṭa theory of indirect signification that would compromise the epistemological status of Vedic injunctions. In this chapter, Śālikanātha explained in detail how the linguistic competence is acquired through a complex cognitive process that involves the repeated observation of the linguistic interaction among competent speakers. Śālikanātha, as other Mīmāṃsā authors, thought that sentences have an exhortative and not a descriptive content. Sentences, beginning from the sentences of the sacred scripture, the Veda, convey a duty and are not concerned with the existence of things, independently of the subjective role of the listener. In the second chapter, Śālikanātha discusses the compelling issue of language acquisition (vyutpatti) of such deontic meanings of sentences, which are by definition atemporal, unperceivable, and thus unknown. While Prābhākara authors claim that every sentence conveys a duty, Bhāṭṭa authors argued that this analysis applies to Vedic language but not to ordinary language, which rather conveys descriptions of states of affairs. Their thesis has important epistemological fallouts, because it implies that the Veda is an instrument of knowledge only in relation to duties. By contrast, sense

250

Introduction

perception, inference, and so on only convey knowledge about what exists. Thus the knowledge of duties conveyed by Vedic sentences cannot be falsified by sense perception or inference, which conveys a completely different set of contents. Mīmāṃsā authors developed their theories also in opposition to Vedānta authors, who claimed that Vedic sentences convey descriptive contents about the Absolute and the self (Chapter 15). The Vedānta author Prakāśātman craftily developed a linguistic theory of unitary meaning (akhaṇḍārtha), elaborating upon Bhartṛhari’s views and Mīmāṃsā theories. Prakāśātman faced a problem similar to the one tackled by Śālikanātha (Chapter 14): how is it possible for language, in the form of scripture, to produce knowledge about an object—namely the absolute, eternal, indivisible Brahman—that cannot be described in sentences without losing its essence? Prakāśātman’s work has a specific historical importance, because it contains a record of the main linguistic theories circulating in the tenth century, and also because it establishes a new autonomy of Vedānta in the linguistic field (Chapter 16). Navyanyāya authors argued that a necessary condition for the knowledge of a sentence meaning is the listener’s awareness of the speaker’s intention. One reason behind this specification is that a listener may not be able to disambiguate the meaning of a polysemous word without understanding the speaker’s will. But when we understand the correct statement that unintendedly slips out of a deceiver’s mouth, are we not entitled to say that we have understood the truth from the sentence? How do we explain the cognition caused by a parrot’s utterance? In Chapter 17 Gaṅgeśa’s solution to this conundrum is presented.

13

Śālikanātha’s “Introduction” to His “Fundamentals of Sentence Meaning” Andrew Ollett

1 Introduction Śālikanāthamiśra’s Fundamentals of Sentence Meaning (Vākyārthamātṛkā), perhaps composed in the early ninth century, is one of the most important contributions to the debate on sentence meaning in premodern South Asia.1 It is a two-part essay that is included in a set of twelve other writings, under the title of Commentarial Essays (Prakaraṇapañcikā). The first part, which is shorter than the second and framed as an introduction to it, develops a general theory of sentence meaning, according to which the words convey the sentence meaning directly by expressing their proper meanings in relation to each other. The second part examines the meaning of injunctive sentences in particular, and argues that the primary meaning of such sentences is “something to be done” (kāryam), which in the context of the Veda takes the form of what Mīmāṃsakas call “something previously unknown” (apūrvam).2 In both parts, Śālikanātha claims to be developing the insights of Prabhākara, the earlier scholar of Mīmāṃsā whom he identifies throughout his writings as his teacher. Both parts are written in the form of a set of verses—twenty-two in the first part, and around fifty in the second—accompanied by a commentary. The commentary is mostly prose, although it includes, besides verses quoted from other authorities, a number of summary verses (saṅgrahaślokāḥ) that summarize Śālikanātha’s position on various topics. The verses, when read on their own, present a relatively streamlined and coherent argument, and it is the verses of the introductory essay that I will focus on in this chapter. Śālikanātha did write several other essays in the form of a set of verses without any surviving commentary. The commentary generally provides additional context

252

Andrew Ollett

for the verses, often by discussing alternative views at length. He thus probably composed the verses first and added a commentary “as a favor to the virtuous,” as he says at the end of the essay.3 From the fact that Śālikanātha refers to the Fundamentals of Sentence Meaning several times in his Straightforward and Lucid (Ṛjuvimalā) on Prabhākara’s commentary, it appears that he composed the essay before undertaking his ambitious commentarial project.4 The commentary of the Fundamentals of Sentence Meaning includes quotations from other authorities, from which we can gain an idea not just of who his interlocutors were, and how he thought of them, but also what kind of intellectual project he was engaged in. Tellingly, Śālikanātha barely quotes his teacher Prabhākara. The authority that he quotes most often is Kumārila Bhaṭṭa, the most renowned Mīmāṃsaka of his time, and perhaps of all time. Because Śālikanātha is concerned to develop and defend Prabhākara’s insights, Kumārila’s views often come in for criticism. But Śālikanātha clearly held Kumārila in enormous esteem, and he borrows or builds upon Kumārila’s ideas as often as he criticizes them. Indeed, the theory that Śālikanātha develops in the Fundamentals of Sentence Meaning, although presented as Prabhākara’s theory, can be seen as a slight modification of Kumārila’s theory in order to take account of Prabhākara’s insights regarding the importance of context. To Śālikanātha, Kumārila was not just as an intellectual sparring partner, but also a kind of role model: Kumārila was a system builder, indeed the most successful system builder in the history of Mīmāṃsā, and Śālikanātha’s overarching goal was to build a system out of Prabhākara’s often obscure teachings. Śālikanātha quotes from Kumārila’s Explanation in Verse (Ślokavārttikam) and Explanation of the System (Tantravārttikam), as well as verses that are not found in either text and are thus likely to have been drawn from his Extensive Notes (Bṛhaṭṭīkā), which is now lost.5 These verses are often programmatic: they discuss, for example, the role of Mīmāṃsā as a system of thought and its distinctive epistemological and hermeneutical positions.6 Hence, although Śālikanātha created a system of “Prābhākara Mīmāṃsā” that would later come to be thought of, and indeed defined, as an exclusive alternative to the “Bhāṭṭa Mīmāṃsā” system created by Kumārila, Śālikanātha’s importance derives in part from his role as witness to, and interpreter of, Kumārila’s thought. Śālikanātha made two significant interventions in the first part of the Fundamentals of Sentence Meaning, one philosophical and one doxographic. The philosophical intervention was his presentation of what he called “the expression of relational meanings” (anvitābhidhānam). In one of his typical maneuvers, he took this relatively obscure phrase from the writings of

Śālikanātha’s “Introduction” to “Fundamentals of Sentence Meaning”

253

Prabhākara and developed it into a comprehensive theory.7 This theory was both compositional and contextualist.8 The principle of compositionality holds that the meaning of larger units of language somehow arises from the meanings of the smaller units out of which they are composed. This principle, or some version thereof, was widely accepted in ancient Indian thought. Within Mīmāṃsā, in particular, it was never seriously doubted that the meaning of a sentence arose from the meanings of its constituent words, although there was an extended debate over precisely how it so arose. The principle of contextualism holds that the meaning of any given expression depends on the context in which that expression appears, or alternatively, that “the meaning of a word in a given sentence is the contribution it makes to the meaning of the sentence” (Janssen 2001, 116). While all Mīmāṃsakas did accept some version of this principle— indeed, much of Mīmāṃsā can be described as interpretation in the light of context—no Mīmāṃsaka had been as radical a contextualist as Śālikanātha. He claimed that the same word expressed a different relational meaning in all of its different contexts of use. It is worth noting here that compositionality and contextualism are generally considered to be in tension (Janssen 2001, 117; Recanati 2012), if not in outright conflict: “Clearly it can’t be,” wrote Fodor (2003, 97), “that both are true.” The more the context seems to determine the meanings of the constituent words, the less those meanings appear to contribute to the overall sentence meaning. The interplay of compositionality and contextualism has been a major issue in contemporary philosophy of language, and Śālikanātha’s attempt to reconcile these two principles has ensured the philosophical relevance of his Fundamentals of Sentence Meaning up to the present day. Śālikanātha’s doxographic intervention was arguably even more important. As noted above, he called his theory “the expression of relational meanings” (anvitābhidhānam), borrowing the phrase from Prabhākara. The alternative to this theory was “the relation of expressed meanings” (abhihitānvayaḥ), which does not appear in Prabhākara’s writings. Śālikanātha appears to have invented this term, for although Śālikanātha describes Kumārila’s position with it, Kumārila himself does not use the phrase. Kumārila’s earliest commentator, Umbeka, uses the phrase “expression of relational meanings,” which he associates with Prabhākara, but never uses the complementary phrase “relation of expressed meanings,” at least in the surviving portion of his commentary.9 Indeed, the concept of a “relation” (anvayaḥ) between word meanings is characteristic of Prabhākara’s thought, and hardly appears at all in Kumārila’s work. In sum, Śālikanātha framed the debate on sentence meaning as a debate

254

Andrew Ollett

between two complementary positions, “the expression of relational meanings” and “the relation of expressed meanings.” These positions differed precisely according to how they accounted for the “relation” between word meanings, although historically only the first position offered such an account. Hence, like most doxographers, Śālikanātha had framed the debate in such a way as to favor his own position. Nevertheless his framing would prove to be enormously influential. Almost immediately, everyone who spoke about sentence meaning in South Asia spoke in the chiastic terms that Śālikanātha provided. The first to do so, it seems, was Bhaṭṭa Jayanta in the late ninth century CE, who reviewed “the expression of relational meanings” and “the relation of expressed meanings” in his discussion of sentence meaning in his Racemose Reasoning (Nyāyamañjarī). There is, however, no clear evidence that Jayanta had read Śālikanātha, which raises the possibility that both Śālikanātha and Jayanta had borrowed the chiastic framing from earlier followers of Prabhākara whose works are now lost. Subsequently, the theory of sentence meaning became a topic of intense debate in Mīmāṃsā, Nyāya, and Alaṃkāraśāstra, and the positions were precisely those that Śālikanātha had staked out.10

2  Sentence Meaning as a Determinate Particular To understand Śālikanātha’s argument in the Fundamentals of Sentence Meaning, we must begin with the notion of a “sentence meaning” that gives the essay its title. This phrase translates the Sanskrit compound vākyārthaḥ. The word arthaḥ has a notoriously broad range of meanings.11 Most pertinent to this discussion, it means the “meaning” of a particular linguistic expression, the “signified” corresponding to a given signifier. But it also means a “thing” that could be designated by a linguistic expression and, more broadly, to a “state of affairs.” The coincidence, or lack thereof, between these two senses was a long-standing topic of debate. One of Śālikanātha’s other essays, the Path of Reason (Nītipathaḥ), adopts a position midway between realism and mentalism: the “meaning” of a linguistic expression is a veridical “state of affairs” by default, because of its inherent power to express such a meaning, but in the case of everything except the Veda we can never be certain of this correspondence unless we have recourse to other sources of knowledge (e.g., by verifying through perception that the meaning of a particular statement is a real state of affairs). The inspiration for such a position certainly came from Prabhākara, who accepted that linguistic expressions are sources of knowledge

Śālikanātha’s “Introduction” to “Fundamentals of Sentence Meaning”

255

in and of themselves only when they occur in the Vedas (Jha 1978, 63–66). The third shade of meaning of arthaḥ is a “purpose,” that is, something sought after. This sense was particularly important for Mīṃāṃsakas, who generally thought of the unity of a text—whether a sentence, or a larger passage, or indeed of an entire work—as a teleological unity, in which all of the elements take their place in relation to a single overarching purpose.12 One of Jaimini’s sūtras says that “a sentence is one on account of there being just one arthaḥ,” and Śabara understands this latter word not simply as “meaning,” but also as “purpose” (prayojanam).13 The sūtra comes in the context of determining the boundaries of sentences in the Veda, though Kumārila noted that it will apply just as well to sentences outside of the Veda.14 Śālikanātha’s view of sentence meaning involves all three of these shades of arthaḥ: it is what is signified by a sentence; it is a state of affairs; and it possesses a teleological unity. Mīmāṃsakas agreed that there was a qualitative difference between “sentence meaning” and “word meaning.” It was not, in other words, simply that a sentence had a greater number of signifying elements in it than a word. For Mīmāṃsakas, word meanings were classes (jātiḥ): the meaning of the word “cow” is “cowness,” a class that inheres in, and is instantiated by, every particular cow. Sentence meanings, by contrast, were particulars (vyaktiḥ, viśeṣaḥ). As Śālikanātha notes, quoting from one of Kumārila’s lost works, the determination of the “sentence particular” (vacanavyaktiḥ) from a given sentence, on the basis of a set of rational principles, is precisely what Mīmāṃsā does.15 At this point, however, we must think about what it means for the meaning of a sentence to be a “particular.” For certain kinds of sentences, such as “the pot is on the ground,” it is easy to imagine that the particular expressed by the sentence is the state of affairs in which the pot is on the ground. From this perspective, the sentence particular seems similar to a proposition. Propositions, however, are commonly thought to be bearers of truth-values, and most of the sentences with which Mīmāṃsakas are concerned are injunctive, and therefore not capable of being evaluated as either true or false, at least not on the surface. What would be the proposition corresponding to a sentence such as “one should perform the agnihotram oblation as long as one lives”? This is not, of course, an insuperable problem, since many solutions have been offered for relating the propositional content of a sentence to other putative features of the sentence, such as its modality, force, and so on. For example, we could think of the aforementioned sentence as presenting some propositional content (“x performs the oblation as long as he lives”) as well as some additional elements that account for both

256

Andrew Ollett

the statement’s deontic force (what Mīmāṃsakas call the “injunctive meaning,” vidhyarthaḥ) and the binding of variables to agents in the “real world” (what Mīmāṃsakas call “eligibility,” adhikāraḥ). There is another puzzle related to the sentences with which Mīmāṃsakas are concerned, and that is their timelessness. When we speak of “sentence meanings” in the world, we almost always refer to sentences that are uttered by speakers in time. The context against which their meaning is determined and evaluated is thus a historical context. Indeed, the very name of “contextualism” is now largely associated with a movement in the history of ideas, and in particular with the work of Quentin Skinner.16 By contrast, Mīmāṃsakas emphatically denied that the sentences of the Vedas could be understood with reference to a historical context. Although this denial was motivated by considerations that we might call dogmatic—the imperative to preserve the Vedas as the unique and infallible source of knowledge about dharma—it requires an ahistorical contextualism. This is an interesting requirement from a theoretical perspective, since it implies that there is no privileged “context of utterance” against which the sentence meaning can be evaluated. Indexical terms such as “I” or “now,” which are typically taken as pointing to a particular context of utterance, are instead taken as pointing to an infinite series of such contexts. This amounts to a theory, albeit an implicit one, of how indexical terms are bound to their referents—of the “character” of such terms, in Kaplan’s terminology (Kaplan 1989). From a hermeneutical perspective, this requirement forces us to think of the “context” against which the meaning of a sentence can be evaluated not in historical terms, but in textual and logical terms: our interpretation of any given sentence, in other words, is constrained by our understanding of the text of which it forms a part and by our knowledge of what is logically possible in the world. This is not necessarily the case for sentences outside of the Vedas, where situational factors might enter into the context that is relevant for determining their meaning. Śālikanātha once characterizes sentence meaning as a “determinate particular” (nirdhāritaviśeṣaḥ). He does so in response to the argument that a sentence meaning is a general term (sāmānyam), since it is composed out of general terms, and that we arrive at a specific meaning (viśeṣaḥ) by implication (ākṣepaḥ) from the general term.17 The opponent compares this process to the implication of an individual from the class: the class “cowness” inheres in every individual cow, and indeed would not exist if some individual cow did not exist, and hence we can cognize “some individual cow” simply on the basis of the cognition of the class. Śālikanātha says, however, that the cognition

Śālikanātha’s “Introduction” to “Fundamentals of Sentence Meaning”

257

of a determinate particular is qualitatively different from the cognition of an implicated particular. Although Śālikanātha does not go into details here, we can think of two meanings for an expression like “some cow,” one non-determinate (“any cow at all”) and another determinate (“a certain cow”).18 Sentence meaning is determinate and thus cannot be known by implication from non-determinate general terms. We may wonder whether the distinction Śālikanātha had in mind corresponds in some way with the distinction in modern philosophy of language between intensions and extensions: it certainly seems as if the problem with the opponent’s proposal is that there is no direct path from general terms, which are intensional, to the extension expressed by a sentence. Yet it is not clear what it would mean for sentence meanings to be extensional, especially since propositions, which we had tentatively identified with sentence meanings above, are often taken to be intensions. Indeed, the distinction between intensions and extensions, which makes sense when speaking of individual terms, is harder to maintain when talking about complex states of affairs. Thus the theory of sentence meaning has to satisfy two requirements, compositionality and particularism. The meaning of the sentence should be a function of the meanings of the smaller expressions within it, and the final sentence must be a “determinate particular.” For Mīmāṃsakas, who hold that word meanings are classes, it will be difficult to satisfy both requirements simultaneously. At what point does the meaning of an expression make the qualitative shift from a class to a particular? Kumārila invoked secondary meaning (lakṣaṇā) to solve this problem (see Chapter 18 in this volume).19 If word meanings are classes, then a complex of word meanings is simply the class that represents their mutual intersection. We know, however, that sentences are used with reference to particular states of affairs. Hence, primary meaning fails to get us to the final meaning of the sentence, which is a particular. This failure triggers the process of secondary meaning, wherein we attempt to recover a meaning that is distinct from, but nevertheless related to, the primary meaning. If the primary meaning is a class, the secondary meaning could be an individual that belongs to that class—something that is both related to the primary meaning and fulfills our requirement that the meaning of the sentence is a particular. For Śālikanātha, this solution is inelegant, because it implies that we can never understand a sentence meaning without first failing to understand it. Moreover, as Lawrence McCrea has noted, the failure of primary meaning is a serious problem for Prabhākara and Śālikanātha, who are committed to the more general epistemological principle that veridical cognitions—such as the cognition of word meanings on the model of sentence meaning under

258

Andrew Ollett

consideration here—can never really be invalidated.20 Śālikanātha thus needs to find another way of resolving the tension between compositionality and particularism.

3  The Mutual Relation of Meanings Śālikanātha’s solution to the problems outlined above—the tension between contextualism and compositionality, and between compositionality and particularism—is relation (anvayaḥ). In relation to the amount of work that this concept does for him, it is rather underdetermined. The essential insight, however, comes from Prabhākara.21 In the context of a sentence, the meaning that an individual word actually expresses is related to the meanings of the other words in that sentence. Thus, in the sentence “bring the cow,” the word “cow” does not express merely the class of cows, but the class of cows in relation to all of the other meaning-bearing elements in the sentence. Śālikanātha does not explain what the precise nature of this relation is. It is clear, however, that his model of sentence meaning is the hierarchical organization of elements in a sacrifice (viniyogaḥ). Hence every meaning-bearing element is integrated into the hierarchical unity of the sentence meaning through its being primary or subordinate to other elements.22 This concept of relation allowed Śālikanātha to distinguish between two kinds of meanings, relational (anvita-) and nonrelational (ananvita-). Although the “proper meaning” (svārthaḥ) of a word can be considered in both its relational and non-relational aspects, each is the target of different cognitive processes, with different epistemological consequences. Śālikanātha calls the non-relational meaning the “proper form” (svarūpam). Although it is technically a class, we can think of it nontechnically as the “dictionary definition” of a word. This is what is expressed by a word in Kumārila’s view. In Śālikanātha’s view, however, the proper form is not expressed at all. It is merely “called to mind,” and this act of recollection— like any act of recollection—is not a veridical cognition in itself. Once it is called to mind, however, it enters into relations with other such elements. It is this relational meaning that is expressed by the word. To truly understand a relational meaning, according to Śālikanātha, is also to understand the relations in which it is embedded, and hence to understand the sentence meaning.23 Kumārila’s account of sentence meaning is often considered to have two stages. First, the individual words express their proper meanings, and then these meanings combine with each other to produce a sentence meaning.24 Prabhākara’s

Śālikanātha’s “Introduction” to “Fundamentals of Sentence Meaning”

259

account, by contrast, can be thought of as having a single stage, in which the individual words express their proper meanings in relation to each other, and in so doing convey the sentence meaning. Yet, in elaborating Prabhākara’s account, Śālikanātha comes very close to Kumārila’s two-stage model. The principal difference would seem that Kumārila applies the term “expression” to the first stage, in which each word produces an awareness of its non-relational meaning, whereas Śālikanātha applies the term to the second stage, in which the word produces an awareness of its relational meaning. And indeed, Śālikanātha seems to see his theory as a revision of Kumārila’s account, based on Prabhākara’s insights, to make it more theoretically streamlined and less open to the kinds of criticism that Kumārila himself had anticipated. Nevertheless, the concept of “expression” ends up looking very different in Śālikanātha’s account than it does in Kumārila’s. For Kumārila, expression really stops at the word meaning. Although other cognitive processes fill in the gap between these word meanings and the particular sentence meaning, the comprehension of sentence meaning is “verbal” because of the essential role that the expression of word meanings plays in it.25 It is important for this comprehension to be “verbal” (śābda-), that is to say, based on language as a source of valid knowledge (śabda-), because otherwise the knowledge of dharma with which the Vedas alone are supposed to provide us, and which consists in the knowledge of what the sentences of the Vedas mean, would no longer be based on a source of valid knowledge. For Śālikanātha, by contrast, expression stops at the sentence meaning and conveys not only the word meanings but the specific relations they have to each other. To say that a linguistic expression “expresses” (abhidhatte) a meaning is to say that it possesses a power (śaktiḥ) to generate a cognition of that meaning in a competent speaker of the language. Thus the discussion about the nature and limits of expression can be framed as a discussion about the quantity and quality of such powers. When framed in this way, the issue becomes a bit clearer, since powers are the types of things that can only be postulated—we cannot directly perceive them and, for the same reason, cannot infer their existence—and a postulation is only warranted when the phenomenon we are trying to account for cannot be accounted for otherwise. This is the basis on which Śālikanātha argues for compositionality. If we need to postulate an additional power for every element of meaning we encounter, we can economize on postulations by associating such powers with individual words rather than larger expressions, up to and including sentences. As an example, Śālikanātha presents a small subset of the Sanskrit language, consisting of seven words: four terms of address (“boy,” “child,” “son,” and “kid”), two imperative verbs (“bring” and “tie”), and

260

Andrew Ollett

one direct object (“cow”). From this subset, eight (4 × 2 × 1) different sentences can be formed. On a non-compositional account of sentence meaning, we must postulate a separate power to express a meaning for each of the eight sentences. On a compositional account, by contrast, we only need to postulate seven powers for each of the seven words. The relative economy of the compositional account increases exponentially as more words are added to the language.26 How can this kind of strategy help to establish Śālikanātha’s contention that the meanings expressed by words are relational? Both positions—Śālikanātha’s and Kumārila’s—involve postulation, namely, the postulation of a power to express a meaning on the part of a linguistic expression. Now we must ask precisely what kind of power each position requires us to postulate, and whether it solves the problem of expressivity that required us to postulate it in the first place. Prabhākara and Śālikanātha put forward the language acquisition argument, discussed below, to prove that words are only expressive of their meanings in some discursive context. If we take this claim seriously, then the only power that can solve the expressivity problem is a power that, as Śālikanātha says, “extends as far as the relation.”27 In Kumārila’s account, by contrast, the expressive power extends only as far as the non-relational meaning, and other cognitive processes then fill in the gap between these non-relational meanings and the overall sentence meaning. Śālikanātha claims that if we are going to postulate an expressive power in the first place, we might as well postulate the kind of power that actually accomplishes what we mean by “expression.” Kumārila might have argued that such an argument goes too far: postulation is only warranted when we cannot otherwise account for a given phenomenon, and in this case, we can account for our cognition of a relation between word meanings through other means. What means exactly? Śālikanātha claims that the three criteria of compatibility (yogyatā), proximity (saṃnidhiḥ), and dependency (ākāṅkṣā) are present whenever we cognize the relation which is, on his view, constitutive of sentence meaning.28 Both Kumārila and Prabhākara make reference to these three criteria in their discussion of how an incomplete sentence in the Vedas can be completed by carrying over (anuṣaṅgaḥ) elements from a nearby sentence.29 In fact, Kumārila’s argumentation here comes quite close to contextualism, since he claims that these criteria allow one to understand a relation (sambandhaḥ) between the part of the sentence that is expressed but incomplete and the part that is not expressed, entailing—contrary to the position that he develops elsewhere—that the expressive power of language depends on one’s knowledge of a relation between word meanings. Śālikanātha could therefore be implying that Kumārila had already hit upon the “expression of relational meanings” in

Śālikanātha’s “Introduction” to “Fundamentals of Sentence Meaning”

261

his discussion of this particular question, of carrying over elements from one sentence into another in order to arrive at a complete sentence, but had failed to generalize the model beyond this case. Let us review Śālikanātha’s model so far. What we end up with is a cognition of a “sentence particular.” We can think of this sentence particular as a fully specified blueprint of a particular state of affairs. In Mīmāṃsā, this state of affairs will generally be the performance of a sacrifice. It is fully specified in the sense of comprising elements of meaning that are totally integrated into a hierarchical structure called a “relation” by Śālikanātha; these elements of meaning are, from this perspective, “relational.” On the end of the output, so to speak, he is in complete agreement with the Mīmāṃsakas who preceded him. The same is true on the other end, that of the input. All Mīmāṃsakas agreed that sentence meaning is compositional and specifically that the expressive elements in a sentence are the individual words that comprise it. Mīmāṃsakas also agreed that the proper meaning expressed by a word was a universal, partly because it helped them to make the argument that the relationship between words and meanings was not subject to change (see Chapter 9 in this volume). Śālikanātha attempted to answer the following question: How do we integrate these two pictures, one of words that individually express abstract universals, and the other of sentences that express actionable particulars? Unless we appeal to other cognitive processes to fill in the gap—which would introduce the epistemological liability of making the cognition of sentence meaning partly inferential—we have two options. We can identify some feature of the input that continues to be present in the output. This was apparently Kumārila’s solution when he said, without much explanation, that the sentence meaning we cognize is “colored by” the word meanings.30 Alternatively, we can identify some feature of the output that is present, if only in a latent or implicit form, in the input. Such a feature, according to Śālikanātha, is the relation between word meanings. If he can establish that words express relational meanings in the first place, then the cognition of a sentence meaning, conceived as a hierarchical relation of word meanings, will follow unproblematically from the cognition of word meanings.

4  Language Acquisition and Language Structure Śālikanātha attempts to prove his hypothesis with an argument from language acquisition: when we learn the meaning of a word, we actually learn a relational meaning, and hence it is a relational meaning that the word has the power to

262

Andrew Ollett

express. All Mīmāṃsakas conceived of learning the meaning as a word in terms of grasping a connection between the word and its meaning that has always existed. But there are several ways in which this can occur. One way is through what Wittgenstein (2001, I §27) called “ostensive definition”: one person points at a cow and says “cow,” and the other person learns the connection. Prabhākara, however, had emphasized that the way we actually learn language is through careful observation of the discursive practice (vyavahāraḥ) of our elders (Jha 1978, 61–62). People do not point at things and shout their names in the real world. They engage in different types of activity through the mediation of language. The associations that we actually form are between sentences, such as “bring the cow,” and particular states of affairs, such as someone bringing a cow. It is by comparing such sentences as “bring the cow” and “bring the horse” with their corresponding states of affairs that we differentially establish the meanings of words such as “cow” and “horse.” The meanings we establish are, in every instance, relational.31 That is, we understand the meaning of “cow” as being related to the act of bringing, and we understand the meaning of “bring” as being related to the cow, its object.32 As intuitive as this account may be, there is still a serious problem on the horizon: a single word may be used in an infinite number of different sentences, and therefore it has an infinite number of relational meanings. We are simply not able to learn all of these relational meanings one by one; to do so, in any case, would be to sacrifice the principle of compositionality. Further, how would we determine which among the infinite number of relational meanings a word has in a given instance? In the course of language acquisition, we are assisted by our perceptual access to the states of affairs expressed by each individual sentence. The relation between word meanings is put before our eyes. What happens when we have to rely on language alone, without the assistance of other sources of knowledge, to know what it is that a given sentence means? What accounts for a word expressing a particular relational meaning? Or what accounts for it not expressing an infinite number of other relational meanings? Śālikanātha’s imagined opponent makes it clear that it will not do to simply say that a word expresses its proper meaning in relation to other meanings in general, that is, without determining precisely what those other meanings are. One of the phenomena that the “expression of relational meanings” was meant to account for in the first place is the fact that sentence meanings are fully determinate particulars.33 The burden on Śālikanātha is therefore to explain how, in a given instance, a word can express not just a meaning that is related to the meanings of all of the

Śālikanātha’s “Introduction” to “Fundamentals of Sentence Meaning”

263

other words in the sentence—that would be relatively simple—but also express the specific set of relations that unites all of the word meanings into a single sentence meaning. The word meanings are like the pieces of a puzzle for which there is only a single solution: their “proper meanings” are like the images found on each piece, and their mutual relations are like their shapes. This is a much richer and more complex notion of word meaning than offered by previous thinkers. The problem is whether Śālikanātha can reasonably claim that the words themselves possess a power to generate an awareness of such a meaning. Śālikanātha sees no reason not to postulate an expressive power for words that “extends as far as the relation,” given that we must postulate some such power in any case. But it is not the case that words express their relation in which their proper meanings are embedded all by themselves: they must always be accompanied by the aforementioned three criteria of compatibility, proximity, and dependency, without which the relation between word meanings cannot be determined. Śālikanātha calls these three factors “conditions” (upādhiḥ) and “secondary characteristics” (upalakṣaṇam). A condition is something that must be present for a cause (in this case, the word) to produce its effect (in this case, a cognition of its relational meaning), the way that fire only produces smoke in the presence of wet fuel. For Mīmāṃsakas, words are expressive precisely because they have a natural power to produce an effect, namely, the cognition of their proper meanings, but the production of the effect always depends on a number of conditions, including, for example, the articulation of the word in speech. Śālikanātha thus means that the aforementioned triad, although, strictly speaking, extrinsic to a word’s expressive power, is nevertheless necessary for a word to express its meaning in any given instance. As noted above, Śālikanātha was not the first thinker to invoke this triad as a way of determining the relation between word meanings. He was, however, the first to make it an indispensable part of understanding a sentence. The precise characterization of these three conditions was debated for many centuries, and the debate continues in current scholarship (Kunjunni Raja 1969, 157–69). It is essential to bear in mind, however, that they pertain to word meanings, not to words, and that they are present if and only if those meanings are integrated into a determinate relational structure. Thus, for all three conditions, they are “semantic” in that they refer to relations between elements of meaning, and they are “syntactic” in that they characterize the hierarchical structure in which those meanings are embedded, but they do not fit comfortably within syntax or semantics, if what we understand by these terms concerns more or less separate domains of language structure.

264

Andrew Ollett

Śālikanātha defines all three of them in relational terms, that is, as ways of relating one word meaning to one or more of its correlates (pratiyogi). Dependency is when a listener stands in need of knowing the correlate of a given word meaning, in order for either the expressed meaning to make sense at all or the word to fully express its relational meaning.34 His example is someone uttering the word “door!” (dvāram). Because the word has an accusative ending, it stands in need of a transitive verb in order to express a relational meaning. In this case, the relational meaning is a door related to some verbal action, such as opening or closing, as the patient of that action.35 Śālikanātha’s introduction of relational meaning into the definition of dependency is important, because it excludes a psychological characterization of the concept. If dependency were merely the listener’s desire to know, then it would not have a natural termination: “bring a cow” could generate a desire to know any number of things on the part of a listener (bring a cow of what color? bring a cow with what instrument? etc.).36 For Śālikanātha, by contrast, dependency is fulfilled by the expression of a relational meaning.37 Since every element of meaning within a sentence enters into a relational meaning, there is dependency on the part of the meaning of the word “cow” for the meaning of the word “white” in “bring the white cow,” since the meanings of these terms are related in the sentence as a whole, but no such dependency ensues when it occurs in the sentence “bring the cow.” Śālikanātha’s notion of dependency comes close to the notion of dependency in linguistics, which is, however, more closely associated with words than with word meanings, and hence with syntax rather than semantics. Nevertheless, some frameworks posit a distinct unit of structure that represents relations of dependency more or less exactly. In Lexical Functional Grammar, for example, the “functional structure” includes abstract semantic relations such as agent and patient (Dalrymple 2001, 3). Śālikanātha follows Kumārila closely in defining proximity rather broadly as the presence in the listener’s mind of a word meaning when another word meaning is already present.38 At first glance, this seems to imply little more than that, in order to come into relation with each other, meanings must be expressed by words that are pronounced in immediate succession. But he takes care to note that proximity is not exclusively “born of linguistic expressions,” that is, he rejects the idea that the proximity of word meanings in the listener’s mind simply mirrors the proximity of words in a sentence. He does think of proximity as sequential and explicitly states that the sequence of proximate meanings depends on the sequence of words that “put the meanings into proximity” with each other.39 Yet here, as elsewhere, Śālikanātha insists on the independence of

Śālikanātha’s “Introduction” to “Fundamentals of Sentence Meaning”

265

meanings from the words that express them.40 A word can “put another meaning into proximity” with its proper meaning, that is, present this other meaning to the listener’s mind, without necessarily calling to mind a word to express it. Hence the meanings expressed by “shut” and “the door” in the sentence “shut the door” are proximate, not only because they happen to be pronounced in immediate succession but because each word puts the meaning of the other into proximity with its own meaning. Indeed, it is under the topic of proximity that Śālikanātha discusses ellipsis completion: if we only hear the phrase “the door,” we can nevertheless come to an understanding of an entire sentence meaning, because the meaning of “the door” is often found in proximity with a relatively small set of correlate meanings, such as closing and opening. Although Śālikanātha does not exactly say it, this example makes it clear that proximity refers to the listener’s expectations regarding the sequential association of word meanings formed through repeated exposure. Proximity, insofar as it encompasses the conditional probability of one meaning given another, is thus quite close to the statistical language models that are widely used in computational linguistics. Śālikanātha finally defines compatibility as the ability to be connected with something else.41 This, too, refers to the listener’s expectations. If we hear the word “sprinkle,” we expect the act of sprinkling to be connected with a liquid. According to Śālikanātha, we rely on this general kind of compatibility to determine whether two particular word meanings are in fact compatible with each other, such as sprinkling and water, and therefore in the determination of relational word meanings. Chomsky’s famous sentence, “colorless green ideas sleep furiously,” does not express a sentence meaning principally because of the absence of compatibility between its word meanings. Compatibility is generally thought of as extralinguistic: we simply rely on our knowledge of the world, independently of our training in a particular language, in order to determine whether two meanings are compatible. Yet here, too, Śālikanātha’s claim that compatibility is something we rely on in language acquisition and bring to the understanding of meaning—and especially his claim that meanings are known to be compatible with certain general types of other meanings— recalls the concept of “subcategorization” in lexical semantics. This refers to features of our mental representations of lexical items that lead us to expect them to enter into combination with some other lexical items in specific grammatical configurations. To say, for instance, that certain verbs can take an “instructional imperative” construction (e.g., “bake the cake” in a recipe) while others generally can’t (*“like the cake”) may be to say that relational meanings, in this case baking and liking in relation to the imperative mood,

266

Andrew Ollett

are differentially compatible with other meanings (Levin 1993, 40). This kind of compatibility is linguistic, since it would be difficult to account for the differences in possibilities of combination simply on the basis of our knowledge of the world. These three conditions help Śālikanātha to offer a revised, and thoroughly contextualist, picture of the connection between word and meaning. Prior to Śālikanātha, Mīmāṃsakas considered each individual word to have a fixed connection to the meaning it expressed. Language acquisition was more or less a matter of learning this connection, that is, to know what is expressed by the words “cow,” “white,” “bring,” and so on. Once this was done, hearing the word “cow” in any sentence will convey the associated meaning. Śālikanātha instead claims that the word meanings in any sentence are in fact relational, in the sense that they are embedded in the hierarchical structure that he calls “relation” and which he identifies with the sentence meaning. This is a claim about the structure of language. Correspondingly, when we learn the meanings of words, we learn them in context. We come to know that a given word expresses its proper meaning only in relation to other meanings that stand in relationships of compatibility, proximity, and dependency with them. This is a claim about language acquisition. In effect, it says that in learning language we learn to situate word meanings in a relational structure and, moreover, that in learning word meanings we do not merely learn “dictionary definitions,” but build up complex mental representations that includes information about their relational possibilities. Finally, after the stage of language acquisition, when we encounter sentences, we understand their meanings by determining the relational meanings of their constituent words. In doing so, we rely on our mental representations of those meanings, including our expectations about how they might connect to the other meanings in the sentence through compatibility, proximity, and dependency. This is a claim about language comprehension, or alternatively about the expressivity of words. One way of seeing Śālikanātha’s intervention is to see him as replacing a relatively simple picture of language acquisition, in which words come to be associated with their meanings, with a more sophisticated picture, in which we also cognize relations between word meanings along with the meanings in their “proper form.” One of his key claims in the Fundamentals of Sentence Meaning is that the criticism that each word would be connected to an infinite number of relational meanings, and must fail to express all but one of the meanings that are in fact associated with it, would simply disappear if only we accept his picture of language acquisition.42 This must be because language acquisition sufficiently

Śālikanātha’s “Introduction” to “Fundamentals of Sentence Meaning”

267

equips us with the expectations, mental representations, and principles that we need in order to cognize relational meanings from the words of a sentence. He admits that the number of relational meanings for any word is infinite, or at least very large.43 If, however, we know that we must ultimately arrive at a single, unified configuration of word meanings, we know that the configuration implied by a particular word meaning—the shape of the puzzle piece, to return to our earlier metaphor—must match the configuration implied by all of the other word meanings. Recognizing the solution to this puzzle might be a problem for other theorists, but for Śālikanātha it is no problem at all, since the way we acquire language in the first place is by starting with the solution, so to speak, and then ascertaining the shape and content of the pieces that comprise it. We avail ourselves of the same properties of compatibility, dependency, and proximity in determining the relations between word meanings, both at the stage of acquisition, that is, when faced with a sentence whose meaning we already know, and subsequently, that is, when faced with a sentence whose meaning is not already presented to us through another source of valid knowledge.

5  From Utterance to Recollection to Expression As noted above, despite the fact that Śālikanātha figures his theory, “the expression of relational meanings,” as the exact antithesis of Kumārila’s, which he calls “the relation of expressed meanings,” Śālikanātha has taken many of Kumārila’s insights on board. Both consider the expression of sentence meaning to proceed in two stages, one in which word meanings are presented to the mind and another in which they are combined into a sentence meaning. And in the second stage both make the ontological leap from the abstract class, expressed by the words, to the determinate particular, expressed by the sentence as a whole. Let us now consider how Śālikanātha connects the two stages, why he considers “expression” to be an appropriate designation of the second stage alone, and what problems he believes he has resolved with his model. At the beginning of the process, one hears a collection of words.44 Then, each of the words individually calls to mind the “proper form” of its meaning, which is non-relational. By the application of a set of rules or principles, the listener integrates these meanings into a sentence particular. Once this has been carried out successfully, the listener has a cognition of the unique relation in which that sentence particular consists.45 The contribution that each of the words makes

268

Andrew Ollett

to this cognition is their relational meaning, and it is this meaning that they are properly said to express. Whereas Kumārila would have said that the word “expresses” its meaning, Śālikanātha maintains that, on its own, it can only call the meaning to mind. When one hears the word “cow” outside of the context of a sentence, one will certainly have some idea of a cow, but it will be completely non-relational. One could think of situations wherein “cow” on its own does in fact convey a relational meaning (for instance, if we can reasonably expect, on the basis of past experience, that it is a command to look at a cow nearby), but these are exceptions, and they are handled by the concept of proximity. These kinds of associations are parallel to, for example, the association between a certain smell and a certain place. As mere associations, they cannot be considered sources of valid knowledge, which axiomatically make one aware of something of which one was not previously aware.46 At the earliest stage of understanding a sentence meaning, words merely call their non-relational meanings to mind in precisely the same way that they do outside of the context of a sentence. Once populated by such non-relational meanings, the mind then sets to work in ascertaining a relation in which all of those meanings can be embedded. Although he does not mention it in this context, the requirement of mutual compatibility, proximity, and dependency among the meanings almost certainly guides this process. At the very least, these three conditions constitute a test that any proposed relation between word meanings must pass. Indeed, they have been leveraged in precisely this way in recent natural language processing work: if a number of different parses are available for a sentence simply on the basis of its grammatical structure, some of those parses may still be disregarded because of their violation of the requirements of compatibility, for example (Panchal and Kulkarni 2018). They may, however, have a more active role to play in determining the relation, in the sense that they provide the listener with expectations that lead her immediately to viable candidates. For Śālikanātha, moreover, it is quite clear that the context which the listener takes into consideration at this stage is not merely the individual sentence, but any relevant information that is present in the listener’s mind. This includes, for example, what we may call discursive context, or the horizon of expectations constituted by sentences that have previously been encountered and understood in a given discursive unit (prakaraṇam), as well as the context of utterance, or the practical situation in which the listener has encountered the sentence. Mīmāṃsakas recognized that both types of context make enormous contributions to the meaning of a sentence. When ascertaining the meaning

Śālikanātha’s “Introduction” to “Fundamentals of Sentence Meaning”

269

of sentences of the Vedas, however, Mīmāṃsakas largely ignored the second type, since they maintained that the meanings of those sentences must remain constant across every single utterance. Śālikanātha is largely silent about how it is, precisely, that these types of extra-sentential context contribute to the meaning of a sentence. One possibility, however, which would be consonant with the principles of Mīmāṃsā, is that progressively larger expressive units constitute progressively more complex relations. Thus in a discursive unit comprising several sentences, the meaning of each sentence must be integrated into a single set of relations with each of the others. And when the context of utterance contributes to the meaning of a sentence, the listener’s awareness of that context provides elements of meaning—which are not, however, word meanings, because they are not expressed by any words—that are similarly integrated into the set of relations that constitute the sentence meaning. Perhaps one of the most pregnant phrases in the Fundamentals is Śālikanātha’s claim that the complex of words establishes the sentence particular through the operation of “principles” (nyāyas).47 This means that the input-output relation between the words and the sentence meaning is not arbitrary or opaque; there are principles that guide this process, and we can think of them as constraining certain relationships between the input and the output. Some of these principles are probably so obvious that they do not need to be spelled out—for instance, that the meanings called to mind by the words in the first stage should generally be present in the second stage, which is a version of the compositionality principle. Others, however, are less obvious. These are the “Mīmāṃsā principles,” that is, the principles that have been formulated in the Mīmāṃsā system. Although Śālikanātha quotes Kumārila’s description of Mīmāṃsā as accounting for how it is that people perform sacrifices on the basis of the texts of the Vedas, he nevertheless states that the principles formulated by Mīmāṃsā apply equally to language outside of the Vedas.48 These principles allow us to determine the “local” relations between word meanings, which can then be integrated into a single “global” relation. For every element of meaning, Śālikanātha says, we need to know, for example, whether it is the topic of the sentence (anuvādyam) or the comment (vidheyam), whether it is primary (pradhānam) or secondary (guṇabhūtam) in relation to another element of meaning, and whether the meaning that has been called to mind is actually intended (vivakṣitam), in the sense that it ends up being incorporated into the final sentence meaning, or unintended (avivakṣitam).49 Mīmāṃsā attempts to discover the principles that allow us to answer these questions in a systematic and consistent way. For instance, a well-known principle of Mīmāṃsā holds that an element of meaning

270

Andrew Ollett

is unintended if it is a secondary qualification of the topic of the sentence (see Yoshimizu 2006). When a listener applies these principles in order to ascertain the sentence particular, she will likely not be aware that she is doing so. In some sense, then, the “Mīmāṃsā principles” are like the grammatical rules which a speaker of a language internalizes and applies without a conscious awareness of the rules themselves. Interestingly, however, Śālikanātha comments that there are cases in which we do have to apply these principles consciously and deliberately in order to arrive at a suitable sentence meaning, and he contrasts these cases to sentences in which we are “very well practiced,” where the comprehension of meaning is immediate.50 Since what is at stake here is the ascertainment of a relation between word meanings, this statement suggests that at least some of the information needed for this ascertainment is stored in the listener’s internal “language model,” and not collected and processed at the moment of hearing. Mīmāṃsakas consider language, at least the language of the Vedas, to be a source of valid knowledge. If “expression” is the word for the production of a cognition of a meaning, and if this cognition is to count as a source of valid knowledge, then we should really only speak of sentence meanings as “expressed.” For word meanings cannot constitute sources of valid knowledge in themselves, since by definition we can only cognize a meaning from a word if we have already cognized the same meaning before. Kumārila is aware of the problem: on the one hand, he has designated words as “expressive,” although they do not in themselves allow us to cognize particular states of affairs; on the other, sentences do allow us to cognize particular states of affairs, but he cannot refer to this as “expression,” since that process has culminated in the word meanings.51 He ends up arguing for a transference of the expressive power of words to the sentence. Śālikanātha avoids this double bind. If words are expressive of their proper meanings only when they help us to understand a qualified sentence meaning, then we have no reason to think of individual words, outside of or isolated from a sentence, as expressive at all. Similarly, if the sentence meaning is simply the relation in which the individual word meanings find their place, the gap between the expression of word meanings and the expression of a sentence meaning disappears completely. Finally, Śālikanātha makes the ontological jump from abstract classes to determinate particulars by having the listener rely on a set of principles for processing the non-relational meanings of words into a determinate relational framework. Relational meanings are particulars.52 This allows him to avoid saying, as Kumārila did, that sentence meaning is only ever expressed indirectly.

Śālikanātha’s “Introduction” to “Fundamentals of Sentence Meaning”

6  Text and Translation of the Verses The notion of the sentence meaning emerges from the words themselves: that is how we represent the thought of Prabhākara Guru: (1)

padebhya eva vākyārtha­­-  pratyayo jāyate yathā / tathā vayaṃ nibadhnīmaḥ prabhākaraguror matam // 1 //

If it is through words, which use up their power in expressing their proper meanings in relation to each other, that those meanings are cognized, then, when that is the case, the sentence meaning is cognized as well as cog­ nized in the same way. (2)

padair evānvitasvārtha­ mātropakṣīṇaśaktibhiḥ /   svārthāś ced bodhitā buddho vākyārtho ’pi tathā sati // 2 //

The wise proclaim that sentence meanings are nothing other than word meanings that have acquired a mutual relation in terms of being primary and secondary. (3)

pradhānaguṇabhāvena labdhānyonyasamanvayān /   padārthān eva vākyārthān saṅgirante vipaścitaḥ // 3 //

Although words have a large number of proper meanings taken individually, on ac­­ count of there being one motive, they express a single sentence meaning. (4)

bhūyāṃso yady api svārthāḥ   padānāṃ te pṛthak pṛthak /   prayojanatayā tv eka-  vākyārthaṃ saṃpracakṣate // 4 //

The sentence, too, is said to be single, since the understanding of that is its single effect. For the primary element is the single motive for conveying the secondary elements. (5)

tatpratītyekakāryatvād   vākyam apy ekam ucyate /   pratipattir guṇānāṃ hi pradhānaikaprayojanā // 5 //

The powers of words are ascertained differen­ tially, by addition and subtraction, when the practical activity of adults arises upon their hearing sentences. (6)

vyavahāreṣu vṛddhānāṃ vākyaśravaṇabhāviṣu / āvāpoddhārabhedena   padānāṃ śaktiniścayaḥ // 6 //

What are added and subtracted are proper meanings in relation to each other. From this one understands that it is only in regard to relational meanings that words have an expressive capacity. (7)

opyante coddhriyante ca   svārthā anvayaśālinaḥ /   anviteṣv eva sāmarthyaṃ padānāṃ tena gamyate // 7 //

If the basis of language acquisition is this— that words express their proper meanings in relation with other meanings that are com­ patible with them, and which have attained dependency and proximity with them (8)

ākāṅk​ṣāsan​nidhi​prāpta­-­ yogyārthā​ntara​saṅga​tān /   svārthān āhuḥ padānīti   vyutpattiḥ saṃśritā yadā // 8 //

—then there will be no problem at all on ac­­count of infinitude or lack of invariable con­­ comitance. Even with respect to word mean­ ings, this same causal complex is present when one becomes aware of a relation. (9)

ānantyavyabhicārābhyāṃ   tadā doṣo na kaścana / padārtheṣv api caivaiṣā   sāmagry anvayabodhane // 9 //

271

272

Andrew Ollett

Their power, however, is not observed through any other source of knowledge. Since there is no other way to account for it, it must be postulated. It is brought about by contact with words that aim at a qualified meaning. (10)

kiṃ tu teṣām adṛṣṭaiṣā   śaktir mānāntarād gatau /   kalpyā viśiṣṭārthapara­ padasaṃsparśabhāvitā // 10 //

Because of their priority, because of their ex­­ pressivity, and because we must accept in any case that they are aimed at that, it is better to accept that that power belongs exclusively to the words themselves. (11)

prāthamyād abhidhātṛtvāt tātparyopagamād api / 53 padānām eva sā śaktir   varam abhyupagamyatām // 11 //

The whole collection of words, once it is heard, calls to mind meanings that are not re­­ lated. Afterwards, it brings about an aware­ ness of the sentence meaning by establishing the particular through exegetic principles. (12)

padajātaṃ śrutaṃ sarvaṃ smāritānanvitārthakam / nyāyasaṃpāditavyakti   paścād vākyārthabodhakam // 12 //

Even during the expression of a relational meaning, the proper form continues to be present. Therefore, a linguistic expression can generate a recollection regarding the proper form of its meaning on its own. (13)

anvitasyābhidhāne ’pi   svarūpaṃ vidyate sadā /   tena svarūpamātre ’pi   śabdo janayati smṛtim // 13 //

Just as the word is sometimes recalled by a thing that is not itself a source of valid knowl­ edge, in the same way, the meaning will be recalled by a word, even though the word is not a source of valid knowledge. (14)

yathārthenāpramāṇena   svapadaṃ smāryate kvacit / padenāpy apramāṇena   tathārthaḥ smārayiṣyate // 14 //

There is no circular reasoning, because a word expresses its proper meaning in its entirety, which is related in this way to the other mean­ ings that are proximate in recollection. (15)

smṛtisannihitair evam   arthair anvitam ātmanaḥ / artham āha padaṃ sarvam   iti nānyonyasaṃśrayaḥ // 15 //

Because both the meanings and the relation are grasped when the words, through their powers, create an awareness of the relational meanings in this way, there is no need of a separate power. (16)

anviteṣu padair evaṃ bodhyamāneṣu śaktibhiḥ / anvayārthagṛhītatvān   nānyāṃ śaktim apekṣate // 16 //

For a man, in understanding the relation, can understand that to which it is related, just as he can understand the class of something in understanding the particular. This is a wellknown principle. (17)

pratīyann anvayaṃ yasmāt pratīyād anvitaṃ pumān / vyaktiṃ jātim ivārthe ’sāv   iti saṃparikīrtyate // 17 //

How could those who explain that what words express is something related to a gen­ eral term account for their relation to a re­­ stricted particular? (18)

sāmānyenānvitaṃ vācyaṃ padānāṃ ye pracakṣaṭe / niyatena viśeṣeṇa   teṣāṃ syād anvayaḥ katham // 18 //

Śālikanātha’s “Introduction” to “Fundamentals of Sentence Meaning” It may be that the particular is implicated, just as in the case of the individual and the class. Nevertheless a determinate particular is not understood, exactly as in that case. (19)

yady apy ākṣipyate nāma   viśeṣo vyaktijātivat / nirdhāritaviśeṣas tu   tadvad eva na gamyate // 19 //

Although it is a particular that is dependent, compatible, and heard in proximity, never­ theless that is not grasped in the absence of something that makes one aware of the rela­ tion. (20)

yady apy ākāṅkṣito yogyo viśeṣaḥ sannidhau śrutaḥ / saṃbandhabodhakābhāve gṛhyate na tathāpy asau // 20 //

If one takes the view that the relation is with a particular, however, then the criteria such as compatibility come to act as conditions for an awareness of the relation at the stage of language acquisition. (21)

sambandhabodhe vyutpattāv upādhitve samāviśat / viśeṣānvayavāde tu   yogyatvādy upakārakam // 21 //

Moreover, since the connection with the gen­ eral term is established simply on the strength of the way things are, those who want that to be what is expressed have subjected language to needless exertion. (22)

kiṃ ca vastubalenaiva   siddhe sāmānyasaṅgame /   tasya vācyatvam icchadbhir   vṛthā śabdaḥ prayāsitaḥ // 22 //

273

Notes 1 When and where Śālikanātha lived is not known with certainty. His Sanskrit style makes it unlikely that he was from South India (as K. Kunhan Raja guesses in his introduction to Prabhākara’s Long Commentary). 2 For the different views that Mīmāṃsakas have taken on apūrvam, see Yoshimizu 2000. For the second part of the Fundamentals of Sentence Meaning, see Chapter 14 in this volume. 3 racitā saccaritānām anugrahārthaṃ kartukāmena / vākyārthamātṛkāyā vṛttir iyaṃ śālikenaiva // (Fundamentals of Sentence Meaning, p. 450). 4 See Fundamentals of Sentence Meaning, Part 2, p. 7 (a reference to v. 12); Part 3, p. 268; part 4, p. 709 (a reference to v. 4) and p. 767. 5 For (rather speculative) reflections on Kumārila’s lost work, see Frauwallner 1962. 6 See the quotations on pp. 396, 402, and 404. 7 The phrase occurs, for example, in Prabhākara’s Long Commentary, p. 395. 8 These two principles, contextualism and compositionality, have both been attributed to Frege. For their history, see Janssen 2001. 9 Purport of the Explanation in Verse, pp. 61, 106. 10 “The expression of relational meanings” and “the relation of expressed meanings” are discussed by Mukulabhaṭṭa, Abhinavagupta, and Bhoja. The latter quotes entire verses of both Bhaṭṭa Jayanta and Śālikanātha (see Kunjunni Raja 1969, 199, 213,

274

Andrew Ollett

215). We know that there were earlier followers of Prabhākara, who are referred to as “Old Prābhākaras” (jaratprābhākaraḥ) in contrast to Śālikanātha. 11 See Kunjunni Raja 1969 and Ganeri 2011, both of whom deal primarily with the various senses of the word arthaḥ in the realm of signification. 12 The most eloquent exposition of this insight can be found in McCrea 2008. 13 See Mīmāṃsāsūtra, 2.1.46, and Śabara’s commentary thereon. 14 Explanation of the System, p. 445: lokamantrabrāhmaṇeṣv avyabhicāry etad vākyalakṣaṇam iti sthitam. 15 Fundamentals of Sentence Meaning, p. 404. My translation of vacanavyakti- as “sentence particular” follows McCrea 2008. 16 See Skinner’s programmatic essay, “Meaning and understanding in the history of ideas” (Skinner 2002) and Ganeri’s application of it to Indian intellectual history (Ganeri 2008). 17 Fundamentals of Sentence Meaning, v. 19. 18 Ganeri (2011, 10) calls the last two meanings “indefinite non-generic” and “definite non-generic,” respectively. 19 The verse cited by Śālikanātha (Fundamentals of Sentence Meaning, p. 393) is presumably from Kumārila’s Extensive Notes: “our position is that sentence meaning is always based on secondary meaning” (vākyārtho lakṣyamāṇo hi sarvatraiveti naḥ sthitiḥ). 20 McCrea 2013, 139–40. This epistemological principle is most evident in Prabhākara’s theory of error (explained by Śālikanātha in one of his Commentarial Essays called Byroad of Reasoning, Nayavīthiḥ), according to which cognitions that are apparently invalidated by subsequent cognitions are actually composite cognitions whose component parts retain their validity in spite of being synthesized incorrectly. For a review of Prabhākara’s theory, and a wider discussion of South Asian theories of error, see Schmithausen 1965, 205–12. 21 See Long Commentary, on 1.1.26, pp. 394–95. 22 Fundamentals of Sentence Meaning, v. 3. 23 Fundamentals of Sentence Meaning, vv. 16–17. 24 See the Sentence section of Kumārila’s Explanation in Verse, v. 228. 25 See the Sentence section of Kumārila’s Explanation in Verse, v. 233. 26 Fundamentals of Sentence Meaning, p. 379. 27 Fundamentals of Sentence Meaning, p. 400. tasyā (sc. abhidhānaśakteḥ) evānvayaparyantatā kalpayituṃ sukarā. 28 Fundamentals of Sentence Meaning, v. 9. 29 Explanation of the System, p. 455. Since Kumārila’s interpretation self-consciously differs from earlier interpretations of this section, I consider it likely that the idea is his. 30 Explanation in Verse, Sentence section, v. 233. 31 Fundamentals of Sentence Meaning, v. 7.

Śālikanātha’s “Introduction” to “Fundamentals of Sentence Meaning”

275

32 Fundamentals of Sentence Meaning, p. 403: dvāv etāv arthau, yad ānayānvitaṃ gotvam, gavānvitaṃ cānayanam. 33 Fundamentals of Sentence Meaning, p. 382: sāmānyānvayābhidhānaṃ ca nāśaṅkanīyam eva, vākyebhyo viśeṣānvayāvagamāt. 34 Fundamentals of Sentence Meaning, p. 388: anvitasyābhidhānārtham uktārthaghaṭanāya vā / pratiyogini jijñāsā yā sākāṅkṣeti gīyate // 35 Keating (2017) discusses this example, which the followers of Kumārila understood as postulation (arthāpattiḥ), in contrast to Śālikanātha (see footnote 40). 36 Fundamentals of Sentence Meaning, p. 385. 37 Fundamentals of Sentence Meaning, p. 388: anvitasyābhidhānasiddhyartham evākāṅkṣā. 38 Fundamentals of Sentence Meaning, p. 389: arthasya śravaṇānantaram ākāṅkṣāyogyatābhyām arthāntare buddhiviparivṛttiḥ. Cf. Explanation in Verse, p. 455: saṃnidhir iti buddhau viparivṛttiḥ. 39 Fundamentals of Sentence Meaning, p. 390: saṃnidhāpakasāmagrīkrameṇa kramavān asau. 40 At this point in the Fundamentals, he enters into an argument against “postulation on the basis of what is heard” (śrutārthāpattiḥ). Whereas Kumārila says that when a statement that is heard does not make sense otherwise (e.g., “fat Devadatta does not eat during the day”) we postulate a linguistic expression (“he eats at night”), Śālikanātha says that we postulate only the meaning. See his Review of the Pramāṇas (Pramāṇapārāyaṇam) in his Commentarial Essays, p. 279. 41 Fundamentals of Sentence Meaning, p. 390, yat sambandhārham. 42 Fundamentals of Sentence Meaning, vv. 8–9ab. 43 Fundamentals of Sentence Meaning, v. 4. 44 In fact the process begins with hearing sounds, which are recognized as phonemes, and then recognized as discrete words. For more on this process, see Chapter 2 in this volume. 45 Fundamentals of Sentence Meaning, v. 12. 46 Fundamentals of Sentence Meaning, v. 13. 47 Fundamentals of Sentence Meaning, v. 12. 48 Fundamentals of Sentence Meaning, p. 404: lokavyavahāravartibhir nyāyaiḥ. The quotation, presumably from Kumārila’s lost Extensive Notes, is dharme pramīyamāṇe hi vedena karaṇātmanā / itikartavyatābhāgaṃ mīmāṃsā pūrayiṣyate // 49 Fundamentals of Sentence Meaning, p. 404. 50 Fundamentals of Sentence Meaning, p. 404: nanu loke drāg eva vākyārthāvagatir neyatīṃ sāmagrīm apekṣate. ucyate: atyantābhyasteṣu vākyeṣu syād evam. 51 Explanation in Verse, Sentence section, vv. 229–30. 52 Fundamentals of Sentence Meaning, v. 20. 53 Subrahmanya Sastri’s edition reads -āvagamād.

276

Andrew Ollett

Primary Sources Commentarial Essays

Subrahmanya Sastri, ed. 1961. Prakaraṇapañcikā Mahām​ahopa​dhyāy​aśrīma­ cchā​likan​ātham​iśrav​iraci​tā. Benares: Benares Hindu University.

Explanation in Verse

Tailanga, Ramasastri, ed. 1898. The MîmânsâŚloka-Vârtika of Kumârila Bhatta. Benares: Vidya Vilas Press.

Explanation of the System

Tantravārttikam of Kumārila. In Mīmāṃsādarśana.

Fundamentals of Sentence Meaning

Vākyārthamātṛkā of Śālikanātha. In Commentarial Essays.

Long Commentary

Ramanatha Shastri, S.K., and S. Subrahmanya Shastri, eds. 1934–67. Bṛhatī of Prabhākara Miśra [on the Mīmāṃsāsūtrabhāṣya of Śabarasvāmin] with the Ṛjuvimalapañcikā of Śālikanātha. 5 vols. Madras: University of Madras.

Mīmāṃsādarśana

Subbāśāstrī, ed. 1929–34. Śrīmajjaimi nipraṇītaṃ Mīmāṃsādarśanam. Vol. 6. Anandashrama Sanskrit Series 97. Pune: Anandasramamudranalayah.

Mīmāṃsāsūtra

In Mīmāṃsādarśana.

Purport of the Explanation in Verse

Ramanatha Sastri, S.K., ed. 1971. Ślokavārtikavyākhyā Tātparyaṭīkā of Uṃveka Bhaṭṭa. Revised by K. Kunjunni Raja and R. Thangaswamy. Madras: University of Madras.

Secondary Sources Dalrymple, Mary. 2001. Lexical-Functional Grammar. San Diego: Academic Press. Fodor, Jerry. 2003. Hume Variations. Oxford: Clarendon Press. Frauwallner, Erich. 1962. “Kumārila’s Bṛhaṭṭīkā.” Wiener Zeitschrift für die Kunde Süd- und Ostasiens 6: 78–90. Ganeri, Jonardon. 2008. “Contextualism in the Study of Indian Intellectual Cultures.” Journal of Indian Philosophy 36 (5/6): 551–62. doi:10.1007/s10781-008-9039-7. Ganeri, Jonardon. 2011. Artha. Meaning. Foundations of Philosophy in India. New Delhi: Oxford University Press.

Śālikanātha’s “Introduction” to “Fundamentals of Sentence Meaning”

277

Janssen, Theo M.V. 2001. “Frege, Contextuality and Compositionality.” Journal of Logic, Language and Information 10 (1): 115–36. Jha, Ganganatha. 1978. The Prābhākara School of Pūrva Mīmāṃsā. Original edition 1911. Delhi: Motilal Banarsidass. Kaplan, David. 1989. “Demonstratives.” In Themes from Kaplan, ed. by Joseph Almog, John Perry, and Howard Wettstein, 481–563. New York: Oxford University Press. Keating, Malcolm. 2017. “(Close) the Door, the King (is Going): The Development of Elliptical Resolution in Bhāṭṭa Mīmāṃsā.” Journal of Indian Philosophy 45: 911–38. doi:10.1007/s10781-017-9328-0. Kunjunni Raja, K. 1969. Indian Theories of Meaning. Adyar Library Series 91. Second edition. Madras: Adyar Library and Research Centre. Levin, Beth. 1993. English Verb Classes and Alternations: A Preliminary Investigation. Chicago: University of Chicago Press. McCrea, Lawrence J. 2008. The Teleology of Poetics in Medieval Kashmir. Harvard Oriental Series 71. Cambridge, MA: Department of Sanskrit / Indian Studies, Harvard University. McCrea, Lawrence J. 2013. “The Transformations of Mīmāṃsā in the Larger Context of Indian Philosophical Discourse.” In Periodization and Historiography of Indian Philosophy, ed. by Eli Franco, 127–43. Vienna: Verein Sammlung De Nobili. Panchal, Sanjeev, and Amba Kulkarni. 2018. “Yogyatā as an Absence of NonCongruity.” In Computational Sanskrit and Digital Humanities: Selected Papers Presented at the 17th World Sanskrit Conference, ed. by Amba Kulkarni and Gérard Huet, 59–82. New Delhi: DK Publishers. Recanati, François. 2012. “Compositionality, Flexibility, and Context Dependence.” In The Oxford Handbook of Compositionality, ed. by Markus Werning, Wolfram Hinzen, and Edouard Machery, 175–91. Oxford: Oxford University Press. doi:10.1093/oxfordhb/9780199541072.013.0008. Schmithausen, Lambert. 1965. Maṇḍanamiśra’s Vibhramavivekaḥ, mit einer Studie zur Entwicklung der indischen Irrtumslehre. Österreichische Akademie der Wissenschaften. Philosophisch-Historische Klasse. Sitzungsberichte 247, 1. Veröffentlichungen der Kommission für Sprachen und Kulturen Süd-und Ostasiens 2. Vienna: Böhlau. Skinner, Quentin. 2002. “Meaning and Understanding in the History of Ideas.” In Visions of Politics, Volume 1: Regarding Method, 57–89. Cambridge: Cambridge University Press. Wittgenstein, Ludwig. 2001. Philosophische Untersuchungen/Philosophical Investigations. Original edition 1953. Malden, MA: Blackwell. Yoshimizu, Kiyotaka. 2000. “Change of View on Apūrva from Śabara to Kumārila.” In The Way to Liberation: Indological Studies in Japan, ed. by Sengaku Mayeda, 149–65. New Delhi: Manohar. Yoshimizu, Kiyotaka. 2006. “The Theorem of the Singleness of a Goblet (grahaekatva-nyāya): A Mīmāṃsā Analysis of Meaning and Context.” Acta Asiatica 90: 15–38.

14

Śālikanātha on Language Acquisition: A Study of “Vākyārthamātṛkā” II Kei Kataoka

1 Introduction A Vedic injunction, as an independent source of valid cognition, conveys a piece of new information that is otherwise unknown, that is, not known through other means of valid cognition. For example, the Vedic injunction “One who desires heaven should sacrifice” conveys a command (niyoga) to sacrifice that is otherwise unknown to the listener. The characterization of valid cognition as a new experience, however, yields a problem when applied to Vedic commands. How can one learn a language if it conveys a referent that cannot be known through any other means of valid cognition? The usual method of confirming a referent through perception looks impossible, because the object of a Vedic sentence, in particular the command (niyoga), is not perceivable. In the latter half of the Vākyārthamātṛkā (“The Source of Sentence-Referents”), Śālikanātha discusses the issue of language acquisition (vyutpatti), in particular, with regard to apūrvakārya, that is, an unknown obligation. He begins the chapter as follows: “How do the prescriptive endings (liṅādi) convey an obligation (kārya, literally ‘what is to be done’) that cannot be known through other means of valid cognition, if language requires learning in order to convey its referent?”1

2  The Circumstances of Language Acquisition Śālikanātha introduces a situation in which an observer learns language through listening to the conversation of two elders. Caitra is directed by Devadatta, his elder, to bring water: “Caitra, bring water!” Caitra then brings water to Devadatta. Then Yajñadatta, the

Śālikanātha on Language Acquisition

279

observer who desires to learn the language, tries to understand the process.2 (1) Yajñadatta reflects as follows: “Caitra’s activity must be preceded by a cognition, just as my activity is [always] preceded by a cognition.”3 (2) “I commenced an activity after having cognized something. Caitra, too, must have cognized the same thing [before he commenced an activity].”4 (3) “But in Caitra’s case it was known through Devadatta’s utterance, while in my case it was known through another means of valid cognition.”5 (4) “What then is the thing after knowing which I started my activity at that time?”6 This is the basic scenario described by Śālikanātha. At this point, Yajñadatta is searching for the thing that prompts him to start an activity. In other words, he seeks the cause of an activity (pravṛttikāraṇa) that is the referent of a prescriptive sentence and that was the object of his past cognition.

3  The Obligation as Distinct from a Mere Action and So On Yajñadatta continues to reflect as follows: (5) “The cause of the activity cannot be just an action (kriyā).”7 “Neither is it a mere fruit (phala).”8 “Nor is it the mere relationship between an action and a fruit (kriyāphalasaṃbandha).”9 (6) “Rather, it must be the kāryatā, the property/aspect of what is to be done. After knowing this [kāryatā], I started an activity.”10 (7) “Never mind mundane actions like coming and going, ultimately even satiating actions such as sucking the breast and so on [which I did when I was a baby] were never commenced by me before it was determined: ‘This is what I have to do.’”11 As will be explained later, Śālikanātha regards the command (niyoga) as the main thing to be done. An action denoted by a verbal root is not the main thing to be accomplished. An action, that is, the object of a verbal root (dhātvartha), is subservient to a command inasmuch as it is merely the object (viṣaya) of the commanded. A fruit maintains a subservient status in the Prābhākara system. It cannot be the real prompter of one’s activity (pravṛttikāraṇa). The last of the three candidates rejected above, that is, the relationship between an action and its fruit, is the referent of the prescriptive endings according to Maṇḍana Miśra.

280

Kei Kataoka

Therefore, Śālikanātha sets forth, in the following section, a separate argument to refute Maṇḍana’s view.12

4  The Refutation of Maṇḍana’s View of Instrumentality The observer continues to reflect on kāryatā as follows: (8) “Phalasādhanatā, i.e., the aspect of a thing as a means of attaining a fruit, is not identical with kāryatā, i.e., the aspect of what is to be done.” “Kāryatā is kṛtisādhyatva, i.e., the aspect of a thing as something to be accomplished by an effort.” “Phalasādhanatā, on the other hand, is that aspect of a thing which is a means of producing a fruit.” “They are different from each other.”13 (9) “In some cases, however, an action which by itself is of the nature of suffering may be what is to be done (kārya). In that case its instrumentality functions as the cause of its being what is to be done.14 To explain further, one considers the action as something to be done because it is a good means of attaining the fruit.15 Therefore, the aspect of what is to be done (kāryatā) is always understood as if identical with instrumentality, because it has a causal relationship with instrumentality.”16 An action has two properties, that is, two properties inhere in an action (Figure 14.1). (10) “The distinction between the two aspects can be clearly observed in the case of pleasure (sukha): Pleasure itself has the aspect of a thing to be done although it is not an instrument.”17 (11) “Therefore, my activity took place by cognizing the aspect of what is to be done, which goes beyond, i.e., is distinct from, instrumentality.”18 Figure 14.1  Two Properties of an Action instrumentality

to be done

action designates samavāya or inherence designates causality

Śālikanātha on Language Acquisition

281

(12) [Thus having ascertained that the aspect of what is to be done as distinct from instrumentality is the thing by understanding which he started an activity, Yajñadatta infers as follows:] “Caitra, too, must have started an activity after having cognized what is to be done.”19

5  Language Acquisition through Insertion and Extraction Devadatta utters a full sentence (“Caitra, bring water!”) and not just the imperative ending. How can one extract the referent of the imperative ending from a complete sentence? Presupposing this sort of question, Śālikanātha states: “Just as other words are learned with respect to their own referents through different insertions (āvāpa) and extractions (udvāpa), so the prescriptive endings (i.e., optative, imperative, gerundive, etc.) are learned with respect to what is to be done.”20

6  Obligation as a Plurality of All Kinds of Imperatives What is then extracted as the pure referent of the prescriptive endings? When an older speaker directs a younger listener to do something, the imperative is called “impelling” (praiṣa).21 When they are of the same age, it is called “inviting” (āmantraṇa).22 When a younger speaker asks an older person to do something, the imperative is called “solicitation” (adhyeṣaṇa).23 Here the thing to be done (kārya) is variously called “impelling,” “inviting,” and “soliciting” due to the position of the speaker, that is, whether he is older, of the same age, or younger in comparison to the listener to be activated (Figure 14.2).24 In all cases, obligation is the single cause of activity. Thus, obligation is ascertained as the referent of Figure 14.2  The Pure Referent of the Prescriptive Endings kārya

Impelling

Inviting

Soliciting

282

Kei Kataoka

the prescriptive endings, because it never deviates from the activity,25 that is, it is consistently the cause of activity (whereas impelling, inviting, and soliciting are not).26

7  Obligation Is Known through Perception and Inference We now come back to the original question. How can one learn an obligation that cannot be known by other means of valid cognition? Śālikanātha gives the following explanation: The main thing accomplished by an effort (kṛtisādhya) is [by definition] called what is to be done. And it (the main thing accomplished by an effort) is knowable even through other means of valid cognition, just as [it can be known through other means of valid cognition] in the case of cooking rice.27

The Vṛtti expounds further as follows: One can internally perceive one’s own mental effort (kṛti/prayatna) and therefore one can perceive its having a particular target (viśiṣṭaprayojanatā), too.28 And the causal relationship between a mental effort (kṛti) and its target (sādhya) is inferred through their co-existence and co-absence.29 Thus, what is to be done (kārya), i.e., the target to be accomplished by a mental effort, can be known through perception and inference.30

Perception (→) grasps an effort and its target. Inference (→) informs of the causality (⇒) between an effort and its target (Figure 14.3). In this way, the prescriptive endings can be learned even with regard to a referent that has the nature of an obligation. Similarly, other words, too, can be learned with regard to their own respective referents that are connected with an obligation.31 Figure 14.3  Obligation is Known Through Perception and Inference perception

effort

inference

perception

target e.g., cooking

designates a cognitive process designates causality

Śālikanātha on Language Acquisition

283

That the cooking of rice (i.e., softening) presupposes a mental effort can be ascertained through perception and inference. Here, it is to be noted that although an obligation overlaps with an action in mundane cases, it is not ultimately identical with and therefore to be distinguished from an action. Śālikanātha therefore commences, in the following section, a discussion on how to distinguish an obligation from an action in conformity with the Vedic usage.

8  The Structure of the Vedic Injunction In order to understand Śālikanātha’s discussion of Vedic language, however, we need to understand in advance the structure of the Vedic injunction: “One who desires heaven should sacrifice (svargakāmo yajeta).” The finite verb yajeta (should sacrifice) consists of the verbal root yaj and the optative ending (liṅ) which respectively denote a sacrifice (yāga) and a command (niyoga).32 The command is the main element (pradhānakārya) to be done (kārya). It is new (apūrva), that is, it is not known through other means of valid cognition (mānāntarāvedya). On the one hand the sacrifice is regarded as the object (viṣaya) of the command. On the other hand, the person who desires heaven is regarded as the niyojya, that is, the person to be commanded. Unlike the Bhāṭṭas, the Prābhākaras do not regard the fruit as the main element to be accomplished. Rather, the fruit remains subservient. In other words, heaven is a subservient element that contributes to the qualification of the person in question.33 A person who desires heaven is qualified to perform the commanded sacrifice and he is regarded as the adhikārin, the person who has the qualification. Thus, the basic structure of the injunction (Figure 14.4) consists of a command (niyoga) as the main thing to be done, a sacrifice as the command’s object (viṣaya), and a person who is to be commanded (niyojya). Note how this differs from the Bhāṭṭa view. According to the Prābhākaras, the command, and not the fruit, is the main element to be achieved. This is the Prābhākaras’ most important claim.34 The command requires someone who

Figure 14.4  The Structure of a Vedic Injunction “should”

“sacrifice”

command

object

“one who desires heaven” a commanded qualified person

284

Kei Kataoka

Figure 14.5  The Command as the Main Element To Be Achieved sacrifice

command

commanded person

heaven

qualified

who desires heaven

carries it out by performing the commanded sacrifice. Heaven is introduced in order to qualify this person. As Śālikanātha points out, this is comparable to a case in which a master benefits his servant (e.g., by giving him a salary) and it thus looks as if the master were subservient to the servant. But in fact the master assists the servant for his own sake, to retain the servant, and not for the sake of the servant. The master never loses his priority. Similarly, a command never loses its priority, although it benefits the commanded person by bestowing a fruit.35 A person who desires heaven is qualified by his desire for heaven and commanded by the command as a person to achieve the command of a sacrifice (Figure 14.5). In different contexts, the command is referred to in different terms. It is the main element (pradhāna) in relation to other, subservient elements (guṇa/śeṣa/ aṅga). It is therefore regarded as the referent of a sentence (vākyārtha) when it is accompanied by subservient elements. Simply put, a command is accompanied by the command’s object and the commanded person. It is what is to be done (kārya/kartavya). It is new (apūrva), that is, not known through other means of valid cognition (mānāntarāvedya).

9  Obligation as Distinct from Action Although the mere “something to be done” that one learns in the process of language acquisition is known through other means of valid cognition and always overlaps with an action, ultimately it should not be considered to be of the nature of an action, because what is to be done that is revealed by the Veda cannot be something knowable through other means of valid cognition. In the following section, Śālikanātha presupposes two steps to learning Vedic injunctions: first, one learns the general referent of the prescriptive endings as a

Śālikanātha on Language Acquisition

285

mere “what is to be done” (kārya); second, one specifies that it is of the nature of something previously unknown (apūrva). In mundane cases, what is to be done overlaps with an actual action and therefore cannot easily be distinguished from it, whereas in Vedic cases it can be clearly distinguished from an action. Thus, Śālikanātha states: In this context, however, the following needs to be determined: whether what is to be done is of the nature of an action or distinct from it.36

Śālikanātha admits that what is to be done would be determined to be of the nature of an action if we just followed the example of mundane cases. As he states: Among them, if we follow mundane cases, it is appropriate to say that what is to be done is an action known through other means of valid cognition. But …37

Śālikanātha’s argument continues as follows: A person who desires heaven is a person who is commanded (niyojya), and he is the person designated by the word “svargakāmaḥ” in the Vedic sentence “svargakāmo yajeta.” The person to be commanded is the one who cognizes the thing to be done as his own task. It is naturally understood that heaven is the result to be accomplished, because heaven is connected with the desire, as shown in the expression “svarga-kāmaḥ.” Thus, the person who is qualified by the desire of attaining heaven is able to understand what is to be done precisely as a means to the desired target, because otherwise, i.e., if the optative ending were to convey what is to be done merely as an action [and not as a command], then the connection of what is to be done with the commanded person would be lost. An actual action perishes momentarily and therefore cannot produce heaven, which is attained in the future. Such a momentary action cannot be connected with the person to be commanded, who desires the fruit, because it does not bring about the desired end. Therefore, only what is to be done which is connectable with the person to be commanded, i.e., what is to be done which endures for a sufficient amount of time and which is distinct from an action, is denoted by the optative ending that expresses an injunction. What is to be done is capable of bringing about heaven because it endures for a while, and it can be connected with the desirous person who is to be commanded. What is to be done which is distinct from an action and so on and which is not known through other means of valid cognition is called “new/unprecedented” (apūrva), because it is not previously known through other means of valid cognition. It is also called “command” by the specialists of Mīmāṃsā-investigation, because it urges the person who is to be commanded toward itself as the thing to be done.38

286

Kei Kataoka

To put it briefly, a mere action cannot fulfill the role delineated by the Veda, because it is momentary, whereas the obligation previously unknown (i.e., the command) can, because it lasts in time. Thus, by referring to Vedic usage it is ascertained that an enduring obligation (more specifically an unknown command) distinct from an action is the primary referent of the prescriptive endings.

10  The Obligation as the Main Element Because the obligation is the sole main element, it is fit to be the sentencereferent. For a (Vedic) sentence states only what is to be done accompanied by a commanded person and an object of the command.39 It is to be remembered that a command to be done is subservient neither to the fruit nor to a person commanded, because it ultimately brings about heaven for the sake of the command. Śālikanātha clarifies this as follows: “The obligation alone, which brings about even heaven and so on to establish the person to be commanded who contributes to the obligation’s accomplishment, is the main element for us.”40

11  The Two-Step Approach of Language Acquisition Thus, one learns that something to be done distinct from a mere action is the main element denoted by the prescriptive endings. Although Śālikanātha’s description of this learning method is based on a specific Vedic passage calling for an optional type of rite (kāmya), that is, a ritual optionally performed on the basis of one’s desire for a particular fruit, it is applicable to all sentences, including a sentence calling for an obligatory rite (nitya) that has no fruit, that is, a ritual to be performed regularly, for example, daily. Śālikanātha makes the following claim: “In this way, the language acquisition that arises from observing the referent in the case of an optional rite is the foundation for the understanding of all sentences.”41 Distinguishing the two stages of learning, that is, first learning mundane usage and then learning the Vedic usage, Śālikanātha further states: “Here language acquisition with respect to what is to be done follows mundane usage. The process of specifying [it as previously unknown], however, is based upon a reflection on Vedic sentences.”42 Here, Śālikanātha explicitly states that proper language acquisition, in particular regarding the prescriptive endings, is possible only by also considering Vedic usage and not just mundane usage.

287

Śālikanātha on Language Acquisition Figure 14.6  The Two-Step Approach of Language Acquisition mundane usage

“yava”

millet

barley

Vedic sentence

The Vṛtti refers to a section of the Mīmāṃsāsūtra.43 The word “yava” is seen to be used in reference to both barley (dīrghaśūka) and millet (priyaṅgu) in mundane usage. Therefore, there is uncertainty regarding the primary referent of the word. The final conclusion, that it directly denotes only barley and not millet, is achieved by considering its fitness to be connected with a supplementary Vedic sentence (transmitted by the learned specialists) (Figure 14.6).44 Similarly, the prescriptive endings are properly determined as denoting an unknown obligation (apūrvakārya) (and not a mere action to be done [kriyākārya]), because (only) the former is fit to be connected with the person commanded in the Veda.45

12  An Action To Be Done and a Previously Unknown Obligation One might object that language acquisition with regard to what is to be done, that which cannot be known by other means of valid cognition and that which is distinct from a mere action, is solely based on mundane usage,46 and not also on a Vedic passage such as “One who desires heaven should sacrifice.” Anticipating this objection, Śālikanātha refutes it as follows: How could an observer understand, before [he learns] language, that [Caitra’s] activity is caused by an obligation if a given obligation could not be known by other means of valid cognition? Without understanding [Caitra’s] activity [as such], how could he infer [Caitra’s] cognition of what is to be done, i.e., the cause of the activity? Still more, how could he infer the denotative power of the word? Only [Caitra’s] cognition of a mere “something to be done” can be inferred from his activity. But from the mundane perspective a mere action is ascertained to be the thing to be done.47

In other words, if one followed only mundane usage, one would learn that a mere action is the obligation and not that the obligation is previously unknown (Figure 14.7).

288

Kei Kataoka

Figure 14.7  An Action To Be Done and a Previously Unknown Obligation

mundane usage (A) mundane usage (B)

obligation action

unknown

KK

AK

Vedic sentence

Therefore, a mere obligation, that is, a mere unspecific “something to be done” (kāryamātraka), is learned by following mundane usage (A), but its being of the nature of something previously unknown is learned by following the Vedic sentence.48 One might object that this view contradicts mundane usage (B) because in all mundane cases (only) an action is learned as the thing to be done (and never something previously unknown).49 But this is not the case, because the condition in which something previously unknown is what is to be done (AK: apūrvakāryatva) is not contradictory to the condition in which an action is what is to be done (KK: kriyākāryatva). When something previously unknown is accomplished by an action, it is necessary for this action, as a means of accomplishing something previously unknown, to have been already accomplished. Therefore, that something previously unknown is to be done (AK) is harmonious with the understanding that an action is to be done (KK).50 In a mundane setting, people intend to express with a word a thing that can be known by other means of valid cognition. Therefore, mundane usage, solely with respect to an action to be done, is accompanied by secondary usage (lakṣaṇā).51 In other words, Śālikanātha claims that AK ascertained through the Vedic passage is the primary referent of the prescriptive endings, whereas KK ascertained through mundane usage is just secondary. Usually, people do not notice that their usage (of the prescriptive endings with respect to an action to be done) is in fact secondary, because they have not ascertained the primary referent (i.e., a previously unknown obligation). But those people who are skillful in terms of the primary referent know that (the action to be done) is a secondary referent.52 Prescriptive endings primarily denote an unknown obligation, from which an action to be done is secondarily understood (Figure 14.8).

Śālikanātha on Language Acquisition

289

Figure 14.8  An Action To Be Done as the Secondary Referent prescriptive endings

unknown obligation

action to be done

(primary)

(secondary)

13 Conclusion What is to be done (kārya), that is, for example, the primary referent of the optative ending (liṅ), is the main element of a sentence-referent. Although only a sentence is uttered in actual usage and not a word in separation, one can extract the referent of a word by the method of insertion and extraction (āvāpa and udvāpa). But how can one learn the referent of the prescriptive endings, which functions as the main element of a sentence-referent? First, in mundane usage, one learns, for example, that the referent of the imperative form is a thing to be done (kārya) and not a mere action. Although in mundane usage a thing to be done (kārya) always overlaps with an action and therefore looks identical to it, it can be distinguished from an action. Furthermore, a learner can distinguish two aspects of an action: its aspect of being a means of attaining a fruit (phalasādhanatā) and its aspect of being a thing to be done (kāryatā). One learns that the latter aspect, that is, kāryatā, is the sole real factor that functions as the cause of an activity, that is, the referent of the optative ending, and it is by understanding the kāryatā that one starts an activity. This internalized imperative which is understood as “This is what I have to do” (like an instinctive impulse) is variously called “impelling,” “invitation,” or “solicitation” according to the relationship between the speaker and the listener. But the only pervasive element is “a thing to be done.” After having learned a mere kārya, that is, a mere “something to be done,” as the referent of the prescriptive endings, in mundane usage, one is later taught the injunction for Vedic usage, in particular the injunction for an optional rite (kāmya). This helps one specify its referent as something unknown to be done (apūrvakārya), distinct from a (knowable and momentary) action to be done (kriyākārya). It becomes clear in a Vedic sentence that an enduring command (niyoga) not known by other means of valid cognition (mānāntarāvedya) is the thing to be done, because otherwise a Vedic sentence which promises a future fruit would not make sense. It is necessary that this Vedic command be accompanied by an object (viṣaya) and a person commanded (niyojya). Thus, a Vedic sentence denotes a command accompanied by a viṣaya and a niyojya.

290

Kei Kataoka

It is not the case that language acquisition takes place only in a mundane context. Both mundane and Vedic sentences are necessary for understanding Vedic injunctions, because otherwise people would understand only that an action to be done is the primary referent of the prescriptive endings. An unknown thing to be done (apūrvakārya) is in fact the primary referent, and a knowable action to be done (kriyākārya) is the secondary referent of the prescriptive endings, although people usually do not notice this. A mundane sentence intends to convey a knowable action to be done, whereas an unknown thing that is to be done is the primary referent of a sentence. Thus, a man-made sentence cannot be an independent source of information inasmuch as it does not independently convey an unknown thing to be done (apūrvakārya).

Notes 1 PrP3, v. 1, 417,5–6: nanu vyutpattyapekṣeṣu śabdeṣv arthābhidhāyiṣu / kathaṃ mānāntarāvedyaṃ kāryam āhur liṅādayaḥ //. The following explanation is based on my own preliminary edition of the verses of Vākyārthamātṛkā II (VM II), collated with the three published editions (PrP1, PrP2, and PrP3) and two manuscripts from Wai (W1) and Kerala (T1). The Vṛtti part is used only when necessary to elucidate the contents of the verses. New verse numbers are introduced because some verses have not been recognized as such in the previous editions. W1 also contains a separate part for the mūlapāṭha, that is, the section of verses only, after the entire Vṛtti ends. In other words, it has two sets of readings for the mūla verses. This mūlapāṭha has helped me to identify the verses not recognized in the previous editions. In the following footnotes, for the sake of conciseness, only the adopted readings are reported; variant readings are not included. 2 VM II, vv. 3–4ab (PrP3, 418,8–10): atrocyate, yadā nāma vṛddhenaikena bhāṣite / jalaṃ caitrāharasveti, caitra āharate jalam // tadā vyutpitsamāno ’nyas tatraivam avagacchati / 3 VM II, v. 4cd (PrP3, 418,11), not recognized as a verse: buddhipūrvā mamevāsya pravṛttir iyam īdṛśī // 4 VM II, v. 5ab (PrP3, 418,14), not recognized as a verse, with my emendation: ahaṃ buddhvā pravṛtto yat tad eṣo ’pi. 5 VM II, v. 5bcd (PrP3, 418,21), not recognized as a verse: tad asya tu / śabdena bodhyate ’nena, mama mānāntareṇa tu // 6 VM II, v. 6ab (PrP3, 419,5): tatra buddhvā pravṛtto ’haṃ kiṃ tāvat svayam anyadā / 7 VM II, v. 6c (PrP3, 427,4), not recognized as a verse, with my correction: na kriyāmātrakaṃ tāvat.

Śālikanātha on Language Acquisition

291

8 VM II, v. 6d (PrP3, 427,4), not recognized as a verse, with my correction: na cāpi phalamātrakam // 9 VM II, v. 7ab (PrP3, 427,4), not recognized as a verse, with my correction: na kriyāphalasaṃbandhamātraṃ vā. 10 VM II, v. 7b (PrP3, 427,8), not recognized as a verse: kiṃ tu kāryatām / 11 VM II, vv. 7cd–9ab (PrP3, 427,11–14), with my correction: āstāṃ tāvat kriyā loke gamanāgamanādikā // antataḥ stanapānādis tṛptikary api yā kriyā / sā yāvan mama kāryeyam iti naivāvadhāryate // tāvat kadāpi me tatra pravṛttir abhavan na hi / 12 To our surprise, the observer who is still in the process of language acquisition essentially knows how to distinguish the Prābhākaras’ view from Maṇḍana’s view expounded in the Vidhiviveka. 13 VM II, vv. 9cd–10d (PrP3, 428,13–15): phalasādhanatā nāma yā sā naiva ca kāryatā // kāryatā kṛtisādhyatvaṃ, phalasādhanatā punaḥ / karaṇatvaṃ phalotpāde, bhidyete te parasparam // 14 VM II, v. 11 (PrP3, 429,2–3): kiṃ tu svayaṃ kleśarūpaṃ karma yat kāryatāṃ   vrajet / phalasādhanatā tatra kāraṇam. 15 PrP3, Vṛtti, 429,11–12: ataḥ karmasu kāryatvāvagamaḥ phalasādhanatāvagama­ nibandhanaiti. 16 VM II, vv. 11d–12ab (PrP3, 429,3–4), with my correction: tena kāryatā // tadbhāvabhāvini nityaṃ tadātmeva pratīyate. 17 VM II, v. 12cd (PrP3, 429,13–430,1), recognized as a verse by Kuroda (1978): tathā cāsādhanasyāpi sukhasyaivāsti kāryatā // 18 VM II, v. 13abc (PrP3, 430,2), not recognized as a verse, with my correction: tena sādhanatottīrṇakāryatāvagamena me // pravṛttir iti. 19 VM II, v. 13cd (PrP3, 430,3), not recognized as a verse, with my emendation; the problem was pointed out by Kuroda (1978), but without a reconstruction: caitro ’pi kāryabodhāt pravartate // 20 VM II, v. 14 (PrP3, 430,10–11): śabdāntarāṇi svārtheṣu vyutpadyante yathaiva hi / āvāpodvāpabhedena tathā kārye liṅādayaḥ // 21 PrP3, Vṛtti, 430,21–431,1: pravartyapuruṣāpekṣayā jyāyasā vaktrā pratipādyamānaṃ kāryam “praiṣaḥ” iti vyapadiśyate. 22 PrP3, Vṛtti, 431,1: samena “āmantraṇam.” 23 PrP3, Vṛtti, 431,1: hīnena “adhyeṣaṇam” iti. 24 VM II, v. 15 (PrP3, 430,19–20): kāryam eva hi vaktṝṇāṃ jyāyaḥsamakanīyasām / pravartyāpekṣayā bhedāt praiṣādivyapadeśabhāk // 25 VM II, v. 16 (PrP3, 431,4–5): kāryam eva hi sarvatra pravṛttāv ekakāraṇam / pravṛttyavyabhicāritvāl liṅādyartho ’vadhāryate // 26 In Maṇḍana’s Vidhiviveka different kinds of imperative such as praiṣa are specifically called upādhi, that is, a nonessential, incidental condition. 27 VM II, v. 17 (PrP3, 431,13–14): kṛtisādhyaṃ pradhānaṃ yat tat kāryam abhidhīyate / tac ca mānāntareṇāpi vedyam odanapākavat //

292

Kei Kataoka

28 PrP3, Vṛtti, 431,16–17: prayatnaś ca kṛtiḥ. sa ca mānasapratyakṣavedya iti viśiṣṭaprayojanatāpi prayatnasya pratyakṣavedyaiva. 29 PrP3, Vṛtti, 431,15–16: kṛtau satyāṃ bhāvād asatyāṃ cābhāvād anumānataḥ kṛtisādhyatā tāvad avagamyate. 30 PrP3, Vṛtti, 432,1: tena pratyakṣānumānābhyāṃ kāryam avagamyate. 31 VM II, v. 18 (PrP3, 432,4–5): evaṃ kāryātmake ’py arthe vyutpadyante liṅādayaḥ / tadanviteṣu svārtheṣu tathā śabdāntarāṇy api // 32 I do not take into consideration here the effort (kṛti) that is denoted by the ākhyāta. The Prābhākara notion of kṛti roughly corresponds to bhāvanā of the Bhāṭṭa school. However, kṛti is not the main element for the Prābhākaras; rather, the command is the main thing to be accomplished by the effort (kṛtisādhya). 33 In other words, heaven is the object of desire (kāma/icchā) for the person who desires heaven. Here the desire for heaven functions as a condition of the qualification (adhikāranimitta). 34 One can perhaps compare the two views with the designations utilitarian and deontic. 35 Cf. PrP3, Vṛtti, 443,23–24: yathātmana eva saṃvidadhānaḥ svāmī garbhadāsasyopakaroti, tathā niyogo ’pi niyojyasyeti na prādhānyapracyutiḥ. 36 VM II, v. 19abc (PrP3, 432,8–9): saṃpradhāryam idaṃ tv atra, tat kāryaṃ kiṃ kriyātmakam / yad vā tadvyatirekīti. 37 VM II, vv. 19d–20ab (PrP3, 432,11–12): tatra lokānusārataḥ // pramāṇāntaravijñeyā kriyā kāryeti yady api / 38 VM II, vv. 21–30 (PrP3, 433,7–441,9): svargakāmādayaḥ kārye niyojyatvena saṃmatāḥ / svargakāmādibhiḥ śabdair vaktavyā ity avasthitam // niyojyaḥ sa ca kāryaṃ yaḥ svakīyatvena budhyate / svargādiḥ kāmayogāc ca sādhyatvenaiva gamyate // tena sādhyatvaparyantasvargādīcchāviśeṣitaḥ / tad eva śaknuyāt kāryaṃ boddhuṃ yat kāmyasādhanam // liṅādis tatra kāryaṃ cet kriyām evāvabodhayet / samanvayo niyojyena tadānīm avahīyate // kriyā hi kṣaṇikatvena na kālāntarabhāvinaḥ / svargādeḥ sādhyamānasya samarthā jananaṃ prati //   iṣṭasyājanikā sā ca niyojyena phalārthinā / kāryatvena na saṃbandham arhati kṣaṇabhaṅginī // tasmān niyojyasaṃbandhasamarthaṃ vidhivācibhiḥ /   kāryaṃ kālāntarasthāyi kriyāto bhinnam ucyate // tad dhi kālāntarasthānāc chaktaṃ svargādisiddhaye / saṃbandho ’py upapadyeta niyojyenāsya kāminā // kriyādibhinnaṃ yat kāryaṃ vedyaṃ mānāntarair na tat / ato mānāntarāpūrvam apūrvam iti gīyate // kāryatvena niyojyaṃ ca svātmani prerayann asau / niyoga iti mīmāṃsāniṣṇātair abhidhīyate // 39 VM II, v. 31 (PrP3, 441,11–12): kāryasyaiva pradhānatvād vākyārthatvaṃ ca yujyate / vākyaṃ tad eva hi prāha niyojyaviṣayānvitam // 40 VM II, v. 32 (PrP3, 443,17–18): ātmasiddhyanukūlasya niyojyasya prasiddhaye / kurvat svargādikam api pradhānaṃ kāryam eva naḥ // 41 VM II, v. 35 (PrP3, 45,15–16): evaṃ kāmādhikārārthaparyālocanayotthitā / vyutpattiḥ sarvavākyārthapratipattinibandhanam //

Śālikanātha on Language Acquisition

293

42 VM II, v. 36 (PrP3, 446,21–22), with my correction: vyutpattir iha kārye ’rthe vyavahārānusāriṇī / kiṃ tu nirdhāraṇāmātraṃ vedavākyavimarśajam //. This key verse of the siddhāntin should be placed before v. 37 (i.e., v. 32 in PrP3), which introduces a pūrvapakṣin’s view beginning with vyavahārata evāhur as witnessed in two editions and two manuscripts, and not after it, although it is incorrectly placed after it in PrP3 and given number 33, and therefore looks as if it were stated by an opponent. 43 PMS, Yava-varāha-adhikaraṇa, 1.3.8–9. 44 PrP3, Vṛtti, 446,6–7: tena vaidikavākyaśeṣānvayārhatālocanena dīrghaśūkādyarthatvam eveti rāddhāntaḥ. A Vedic sentence determines the referent of the word yava, which is ambiguous in mundane usage. 45 PrP3, Vṛtti, 446,9–10: vaidikaniyojyānvayayogyatayā liṅādīnām apūrvakāryābhidhāyakatvanirṇayo nānupapannaḥ. 46 VM II, v. 37 (PrP3, 446,15–16): vyavahārata evāhur vyutpattim apare punaḥ / kārye mānāntarāvedye kriyādivyatirekiṇi // 47 VM II, vv. 39–41 (PrP3, 447,7–22): kārye mānāntarāvedye pārśvasthas tannibandhanam / vyavahāraṃ kathaṃkāraṃ śabdāt prāg avabudhyatām //   vyavahāram avijñāya tannibandhanatadgatā / pratipattiḥ kathaṃ jñeyā, śabdaśaktiḥ kathaṃtarām // kāryapratītimātraṃ ca pravṛtter anumīyatām / kiṃ tu kāryaṃ kriyaiveti lokadṛṣṭyāvasīyate // 48 VM II, v. 43 (PrP3, 448,19–20): tasmāl lokānusāreṇa vyutpattiḥ kāryamātrake / tasya tv apūrvarūpatvaṃ vedavākyānusārataḥ // 49 VM II, v. 44 (PrP3, 448,22–23): nanu lokavirodhitvaṃ pakṣe ’sminn api dṛśyate / sarvathaiva yato loke kriyā kāryeti gamyate // 50 VM II, v. 46 (PrP3, 449,4–5): apūrvaṃ hi kriyāsādhyaṃ, sādhitā sādhanaṃ kriyā / tasmād apūrvakāryatvaṃ kriyākāryatvasaṃgatam // 51 VM II, v. 47 (PrP3, 449,6–7): pramāṇāntaragamyaṃ hi lokaḥ śabdair vivakṣati / kriyākāryatva evātaḥ prayogo lakṣaṇānvitaḥ // 52 VM II, v. 48 (PrP3, 449,17–18): lakṣaṇānabhimānas tu mukhyārthānavadhāraṇāt / ye tu mukhyārthakuśalās teṣāṃ lākṣaṇikatvadhīḥ //

Printed Sources PMS

Subbāśāstrī, ed. 1929–34. Śrīmajjaiminipraṇītaṃ Mīmāṃsādarśanam. Seven parts. Pune: Anandasramamudranalaya.

PrP1

Bālaśāstrī, ed. 1870. “Prakaraṇa-pañcikā.” The Pandit (Benares) 5:113–17, 139–43. On page 113 Bālaśāstrī noted that ff. 79–99 were lost. Therefore, the edition does not contain the beginning part of Vam II. It begins from liṅādiśabda eva pramāṇam iti sāhasam […], which corresponds to PrP3, 420,15.

294

Kei Kataoka

PrP2

Mukunda Śāstrī, Aḍkar, and Lakṣmaṅaśāstri Drāviḍa, eds. 1904. Prakaraṇapañcikā nāma Prabhākaramatānusāri-Mīmāṃsādarśanam. Chowkhamba Sanskrit Series 17. Benares: Vidyavilasa Yantralaya. Vam II in the PrP2 suddenly ends in the middle of the Vṛtti ad v. 3 at p. 22, which corresponds to PrP3, 420,14; it restarts the remainder, however, at page 172.

PrP3

Subrahmanya Sastri, A., ed. 1961. Prakaraṇa Pañcikā of Śālikanātha Miśra with the Nyāya-Siddhi of Jaipuri Nārāyaṇa Bhaṭṭa. Banaras Hindu University Darśana Series 4. Varanasi: Benares Hindu University.

VM II

Kataoka, Kei, ed. In preparation. “Vākyārthamātṛkā II.”

Manuscript Sources T1 Thiruvananthapuram, University Manuscripts Library, University of Kerala, Serial No. 16220, Manuscript no. 685. 98 pages, Paper, Devanāgarī script. W1 Wai, Prajñāpāṭhaśālā, Serial No. 6285, L. No. 6–5/411. 56 folios, Paper, Devanāgarī script.

15

The Deontic Nature of Language in Mīmāṃsā and Vedānta Schools1 Elisa Freschi

1  Introduction: What Is the Meaning Conveyed by a Text? The authors of the Nyāya school are upholders of a correspondence view of language, akin to the one found in L. Wittgenstein’s Tractatus. According to them, language reproduces ontologically given states of affairs in an epistemically valid way. In other words, a sentence like “The cat is on the mat” simply describes the state of affairs of a feline lying on a piece of soft textile. This view might appear to be the obvious option to Euro-American readers, but it was by no means the only one in the history of early Indian reflections on language. This contribution will discuss the idea that language communicates a deontic meaning, that is, a duty, something to be done (kārya). For simplicity’s sake, let me call “descriptive view” the view that language describes states of affairs, and “prescriptive view” that according to which language conveys things to be done. The descriptive view has no problem explaining the meaning of descriptive sentences like “The cat is on the mat,” but can encounter problems with prescriptive sentences like “Do your homework!.” In fact, if all sentences just describe an actual state of affairs, what is the state of affairs described by this sentence? The homework to be done does not yet exist in the external world, so in this sense the descriptive solution seems inadequate. A Nyāya author would still claim that the sentence describes a state of affairs in which the addressed person is characterized by the fact of having to do her homework. In current philosophy of language or deontic logic one could say that such a sentence describes an ideal world in which the addressed person would have done her homework. These explanations, however, appear

296

Elisa Freschi

as counterintuitive, because they do not correspond to what people usually understand as the message of prescriptive sentences. Normally, when we receive a command, we do not think of ideal worlds being described, and neither do we understand it as describing something about us. Therefore, both the above interpretations seem rather a posteriori ways of avoiding the deontic element intrinsic in commands. In fact, linguists working on the semantics of imperative sentences still discuss about how to determine their meaning and how to apply truth conditions to them, without postulating that imperatives are identical to modal sentences.2 By contrast, the prescriptive view has no problem explaining the meaning of prescriptive sentences but requires more effort to explain descriptive sentences as parasitical on implicit or explicit prescriptive ones. This chapter describes and compares the interpretations of prescriptive language in ordinary communication and in the sacred texts in the two schools of Mīmāṃsā and Vedānta.

2  The Veda, Mīmāṃsā, and Vedānta 2.1  What Is the Veda? The term “Veda” has meant different things for contemporary scholars and for most Mīmāṃsā and Vedānta authors. Contemporary scholars tend to call only the so-called Saṃhitās, namely the Ṛgveda-, the Sāmaveda-, the Yajurveda- and possibly the Atharvavedasaṃhitā, “Veda.” To each of these four compositions (saṃhitā), the tradition attaches further texts divided into different genres: the Brāhmaṇas, the Āraṇyakas, and the Upaniṣads. The Brāhmaṇas contain ritual prescriptions, as well as some etiological and mythical material loosely related to sacrifices. The Āraṇyakas contain speculations about the interiorization of sacrifice, and the Upaniṣads progress further on the path of the Āraṇyakas by offering philosophical speculations, which usually originate in the interiorization of sacrificial elements. All these texts are considered to be Veda by Indian authors. For Mīmāṃsā authors, however, “Veda” is primarily used in reference to the sacrificial texts, the Brāhmaṇas, while the Upaniṣads and the Saṃhitās are considered from the standpoint of the Brāhmaṇas. In the Mīmāṃsā view, the Saṃhitās supply the ritual formulas (mantra) to be used in the sacrifices prescribed in the Brāhmaṇas, while the Upaniṣads provide supplementary sentences about sacrificial elements. For instance, the speculations around the self (ātman) contained in the Upaniṣads should be understood as eulogies of

The Deontic Nature of Language in Mīmāṃsā and Vedānta

297

the agent of sacrifices. By contrast, Vedānta authors look at the Veda from the standpoint of the Upaniṣads.

2.2  The Mīmāṃsā School The Mīmāṃsā school recognizes as its basic text Jaimini’s Pūrvamīmāṃsāsūtra (henceforth PMS), possibly composed in the last centuries BCE. The PMS has been commented upon several times, but Śabara’s Bhāṣya (henceforth ŚBh), possibly composed in the fourth or fifth century CE, soon became the standard commentary and is therefore the only extant one, having superseded previous ones. In the sixth or seventh century two authors, Kumārila Bhaṭṭa and Prabhākara Miśra, offered divergent sub-commentaries on the ŚBh and thus created two Mīmāṃsā subschools, known as the Bhāṭṭa and the Prābhākara ones, respectively.3 The initial concern of the Mīmāṃsā school is the exegesis of Vedic texts, and especially of their prescriptive portion, the Brāhmaṇas. Since Mīmāṃsā authors consider the Vedas as authorless, they developed special hermeneutic rules for the interpretation of Vedic texts, independently of any author’s intention. Despite this particularity, for many Mīmāṃsā authors and especially for those of the Prābhākara school the Vedas remain the paradigmatic case of language, although they always take into consideration also worldly (laukika) usages of language, along with the Vedic ones.

2.3  The Vedānta School The Vedānta school is a philosophical school primarily focusing on the exegesis of the Upaniṣads. Vedānta, as we currently know it, gained recognition in the philosophical arena of South Asia only in the latter half of the first millennium CE; hence Jaimini, Śabara, Kumārila, and Prabhākara did not need to address or reject its claims. One of the first authors who made Vedānta into a full-fledged philosophical school is Śaṅkara (possibly eighth century CE), who is traditionally recognized as the founder of the Vedānta school of nondualism (advaita). Śaṅkara is not keen to establish the harmony of Mīmāṃsā and Vedānta, and instead affirms that the study of Vedānta can be autonomous and does not presuppose the study of Mīmāṃsā. The situation becomes more complex for further Vedānta authors. Rāmānuja (eleventh century), who is traditionally credited with the foundation of the Vedānta school of qualified nondualism (viśiṣṭādvaita), is in fact actively campaigning for the unity of Mīmāṃsā and Vedānta, which he respectively calls

298

Elisa Freschi

Prior (pūrva) and Subsequent (uttara) Mīmāṃsā. The doctrine of the unity of the Mīmāṃsā will be further developed by a later thinker of the Viśiṣṭādvaita Vedānta school, namely Veṅkaṭanātha (thirteenth to fourteenth century), who tries to synthesize a unitary Mīmāṃsā teaching.4

2.4  Views on Deontic Language in Mīmāṃsā and Vedānta The main difference between the two sub-schools of Mīmāṃsā lies in the fact that Prābhākara authors consider worldly language to be structurally similar to Vedic language, insofar as both convey only something to be done. In other words, sentences do not convey the existence of something, but rather that something is to be done. Sentences which look as if conveying a descriptive statements should be interpreted as supplementing a prescriptive statement, which might be explicitly expressed, but can also be implicit. For instance, in day-to-day language, “It is hot in here” should be understood as a supplement of “Please, open the window,” just like in Vedic language “Vāyu is the swiftest deity” is a supplement of “One should sacrifice to Vāyu.” Bhāṭṭa authors, by contrast, claim that only Vedic language conveys something to be done, whereas they interpret worldly language according to the descriptive view outlined above (Section 1). Otherwise said, Prābhākara authors uphold that in all cases language conveys something to be realized (sādhya), whereas Bhāṭṭa authors state that worldly language conveys something already realized (siddha). Vedānta authors, some of whom emphasize their connection to the Mīmāṃsā, either claim that both in the Veda and in worldly usage all sentences convey descriptive meanings or try to find a way to account for both this thesis and the Mīmāṃsā view of Vedic prescriptions. These differences are rooted in a more general relation of Vedānta with Mīmāṃsā: authors who are closer to Mīmāṃsā try to embed its approach to Vedic language into their system, as will later be explained in Section 5.2.

3  Inner-Mīmāṃsā History of Deontic Language According to the Mīmāṃsā, each instrument of knowledge is defined by its capacity to make one understand something new, that is, previously unknown (apūrva). Sense perception is an instrument of knowledge because it allows one to access what exists hic et nunc. By contrast, the Veda is an instrument of knowledge because it conveys dharma, that is, one’s individual duty, which

The Deontic Nature of Language in Mīmāṃsā and Vedānta

299

Table 15.1  The Early Mīmāṃsā and Prābhākara Model Instrument of knowledge

Field of knowledge

Sense perception, etc.

What exists

Veda

What should be done (dharma)

cannot be known through sense perception or other instruments of knowledge since these all depend on perceptual data. In fact, except for the Veda, all the instruments of knowledge, beginning with inference, can only elaborate the raw data delivered by perception. Therefore, in order to safeguard the validity of the Veda as an instrument of knowledge, dharma needs to be drastically distinguished from the senseperceivable aspect of sacrifices, such as the sacrificial substances and acts, and needs to be radically inaccessible to sense perception. The nonavailability of dharma to sense perception is therefore a necessary presupposition in Mīmāṃsā, because it is the basis for the primacy of the Veda. Consequently, the Veda needs to be prescribing duties, since otherwise the Veda would not be conveying anything new and specific. In this connection, there is a fundamental divergence in the epistemology of the two schools. Prābhākara authors uphold that there is a radical difference between Vedic and worldly communication, insofar as only the Veda conveys something new and is thus an instrument of knowledge (pramāṇa). By contrast, Bhāṭṭa authors maintain that in both cases linguistic communication is an instrument of knowledge, although only in the case of Vedic language it is a unique source for its content. Thus, according to the Bhāṭṭa view, ordinary language can be about existing states of affairs, which one could have known also through other instruments of knowledge, if one had tried hard enough. For instance, I know that Siberian tigers or Tasmanian emus exist just because of the reports of other people, but could also imagine traveling to Siberia or Tasmania and trying to see them. Even if it does not convey an exclusive type of knowledge, ordinary language is still useful, because it makes it possible for me not to have to check for myself everything in the world.

3.1  Jaimini and Śabara on Deontic Language 3.1.1  Jaimini on artha Deontic elements played a role since the early history of Mīmāṃsā, as shown also by the complex semantics of the term artha. This plays a key role since

300

Elisa Freschi Table 15.2  The Bhāṭṭa Mīmāṃsā Model Instrument of knowledge

Field of knowledge

Sense perception, etc.

What exists

Worldly language

What exists

Veda

What should be done (dharma)

the beginning of the PMS, but has multiple meanings in Sanskrit.5 In Sanskrit philosophical texts, it could mean any of the following: 1. 2. 3. 4.

the meaning of a linguistic expression; the content of any cognitive act; its corresponding referent; or a goal or purpose.

The confluence of these senses permeates Mīmāṃsā texts and needs to be constantly kept in mind when readers encounter the term artha in Mīmāṃsā literature.6 This is not due to a lack of awareness, but rather to the Mīmāṃsā approach to language, according to which the meaning of language is chiefly deontic, that is, something to be realized. Prābhākara authors emphasize duty (kārya), whereas Bhāṭṭa authors stress the aspect of purpose. In both cases, however, the criterion of having a single meaning (ekārthatā) identifies a sentence and entails the fact of having a single meaning which is a purpose or something to be done.7 It is noteworthy that at least the meanings (1) and (4) are present already in Jaimini’s use of the term artha. PMS, 1.1.2, most significantly, defines the object of the Mīmāṃsā investigation, dharma, as follows: “Dharma is the artha conveyed by Vedic prescriptions (codanālakṣaṇo ’rtho dharmaḥ).” What does artha mean here? The aphorism cannot mean that dharma is the content or meaning of a Vedic prescription, because this would be tautological and it is a fixed convention in philosophical sūtras that all words must be interpreted as having a distinct purpose. If this were the case, either of the words “artha” or “conveyed by” (lakṣaṇa) could just be left out, without loss of meaning. Therefore Frauwallner here translates “artha” as “useful thing” (“etwas Nützliches,” Frauwallner 1968, 17). Yet, Frauwallner’s solution is not fully satisfactory. When Śabara explains that the mention of “artha” in the aphorism is necessary, he discusses in this connection the conundrum of alleged non-arthas (anarthas) that appear to be conveyed by some Vedic prescriptions, for example, the malefic sacrifice called śyena through which one can harm one’s enemy. Śabara, and later Kumārila,

The Deontic Nature of Language in Mīmāṃsā and Vedānta

301

ask whether such a non-artha can legitimately be called dharma just because it is conveyed by a Vedic prescription. The problem with the śyena sacrifice is not that it is in itself not useful, but that its goal, namely harming one’s enemy, is in contradiction with further Vedic duties, most notably with the prohibition to harm living beings. The śyena is certainly useful, if one wants to harm one’s enemy, and could even be one’s own goal, as in meaning (4) above. But it is defined as a non-artha because it is in conflict with other duties conveyed by the Veda, and so it cannot be prescribed on their same level. Therefore it is not a principal, unconditioned duty, and can at most become obligatory if one is in a subideal situation.8 Moreover, a translation of artha that focuses on its descriptive aspect, such as “useful thing,” certainly does not do justice to the nature of dharma, which is defined as something to be realized, and not as something already existing. Let me now check this interpretation against the background of another important aphorism, namely PMS, 1.1.5: The relation between language and artha is original. The knowledge of dharma is the teaching. And this is infallible in regard to an imperceptible artha. Therefore it is an instrument of knowledge, according to Bādarāyaṇa, because it is independent.9

As explained above, it is important to highlight how in order for the Veda to be an instrument of knowledge its artha needs to be unperceived (anupalabdha). This would not be the case if an artha were just temporarily unavailable, for example, because its vision is obstructed by a wall. Consequently, there must be something in the nature of the proper artha of the Veda that makes it unsuitable to be directly perceived. In the context of Jaimini’s aphorisms,10 this intrinsically imperceptible artha is duty, dharma. This runs against the simplistic interpretation of artha as an external referent.

3.1.2  Śabara’s Seemingly Contradictory Statements Even admitting that Jaimini understands language as conveying a deontic purpose, one might object that this is not Śabara’s standpoint. In fact, in his commentary on PMS, 1.1.5, Śabara explicitly states that the meaning of words is a universal, and that this universal is perceptible. He even opposes the condition of being perceptible to that of being something to be realized (Vṛttikāragrantha, 40): [Q:] What is the meaning of the word “cow”? [A:] We say that this is the universal specified by having a dewlap, etc.

302

Elisa Freschi

[Q:] But is the universal something to be realized or not? [A:] If something is directly perceptible, it should not be something to be realized.11

Does this interpretation of words signifying perceptible universals hold for all words? Elsewhere, Śabara himself states that sacrifices are instead something to be realized (ŚBh, ad PMS 3.6.43).12 These two statements could be harmonized, assuming that for Śabara some words denote something already established and some denote something to be realized. But what about sentences? Śabara’s states that “a word signifies a universal, a sentence an individual” (ŚBh, ad PMS 1.1.24).13 This phrase at first sight leaves no space for artha as a purpose to be accomplished, since it seems to distinguish the meaning of words and that of sentences only on the basis of their being more or less specific. Śabara leaves the problem of the seeming contradiction between his and Jaimini’s standpoints unresolved. Most significantly, he uses the prescriptive expression “One who desires heaven should sacrifice” (svargakāmo yajeta) as if it were interchangeable with the description “Through sacrifice, heaven occurs” (yāgāt svargo bhavati),14 thus implicitly presupposing that prescriptive sentences can be reinterpreted as descriptive ones, describing how the prescribed action is the instrument to realize a desired goal. However, Śabara states in his commentary on Jaimini’s definition of sentence, “what has a single artha” (ŚBh, ad PMS 2.1.46), that in the case of a universal there is a seen meaning (dṛṣṭa artha), whereas in the case of the sentence he rather speaks of a purpose (prayojana). This suggests that by “a word signifies a universal, a sentence an individual” Śabara did not want to indicate an ontological correlate for words as well as for sentences. A seen meaning could be interpreted as indicating an established meaning (siddha artha) and this could be in turn equated to an external object only in the case of some nouns and not in the case of nouns expressing sacrifices. Nonetheless, the identification of the sentence meaning as purpose clearly rules out the possibility of identifying artha with an external object. Śabara’s sub-commentator Kumārila was forced to try to offer a solution to Śabara’s implicit ambiguity. Kumārila kept Śabara’s ontological commitment to substantialism in the case of words, and more explicitly of nouns, insofar as he identified the meaning of words as the universal (ākṛti). Kumārila then explained that at the level of sentences every element of a sentence concurs toward a sentence meaning, which is an action and thus something to be accomplished. In this way, Śabara’s universal meaning (sāmānya) is confirmed as something

The Deontic Nature of Language in Mīmāṃsā and Vedānta

303

already established. By contrast, his individual meaning (viśeṣa), which is conveyed by a sentence, becomes with Kumārila more of a goal, rather than something already established. This position is even stronger in Prabhākara’s sub-commentary on Śabara, due to the interpretation of the sentence meaning as something to be done (kārya). In sum, sentences are characterized as conveying a single artha in both Kumārila’s and Prabhākara’s accounts, but the artha of sentences is respectively considered a purpose or something to be done.

3.2  Prābhākara Mīmāṃsā on Prescriptive Language as the Norm Bhāṭṭa authors differ from the Prābhākara analysis of non-prescriptive statements in ordinary and Vedic language—outlined above through the examples of “It is hot in here” and “Vāyu is the swiftest deity” (Section 2.4)—because according to them ordinary language can also convey the description of states of affairs. Nonetheless, they agree that Vedic sentences only convey a prescriptive meaning. This asymmetry between Vedic and ordinary language is due to the abovementioned fact that the Veda, in order to be an instrument of knowledge, needs to have a specific field of application, which cannot be shared with the other instruments of knowledge. Conversely, Prābhākara authors did not want to have two distinct accounts for what is conveyed by Vedic and ordinary language; hence they asserted that, in general, each kind of language conveys duties. They are thus confronted with the problem of explaining how even in common language seemingly descriptive statements do not in fact denote any descriptive meaning. Two counterfactual examples advanced by Prābhākara opponents are “Your wife just begot a son” and “Your unmarried daughter is pregnant.” These sentences clearly convey a descriptive content and they are certainly understood by their hearer, who as a consequence experiences happiness and anger, respectively. How can this be explained within the Prābhākara theory? Prābhākara authors answer by arguing that the hearer’s reaction is not due to the understanding of a descriptive meaning, which would contradict their theory. Rather, the hearer reacts because he infers that something extremely good or bad must have happened, after noticing the facial expression of the speaker. A present-day Prābhākara might argue in favor of the prescriptive view differently, perhaps by construing the above sentences as supplements to an implicit prescription, for example, “Go back home and perform the

304

Elisa Freschi

prescribed rituals!” and “Punish your daughter!.” They might in this way stress the fact that language is used in order to perform illocutionary or perlocutionary speech acts and that it is hardly the case that it represents a pure exchange of information. Even in the extreme case of chatting or gossiping, the exchange of information can be better understood as a supplement to an implicit prescription. “X bought a new car” would therefore be a supplement of “You should also buy one” or of “You should blame them for their choices.” Alternatively, a present-day Prābhākara proponent might argue that Vedic language is prescriptive in a different sense than ordinary language. In favor of this view, he could mention the fact that according to Prābhākara Mīmāṃsā ordinary language is not an independent instrument of knowledge, since one understands its meaning via the understanding of the intention in the utterer’s mind. This is not the case with Vedic texts, since they are authorless and therefore convey their meaning independently of the intrusion of an utterer’s intention.

4  Epistemological and Logical Consequences The thesis that all meanings are prescriptive has important epistemological and logical implications. Epistemologically, it implies that the Veda is an instrument of knowledge only in relation to duties. By contrast, sense perception or inference only conveys knowledge about what exists. Therefore, the knowledge of duties conveyed by Vedic sentences cannot be falsified by other instruments of knowledge, which convey a completely different set of contents. This leads to the conclusion that sense perception and the Veda have two radically different precincts of application. In this way artha can also mean an epistemologically sound content of knowledge, as in (2) listed in Section 3.1.1. The Mīmāṃsā also upholds the theory of intrinsic validity of cognitions (svataḥ prāmāṇya), according to which any cognition is valid unless and until it is proven to be false.15 The conjunction of the prescriptive theory with that of intrinsic validity leads to the conclusion that, since the Veda teaches us about duties and since duties cannot be known through any other instrument of knowledge but the Veda, the Veda remains unfalsifiable. Further, the prescriptive nature of meanings contradicts the idea that language simply corresponds to states of affairs, without any intermediate mental step, since prescriptive contents just do not exist as external and subject-independent states of affairs.

The Deontic Nature of Language in Mīmāṃsā and Vedānta

305

Recapitulating, of the four possibilities listed in Section 3.1.1, the artha of the Veda is its meaning (1), while being also an epistemologically soundly conveyed content (2), and a goal to be achieved through one’s ritual actions (4). The only acceptation that needs to be restricted is the descriptive one (3), which applies only to single word-meanings. Thus, artha in the Veda comes closer to something to be done, a duty. According to this interpretation of PMS, 1.1.2, dharma is a duty conveyed by the Veda and a malicious sacrifice such as the śyena would be a non-artha because it is not an imperative duty (Section 3 above). This further suggests that, in order to defend their prescriptive views, Mīmāṃsā authors needed a theory of language, an epistemology, and a logic that is able to accommodate nondescriptive items. They developed a structured theory regarding duties and their hierarchy, distinguishing among primary and secondary ones, among prohibitions and prescriptions, and adding further conditions of application for each of them in order to rule out possible deontic conflicts. In other words, they developed a way to think of knowledge and of the sentences conveying it, which implied the prescriptive dimension much like in medieval and contemporary modal and deontic logics.16

5  Deontic Logic and Its Linguistic Consequences The Mīmāṃsā analysis differentiates between prescriptions (vidhi) and prohibitions (niṣedha). These are distinguished because the performance of the content of prescriptions generally leads to results that are independently desirable, such as sons, cattle, wealth, and happiness. By contrast, abiding by a prohibition does not lead to anything desired, whereas transgressing it leads to negative consequences.17 Prescriptions and prohibitions, therefore, are not identified just on the basis of their superficial linguistic form. Although generally prohibitions include an explicit negation, while prescriptions do not, a sentence may present a negation and still be a prescription, if it leads to a result. This means that, just like in the case of the identification of verbs expressing a prescription based on their meaning rather than on their form, semantics prevail over morphology. For instance, the utterance “One should not look at the rising sun” can in this sense be reinterpreted as enjoining “One should form the resolution not to look at the raising sun,” if it entails a desirable result. Conversely, exhortative statements which do not contain a negative particle must be interpreted as prohibitions, if they entail a sanction.

306

Elisa Freschi

At times, no sanction is mentioned, but still an exhortative statement containing a negative particle needs to be interpreted as a prohibition. For instance, the utterance “One should eat the five types of five-nailed animals” in fact conveys a prohibition, namely, the prohibition to eat any other kind of animal. Why so? Because, as outlined above, each instrument of knowledge is defined by the fact that it conveys something new (apūrva), and the Veda is an instrument of knowledge for this very reason; a prescription telling one that she should eat animals would not convey anything new, since appetite for edible meat is a natural instinct that does not require instructions. By contrast, what is new in the statement is its prohibitive component, namely that animals others than the five mentioned should not be eaten. Alongside novelty, another criterion is the principle of singularity, namely, the idea that Vedic prescriptions should convey a single duty. Sentences are accordingly interpreted following a strong teleological approach, in which everything should ultimately be functional to the communication of a single duty. Consequently, whatever does not directly contribute to this purpose is interpreted as indirectly contributing to it, for instance, as a praise of elements connected with the duty. This is how commendatory statements (arthavādas) are interpreted as supplements to a prescription. In other words, they are descriptive statements that have a meaning exactly because they supplement a prescription (see Section 2.4 above). A further deontic criterion guiding the interpretation of sentences is that prescriptions should be deontically feasible. The prescribed obligations need to be fulfillable. For instance, the prescription “One should sacrifice as long as one is alive” could in principle be interpreted in two ways: (a) one should perform a sacrifice that lasts the span of one’s life, and (b) one should sacrifice, under the condition of being alive. In other words, the phrase “as long as one is alive,” could be interpreted as specifying either the act or the agent. The former interpretation (a) would lead to several problems, insofar as it is impossible to perform a single sacrifice that lasts exactly the extent of one’s life, considering that one cannot predict her life span. The sacrifice would invariably be too long or short. If one were to complete it and then repeat it, in order to fill the reach of one’s whole life, one would add an element which is not directly enjoined in the original prescription, namely the sacrifice’s repetition. By contrast, the latter interpretation (b) does not require any further addition and is completely compatible with what is directly enjoined, since the specification “as long as one is alive” is interpreted as expressing a condition, and not a precise time span.18

The Deontic Nature of Language in Mīmāṃsā and Vedānta

307

Summing up, the three criteria of novelty, teleology and feasibility guide one’s interpretation and constitute some of the rules (nyāya) to be applied before the final ascertainment of the sentence meaning.

5.1  Vedānta Later Mīmāṃsā authors defend the prescriptive theory also against the Vedānta view, according to which the main function of the Veda is to convey descriptive contents about the Absolute (brahman) and the self (ātman). Vedānta authors concur that the Veda needs to have a specific precinct of application, inaccessible to sense perception. However, they add that this unique content is not (or is not only) prescriptive, since the main purpose of the Veda is, according to them, letting us know about the brahman, which is supra-sensorial and can only be known through the Veda. Their next task, therefore, is to decide whether the Veda has only one main content—namely, brahman—or rather two—namely, brahman and dharma—and in the latter case how the two are connected. This means that the tables discussed above (Table 15.2) remains valid, but the identification of the specific field of the Veda may change (Table 15.3). In Advaita Vedānta, Śaṅkara was ready to admit that the Brāhmaṇas do in fact convey prescriptive meanings (sādhya artha), whereas the Upaniṣads convey descriptions of states of affairs (siddha artha), that is, they describe the reality of the self (ātman) and of the brahman. Thus, for Śaṅkara two different hermeneutic strategies are justified in the case of Brāhmaṇas and Upaniṣads, since they are different classes of texts and are addressed to different kinds of people. Rāmānuja and Veṅkaṭanātha, who try to synthesize a unitary Mīmāṃsā teaching, are confronted with the problem of taking seriously the unity of the Vedic texts and of their exegetical strategies. Accordingly, Rāmānuja supports the underlying unity of what he calls the Prior and Subsequent Mīmāṃsā, that is, Mīmāṃsā and Vedānta. He needs, therefore, to solve the apparent contradiction of two seemingly different hermeneutic strategies for what should be parts of the same text (the Veda). Thus, Rāmānuja cannot just use the strategy applied by Table 15.3  The Vedānta Model Instrument of knowledge

Field of knowledge

Sense perception, etc.

What exists in the world

Veda

brahman and/or dharma

308

Elisa Freschi

Śaṅkara to the Upaniṣads and extend it to the Brāhmaṇas. Moreover, he cannot just deny that actions are undertaken upon hearing prescriptive statements such as those expressed in the Brāhmaṇas, so a theory of signification which does not account for this phenomenon would not be adequate. Consequently, Rāmānuja explains that actions are, in fact, undertaken upon hearing the prescriptions of the Brāhmaṇas, but only as a consequence of the fact that they convey descriptive meanings. In a simile by Rāmānuja, only once one knows that there is a treasure in a house does one undertake action to take hold of it. Analogously, prescriptions are parasitical upon descriptive statements, insofar as prescriptive meanings are conveyed only once the descriptive meaning has been understood. It could not work in any other possible way, according to Rāmānuja and Veṅkaṭanātha, because one would not undertake any action unless one had cognized something as the goal to be attained by one’s action.19

5.2  From Words to Meaning according to Viśiṣṭādvaita Vedānta In his Seśvaramīmāṃsā commentary to the PMS, 1.2.1, Veṅkaṭanātha writes: The Veda is valid because it conveys an artha which has a purpose (vedasya […] saprayojanārthabodhanena prāmāṇyam).

As seen above, in Jaimini’s aphorisms artha has several acceptations, which all share the meaning of purpose. The above statement shows that in the Seśvaramīmāṃsā, unlike in the PMS, artha must be different than purpose, because it possesses a purpose (saprayojana-artha). Thus defined, it cannot be the same as purpose. Thus, Viśiṣṭādvaita Vedānta authors like Veṅkaṭanātha split again the artha of PMS into two aspects, so that linguistic expressions convey meanings and these meanings are purposeful toward a given result, which becomes therefore a second step in the signification process. Accordingly, since the Veda is a valid instrument of knowledge because it conveys an artha which has a purpose, the purpose is achieved through the artha. This might look like a small change in comparison to the position of Prior Mīmāṃsā, but in fact it leads to a different appreciation of the epistemological role of artha. In fact, according to Rāmānuja (see Section 5.1 above), only if the artha is true one can fulfill the purpose. The processes of linguistic communication analyzed in the Bhāṭṭa and Prābhākara schools of Mīmāṃsā (see Chapter 9 in this volume) remain in place, but with the addition of the intermediate step of communicating an established meaning (siddha artha). Does this mean that the

The Deontic Nature of Language in Mīmāṃsā and Vedānta

309

Viśiṣṭādvaita Vedānta authors see sentences as conveying an established meaning, like in the Nyāya school? Not exactly, since it is not the case that one directly moves from a linguistic expression (śabda) to an established meaning (siddha artha), as claimed by the Nyāya school. Rather, Veṅkaṭanātha appears to assume that first the individual words of a sentence communicate their meanings, as with the theory of the Bhāṭṭa school, and then these meanings combine into a sentence meaning expressing something established. Through this established meaning, a purpose is then conveyed. In this model, descriptive passages of the Veda such as eulogies (arthavāda) are no longer to be understood just as supplements of prescriptions (see Section 2.4 above); on the contrary, prescriptions are considered as a later output of one’s understanding of descriptive statements. Therefore, Rāmānuja and Veṅkaṭanātha seem to have adopted the Bhāṭṭa theory of sentence meaning, but to have widened it with the addition of a distinct step of the cognition of the purpose.20

6 Conclusion Indian philosophical schools dedicated much energy to the issue of prescriptive language. The most striking case is that of the Prābhākara Mīmāṃsā authors. According to Mīmāṃsā, epistemically valid knowledge necessarily is knowledge of something new. In the specific case of language, its capacity to convey something new is directly connected by Prābhākara authors to its capacity to convey a duty. Language which does not convey duties only repeats contents and is therefore not an independent instrument of knowledge. The Bhāṭṭa sub-school of Mīmāṃsā adopts a less extreme view, insofar as it distinguishes the descriptive use of language from the prescriptive one, and recognizes that both uses can have epistemic validity. Lastly, the Vedānta schools, and especially Viśiṣṭādvaita Vedānta, face the additional challenge of having to deal with the prescriptive portions of the Vedas without abandoning their descriptive interpretation of the Upaniṣads. I have shown how this leads to an interpretation of prescriptive language as parasitic on descriptive meanings, and as leading to the cognition of a goal. In logical terms, one could think of a parallel with the modern debate between interpreters suggesting an imperative logic—not subject to truth tables—and scholars using deontic logic—subject to truth tables, although in an ideal condition of fulfillment.

310

Elisa Freschi

Notes 1 Research for this chapter has been supported by the WWTF project MA 16_028. I could read and discuss the texts upon which this chapter is based during several workshops organized at the IKGA (see https​://mi​mamsa​langu​age.w​ordpr​ess.c​om/ pa​ge/).​ I am grateful to Marco Lauri for improving the language. 2 One can get an idea of the complex problem of the semantics of imperatives by reading Charlow 2014. I am grateful to Jennifer Nagel for having pointed it out to me. 3 For an overview on the above dates, see Kataoka 2011, Introduction. 4 Rāmānuja’s and Veṅkaṭanātha’s position might have had sociological reasons, insofar as they might have wanted to profit from the prestige enjoyed by the Mīmāṃsā school. This thesis is suggested as a reason for the imitation of the Mīmāṃsā methodology by Vedānta authors in Bronkhorst 2007. In the following, I will rather focus on the intrinsic philosophical value and content of their ideas. 5 See Freschi and Keidan 2017 for a discussion of the semantic values of artha. 6 Focusing on Jaimini, already Scharf (1996, chapter 3.2) observes that the term artha conveys at the same time the idea that dharma is the meaning of Vedic prescriptions and that it is a purpose, that is, something to be achieved. 7 See PMS, 2.1.46 and ŚBh thereon. 8 See Freschi et al. 2017 and Freschi et al., forthcoming for a deontic analysis of the śyena conundrum. 9 autpattikas tu śabdasyārthena sambandhaḥ. tasya jñānam upadeśaḥ. avyatirekaś cārthe ’nupalabdhe tat pramāṇaṃ bādarāyaṇasyānapekṣatvāt. 10 See, especially, PMS, 1.1.2 and the commentary thereon. 11 atha gaur ity asya śabdasya ko ’rthaḥ? sāsnādiviśiṣṭā ākṛtir iti brūmaḥ. nanv ākṛtiḥ sādhyāsti vā na veti? na pratyakṣā satī sādhyā bhavitum arhati. 12 sādhyaś ca jyotiṣṭomaḥ. 13 sāmānye hi padaṃ pravartate, viśeṣe vākyam. 14 See (ŚBh, ad PMS 3.4.39): asti jyotiṣṭomaḥ, jyotiṣṭomena svargakāmo yajeteti. tatra dīkṣaṇīyādayaś ca yāgā vidyante, sautye cāhani somayāgaḥ. tatra saṃdehaḥ, kim atra yāgamātra pradhānam, uta somayāga iti. kiṃ prāptam? jyotiṣṭome tulyāni sarvāṇi bhaveyuḥ. kutaḥ? aviśiṣṭaṃ hi kāraṇam, yāgāt phalaṃ śrūyate. Similar paraphrases can also be found in (ŚBh, ad PMS 3.4.40 and 6.1.3). See Kataoka 1995 for a comprehensive discussion of this topic. 15 On the svataḥ prāmāṇya theory, see Taber 1992; Kataoka 2011. 16 See Freschi, Ollett, and Pascucci, forthcoming for an introduction to the deontic logic presupposed by Mīmāṃsā authors. 17 But the distinction between the prescriptions and prohibitions is more complex. For further details, see Freschi and Pascucci, forthcoming.

The Deontic Nature of Language in Mīmāṃsā and Vedānta

311

18 This topic is discussed in PMS, 3.4.3, as well as in the ŚBh and Tantravārttika thereon. 19 This argument and the treasure example are discussed in more detail in Freschi, forthcoming. 20 Theoretically the same scheme could also be applied to the Prābhākara analysis, but I don’t know whether it ever was applied to it.

Primary Sources Mīmāṃsādarśana

Abhyankar, Kashinath Vasudev, and Ganesh Shastri Ambadas Joshi, eds. 1970–1974. Mīmāṃsādarśana. Anandasramasamskrtagranthavali 97. Pune: Anandasrama.

PMS

Pūrva Mīmāṃsā Sūtra of Jaimini. In Mīmāṃsādarśana.

ŚBh

Śābarabhāṣya of Śabara. In Mīmāṃsādarśana.

Seśvaramīmāṃsā

Viraraghavacharya, Uttamur T., ed. 1971. Seśvaramīmāṃsā and Mīmāṃsāpaduka of Veṅkaṭanātha. Madras: Ubhaya Vedanta Granthamala.

Tantravārttika

Tantravārttika of Kumārila Bhaṭṭa. In Mīmāṃsādarśana.

Vṛttikāragrantha

Frauwallner, Erich, ed. Vṛttikāragrantha. In Frauwallner 1968.

Secondary Sources Bronkhorst, Johannes. 2007. “Vedānta as Mīmāṃsā.” In Mīmāṃsā and Vedānta: Interaction and Continuity, ed. by Johannes Bronkhorst, 1–91. Delhi: Motilal Banarsidass. Charlow, Nate. 2014. “The Meaning of Imperatives.” Philosophy Compass 9 (8): 540–55. Frauwallner, Erich. 1968. Materialien zur ältesten Erkenntnislehre der Karmamīmāṃsā. Veröffentlichungen der Kommission für Sprachen und Kulturen Süd- und Ostasiens 6. Wien: Österreichische Akademie der Wissenschaften. Freschi, Elisa. Forthcoming. “Descriptive and Prescriptive Language in the Vedas: The Pūrva Mīmāṃsā Interpretation of the Vedas from Śaṅkara to Rāmānuja and after Him.” In The World of Ramanuja: Tradition and Thought in History and Today, ed. by Godabarisha Mishra and Chakravarthi Ram-Prasad. Madras: University of Madras. Freschi, Elisa, and Artemij Keidan. 2017. “Understanding a Philosophical Text: A Multi-Disciplinary Approach to the Problem of ‘Meaning’ in Jayanta’s Nyayamañjarī, Book 5.” In Reading Bhaṭṭa Jayanta on Buddhist Nominalism, ed.

312

Elisa Freschi

by Patrick McAllister, 251–90. Wien: Verlag der Österreichischen Akademie der Wissenschaften. Freschi, Elisa, Andrew Ollett, and Matteo Pascucci. 2019. “Duty and Sacrifice. A Logical Analysis of the Mīmāṃsā Theory of Vedic Injunctions.” History and Philosophy of Logic. 40 (4): 323–54. Freschi, Elisa, and Matteo Pascucci. Forthcoming. “Deontic concepts and their clash in Mīmāṃsā: Towards an interpretation.” Freschi, Elisa, et al. 2017. “Understanding Prescriptive Texts: Rules and Logic Elaborated by the Mīmāṃsā School.” Journal of World Philosophies 2 (1): 47–66. Kataoka, Kei. 1995. “Naraseru ho Raishaku-gaku (Mīmāṃsā Theory of Causal Action: Śabara’s Concept of Bhava, Kriyā and Bhāvanā) [In Japanese with an English Summary].” Indo-Tetsugaku Bukkyo-gaku Kenkyu 3: 47–60. Kataoka, Kei. 2011. Kumārila on Truth, Omniscience, and Killing. Part 2. An Annotated Translation of Mīmāṃsā-Ślokavārttika ad 1.1.2 (Codanāsūtra). Beiträge zur Kultur- und Geistesgeschichte Asiens 68. Wien: Verlag der Österreichischen Akademie der Wissenschaften. Scharf, Peter M. 1996. The Denotation of Generic Terms in Ancient Indian Philosophy: Grammar, Nyāya, and Mīmāṃsā. Transactions of the American Philosophical Society Held at Philadelphia for Promoting Useful Knowledge 86.3. Philadelphia: American Philosophical Society. Taber, John. 1992. “What Did Kumārila Bhaṭṭa Mean by svataḥ prāmāṇya?” Journal of the American Oriental Society 112 (2): 204–21.

16

Speaking of the Individual: Prakāśātman’s Akhaṇḍārthavāda and the Beginnings of a Theory of Language in Classical Advaita-Vedānta1 Hugo David

1 Introduction Vedānta (lit. “the end of the Veda”) is best known as one of the many “views” (darśana) allegedly structuring the field of classical Indian philosophy, and as one of the most successful religious-philosophical movements originating in medieval India. Spreading its roots deep into the latest strata of Vedic and immediately post-Vedic literature—the Upaniṣads—Vedānta extends its influence far beyond the first millennium well into the second, giving most Hindu religious currents of that period their common language before becoming, in the nineteenth century, a major influence for modern Hindu reformist movements assembled under the loose banner of “Neo-Vedānta.” Yet, our first testimonies of Vedānta as an independent textual tradition, dating back as early as the sixth century CE,2 normally prefer to define it as a mere “exegesis of the sentences of the Vedānta” (vedāntavākyamīmāṃsā)3 and its adherents as just interpreters of the Vedic text. The word “Vedānta” itself is not originally the name of a “school,” as it will become only much later, but of a category of Vedic statements considered to be functionally distinct from the two basic textual categories of classical Vedic exegesis (Mīmāṃsā): “formulas” (mantra) and “explanations” (brāhmaṇa), the latter encompassing both ritual prescriptions and additional comments (mythology, semantic analysis, etc.).4 The mere analytical acknowledgement of “sentences of the Vedānta” (vedānta or vedāntavākya) as a separate category—a comparatively

314

Hugo David

late event in the history of Indian hermeneutics—5suffices to define its proponents as a dissident branch of Mīmāṃsā rather than a specialized subbranch, a “Latter Exegesis” (Uttara-Mīmāṃsā) as it will be called later on to distinguish it from Jaimini’s “Former Exegesis” (Pūrva-Mīmāṃsā). This “new” exegesis isolates the Upaniṣads from the rest of the Vedic corpus, introduces the idea of an Upaniṣadic canon by proclaiming their consistency (samanvaya, lit. “convergence”), and practically inaugurates the genre of Brahmanical scriptural commentary. More importantly, it claims the theoretical, pedagogical, and soteriological value of Vedāntic statements—and thereby of the Veda as a whole—whereas classical Mīmāṃsā denied it, all scriptural statements being, according to its dominant view, either merely evocative or prescriptive of some ritual action. Understanding Vedānta, therefore, requires that we take seriously its fundamental claim to represent, beyond the diversity of philosophical “views,” the very thought of the Veda. Given this central preoccupation with the interpretation of Vedic texts, one would certainly expect Vedānta to develop a keen interest in questions of language, especially those related to sentence analysis, so prominent in classical Vedic exegesis. Such an expectation, however, is very unequally fulfilled by the Vedāntic tradition. Śaṅkara (second half of the eighth century), often regarded as the founder of the school’s dominant trend before the eleventh century— known as “Vedāntic nondualism” (Advaita-Vedānta) or “pure nondualism” (Kevala-Advaita)—devotes a number of pages to the analysis of particular scriptural statements, but topically deals with questions of language only in a few passages of his commentaries (Bhāṣya) on the Brahmasūtras and Upaniṣads, most famously in the “Chapter on Deities” (devatādhikaraṇa) of his Brahmasūtrabhāṣya, 1.3.26–33. Even there, far from being the great innovator he was sometimes claimed to be,6 Śaṅkara displays a rather conservative attitude by borrowing most of his arguments from the older Mīmāṃsābhāṣya of Śabara (or Śabarasvāmin, fifth century), thus showing surprisingly little awareness of the intense debates that were raging on linguistic issues in Brahmanical circles around his time. This state of affairs, which can be explained in more than one way, often led scholars of Indian philosophy to think that Vedāntins do not have a specific theory of language and that they merely reproduce, whenever needed, positions of earlier specialists of Vedic exegesis (Mīmāṃsaka), especially those of Kumārila. In this, they could—and actually often did—claim support from the Advaitic tradition itself, expressing a similar view in the form of the oftquoted slogan: “As for the world of experience, the view of [Kumārila] Bhaṭṭa [is preferred]” (vyavahāre bhaṭṭanayaḥ),7 claiming that Vedānta differs from

Speaking of the Individual

315

Kumārila with respect to the ultimate, nondual reality, but for worldly purposes simply conforms to his views. I shall argue here that this opinion represents, at best, just one part of the picture and results from the consideration of a very limited number of textual sources. A more thorough examination of the post-Śaṅkara Advaita tradition—often called, somewhat improperly, “late” Advaita—reveals interesting discontinuities and differences in the way Vedāntins dealt with language and the epistemology of verbal knowledge. Such discontinuities and differences need to be explained from a historical, philosophical as well as a hermeneutic point of view. In fact, if philosophy of language—including wordsemantics, sentence-semantics and “regional” semantics such as the analysis of nonassertive statements—could still be considered a relatively minor topic for the first generation of Śaṅkaran commentators in the eighth and ninth centuries (Sureśvara, Padmapāda, etc.), this is no longer the case in the tenth. From that time onwards, and at least up to the fourteenth century, the epistemology of verbal knowledge (śābdabodha) becomes a major preoccupation for Advaitins, as testified to by large chapters of their works devoted to questions of general semantics, and even by entire treatises dealing with the most technical aspects of linguistic theory. This is notably the case of the Śābdanirṇaya (“An Enquiry into Verbal Knowledge”) by Prakāśātman (950–1000), the celebrated author of a commentary called Vivaraṇa on Padmapāda’s Pañcapādikā and the initiator of the eponymous Vivaraṇa-line of interpretation (prasthāna) of Śaṅkara’s Brahmasūtrabhāṣya.8 The aim of this chapter is to propose an explanation for this late and somewhat intriguing entrance of Vedāntins into the Indian arena of debates on verbal knowledge, as well as to outline the role played by Prakāśātman in that process. My main thesis is that the elaboration of a specifically “Vedāntin” theory of verbal knowledge is best understood as a complement to two closely related Advaitic views, both finding their achievement in the tenth century: the idea of “immediate verbal knowledge” (śābdāparokṣajñāna), and that of sentences referring to an “undivided object” (akhaṇḍārtha).9 Though belonging to different areas of philosophical enquiry—the epistemology of perception and verbal knowledge, on the one hand and scriptural hermeneutics, on the other—and forming partly independent conceptual and historical nexuses, these two theses entertain close relations and are to a certain extent the two sides of the same coin.10 I shall argue that it is the new articulation of these two views in the works of Prakāśātman that called for a new theory of expression and verbal knowledge, irreducible to any of the options then available in Mīmāṃsā

316

Hugo David

or other contemporary systems. Starting in Section 2 from Prakāśātman’s idea of scriptural knowledge as a form of cognition having the immediacy of perception while remaining thoroughly verbal, we will, in Section 3, move toward semantics and the several innovations brought by Prakāśātman to a theory of the sentence he inherits from Śālikanātha, before finally dealing in Section 4 with their usage in turning Padmapāda’s understanding of scriptural convergence (samanvaya) into a full-fledged theory of the “undivided object”— the unqualified Brahman—as the specific import of Upaniṣadic statements.

2  Prakāśātman in the History of Indian Linguistic Thought—the Concept of “Immediate Verbal Cognition” The last centuries of the first millennium are a period of extraordinary flourishing for Indian philosophy of language, as amply testified by other contributions in this volume. This development of linguistic thought found its most appropriate means of expression in the form of independent treatises or “monographs” (prakaraṇa) often composed in a typical “mixed” (miśraka) style, in which Sanskrit prose and verse alternate. A prototype of such texts, particularly numerous in the two schools of Vedic Exegesis (Pūrva- and Uttara-Mīmāṃsā) in that period, are the works of the Vedāntin Maṇḍana Miśra (660–720?), who favors this form above any other, and of the Prābhākara-Mīmāṃsaka Śālikanātha (around 900?), whose monographs generally assembled under the title Prakaraṇapañcikā (“A Commentary in the Form of Monographs”) were subject to multiple imitations and rewritings in the following centuries. Prakāśātman’s Śābdanirṇaya, the first Vedāntic treatise entirely devoted to linguistic theory, belongs to this class of “mixed” independent treatises. As such, it appeared to many of its medieval readers as a close imitation of the Vākyārthamātṛkā (“Basic Facts on the Object of the Sentence”), Śālikanātha’s influential monograph on sentence-semantics.11 The very organization of the Śābdanirṇaya in two large sections, one devoted to general semantics, the other to a theory of action and injunction, recalls Śālikanātha’s classical treatment of Prabhākara’s views on the sentence (vākya) and injunction (vidhi). Thus, the Śābdanirṇaya could easily be read as the counterpart on the Prābhākara side (though with strong Vedāntic leanings) of Vācaspati Miśra’s Tattvabindu, also probably composed in the latter half of the tenth century and unconditionally supportive of Kumārila’s positions on sentence-semantics.12 But the opening

Speaking of the Individual

317

verse of Prakāśātman’s treatise, in which he sets the purpose of his endeavor, suggests that he might have had wider ambitions:13 Having bowed to the Brahman, we compose this treatise [to show] how its essence is seen (īkṣyate) through the verbal means of knowledge (śābdān mānāt), for the clarification of our own thought.

By stating that Brahman, the Absolute that is the topic of Vedāntic enquiry, is “seen” (īkṣ-) by means of speech in the form of the Veda, Prakāśātman discreetly alludes to one of his most characteristic theses, discussed later in the same treatise (v. 71 and auto-commentary) and again in his Vivaraṇa on Padmapāda’s Pañcapādikā.14 According to Prakāśātman, one should take the claim that Brahman is seen through the Sacred Text literally, in other words that certain statements—mainly, though perhaps not exclusively, Upaniṣadic statements like “Brahman is awareness, bliss” (vijñānam ānandaṃ brahma) or “I am Brahman” (ahaṃ brahmāsmi)—15are capable of producing, besides belief or knowledge (jñāna) in general, “immediate knowledge” (aparokṣajñāna) of otherwise radically inaccessible entities. Immediate cognition produced from words nevertheless remains fully “verbal,” hence the idea of “verbal immediate knowledge” (śābdāparokṣajñāna).16 The very phrase “verbal immediate knowledge” probably appeared to the ears of most tenth-century Indian thinkers as a powerful philosophical oxymoron, and to many as a plain absurdity. Surely, the perception of a mango, for instance, can rightly be said to be “immediate” (aparokṣa), but not the mere understanding of a statement like “There are mangos on the side of the river,” where mangos appear in an abstract, intangible way, devoid of all the vividness and directness normally associated with the idea of “immediacy.” Most Indian philosophers in Prakāśātman’s time thus accepted a broad division of knowledge into two categories: perceptual, that is, immediate (aparokṣa), on the one hand, and indirect (parokṣa), on the other—the latter encompassing all varieties of non-perceptual cognition: verbal, but also inferential, presumptive, and so on. In practice, most of them would interchangeably use the expressions “perception” (pratyakṣa) and “immediate knowledge” (aparokṣajñāna)—also sometimes called “presentification” (sākṣātkāra). By speaking of “immediate verbal knowledge,” Prakāśātman thus attempted a major breach in Indian epistemological orthodoxy, by shifting the criteria of immediacy from the instrument (karaṇa) of knowledge to its content (viṣaya). Sure enough, no Indian thinker of the time would have found difficulty in applying immediacy to both poles of the cognitive process, namely cognition

318

Hugo David

itself and its content. In the perception of a mango, for instance, both the mango and its perception can rightly be said to be “immediate,” or perhaps more appropriately in the second case, “not beyond sight”—this being the most literal rendering of the Sanskrit word aparokṣa. But the question is: which one comes first? Is the object “in sight” because it is an object of sensitive perception, or on the contrary, is our perception of a given object immediate because that object is, through some intrinsic property, “not beyond sight”? While most of his contemporaries simply take the first option for granted, stressing the immediacy of cognition alone and explaining it in terms of contact (saṃnikarṣa) of an object with the senses, Prakāśātman contends that it is primarily the content of cognition that has immediacy, and that this character is then transferred to the corresponding cognition. A good summary of his views is found in the following lines by the famous South Indian polymath Appayya Dīkṣita, who often substantiates the lasting vitality of Prakāśātman’s ideas in the sixteenth century:17 The immediacy of cognition is the fact that cognition is conducive to the experience of an immediate object. It is constant and natural in [cognition] itself, in pleasure and other [feelings], or in the consciousness of the witness,18 which consists of light and is perpetually manifest. In [cognitive] modifications such as visual [cognition] and the like, it is conditioned by the superimposition of non-difference with consciousness, manifested for this or that [object]. The immediacy of cognitions does not consist in a universal (jāti), or in a condition (upādhi) such as being produced be the senses, etc.

In the same passage of the Parimala, Appayya Dīkṣita invokes as a criteria for the immediacy of the content of cognition “identity with consciousness” (saṃvidabheda), in perfect agreement with Prakāśātman’s characterization of “unity” or “identity with consciousness” (saṃvidaikya or saṃvidabheda) as one of the three criteria allowing one to determine the immediacy of a knowledge-content.19 This explains why feelings of, for instance, pleasure as well as cognition and, of course, the underlying conscious principle itself are “immediate” in the most primary sense, since in their case no difference with consciousness is discernible. The immediacy of the perceived object, then, is achieved only in some roundabout way by means of the temporary suspension of difference between consciousness and its object (the mango, for instance) when the mind undergoes “modification” (vṛtti), literally going out to the “place” of the object, and espouses the form of the object.20 Armed with this new, typically Advaitic concept of immediate verbal knowledge,21 Prakāśātman could tackle one of the most ancient puzzles of

Speaking of the Individual

319

Indian nondualism, as old perhaps as the nondualist school itself: if the object of Vedāntic statements, the quality-less Brahman (nirguṇabrahman), never enters into contact with the senses but is known only by scriptural means, how can one ever reach immediate knowledge of that object? And liberation (mokṣa) presupposes such an immediate knowledge, for the illusion of duality, akin to the cognition in a dream or a perceptual illusion, can only be dispelled by a cognition which is as immediate as the illusion itself, not by a mere statement. Moreover, the founder of the nondualist school explicitly mentioned the acquaintance with the inner Self as something “immediate.”22 Two solutions were proposed to this key problem in post-Śaṅkara Advaita, which mainly differ on the centrality they confer to speech in the process leading to the apprehension of Brahman. A “weaker” solution, generally associated with the name of Prakāśātman’s great contemporary Vācaspati Miśra, attributes the main part to the mind (manas), and a function of mere “preparation” (saṃskāra) to Scripture. Just as one makes a dish fit for consumption by adding spices, or just as a listener becomes apt to appreciate an orchestral piece by musical training, Scripture “prepares” the mind for immediate knowledge of Brahman, but does not produce such knowledge.23 Prakāśātman, on the other hand, rejects this attempt at salvaging epistemological orthodoxy, and rather opts for a “strong” solution in which knowledge of Brahman remains, up to the very end, integrally verbal. The role of mental activity, without disappearing altogether, is confined to the removal of possible obstacles. In this view, the verbal cognition of Brahman is in itself immediate, albeit produced by verbal statements, and the reason for that is precisely the one stated above, namely, the special nature of its content24: Brahman is the material cause of every awareness. Thus, even though the intellection is produced by the verbal means of knowledge when it takes the shape of Brahman, this [Brahman] appears with immediacy in the first place, because it is not different from that [intellection], or because it produces it.

This, of course, is not to say that scriptural statements can never produce a mediate, non-participative, knowledge of the internal Self, a fact that would be hard to deny. Rather, it is the primacy of such mediate knowledge that is contested. The function of Vedāntic Scripture is therefore not to give any new information about a given object, but to discard obstacles to the display of Brahman’s natural immediacy, and thereby to (verbally) unveil Brahman’s essential identity with consciousness. Now, what does it mean to say, as Prakāśātman does in the last quote from the Vivaraṇa, that verbal cognition “takes the shape of Brahman”

320

Hugo David

(brahmākāra-)? And were the linguistic tools developed in Brahmanical schools by Prakāśātman’s time sufficient to accommodate such a revolutionary conception of scriptural knowledge? To answer these questions, we must now turn to the core of Prakāśātman’s linguistic theory, namely his theory of the sentence (vākya), of which his theory of Vedāntic statements (examined in Section 4) is a continuation.

3  Prakāśātman’s Semantics— anvitābhidhānavāda in a Vedāntic Garb The most immediately remarkable feature of Prakāśātman’s linguistic theory as it is developed in the Śābdanirṇaya and the last two sections of the Vivaraṇa is his adoption of the “view of the expression [of objects] in connection” (anvitābhidhānavāda), the emblematic view of the Prābhākara school of (Pūrva-) Mīmāṃsā in the domain of sentence-semantics. As is well known, according to Prabhākara and his followers, words in a sentence express what is already connected or, in other words, directly convey the object of a sentence by their primary power of expression.25 The attention of Prakāśātman’s contemporaries was especially drawn to this surprising alignment of the great Vedāntin on the ideas of Prabhākara (seventh century?) and, above all, of his ninth-/tenth-century disciple Śālikanātha, as is evident from the number of quotations of Prakāśātman’s verses dealing with the topic of sentence-semantics in later Advaitic works.26 In fact, the large section of the Śābdanirṇaya where this thesis is demonstrated (vv. 20–41 and the corresponding svavṛtti) may give the superficial impression of being a mere repetition of the first part of Śālikanātha’s Vākyārthamātṛkā (see above, Section 2), frequently paraphrased by Prakāśātman.27 It seems, at first sight, that Prakāśātman accepts Śālikanātha’s theory of the sentence as far as the general theory of expression is concerned, and only departs from it while determining the nature of what exactly gets “connected” when words express their object “in connection” (anvita). An essential corollary to Śālikanātha’s theory of expression is, in fact, his idea that all sentences express an obligation (kārya, lit. “something to be done”); against this, Vedāntins maintain that connection perhaps can be with an obligation in some cases, but need not always be so, and therefore propose to replace Śālikanātha’s “view of the expression [of objects] in connection with an obligation” (kāryānvitābhidhānavāda) by another view known as the “view of the expression [of objects] in connection with others [in general]” (anyetarānvitābhidhānavāda). More interesting, however,

Speaking of the Individual

321

are the several adaptations of Śālikanātha’s basic theory that are discreetly, and yet unmistakably, introduced by Prakāśātman in the course of his exposition of anvitābhidhānavāda. Two of them are important for our purpose, and need to be outlined in some detail, namely, Prakāśātman’s conception of the cognition of “isolated” objects prior to their expression as connected entities and his theory of indirect denotation (lakṣaṇā).

3.1  The Isolated Object of Words In spite of Prabhākara and Śalikanātha’s insistence on the fact that words in a sentence always refer to their object “in connection,” classical Prābhākara theory nevertheless recognizes, before that specific operation takes place, an intermediate stage in which words refer to their objects as “isolated” (kevala) or “non-connected” (ananvita) entities.28 This preliminary stage is indispensable for the coherence of the theory, for otherwise one could not explain how the first words in a sentence would express their object as connected to something which is not yet available to the mind of the listener. Therefore, the understanding process according to Śālikanātha takes place in two steps: words convey first of all their object “in general,” regardless of their connection, and only then refer to their object as connected to others, conveyed by other words in the sentence. What, then, marks the difference with Kumārila’s idea of an a posteriori connection (anvaya) of objects expressed (abhihita) in isolation is the nature of that first cognition. For, in Śālikanātha’s view, this primary, purely lexical understanding of the object of words is not yet an operation of expression, but just a form of memory (smṛti) happening in accordance with the popular maxim of the elephant and the elephant keeper (hastihastipakanyāya): just as we remember the elephant by seeing the elephant keeper (or vice versa) because both were previously perceived in the vicinity of each other, so when we hear a word (“cow”) we automatically remember its object (the cow), and vice versa, by mere association and without any intervention of the word’s expressive power (śakti). Unexpectedly, Prakāśātman’s discussion of this key aspect of Prābhākara semantics radically departs from Śālikanātha’s view. The alternative he proposes, however, is not entirely consistent in his two main works, although both versions agree in rejecting Śālikanātha’s idea that isolated objects would be known by a mere association of ideas, without any causal involvement of the words’ power of expression. In fact, it is precisely on the role attributed to that power that Prakāśātman hesitates at the various stages of the redaction of his

322

Hugo David

work. The oldest available version may be the one we find in the Vivaraṇa.29 There, Prakāśātman agrees with Śālikanātha in saying that the cognition of isolated objects is a memory, but does not accept that it occurs automatically. For how could one otherwise ensure that what is recalled is actually the object, and not any of the various circumstances in which the word was first heard—the speaker, the persons who were present or other adventitious contextual factors? In order to produce a cognition of the object and of nothing else memory must follow, so to say, the path already traced by the process of learning the language, and conform to the expressive power as it was established when the listener first entered the linguistic community. The Śābdanirṇaya’s version (vv. 34–35 and auto-commentary) is more complex, because there Prakāśātman chooses to abandon the recourse to memory and to reinvest the exegetical concept of “re-statement” (anuvāda), familiar in classical Mīmāṃsā but endowed with a new epistemological value. The difference between memory and re-statement— both cases of “second-hand” cognition—may not be obvious at first sight, but it is fundamental in the present context: while memory takes place by the awakening of a trace (saṃskāra) left by a previous, “first-hand” cognition, a re-statement occurs through that same causal complex which gave rise to the first, “fresh” cognition. Consider someone staring at a picture on a wall, but distracted in his contemplation by a noise in the nearby street. Coming back to the picture after a while, he will have a second apprehension which is neither a memory nor a fundamentally new cognition, for after all the picture he is looking at is still the same, regardless of whether he continuously stares at it or not.30 Lexical knowledge in the context of verbal understanding is precisely of this kind: its cause being the words’ expressive power, the existence of which we learned when we first encountered the word used by elders, it is not a case of memory; but it is also nothing new, and therefore cannot qualify as “valid knowledge” stricto sensu. The process of understanding language as conceived by Prakāśātman therefore implies, before denotation of objects “in connection” can take place, a mere re-statement of isolated objects by means of the words’ very expressive power, which is distinct from memory.

3.2  Indirect Denotation A second major innovation of Prakāśātman with respect to Śālikanātha concerns the status or “indicative reference” (lakṣaṇavṛtti) or indirect denotation,31 another crucial issue in Śālikanātha’s anvitābhidhāna theory.32 It is solved in a comparable manner by the great Prābhākara philosopher: according to him,

Speaking of the Individual

323

the “expression [of objects] in connection” only concerns words whose object is direct (mukhya); figurative words, in turn, do not express any connected object but merely function as reminders of their “isolated” or “non-connected object” (kevalārtha or ananvitārtha) through the process outlined in the preceding paragraphs. Thus, in a classical example of indirect denotation like “The cradles are crying” (mañcāḥ krośanti), standing for “The babies in the cradles are crying,”33 the word meaning “cradle” (mañca) does not express its indirect object—namely, the human beings present in the cradles—in relation to the action of crying. It merely presents to the mind of the hearer the derived object—the babies—as a correlate of the action of crying expressed by the verb krośanti, meaning “are crying” (in Indian terms: it is a “reminder,” smāraka, it is not “expressive,” vācaka). The two causes for this process are the impossibility (anupapatti) for the sentence to make sense without such an indirect denotation and the relation (saṃbandha) of proximity between the cradles and the human beings that are placed in them. The verb, then, is the only word in the sentence that functions in accordance with the model of “expression in connection”; other words are used as mere reminders. This assuredly constitutes from the point of view of Śālikanātha another significant exception to Prabhākara’s principle that words always express their object as connected entities. It also has an immediate consequence, which might have been overseen by Śālikanātha himself but would become central for his Vedāntin readers, namely, that it is impossible, on such premises, to conceive how a meaningful sentence could be composed only of words referring on the mode of lakṣaṇā. For unless at least one word is used in its direct sense it is impossible, following Śālikanātha’s assumptions, that the listener ever reaches the connection of objects among themselves, in other words the particular, complex state of affairs that is the specific object of the sentence. Prakāśātman’s theory of indicative reference is obviously—and as explicitly as is allowed by his elliptic style—built as an alternative to that of Śālikanātha. As a consequence, it mainly strives to suppress this exception, and to account for “expression in connection” even in the case of figurative words:34 In case of an indicative [reference], the idea of the [mutual] association [of word-objects] (saṃsarga) should arise from the word (padāt) itself that says (-vācin) this [mutual association].

Prakāśātman’s intention might not be entirely clear from this verse, but it is made explicit in the auto-commentary on the basis of the example of the crying cradles. Against the expectations of a reader familiar with Śālikanātha’s theory outlined

324

Hugo David

above, the word (pada) intended in Prakāśātman’s verse is not the verb meaning “are crying” (krośanti), but the “figurative” word meaning “cradles” (mañca) “for,” as he explains in the svavṛtti, “it is only insofar as speech conveys it as connected to the action of crying that the object of the word “cradle” (mañca) causes the idea of human beings, [also] connected to [the action of] crying, not when it is taken in isolation.”35 The very process of translation from the cradles to the human beings lying in them would never take place if the word mañca did not already present its direct object (the cradles) as expecting a connection with the action of crying. Rather than the mere fulfillment of an expectation by the mnemonic “bringing near” (saṃnidhāpana) of an unconnected object, Prakāśātman conceives of indirect reference as a series of attempts at producing a meaningful connection, where the first attempts are subject to obstruction (virodha) through some factual or conceptual impossibility. Transfer, then, does not operate at the level of the word, as a mere shifting from an object to another, but at the level of the sentence, as the replacement of an obstructed connection by a more relevant one, the last in a series of attempts at producing a plausible interpretation of the sentence. The immediate—though unstated— consequence of this “improved” theory of indirect denotation in the context of anvitābhidhānavāda is that it makes speech directly responsible for all stages in the comprehension of a figurative statement; by doing so, it also lifts the ban on sentences composed exclusively of words referring in an indirect way, because any word (direct or figurative) becomes capable of referring by its own power to the object of the sentence. Now, how are we to understand these two amendments to Śālikanātha’s theory of the sentence? Are they just punctual innovations, aimed at removing undesired exceptions to the principle of “expression [of objects] in connection,” or is there more to them? The Vedāntic tradition itself may be of little use to answer this question, as neither Ānandabodha, the eleventh-century commentator on Prakāśātman’s Śābdanirṇaya, nor later Advaita authors referring to Prakāśātman’s views (Citsukha, Ānandapūrṇa, Pratyaksvarūpa, etc.) seem to have taken full notice of these important modifications. In the eyes of post-tenth-century Vedāntins, the author of the Vivaraṇa is essentially just one more follower of anvitābhidhānavāda, albeit from their own camp. In the last part of this chapter, however, I will argue for another possibility, namely that Prakāśātman needed this “improved” version of “expression in connection” to serve his own exegetical purpose, which is to ground in a full-fledged semantic theory his interpretation of Upaniṣadic statements as referring to an “undivided object” in the perspective of immediate verbal knowledge.

Speaking of the Individual

325

4  Description and Identification: The “Undivided Object” (akhaṇḍārtha) from Padmapāda to the Vivaraṇa 4.1  The Two Functions of a Sentence According to Padmapāda In order to understand what Vedāntins mean when they speak of sentences expressing an “undivided object” (akhaṇḍārtha),36 we must now take one step back from Prakāśātman, and turn for a moment to his great elder in the tradition of Advaitic commentaries, namely Padmapāda (late eighth century?). Padmapāda is mainly known as the author of an extensive commentary called Pañcapādikā on the initial section of Śaṅkara’s Brahmasūtrabhāṣya, 1.1.1–4; Prakāśātman’s Vivaraṇa is only one of the many sub-commentaries—some of them still unpublished—written throughout the late medieval period on this fundamental Advaitic work. Generally considered to be one of Śaṅkara’s four direct disciples, Padmapāda is, without any doubt, one of the most original thinkers in the early Śaṅkaran tradition.37 The most interesting remarks for our purpose are found in his commentary on the fourth of Bādarāyaṇa’s “Aphorisms on Brahman” (Brahmasūtra): “Rather, it [i.e., Brahman] [is known] through their (?) convergence” (tat tu samanvayāt). This crucial statement, which lays down the very principle of Vedāntic exegesis, gave rise to surprisingly diverse interpretations among early Advaitins, the most striking divergence being probably the one opposing Śaṅkara to his alleged pupil.38 For Śaṅkara, the word samanvaya in this aphorism essentially means “convergence” of sentences of various Upaniṣads on a single object, the unqualified Brahman.39 For Padmapāda, on the other hand, the term should be interpreted in terms of a relation of words within a sentence: words, when they “converge,” become capable of conveying Brahman. But then, what should we understand by “convergence”? Padmapāda’s answer is as follows:40 “Convergence” (samanvaya) [essentially means] “integrated connection” (samyag anvayaḥ). But what is it that makes the connection “integrated”? [We answer:] the connection of words whose objects are not mutually delimited, that do not expect anything else, [referring] merely to the object of a nominal base, unspecified (avyatirikta) and simple (ekarasa).

These are truly groundbreaking lines, that would still be recognized by late Vedāntins like Madhusūdana Sarasvatī (Bengal, sixteenth century) as the most genuine expression of the Vedāntic akhaṇḍārthapakṣa.41 In this passage, Padmapāda considers “convergence” between words in a sentence from two

326

Hugo David

points of view: positive and negative. Negatively, words in a sentence are said to “converge” if their objects differ from an action, a factor of action or their association, and do not have any expectancy (ākāṅkṣā) toward each other. This last element is important, as it exactly contradicts the classical Mīmāṃsaka theory of the sentence, which regards mutual expectancy between the parts as one of the main criteria allowing one to determine the unity of the sentence as a single, organic whole.42 A simple sentence like “Devadatta walks” (devadatto gacchati), for instance, cannot satisfy Padmapāda’s definition of “convergence” as it mainly describes the relation between an agent (Devadatta) and the action of walking, each of these two elements generating in the listener the “expectation” of the other. But “convergence” does not only rule out sentences describing the relation of an action to its factors; it also excludes other types of relation, such as that of simple qualification (viśeṣyaviśeṣaṇabhāva). A sentence like “The lotus [is] blue” (nīlam utpalam), for example, will not satisfy this condition, nor will “definitional” sentences of Scripture like “Brahman [is] truth, awareness, infinite” (satyaṃ jñānam anantaṃ brahma)43 as they are interpreted in earlier Vedāntic tradition, beginning with Śaṅkara’s commentaries on the ancient Upaniṣads.44 What remains, once every relation has been excluded, is the bare, unrelated thing referred to by Padmapāda as “merely the object of the nominal base,” a characterization that alone constitutes a positive description of the referent of that peculiar kind of sentences. Now, all this may remain quite esoteric unless we take into account the properly linguistic dimension of Padmapāda’s reflection in the same chapter of the Pañcapādikā. Another fundamental intuition of Padmapāda is, in fact, that one cannot understand a sentence without also considering the dialogic situation in which it is embedded, an aspect of verbal understanding rarely taken into account in Mīmāṃsaka analyses. Padmapāda probably adapts from Maṇḍana Miśra’s Brahmasiddhi his stock example to illustrate that point: “The moon is the most intense light” (prakṛṣṭaprakāśaś candraḥ).45 How does this sentence function? To answer this question, says Padmapāda, one must not only understand the sentence itself, its lexical components and its syntax, but also consider the set of possible questions to which it constitutes a plausible answer (prativacana); for instance, “How is the moon?,” but also something like “Which one of those stars is the moon?.” Our interpretation of the sentence will completely change according to which of these two questions it is expected to answer. In the first case, the speaker is expected to describe salient features of the moon at a given moment, answering a question like “What is the moon like tonight?.” In the second case, such a description would be vain, as it would

Speaking of the Individual

327

apply to something that is not known to the listener; what is expected from the speaker is rather that he achieves precisely that basic identification of the object prior to any attempt at describing it. Put differently, the interpretation of the very same sentence will fundamentally change according to whether we already know what is the moon or whether we do not know what it is.46 While the second case may occasionally apply to the moon (or, perhaps more likely, to less well-known asterisms), it will most regularly be the case of an object like Brahman, reputed unknowable by any means except through Scripture. From different dialogic situations, two entirely distinct functions of language emerge, corresponding to two radically opposed ways for a sentence to refer to its object: a function of description, on the one hand, in which qualification is central (regardless of whether an action is involved or not) and a function of mere identification, on the other, where language assumes the role of a simple designation, thus becoming somewhat similar to the gesture of physically pointing out a perceivable object in front of us. No qualification is expected in this second case; rather, all words taken together independently contribute in designating the right object by progressively focusing the attention of the listener on a more and more restricted set of things. The seeming “definitions” of the Upaniṣads, like the typical sentence quoted above, “Brahman is truth, awareness, infinite” (satyaṃ jñānam anantaṃ brahma), function precisely in that way: words meaning “truth” (satyaṃ), “awareness” (jñānam), and “infinite” (anantam) are not qualifications of Brahman in any rigorous sense; they are attempts at reducing a formerly open scope to a single point. Together, they contribute to identify and display for the listener a unique, unrelated entity denoted by the formerly “empty” nominal base “Brahman.” In this specific sense, following the provocative word of Prakāśātman in the Vivaraṇa, we can say of such sentences that for them “the object of the word [i.e., “Brahman”] is itself the object of the sentence” (vākyārtha eva padārthaḥ).47

4.2  Prakāśātman on the Undivided Object Padmapāda, however, did not draw all the consequences of these intuitions for linguistic theory, and indeed left many questions unanswered: by which epistemological process do words lead to the knowledge of an “undivided” object? Do they “express” it in any ordinary sense? And if they do, what is the mode of their reference, is it direct or indirect? After Padmapāda, these and similar questions were to be debated for many centuries in Advaitic circles, not restricted to the “Vivaraṇa-” tradition of commentators directly claiming

328

Hugo David

his heritage. Prakāśātman—for our concerns, the earliest commentator on Padmapāda’s work—may, however, be the first philosopher to envisage a systematic answer to this set of questions in the form of an articulate semantic theory. The question of the mode of referring of words of akhaṇḍa statements seems to have had special prominence for him; it is dealt with in the following lines of the Vivaraṇa, discussing Padmapāda’s “moonlight” example:48 [In the sentence “The moon is the most intense light,”] the word “light” refers to a specific individual, by indication (lakṣaṇā), through the expression of the universal [“being-light”]; the word “most intense” refers to a specific light, [also] by indication, through the expression of its intensity.49 This being the case, since neither a quality nor a universal are to be expressed by the word “moon,” both are discarded by means of an indication [operating by] abandoning [the direct object] (jahallakṣaṇā), and what is conveyed as being expressed by the word “moon” is the specific light in which both [the quality and the universal] inhere; this is how we establish that the [three] words “most intense,” “light” and “moon” [together] refer to a single object (ekārthavṛttitā).

What is immediately striking in this description is the centrality of “indicative reference” (lakṣaṇā[vṛtti]). In fact, all words of the sentence function by indication, albeit for different reasons: the reference to an individual (vyakti) via the expression of a quality or a universal, or the synthetic “discarding” of both the quality and the universal in order for the three words to refer to the single, nonqualified individual in which these properties inhere. All three words of the sentence “converge” in that they co-operate in bringing to the mind of the listener a single, non-differentiated individual (the moon), so that he becomes capable of apprehending it immediately through his own eyes: “This is the moon!” The individual moon, indicated by the word “moon” (candra), therefore, constitutes in itself the object of the sentence. For, totally unconnected as it may be to the direct object of other words (the universal “being-light” and the quality of “intensity”), it could never be referred to by the word “moon” taken in isolation, but only by this word when taken as part of the sentence, as a result of the process of progressive restriction in scope operated by the two other words, “light” and “most intense.” A similar procedure is invoked for Upaniṣadic sentences as well, though Prakāśātman introduces in their case some further refinement in the theorization of the various ways in which Upaniṣadic words refer to their object:50 The [Upaniṣadic] words “knowledge” and “bliss” [directly] convey a specific bliss. The words “one,” “truth,” “infinite” refer to that [specific bliss] by means of

Speaking of the Individual

329

an indication (lakṣaṇā), through the expression of the absence of [respectively] difference, falsity and duality. Words like “omniscient,” “omnipotent,” etc. refer to that [same specific bliss] by means of a condition (upādhi), which is the phenomenal proliferation, indescribable [in terms of being or non-being]. […] Thus, all words referring to the cause of the world [together] convey Brahman, perfectly simple, by means of direct [reference], of indication, or of a condition.”

In spite of the slight complication introduced by Prakāśātman in the case of words like “bliss” (ānanda), allegedly referring to Brahman without any indicative operation, or “omniscient” and “omnipotent,” referring to that blissful being only “under the condition” that the world exists, it should be clear by now that the central mode of reference by which words in an akhaṇḍa sentence designate the individual is indicative. By directly attributing to the power of speech the cognition of the object of the sentence, Śālikanātha’s sentence theory (see above, Section 3) was clearly preferable to Kumārila’s idea of a mental a posteriori connection of objects expressed independently. But the way Śālikanātha’s theory accounted for indirect language was also too restricted to marginal cases of figurative usages to fully explain the kind of global co-operation of words that is at stake in definitional (i.e., akhaṇḍa-, “undivided”) sentences, characterized by the “[simultaneous] co-reference” (ekārthavṛttitā) of all words through various modes of indirect denotation, ending up in the immediate apprehension of a formerly unknown individual. In one of the last verses of the Śābdanirṇaya, Prakāśātman confirms the link we try to establish here between his hypothesis in the field of sentencesemantics—that is, “revised” anvitābhidhānavāda—and the exegetical theory of Upaniṣadic statements about Brahman he inherits from Padmapāda:51 Thus, in our view, [Upaniṣadic] words like “truth,” etc. convey, by the same power by which they express [their] object in connection, a realised object [i.e., the quality-less Brahman], which cannot be known by any other means.

The phrase “by the same power by which they express [their] object in connection” (saṃsṛṣṭārthoktiśaktitaḥ) is an unmistakable reference to the anvitābhidhāna theory advocated throughout Prakāśātman’s treatise. The subordination of linguistic theory to scriptural exegesis, then, is clearly apparent: from Śālikanātha, Prakāśātman borrows the idea of a direct efficiency of speech at all stages of the linguistic process; but he also needs to push this principle one step further, by excluding from that process all elements implying extralinguistic mental operations: memory, association, connection

330

Hugo David

between known objects, and so on. Semantic theory therefore constitutes, for Prakāśātman, the missing link between a theory of immediate verbal knowledge, in which direct apprehension of the Absolute happens by means of speech alone, and an exegesis of “undivided” statements, explaining the positive designation of an unqualified entity.

5 Conclusion It should now be clear that the three thematic lines followed in this chapter contribute together to answer, from different points of view, the same question: How is it possible for language, in the form of Scripture, to produce knowledge about an object that cannot be described in sentences without losing its essence? As we have shown, the Advaitins’ effort to answer this classical question of negative theology and gnoseology motivated—not to say “determined”—their engagement in the linguistic debates of their time. In this sense, one is historically justified to speak of a “Vedāntic” theory of language, and not merely of linguistic theories occasionally developed among Vedāntins. This is not to say, however, that all Vedāntins saw things in this way, and articulated a semantic reflection with larger epistemological and hermeneutical issues. Many authors after the tenth century freely invested the newly opened field and positioned themselves quite independently of all other considerations. “Even if one was to admit,” says the sixteenth-century Advaitin Nṛsiṃhāśrama, “that the power [of words] concerns the mere word-objects,”—a clear reference to Kumārila’s position on sentence-semantics—“there would be no harm for us [Advaitins], who hold the view of the undivided object (akhaṇḍārthavādin); but since the use of a word always implies the desire to know another particular, since this desire is always preceded by cognition of connection in general, and since that knowledge is necessarily produced by speech, [we conclude that] speech expresses [its object] in connection.”52 Writing in South India close to the time of Appayya Dīkṣita, Nṛsiṃhāśrama not only witnesses the remarkable endurance of semantic hypotheses introduced into Vedānta by Prakāśātman some six centuries before him; he also attests this newly conquered autonomy of the linguistic field in Advaita after the tenth century, an autonomy that would be carried out however at the expense of its exegetical roots.

Speaking of the Individual

331

Notes 1 Preliminary drafts of this chapter were presented on various occasions in Leiden, Paris (Société Asiatique), Cambridge and finally Vienna. I thank the auditors of these presentations for their remarks and criticism. I also wish to express my gratitude to Prof. K. Srinivasan (Kuppuswami Sastri Research Institute, Chennai) and S. L. P. Anjaneya Sarma (EFEO, Pondicherry), with whom I had the privilege to read and discuss most of the texts that form the basis of this study during the last ten years. I am, of course, alone responsible for all hypotheses advanced in the course of this chapter. 2 The mention of vedāntavādins (upholders of the Vedānta(s)) by the Buddhist author Bhavya (also known as Bhāviveka or Bhāvaviveka, sixth century) in the Vedāntaviniścaya chapter of his Madhyamakahṛdayakārikās, v. 8.1, is generally considered to be the earliest reference to Vedāntins as a separate school of thought. 3 Śaṅkara, Brahmasūtrabhāṣya, 1.1.1, p. 83, l. 2. 4 See Mīmāṃsāsūtra, 2.1.30–33. 5 See David 2017b, forthcoming(a). 6 See Biardeau 1957, a detailed critique of which is found in Bhattacharya 2001. 7 For an early occurrence, ca. twelfth/thirteenth century, see Citsukha’s Tattvapradīpikā (Citsukhī), p. 155, l. 12–13. 8 Another, somewhat better-known example, is Sarvajñātman’s Saṃkṣepaśarīraka (“Śārīraka [i.e., Vedānta] in a Nutshell”), “an extensive and stunningly developed theory of the (Vedic) sentence” according to its first translator (Vetter 1972, 5). Sometimes considered a disciple of Sureśvara—which would make him a predecessor, or at best a younger contemporary, of Prakāśātman—Sarvajñātman is more likely to postdate the author of the Vivaraṇa by at least one or two centuries. For an extensive, though inconclusive, discussion of Sarvajñātman’s date, see David, forthcoming(b). A close study of the arguments of the Saṃkṣepaśārīraka displays a considerable debt toward Prakāśātman in the field of sentence-semantics, where both authors advocate the same “Vedāntic” version of Prabhākara’s “doctrine of the expression of [objects] in connection” (anvitābhidhānavāda, see below, Section 3, and Chapter 9 in this volume). This remarkable proximity with Prābhākara views on the sentence, which sharply contrast with the idea of the Vedāntins’ exclusive reliance on Kumārila, is shared by a number of later Advaita thinkers (though by no means by all of them), significantly by Ānandabodha (eleventh century), the author of the only commentary, named Nyāyadīpikā, on Prakāśātman’s Śābdanirṇaya, and by Ānandapūrṇa (South-West India, fourteenth century), who largely draws on both Prakāśatman and Ānandabodha while composing his little-studied Nyāyacandrikā. 9 Despite superficial resemblance in designation, the Vedāntic theory of akhaṇḍārtha is totally unrelated to Bhartṛhari’s view of the sentence (vākya) and its object (artha)

332

Hugo David

as “indivisible” (akhaṇḍa) entities (see Chapter 4 in this volume). To the best of my knowledge, the very expression akhaṇḍārtha is not found in Vedāntic sources before Ānandabodha, who repeatedly uses it in his Nyāyadīpikā, for example, p. 387, l. 4 and (presumably later) Nyāyamakaranda, for example, p. 259, l. 4. 10 The thesis of akhaṇḍārtha finds its roots in Padmapāda’s analysis of the concept of “convergence” (samanvaya) in Brahmasūtra, 1.1.4, as we shall see. The idea of śābdāparokṣajñāna, in turn, is not found, to the best of my knowledge, before Prakāśātman, and remained associated with his name in later times. Similarly, while most Advaitins after the tenth century subscribe to the idea of an “undivided object” as the specific import of Upaniṣadic statements, the question whether verbal knowledge can be immediate or not becomes after the tenth century a major point of disagreement between the two lines of interpretation (prasthāna) of Śaṅkara’s Bhāṣya, the so-called Bhāmatī- and Vivaraṇa- lines. Still, I hold for philosophically significant that both theses made their apparition simultaneously, as two aspects of Prakāśātman’s “system” of Vedānta (on this expression, see Cammann 1965). 11 On Śālikanātha’s Vākyārthamātṛkā, see Wicher 1987 (a full translation into German, unfortunately still unpublished), Pandurangi 2004, 348–428, and in this volume Chapters 13 and 14. 12 Vācaspati’s Tattvabindu is one of the rare treatises belonging to that tradition that have been translated into a Western language. See Biardeau 1979. 13 Śābdanirṇaya, v. 4: natvā brahma yathā śābdān mānāt tattattvam īkṣyate / tathā vayaṃ nibadhnīmaḥ svabuddhipariśuddhaye //. Passages from the Śābdanirṇaya are quoted here following my new critical edition of this text (David, forthcoming(b)), which often differs from its editio princeps by T. Gaṇapati Śāstrī (Trivandrum, 1917). Since this edition has not yet come out, passages are referred to by verse (rather than page) numbers. 14 See Prakāśātman, Pañcapādikāvivaraṇa, p. 403, l. 2–p. 410, l. 4 (end of the first varṇaka). English translation in Gupta 2011, 480–83. 15 Respectively, Bṛhadāraṇyakopaniṣad, 3.9.28 and 1.4.10. 16 Surprisingly little attention has been devoted to this central point in the system of Vedānta by historians of the school. See, however, the brief explanations in Suryanarayana Sastri and Raja 1992, 266–68; Cammann 1965, 159–61; Alston 1997, 832–33; Ghosh 2006. 17 Appayya Dīkṣita (Vedāntakalpataru-)Parimala, 1.1.1, p. 55, l. 21–22: jñānasyāparokṣyam aparokṣārthavyavahārānukūlajñānatvam. tat svasya sukhādeś ca prakāśarūpe nityābhivyaktasākṣicaitanye vānugataṃ svābhāvikam, cākṣuṣādivṛttiṣu tatta​dabhi​vyakt​acait​anyāb​hedād​hyāso​pādhi​kam. na tu jātirūpam indriyajanyatvādyupādhirūpaṃ vā jñānānām āparokṣyam. Appayya Dīkṣita himself rejects later on in the Parimala the possibility of immediate verbal

Speaking of the Individual

18

19

20

21

22 23 24

333

cognition; in another Vedāntic work of his, however, the Siddhāntaleśasaṃgraha, Appayya Dīkṣita accepts Prakāśātman’s idea. This major discrepancy in Appayya’s work still awaits historical clarification, but need not concern us here. The expression sākṣicaitanya (“consciousness of the witness”) designates, in a typically Vedāntic way, the conscious principle underlying cognition, which is at least empirically different from the various cognitive events, here simply referred to by the word jñāna (cognition). See Prakāśātman, Śābdanirṇaya, v. 71 and the auto-commentary (… saṃvidaikyād …), Pañcapādikāvivaraṇa, p. 403, l. 2 (… saṃvidabhedād …). The two remaining criteria are not the same in Prakāśātman’s two main works: in the Vivaraṇa they are “the fact of producing the awareness of itself” (svasaṃvijjanakatva) and “the fact of being in contact with the senses” (indriyasaṃprayuktatva), the latter clearly a remnant of earlier theories or a concession to mainstream conceptions of immediacy; the Śābdanirṇaya complicates the picture by introducing an intermediate stage—the “intellection” (saṃvedana)—which is the effect of cognition (jñāna) itself. The immediacy of the intellection is then determined, according to Prakāśātman, by “the fact that it is born from the content” (viṣayotpannatva) or by “the suppression of nescience” (ajñānahāni). Later authors, like Appayya Dīkṣita, prefer to avoid this multiplicity of criteria and to abandon the concept of “intellection” altogether, thereby reducing the set of criteria from three to one, namely “identity with consciousness.” This explanation of the cognitive process in terms of Sāṃkhya psychology— rather than in terms of “contact” (saṃnikarṣa) of the object with the senses—is remarkable, but need not concern us here. Mainly, it allows Prakāśātman to consider sensitive perception a very particular case of immediacy compared to the apprehension of other “naturally” immediate entities like cognition or feelings, accessible without the mediation of the senses. A handy summary of Prakāśātman’s unique and in many ways surprisingly archaic theory of perception is found in the first chapter, on perception, of the Vedāntaparibhāṣā, Dharmarāja Adhvarin’s popular Vedāntic “textbook” (§ 17–22, pp. 12–14). See also Cammann 1965, 132–37. As far as my knowledge goes, no philosopher outside of Advaita favored the idea of immediate verbal knowledge. In any event, all rival schools of Vedānta—first among them the schools of Rāmānuja and Madhva—unanimously rejected it. Śaṅkara, Brahmasūtrabhāṣya, adhyāsa- pp. 38–39: […] aparokṣatvāc ca pratyagātmaprasiddheḥ; “for the acquaintance with the inner Self is immediate.” See Vācaspati Miśra, Bhāmatī, 1.1.1, p. 54, l. 18–p. 58, l. 4. Prakāśātman, Pañcapādikāvivaraṇa, end of first varṇaka, p. 406, l. 1–p. 407, l. 2: brahmaṇa eva sarvasaṃvidupādānatvāt brahmākāraśabdapramāṇajanyasaṃvedane ’pi tadabhinnatayā vā tajjanakatayā vā brahmāpi prathamam evāparok­ṣatayāvabhāsate.

334

Hugo David

25 The main components of the famous controversy between partisans of Kumārila and Prabhākara on the signification of sentences have been exposed many times, and need not be repeated here. See the classical summaries by Kunjunni Raja 1969, 191–227 and Gerschheimer 1996, 88–91, as well as M. Biardeau’s translation of Vācaspati’s Tattvabindu (Biardeau 1979), mainly concerned with the debate between the proponents of anvitābhidhāna and abhihitānvaya. 26 See for instance Pratyaksvarūpa’s Nayanaprasādinī (a fourteenth-century commentary on Citsukha’s Tattvapradīpikā), where vv. 33 and 63 of the Śābdanirṇaya are quoted (p. 147, l. 28–30 and l. 33) along with extracts from Śālikanātha’s Vākyārthamātṛkā (vv. 1.12 and 1.15) and its svavṛtti. 27 Although I could not find any literal quote from Śalikanātha anywhere in Prakāśātman’s work, his paraphrases are in general sufficiently close to the letter of Śālikanātha’s treatises to show without any possible doubt that he was familiar with several of his works. 28 For all aspects of Śālikanātha’s theory developed in this paragraph, see his Vākyārthamātṛkā, vv. 1.12–15 with the auto-commentary. 29 Prakāśātman, Pañcapādikāvivaraṇa, p. 138, l. 3–p. 140, l.4. This assumption is based on my hypothesis that vv. 34–35 of Prakāśātman’s Śābdanirṇaya are not addressed to Śālikanātha, but are rather an auto-critique. See the explanations given in the notes to my translation of these verses (David, forthcoming). 30 This distinction between re-statement (anuvāda) and valid knowledge (pramāṇa) is not universally accepted, even in Mīmāṃsā. Both are treated as one, for instance, in the Nayaviveka, 1.1.2–p. 53, l. 4–6 by the eleventh-century PrābhākaraMīmāṃsaka Bhavanātha. We still know too little, however, on the history of the notion of anuvāda to affirm with any certitude that this difference of opinion is due to a divergence of various schools on this topic, and not just to individual choices. The distinction is, in any case, quite essential to Prakāśātman’s theorization of anvitābhidhānavāda. 31 On lakṣaṇā(vṛtti) and its numerous possible English renderings, see Part Four of the present volume. 32 On Śālikanātha’s theory of indicative reference, see the important passage of the Vākyārthamātṛkā, svavṛtti on 1.12 (p. 405, l. 6–14) translated in Gerschheimer 1996, 417; see also David, forthcoming(b); notes to the translation of v. 39 of Prakāśātman’s Śābdanirṇaya. 33 This is just one of the many possible interpretations of this classical example, that goes back at least to Vātsyāyana/Pakṣilasvāmin’s Nyāyabhāṣya (ad Nyāyasūtra 2.2.62). On the history of the example, as well as on its various interpretations, see Gerschheimer 1996, 451–52. All of them agree at least on the fact that what is “indicated” (lakṣya) by the word mañca (“cradle”) are the human beings (puruṣa) placed in the cradles (mañcastha).

Speaking of the Individual

335

34 Prakāśātman, Śābdanirṇaya v. 39ab: lakṣaṇāyāṃ ca tadvācipadāt saṃsargadhīr bhavet. 35 Prakāśātman, Śābdanirṇaya v. 39 (svavṛtti): krośanasaṃsargitayā śabdāt pratipanno hi mañcapadārthaḥ puruṣe krośanasaṃsargiṇi buddhihetuḥ, na   kevala iti. 36 On the first occurrences of the term akhaṇḍārtha, used here as a concession to its extensive usage in the later Vedāntic literature, see above, note 9. The only study specifically devoted so far to this central aspect of Advaita doctrine is a brief and fairly old article by S.S. Suryanarayana Sastri (Suryanarayana Sastri 1961), that still deserves to be mentioned. 37 For general information on Padmapāda, his work and his relation to Śaṅkara, see the classical study by Hacker (1950, especially pp. 21–27). The Pañcapādikā—the signification of whose title is still unclear—has been translated into English in Venkataramiah 1948. 38 Another interesting divergence in the interpretation of this sūtra is between Śaṅkara and Maṇḍana Miśra (presumably the former’s older contemporary), who interprets “convergence” as “the association of word-objects, [that is] their assignment [to the main word-object]” (padārthānāṃ saṃsargo viniyogas— Brahmasiddhi, 2.182, svavṛtti p. 155, l. 10). By his focus on “the association of word-objects,” Maṇḍana anticipates Padmapāda’s interpretation, without however bringing in the technical innovations that would characterize the latter’s understanding of this key-aphorism. 39 Śaṅkara, Brahmasūtrabhāṣya, 1.1.4: sarveṣu hi vedānteṣu vākyāni tātparyeṇaitasyārthasya pratipādakatvena samanugatāni; “For in all [passages of the] Vedānta, sentences pervasively aim (tātparyeṇa samanugatāni) at conveying that object [i.e., Brahman].” 40 Padmapāda, Pañcapādikā, p. 322, l. 7–p. 323, l. 2: samyag anvayaḥ samanvayaḥ. atha keyaṃ samyaktānvayasya? padānāṃ parasparānavacchinnārthānām ananyākāṅkṣāṇām avyatiriktaikarasaprātipadikārthamātrānvayaḥ. 41 See Madhusūdana Sarasvatī, Advaitasiddhi, p. 663, l. 5–6. 42 See Mīmāṃsāsūtra, 2.1.46, arthaikatvād ekaṃ vākyaṃ sākāṅkṣaṃ ced vibhāge syāt; on the early history of the exegetical theory of the sentence, see David 2017a, 34–39. 43 Taittirīyopaniṣad, 2.1. 44 On Śaṅkara’s interpretation of this sentence from the Taittirīyopaniṣad, based on the distinction between “definitional” and “qualifying” qualification, see Bhattacharya 2001. 45 See Maṇḍana Miśra, Brahmasiddhi, p. 5, l. 6: prakṛṣṭaḥ prakāśaḥ savitā, “The sun is the most intense light.” 46 This interpretation of Padmapāda’s theory in terms of what is known or unknown is borrowed from a later Advaitic work, Amalānanda’s (Vedānta-)Kalpataru, p. 93,

336

47 48

49

50

51 52

Hugo David l. 2–4. It is, however, perfectly consonant with Padmapāda’s earlier description of the same process. Prakāśātman, Pañcapādikāvivaraṇa, p. 717, l. 9. Prakāśātman, Pañcapādikāvivaraṇa, p. 719, l. 3–7: prakāśaśabdaḥ sāmānyābhidhānamukhena lakṣaṇayā vyaktiviśeṣe vartate. prakṛṣṭaśabdaś ca lakṣaṇayā prakarṣābhidhānamukhena prakāśaviśeṣe vartate. tatra guṇasāmānyayoḥ candrapadābhidheyatvābhāvāj jahallakṣaṇayā tad ubhayaṃ vyudasya tatsamavāyiprakāśaviśeṣa eva candrapadābhidheyatayā samarpyata iti prakṛṣṭaprakāśacandraśabdānām ekārthavṛttitā siddhā. In spite of some confusion due to the use of abstracts in English, “intensity” is ontologically speaking a quality (guṇa), and should not be confused with a universal (sāmānya) like “being-light.” Prakāśātman, Pañcapādikāvivaraṇa, p. 699, l. 19–p. 702, l. 3: jñānānandaśabdāv ānandaviśeṣaṃ gamayataḥ. ekasatyānantaśabdā bhedamithyātvadvaitābhāvābhi­ dhānadvāreṇa tatra lakṣaṇayā vartante. sarvajñaḥ sarvaśaktir ityādiśabdāś cānirvacanīyaprapañcopādhitayā tatra vartante […] iti sarve jagatkāraṇaviṣayāḥ śabdā mukhyalakṣaṇopādhibhir ekarasam eva brahma pratipādayanti. Prakāśātman, Śābdanirṇaya, v. 67: ataḥ satyādiśabdās tu saṃsṛṣṭārthoktiśaktitaḥ / mānāntarasyāviṣayaṃ bhūtārthaṃ bodhayanti naḥ // Nṛsiṃhāśrama, Vedāntatattvaviveka, pp. 648–51: yady api padārthamātre śaktyabhyupagame ’pi nākhaṇḍārthavādinām asmākaṃ kā cit kṣatiḥ, tathāpy ekapadaprayoge ’py anvayaviśeṣasya niyamena jijñāsādarśanāt, tasyāś cānvayasāmānyajñānapūrvakatvāt, tajjñānasya ca tadā śabdād anyato ’saṃbhavāt śabda evānvitam abhidhatte.

Primary Sources Advaitasiddhi

Anantakṛṣṇa Śāstrī, N.S., ed. 2005. Advaitasiddhi of Madhusūdanasarasvatī with the Commentaries Gauḍabrahmānandī (= Laghucandrikā of Gauḍabrahmānanda), Viṭṭhaleśopādhyāyī, Siddhivyākhyā of Balabhadra and Critical Summary called Caturgranthī by N.S. Anantakṛṣṇa Śāstrī. Parimal Sanskrit Series 7. Delhi: Parimal Publications.

Bhāmatī

Bhāmatī of Vācaspatīmiśra. In Brahmasūtrabhāṣya.

Brahmasiddhi

Kuppuswami Sastri, S., ed. 1937. Brahmasiddhi by Ācārya Maṇḍanamiśra with [the] Commentary by Śaṅkhapāṇi. Madras Government Oriental Manuscripts Series 4. Madras: Government Press.

Speaking of the Individual

337

Brahmasūtrabhāṣya

Anantakṛṣṇa Śāstrī, M., ed. 1938. Brahmasūtra Śāṅkara Bhāṣya with the Commentaries Bhāmatī, Kalpataru and Parimala. Bombay: Nirnaya Sagar Press.

Kalpataru

(Vedānta-)Kalpataru of Amalānanda. In Brahmasūtrabhāṣya.

Nayanaprasādinī

Nayanaprasādinī of Pratyaksvarūpa. 1986. In Tattvapradīpikā (Citsukhī).

Nayaviveka

Rāmanātha Śāstrī, S.K., ed. 1937. Nayaviveka of Bhavanātha Miśra with the Vivekatattva of Ravideva [Tarkapāda]. Madras University Sanskrit Series 12. Madras: University of Madras.

Nyāyadīpikā

Prabhākaraprasāda, ed. 2003. Śrīma​dānan​dabod​hakṛta­­ śābd​anirṇ​ayadī​pikāy​āḥ saṃpādanam. Delhi: Nāg Publishers.

Nyāyamakaranda

Swâmi Bâlarâma Udaseen Mândalika, N. S. N., ed. 1907. Nyāyamakaranda, a Treatise on Vedānta Philosophy by Śrī Ānandabodha Bhaṭṭārakācārya, with a Commentary by Citsukha Muni, Pramāṇamālā and Nyāyadīpāvalī. Chowkhamba Sanskrit Series, 38, 62, 87 & 117. Benares: Chowkhamba Sanskrit Book-Depot.

Pañcapādikā

Śrīrāma Śāstrī, S., and S.R. Kṛṣṇamūrti Śāstrī, eds. 1958. Pañcapādikā of Śrī Padmapādācarya, with the Commentaries Prabodhapariśodhinī of Ātmasvarūpa and Tātparyārthadyotinī of Vijñānātman and Pañcapādikāvivaraṇam of Śrī Prakāśātman with Tātparyadīpikā of Citsukhācārya and Bhāvaprakāśikā of Nṛsiṃhāśrama. Madras Government Oriental Series 155. Madras: Government Oriental Manuscripts Library.

Pañcapādikāvivaraṇa

Pañcapādikāvivaraṇa of Prakāśātman. 1958. In Pañcapādikā.

Parimala

(Vedāntakalpataru-)Parimala of Appayya Dīkṣita. In Brahmasūtrabhāṣya.

Śābdanirṇaya

David, Hugo. Śābdanirṇaya of Prakāśātman. In David 2020.

Tattvapradīpikā (Citsukhī) Kâshinâth Shâstri, ed. 1986. Tattvapradīpikā (Citsukhī) of Paramahaṃsa Citsukhācārya, with the Commentary Nayanaprasādinī. Vrajajivan Prachyabharati Granthamala 19. Delhi: Chaukhamba Sanskrit Pratisthan.

338

Hugo David

Vākyārthamātṛkā

Subrahmanya Sastri, A., ed. 1961. In Prakaraṇa Pañcikā of Śālikanātha Miśra with the Nyāya-Siddhi of Jaipuri Nārāyaṇa Bhaṭṭa. Banaras Hindu University Darśana Series 4. Varanasi: Benares Hindu University.

Vedāntaparibhāṣā

Suryanarayana Sastri, S. S., ed. 1984. Vedāntaparibhāṣā, by Dharmarāja Adhvarin. Edited and translated. Madras: The Adyar Library and Research Centre. First edition 1942.

Vedāntatattvaviveka

Narayana Swamy Sastri, S., ed. 1955. Vedāntatattvaviveka of Nṛsiṃhāśrami, with the Commentary Tattvavivekadīpana of the author and Pūraṇī Vyakhyā of Agnihotrayajvān. Oriental Research Institute Publications, Sanskrit Series 96. Mysore: University of Mysore.

Secondary Sources Alston, A.J. 1997. Sri Swami Satchidanandendra Sarasvati: The Method of the Vedānta. A Critical Account of the Advaita Tradition. Delhi: Motilal Banarsidass. First edition 1989, Kegan Paul, London. Bhattacharya, Kamaleswar. 2001. “Lakṣaṇa, Lakṣaṇā and Apophaticism in Śaṅkara’s Commentary on Taittirīyopaniṣad II,1.” In Le parole e i marmi: studi in onore di Raniero Gnoli nel suo 70 compleanno, ed. by Raffaele Torella et al., 85–95. Serie Orientale Roma 92. Roma: Istituto Italiano per l’Africa e l’Oriente. Biardeau, Madeleine. 1957. “La définition dans la pensée indienne.” Journal Asiatique 245: 371–84. Biardeau, Madeleine, ed. 1979. Le Tattvabindu de Vācaspati Miśra. Publications de l’Institut Français d’Indologie 3. Édition critique, traduction et introduction. Pondichéry: Institut français d’Indologie. Reprint of the 1956 edition. Cammann, Klaus. 1965. Das System des Advaita nach der Lehre Prakāśātmans. Münchener Indologische Studien 4. Wiesbaden: Otto Harrassowitz. David, Hugo. 2017a. “Les définitions de l’énoncé (vākya) dans la tradition sanskrite: entre grammaire et exégèse.” Langages 205: 27–41. David, Hugo. 2017b. “Les origines du Vedānta comme tradition scolastique. État du problème, nouvelles hypothèses.” Bulletin de l’École française d’Extrême-Orient 102 (2016): 11–46. David, Hugo. 2020. Une philosophie de la parole. L’”enquête sur la connaissance verbale” (Śābdanirṇaya) de Prakāśātman (édition critique, traduction, commentaire, avec une nouvelle édition du commentaire d’Ānandabodha). Paris: École française d’Extrême-Orient.

Speaking of the Individual

339

David, Hugo. Forthcoming. “The Birth of Uttara-Mīmāṃsā. A New Look at the Early History of Vedānta from a Hermeneutic Point of View.” Gerschheimer, Gerdi. 1996. La théorie de la signification chez Gadādhara. Le Sāmāṇyakāṇḍa du Śaktivādavicāra. Publications de l’Institut de Civilisation Indienne 65. 2 vols. Paris: De Boccard. Ghosh, Raghunath. 2006. “The Advaita Theory of Śābdāparokṣatva: Is It Testimonial or Perceptual.” In Śabdapramāṇa in Indian Philosophy, ed. by Manjulika Ghosh and Bhaswati Bhattacharya Chakrabarti, 135–41. University of North Bengal Studies in Philosophy 7. New Delhi: Northern Book Centre. Gupta, Bina. 2011. Consciousness, Knowledge, and Ignorance: Prakāśātman’s Pañcapādikāvivaraṇa “Elucidation of Five Parts,” Section on Inquiry (Jijñāsādhikaraṇa), First Part (Prathama Varṇaka). Introduction, Transliterated   Text, Translation and Critical Explanatory Notes. Treasury of the Indic Sciences Series. New York: The American Institute of Buddhist Studies at Columbia University in New York. Hacker, Paul. 1950. Untersuchungen über Texte des frühen Advaitavāda. 1. Die Schüler Śaṅkaras. Akademie der Wissenschaft und der Literatur in Mainz. Abhandlung der Geistes- und Sozialwissenschaftlichen Klasse 6. Wiesbaden: Franz Steiner Verlag. Kunjunni Raja, K. 1969. Indian Theories of Meaning. Adyar Library Series 91. Second edition. Madras: Adyar Library and Research Centre. Pandurangi, Krishnacharya T. 2004. Prakaraṇapañcikā of Śālikanātha, with an Exposition in English. New Delhi: Indian Council of Philosophical Research. Suryanarayana Sastri, S.S. 1961. “Akhaṇḍārtha.” In Collected Papers of Professor S.S. Suryanarayana Sastri, ed. by T.M.P. Mahadevan. Madras University Philosophical Series 8. Madras: University of Madras. First published in the Journal of Oriental Research 12, 1938. Suryanarayana Sastri, S.S., and C. Kunhan Raja. 1992. Bhāmatī of Vācaspati on Śaṅkara’s Brahmasūtrabhāṣya (Catuḥsūtrī) (edited and translated). Madras: The Adyar Library and Research Centre. First edition 1933. Venkataramiah, Devandahalli. 1948. The Pañcapādikā of Padmapāda (translated into English). Gaekwad’s Oriental Series 107. Baroda: Oriental Institute. Vetter, Tilmann. 1972. Sarvajñātman’s Saṃkṣepaśārīrakam. 1. Kapitel. Einführung, Übersetzung und Anmerkungen. Philosophisch-Historische Klasse. Sitzungsberichte 282, 3. Veröffentlichungen der Kommission für Sprachen und Kulturen Südund Ostasiens 11. Wien: Hermann Böhlaus / Österreichische Akademie der Wissenschaften. Wicher, Irene. 1987. “Vākya und Vidhi. Śālikanātha’s Vākyārthamātṛkā.” Introduction and German Translation with Notes. PhD thesis, Universität Wien.

17

The Role of Intention in Gaṅgeśa’s Non-Communicational Model of Verbal Understanding Yoichi Iwasaki

1  Why Do We Have to Understand Intention? To understand the meaning of a sentence, we have to first understand what the speaker wants to say by each word contained in the sentence—this is a rule commonly accepted by Navya Nyāya philosophers. The reason to formulate this rule is primarily that a hearer cannot disambiguate the meaning of a polysemous word without understanding the speaker’s will. When a model hearer hears a sentence that contains a polysemous word, she guesses what the speaker wants to say by that word, based on the context in which the sentence is given, then identifies the meaning of the word with what the speaker wants to say, and finally gets the whole sentence meaning. One popular example is, “Bring the saindhava.” The word “saindhava,” which literally means any specialty from the Sindhu country, usually refers to either a horse or salt. When this sentence is uttered at a dinner table, the addressee understands from the context that the speaker wants to mean salt with the word, and specifies that the “saindhava” in this case means salt, then accordingly determines the meaning of the whole sentence. Understanding what the speaker wants to say is thus necessary for disambiguation. This is stated in one of the most popular primers of Navya Nyāya called Nyāyasiddhāntamuktāvalī as follows: If the awareness of the intention [viz. the speaker’s will] were not a cause [of verbal understanding], it would not be possible to have [selectively] the awareness of a horse in some cases, and that of salt in other cases, when [the hearer hears the sentence,] “Bring the saindhava,” etc.1

The Role of Intention in Gaṅgeśa’s Model

341

I purposefully avoided using the word “intention” in the above paragraph to refer to what the speaker wants to say, because intention, or “tātparya” in Sanskrit, is primarily a property of words in the Navya Nyāya semantics. This term literally means the state of intending a particular thing, and this state is possessed by words. Udayana, the tenth-century Nyāya philosopher, defines the intended thing (para) as what is referred to by the word (uddeśya), and attributes this “reference” to the word rather than the speaker. He states: “The word that functions with reference to x intends x, because it is so understood by general people.”2 For example, the word “saindhava” in the above example intends salt; in other words, the word has the intention for salt. The clear distinction between the word’s intention and the speaker’s will is a crucial point. What is required to be understood for understanding the sentence meaning is actually the word’s intention, and understanding the speaker’s will is the way to satisfy this requirement. This usage of the noun “intention” and the verb “to intend” might sound odd in current English, because intention is usually attributed to a person. This oddness, however, is significant. There were Navya Nyāya philosophers who transferred this intention (tātparya) from the word to the speaker. According to Gaṅgeśa (fourteenth century), a word’s intention is determined arbitrarily by the speaker. He defines the state of intending x (tatparatva) as having x as the purpose (prayojana), and explains that this “purpose” is determined by the speaker at will.3 “(The state of being the purpose) is the state of being the object of the will of the person who tells (viz., the speaker). When a word is employed by the speaker with the will to convey the sense of x, the word intends x.”4 Later Navya Nyāya philosophers even state that a word’s intention is equal to the speaker’s will: “The intention is said to be the speaker’s will.”5 The word’s intention and the speaker’s will are often mixed up in Navya Nyāya after Gaṅgeśa. You will notice specimens in the passages quoted hereafter that intention is attributed to the speaker. Nevertheless, their distinction must be strictly observed; otherwise, we will face the problem of the speaking parrot.6

2  The Problem of the Speaking Parrot The premise is that a parrot speaks without thought. However, we do sometimes understand its speech, when it is grammatically correct. This causes a problem:

342

Yoichi Iwasaki

How can we understand the parrot’s speech? As said above, we must guess the speaker’s will, which determines the word’s intention, in order to understand the meaning of a sentence, but there is no speaker’s will in the parrot’s speech. Navya Nyāya philosophers who uphold the necessity of understanding the intention in such terms have to give an account of what is going on here. Specifically, they are required to explain how we can disambiguate the meaning of a polysemous word used in the sentence uttered by a parrot. The standard strategy of Navya Nyāya philosophers is to assume that there is God’s will behind the parrot’s speech. Let us look at two instances. The first is provided by Rucidatta (sixteenth century), one of the early commentators of Gaṅgeśa: [Objection:] There is a deviation [of the awareness of the intention from being a cause of the understanding of sentence meaning] in the case of the sentence of a parrot, and so on,7 because of the verbal understanding [of the sentence meaning that is produced] despite that the intention is known to be absent in that case. It cannot be stated that that is not so [viz., there is no deviation] for the reason that there is God’s intention in that case [of parrots’ speech, etc.] as well, because we cannot understand it [viz., God’s intention] in that case [of parrot’s speech etc.]. Nothing helps us to understand it, because the sentence of a parrot, and so on, does not conform to the exegetical rules, context, and so forth, [which help us in other cases].8 [Answer:] Some people hold as follows: No [there is no deviation]. Because in that case, we can understand it [viz., God’s intention] even by the semantic fitness and so on,9 since there would be no understanding of the semantic connection [of word meanings, viz., the sentence meaning] when we do not understand them. Others maintain that [former Nyāya philosophers] have stated the necessity of it [viz., the awareness of the intention] with respect only to other cases [than the sentence of a parrot, etc.].10

Rucidatta above mentions the view that there is God’s will behind the parrot’s speech. This view meets an objection that we could not know His will, even if it were there. Rucidatta quotes the reply from “some people” that we can know it by the semantic fitness (yogyatā) and the like. Semantic fitness is usually defined as a lack of blocking knowledge (bādhaka-pramā-abhāva), which is, in Navya Nyāya, virtually equivalent to correspondence with actuality. This is because a blocking knowledge should be absent not only in the speaker’s mind but in everybody’s, including God’s. That omniscient God does not know the fact that

The Role of Intention in Gaṅgeśa’s Model

343

something blocks x means that x is not blocked, or in other words, correct.11 Rucidatta perhaps wants to say that we can infer God’s will so that the parrot’s speech tells the truth, because, if God is controlling the parrot’s speech, He should intend to say only truth using the parrot’s mouth. However, it is actually not the case that parrots always tell the truth. They sometimes utter untruthful speech, which must not be a product of God’s will, yet we can understand their untruthful speech; this is another problem, which is solved in Nyāyasiddhānta muktāvalī of Viśvanātha (seventeenth century). He also assumes God’s will to be behind the parrot’s speech, but only for truthful speech. When the parrot tells an untruth, according to Viśvanātha, its speech is a product of its trainer’s will. Similarly, in the case of a parrot’s sentence, too, the awareness of God’s intention is a cause [of the understanding of sentence meaning]. As for the parrot’s sentence that disconforms [viz., that is wrong], the only awareness of the trainer’s intention should be said to be the cause. Others say that the awareness of an intention is a cause only in cases such as polysemous expressions. Therefore, in the case of a parrot’s sentence, verbal understanding is produced without an awareness of the intention.12

This is hardly convincing. Who guesses what God wants to say, while hearing a parrot’s speech? Rather than inviting a parrot to participate, let us check this by using an online random sentence generator. There is a web service that returns computer-generated sentences that are grammatically correct and semantically acceptable, but do not carry the speaker’s will, since there are no speakers behind.13 When I used this web service, it returned the sentence, “Millie must have given the box a shake.” I interpreted the polysemous word “box” to mean a square container, though it may also mean a theatre’s ticket counter, to fight by fists, and the like. In this process of verbal understanding, I did not guess God’s will. I just thought that a square container is the default or the most suitable meaning of “box” collocated with other words in the above sentence. If Rucidatta means to say that I unconsciously knew God’s will because of semantic fitness, I cannot object to it because it is perhaps undisprovable. This counterintuitive solution based on God’s will is perhaps a consequence of the confusion between an epistemic cause and a cognitive cause of a veridical awareness. Let me draw a distinction between these two types of causes. This distinction was not recognized by Nyāya philosophers, but its introduction helps us work on their theoretical problems. An epistemic cause is a necessary condition for the resulting cognition to be veridical, and it is called an excellence or “guṇa” in Indian epistemology. A cognitive cause, on

344

Yoichi Iwasaki

the other hand, is necessary for the production of any awareness regardless of whether it is veridical or not. An epistemic cause does not have to be a cognitive cause. For example, sufficient surrounding light, which is a typical epistemic cause of visual perception, is not required for the production of a non-veridical visual perception. As for a verbal understanding, correspondence with actuality is only an epistemic cause and not a cognitive cause, since we can understand a sentence even if it disconforms with actuality. On the other hand, syntactic correctness is a cognitive cause, because any sentence has to be syntactically correct to be understood. God’s will behind a parrot’s speech was known to earlier Nyāya philosophers before Gaṅgeśa as an epistemic cause. We can see their claim in the general epistemology section (prāmāṇyavāda) of Tattvacintāmaṇi, where he asks what the epistemic cause for verbal understanding is, in other words, what makes the resulting verbal understanding veridical. [Gaṅgeśa:] In the cases of the sentence of a deceiver who misunderstands the fact, and the sentence of a parrot, these sentences are [supposed to be] uttered by trustworthy persons, because these are sources of knowledge. Therefore, [these sentences] are [considered] to be generated by God’s veridical awareness of the thing signified by the sentence, just like in the case of the Veda, because He is the producer of all products. [Objection:] If it is the case, then there would be no possibility of a seemingly correct statement [that is actually not a source of knowledge,] because such a statement is also [considered to be] uttered by God. [Reply:] It is not so, because, since the thing signified by that [wrong] sentence actually does not exist [in the actual world], such a thing cannot be the object of the Lord’s awareness. Thus hold the scholars who know the tradition.14

The “scholars who know the tradition” think that it is the speaker’s veridical awareness of the thing meant by the sentence. There arises a problem: We can have a veridical verbal understanding from a parrot’s speech, but what makes it veridical? The “scholars who know the tradition” reply that the truthful speech of a parrot is a product of God’s awareness of the thing stated by the parrot. This reply hints that they want to say that no cognition is coincidentally veridical, and that there must be a cause that makes the cognition necessarily veridical. It should also be noted that for an awareness to be veridical, the awareness of an epistemic cause is not a necessary condition. For example, one who sees something does not have to be aware that there is sufficient light to see it correctly. Similarly, since God’s awareness is an epistemic cause, we

The Role of Intention in Gaṅgeśa’s Model

345

do not have to know that it is there. Indeed, we can never be aware, prior to understanding the sentence meaning, that a parrot’s speech is a product of God’s said awareness. Only after the resulting awareness is verified to be veridical could we retrospectively know that it was a product of God’s awareness. Rucidatta and others might have used this idea in an inappropriate manner for the cognitive theory of verbal understanding. Presumably, since Rucidatta and Viśvanātha were conscious of the difficulty of this view, they quoted alternative viewpoints showing that the awareness of the speaker’s will is not always necessary. However, they did not explain how we can disambiguate the meaning of a polysemous word. I can scarcely justify their solutions to the problem of parrot’s speech. It is more feasible to assume speakerless intention of the parrot’s speech, as Gaṅgeśa did in the section of verbal understanding. He there develops a discussion on the word’s intention that is not tied with the speaker’s will and concludes as follows: As a matter of fact, the intention is a word’s state of aiming at the [hearer’s] cognition of the relation [of its meaning] with the meaning of another word. [This state] is ascertained based on the exegetical rules as to the Veda, and based on the exegetical rules or the context etc. as to secular utterances.15

This account of understanding speakerless intention can explain the cognitive process of understanding sentence meaning of a parrot’s speech. We can no longer insist upon the necessity of understanding the speaker’s will.

3  The Role of Intention in the Non-communication Model of Verbal Understanding The above theory of speakerless intention suggests that the awareness of the speaker’s will is not a cognitive cause of understanding sentence meaning, but this does not reject the epistemic necessity of knowing the speaker’s will. To know a word’s correct intention that is defined by the speaker, we have to know what the speaker wants to say by that word. However, Gaṅgeśa rejects its epistemic necessity for another reason: we do not have to know the correct intention to get knowledge. If the goal of the hearer’s cognitive activity is to know the speaker’s thoughts, then it is crucial to know the correct intention, or, rather, it is the goal itself, but this is actually not our goal. By contrast, the job of the word as a source of knowledge (pramāṇa) is to make the hearer know the world, not to transfer thoughts from speaker to hearer. Therefore, even when

346

Yoichi Iwasaki

the hearer misunderstands the intention and chooses the unintended meaning of a polysemous word, if the resulting cognition is veridical, the hearer would happily obtain veridical awareness, and the wrong yet misunderstood sentence would be regarded as a source of veridical awareness. This standpoint is clearly presented in the following dialogue: [Objection:] If that is the case [viz., if the correct awareness of semantic fitness etc. is the epistemic cause of verbal understanding], when the hearer misunderstands the intention of a polysemous word that actually intends a different meaning [from the one that the hearer understands], or even without it [viz., misunderstanding of the intention], as long as there is the correct awareness of semantic fitness and so forth, the awareness of the different meaning (śakya) that conforms [viz., that is correct], would also be labeled veridical awareness, and consequently, even in that case the [wrong but misunderstood] sentence would be regarded as a source of knowledge. [The sentence is actually wrong, so it must not be deemed a source of knowledge.] [Gaṅgeśa:] No, [there is no fault] because it [viz., the epistemic outcome] is what we expect. As to the object of intention, moreover, the [wrong yet misunderstood] sentence, in that case, does not produce the veridical awareness [of that object], and therefore is not regarded as a source of knowledge.16

Gaṅgeśa recognizes that there is a correct intention of a word as defined by the speaker, but he does not acknowledge any epistemic significance of knowing it. A parrot’s speech does not even have a correct intention, because nobody is qualified to decide what is truly intended. It is, however, indeed necessary to understand the intention of a polysemous word for the awareness of a sentence meaning to arise, in the sense that the hearer must decide on one meaning of a polysemous word. I am not sure why later Navya Nyāya philosophers gave the speaker so much esteem. Gaṅgeśa’s theory is quite feasible, but his successors did not adopt it. I do not understand what advantage they thought their new theories have.

Notes 1 Nyāyasiddhāntamuktāvalī, p. 315: yadi tātparyajñānaṃ kāraṇaṃ na syāt, tadā saindhavam ānayetyādau kva cid aśvasya kva cil lavaṇasya bodha iti na syāt. 2 Nyāyakusumāñjali, p. 620: yaduddeśena yaḥ śabdaḥ pravṛttaḥ sa tatparaḥ, tathaiva lokavyutpatteḥ.

The Role of Intention in Gaṅgeśa’s Model

347

3 Tattvacintāmaṇi, vol. 4 (1), p. 325: tatprayojakatvaṃ tatparatvam. (The state of intending x is the state of having x as the purpose.) The meaning of this definition is elaborated in Iwasaki 2011 along with its historical background. 4 Tattvacintāmaṇi, vol. 4 (1), p. 327: kin tu pratipādakecchāviṣayatvam. yaḥ śabdaḥ vaktrā yadicchayā prayuktaḥ sa tatparaḥ. 5 Nyāyasiddhāntamuktāvalī, v. 84, p. 315: vaktur icchā tu tātparyaṃ parikīrtitam // 6 I have presented the problem of the parrot’s speech in Iwasaki 2018 focusing on its historical development for Sanskritists. Please refer to it for previous studies on intention and more detailed textual analysis. 7 For example, children who repeat elders’ words without knowing meaning also cause the same problem. See Tattvacintāmaṇi, vol. 1, pp. 345–47 for the set of “a parrot, a child, and so on” (śukabālādi). 8 Gaṅgeśa claims that the exegetical rules help to understand Vedic sentences, and the context as well as those rules help in case of ordinary utterances. See Tattvacintāmaṇi, vol. 4 (1), pp. 179–82. 9 Rucidatta was possibly thinking of syntactic correctness (ākāṅkṣā) and temporal unity (āsatti) by “and so on,” which are also causes of the understanding of sentence meaning. 10 Tattvacintāmaṇiprakāśa, p. 12: nanu śukādivākye tātparyavyatirekaniścaye ’pi śābdabodhād vyabhicāraḥ, na ca tatrāpīśvaratātparyasattvān na tathātvam iti vācyam. śukādivākyasya nyāyaprakaraṇādyananurodhitayā grāhakābhāvena   tatra tadgrahasyāśakyatvād iti cet, na, yogyatādināpi tatra tadgrahasaṃbhavāt. tadagrahe tatrānvayabodhāsiddher ity eke. tadatiriktasthala eva taddhetutvam uktam ity anye. 11 For this argument, see Mathurānātha’s Tattvacintāmaṇirahasya, vol. 4 (1), p. 263. 12 Nyāyasiddhāntamuktāvalī, pp. 316ff: itthañ ca śukavākye ’pīśvaratātparyajñānaṃ kāraṇam. visaṃvādiśukavākye tu śikṣayitur eva tātparyajñānaṃ kāraṇaṃ vācyam. anye tu—nānārthādau kva cid eva tātparyajñānaṃ kāraṇam. tathā ca śukavākye vinaiva tātparyajñānaṃ śābdabodhaḥ. 13 http:​//ait​ech.a​c.jp/​~ites​ls/c/​r8.cg​i/sen​tence​s, accessed on November 26, 2018. 14 Tattvacintāmaṇi, vol. 1, pp. 349–50: bhrāntapratārakavākye śukādivākye ca pramāṇaśabdatvenāptoktatvāt vedavad īśvarasyaiva yathārthavākyārthajñānaṃ janakam, tasya kāryamātre kartṛtvāt. śukādivākyasya ca vedatulyatā doṣābhāvavādināpi vācyā. nanv evaṃ śabdābhāsocchedaḥ tasyāpīśvaravaktṛkatvād iti cet, na, tadvākyārthasyāsattvena bhagavajjñānāgocaratvāt. evaṃ liṅgābhāsajanyapramāyām api vahnivyāpyavattvajñānam īśvarasyaiva janakam iti sampradāyavidaḥ. 15 Tattvacintāmaṇi, vol. 4 (1), pp. 179–80: vastutas tu itarapadasya itarapadārthasa­ ṃsargajñānaparatvaṃ tātparyaṃ tac ca vede nyāyād avadhāryate loke nyāyāt prakaraṇāder vā.

348

Yoichi Iwasaki

16 Tattvacintāmaṇi, vol. 1, pp. 351 ff. Tattvacintāmaṇi: nanv evaṃ nānārthād anyaparāt tātparyabhrame taṃ vinaiva vā yathārthayogyatādijñāne sati saṃvādyaparaśakyajñānam api prameti tatrāpi tadvākyaṃ pramāṇaṃ syād iti cet, na, iṣṭatvāt. tātparyaviṣaye ca tadvākyaṃ na tadā pramājanakam iti na pramāṇam.

Primary Sources Nyāyakusumāñjali

Goswami, Mahaprabhulal, ed. 1972. Nyāyakusumāñjali of Udayanācārya. Mithila Institute Series, Ancient Texts 23. Darbhanga: Mithila Research Institute.

Nyāyasiddhāntamuktāvalī

Śukla, Harirāma, ed. 1997. Nyāyasiddhāntamuktāvalī of Viśvanātha Pañchānan Bhaṭṭāchārya. Kashi Sanskrit Series 6. Fifth edition. Varanasi: Chaukhambha Sanskrit Sansthan.

Tattvacintāmaṇi

Kamakhyanath Sharma, Tarkavagish, ed. 1990. The Tattvacintāmaṇi of Gaṅgeśa Upādhyāya. 4 vols. Delhi: Chaukhamba Sanskrit Pratishthan. Original edition Calcutta: Asiatic Society, 1888–1901.

Tattvacintāmaṇiprakāśa

Saha, Sukharanjan, and P.K. Mukhopadhyay, eds. 1991. Rucidatta’s Śabdamaṇi-Prakāśa. Calcutta: Jadavpur University.

Tattvacintāmaṇirahasya

Tattvacintāmaṇirahasya of Mathurānātha. In Tattvacintāmaṇi.

Secondary Sources Iwasaki, Yoichi. 2011. “Some Remarks on Gaṅgeśa’s Argument on Tātparya.” Indogaku Bukkyōgaku kenkyū (Journal of Indian and Buddhist Studies) 59 (3): 1132–36. Iwasaki, Yoichi. 2018. “Problems in Postulating Tātparyajñāna as a Requisite for the Generation of Verbal Understanding.” Nagoya Studies in Indian Culture and Buddhism: Saṃbhāṣā 34: 27–41.

Part Four

Of Implicatures and Figurative Meanings

350

Introduction

Linguistic communication involves unspoken elements which are neither directly expressed nor literally understood. We are here borrowing the Gricean term, “implicature,” to embrace all the variegated nuances of nonliteral meanings. Though such suggestive meanings are most apparent in poetic and figurative expressions, where what is intended is not identical with what is literally said, implicatures are widely spread throughout linguistic practices and are key factors of even the most basic aspects of linguistic interaction. The standard view of sentence-signification developed by Kumārila Bhaṭṭa, and consistently upheld by the Bhāṭṭa tradition that derives from his works, is that sentence meaning is invariably conveyed not by direct signification (abhidhā), but, rather, through an indirect process usually labeled as the secondary or figurative function (lakṣaṇā). Denotation operates only at the level of individual word meanings, and once they have been denoted, these word meanings then lead the hearer to understand the unified sentence meaning through this process of secondary signification (Chapter 18). The Nyāya author Jayanta, after scrutinizing the Bhāṭṭa view, maintained that words have a second causal force beside their referential power, and called it “intentionality” (tātparyaśakti). Jayanta insisted that even poetic meanings can be explained through these two regular causal forces of words: referentiality and intentionality. The close examination of Jayanta’s passages reveals how this notion of intentionality is the fruit of a syncretic work done by Jayanta, who imported with some correctives the Mīmāṃsā theory of sentence meaning (Chapter 19). In poetics the term rasa has been used since antiquity to denote aesthetic emotions, which the connoisseur can experience by reading poetry. Some literary critics—specifically Bhaṭṭa Nāyaka, Abhinavagupta, Dhanika, and Bhoja—have described this rasa as a peculiar kind of “sentence meaning.” The theoretical approach of these authors results in a synthesis of the theory of rasa with the Mīmāṃsā theories of meaning, also influenced by Jayanta’s postulation

352

Introduction

of an intentionality of words. Unlike in Mīmāṃsā and Nyāya, however, the output of verbal knowledge is no longer the awareness of a duty or a true belief but, rather, an aesthetic experience (Chapter 20). The most evident case of the unspoken message of words is poetical language. Ānandavardhana and Abhinavagupta explained this phenomenon by postulating a new power of suggestion (dhvani) as a third-order cause of signification, even beyond indication or figurative language (lakṣaṇā). Mukula Bhaṭṭa responded to the theory of dhvani by explaining every variety of poetical meaning as a subcategory of indirect signification (lakṣaṇā), which is considered by Mukula as all encompassing, inherent in any linguistic expression (Chapter 21).

18

Kumārila on the Role of Implicature in Sentence-Signification Lawrence McCrea

As the most fully developed hermeneutic system in premodern India, the scriptural-interpretive tradition of Mīmāṃsā was naturally closely concerned with how texts express their meanings, and particularly with how words or word meanings combine to generate coherent sentence meanings—to such an extent that Mīmāṃsā came to be widely known among brāhmaṇical intellectuals as the “vākya-śāstra,” or “science of sentences.” The specific character of sentencesignification was an important topic for the Mīmāṃsakas, partly as a matter of general theoretical interest, but also because it impacted many concrete interpretive decisions. In deciding the interpretive and ritual-performative questions that form the bulk of Mīmāṃsā discussions, it is often important to know, for example, which elements of a given sentence meaning are directly expressed and which are only implied or indirectly indicated, which form the subject of the sentence and which the predicate, and so on. Hence both the general nature and the specific mechanics of sentence-level expression— exactly how individual words or word-meanings combine to convey a coherent sentential meaning—are persistent concerns for all Mīmāṃsakas in all periods. As is well known, in time the two main schools of post–Śabara Mīmāṃsā, those founded by Kumārilabhaṭṭa and Prabhākara, came to uphold two contrary theories of sentence meaning. The followers of Kumārila or Bhāṭṭas upholding the “construal of the denoted” (abhihitānvaya) theory—that words each denote their own individual meanings, and these meanings then combine with and qualify one another so as to produce a single coherent sentence meaning. The Prābhākaras, by contrast, adhere to the “construed denotation” theory—that the words comprising a sentence denote their meanings simultaneously and already construed into a unified sentence meaning. Kumārila, however, shows

354

Lawrence McCrea

no knowledge of Prabhākara or his works, and his own extensive discussions of sentence meaning show no awareness of the construed-denotation theory.1 The basic problematic he confronts in his main discussion of sentence meaning in the Vākyādhikaraṇa of his Ślokavārttika is that already set forth by Śabara in his commentary on PMS, 1.1.24: “Even if the word and its connection [with its meaning] are original, i.e. eternal, nevertheless [it cannot be said that] ‘Dharma is a good indicated by a Vedic injunction’ (PMS, 1.1.2), since a Vedic command is a sentence.”2 Even if it is granted (as the Mīmāṃsakas claim already to have proved in the earlier part of the opening chapter of the Mīmāṃsāsūtra) that the words that make up the Veda, the meanings they denote, and the connections between them are all eternal, this would not be enough to establish the Vedic scriptures as a valid source of knowledge regarding dharma. The relation between individual words and their meanings may be shown to be stable over time and hence reliable, but the same cannot be said of sentences. Speakers of any language are able to produce and interpret entirely new sentences from known words, and hearers derive from them meanings they have not previously learned in association with those sentences, which, for the Mīmāṃsakas, is seen as posing a potential threat to the reliability of the knowledge of dharma derived from Vedic scriptures. The eternality of individual words, the meanings they denote, and the connections between them, will be of use in defending the reliability of Vedic sentences only if it can be shown that particular, and potentially novel, sentence meanings can be reliably derived from the stable meanings invariably associated with individual words, and showing how this derivation occurs is the principal business of Kumārila’s discussion in the Vākyādhikaraṇa.

1  Implicature as the Process of Sentence-Signification The standard view of sentence-signification developed by Kumārila in the Vākyādhikaraṇa and elsewhere, and consistently upheld by the Bhāṭṭa tradition that derives from his works, is that sentence meaning is invariably conveyed not by direct signification (abhidhā), but only indirectly, by an indirect process usually labeled as secondary or figurative implicature (lakṣaṇā). Denotation operates only at the level of individual word meanings—once they have been denoted, these word meanings then lead the hearer to understand the unified sentence meaning through a process of secondary signification. Yet, while it is entirely clear and universally agreed upon among traditional Indian critics

Kumārila on the Role of Implicature in Sentence-Signification

355

and doxographers that this was indeed Kumārila’s view, he is surprisingly (and atypically) reticent or hesitant in advancing it. His most sustained discussion of sentence meaning, the Vākyādhikaraṇa section of his Ślokavārttika, contains no straightforward declaration to this effect. When the most articulate and influential critic of Kumārila’s sentence theory, Śalikanāthamiśra, and it’s most concerted defender, Vācaspatimiśra, seek for a straightforward textual source summing up Kumārila’s basic doctrine on sentence meaning, they avoid the Ślokavārttika altogether and quote a half verse drawn, presumably, from Kumārila’s (lost) Bṛhaṭṭīkā: “It is our view that sentence meaning is, in every case, secondarily expressed.”3 Vācaspatimiśra, seeking to pair this with a parallel quotation from the better-known Ślokavārttika, adds the following: Even if the speech sounds directly produce only communication of the word meanings, they do not stop there, as this would be useless. Since they are employed in order that the sentence meaning will be understood, the communication of the word meanings is an intermediate step, as the flaming of fuel sticks is an intermediate step in the process of cooking.4

This clearly makes the point that words do not directly convey sentence meaning, and that reaching this sentence meaning therefore requires a second cognitive step, but the semantic or cognitive process involved in this second step is not spelled out specifically, being accorded only the general label “understanding” (miti). Why should Kumārila be so coy about this? Some later critics of Kumārila’s lakṣaṇā-based theory sentence meaning object that it renders universal a significatory process originally conceived as anomalous and exceptional: If all words always convey their ultimate meanings through secondary signification, what is there to differentiate well-known cases of lakṣaṇā such as “a village on the Ganges” (gaṅgāyāṃ ghoṣaḥ) from (seemingly) non-metonymical expressions such as “a fish in the Ganges” (gaṅgāyāṃ matsyaḥ)? Perhaps it is a sense that he is stretching the bounds of lakṣaṇā as ordinarily understood that makes Kumārila somewhat reluctant to straightforwardly proclaim his central doctrine of sentence meaning in the Ślokavārttika.

2  Śabara on Sentence Meaning However we are to account for Kumārila’s seeming reticence in this matter, the view that sentence meaning is only secondarily conveyed seems to be well-

356

Lawrence McCrea

grounded in the earlier Mīmāṃsā tradition, particularly in Śabara’s Bhāṣya. While the adherents of both the Bhāṭṭa and Prābhākara schools present their views as founded on Śabara’s, it seems quite clear that he upheld a position on sentence meaning very close to that of Kumārila. This is fairly explicitly expressed in a key passage of Śabara’s own commentary on the Vākyādhikaraṇa: “Words each denote their own individual meaning and then cease functioning. Then the word-meanings, once they have been understood, cause one to understand the sentence meaning.”5 This appears to be a straightforward proclamation of the Bhāṭṭa, abhihitānvaya doctrine—that words directly express only their own individual meanings, and that any conjoint or construed meaning they convey when grouped into sentences emerges only subsequently, and only from the word-meanings (padārthas), not directly from the words themselves. Furthermore, it explicitly restricts the process of denotation (abhidhā) to the expression of individual word meanings. However we understand the process by which the word-meanings “cause one to understand” the sentence meaning, and whatever label may be applied to this process, it is clearly asserted here to be distinct from, and subsequent to, the process of denotation by which each word denotes its individual meaning.

3  Sentence Meaning and the Semantics of Universals and Particulars The need to see the expression of sentence meaning as a secondary, nondenotative process is closely linked, for both Śabara and Kumārila, to the preexisting and deeply rooted Mīmāṃsā doctrine that only universals may be directly denoted. This assertion of the universality of all denoted meanings is itself required by the claim of Vedic eternality. If the Vedas are eternal, so, then, must be the words that compose them and the meanings that they express. In defending this claim, the Mīmāṃsakas offer a more general argument regarding the nature of signification. For language to function as a stable system of signs, the meanings we directly associate with any signifying term (i.e., what they denote) must be abstract, universal, and repeatable features, rather than their individual manifestations. Hence a word such as “cow” can literally designate only the universal “cowness” (which is unitary and eternal), rather than any individual cow or cows (which are multiple and ephemeral). Any reference to particular individuals must be indirect, achieved only by some secondary semantic or cognitive process. Śabara himself makes the point quite explicitly:

Kumārila on the Role of Implicature in Sentence-Signification

357

“People figuratively employ (upacaranti) the word ‘domestic animal’ (paśu), which expresses only a universal, with reference to a particular substance, a hairy, four-footed thing with horns and a tail.”6 Moreover, he elsewhere explicitly links the universal/particular dimension of expression to the wordmeaning/sentence meaning dichotomy: In every case, it is only when the word-meaning has been blocked that sentencemeaning is possible, not otherwise. For the word refers to a universal, but the sentence to a particular. Sentence meaning is the fixing of word-meanings, which function universally, in some particular. As has been said “There is the traditional recitation of these words, in just these senses, for the sake of action, since meaning is grounded in them” (PMS 1.2.25). Among these [two, that is, word-meaning and sentence-meaning], word-meaning is perceived, while sentence-meaning is inferred.7

While Śabara is not entirely clear or consistent in his characterization of the process by which an interpreter moves from word- to sentence meaning (describing it as figurative [upacāra], inferential, or simply saying that the wordmeanings “cause one to understand” the sentence meaning), he clearly takes sentence meaning to be arrived at by some indirect, non-denotative process, and to be cognized only subsequent to and proceeding from the denotation of the individual, universal, word-meanings. Kumārila too links the question of sentence meaning to the theory that only universals may be denoted, and also connects the problem of moving from the denoted, universal, word-meaning to the secondarily indicated, particular, sentence meaning to the process of “qualification” (viśeṣaṇa). One point where he does explicitly label the process involved in moving from word- to sentence meaning as lakṣaṇā specifically involves the move from a universal to a particular referent: The universal is understood through that [word], without regard to its existence or non-existence. Therefore even the use of the word “is” is possible even with regard to that [universal]. Nobody intends to refer to the “existence” or “nonexistence” of a universal, since it is eternal; rather these [i.e., existence and nonexistence] are qualifiers of the individual which is figuratively indicated [lakṣyamāṇa] by it.8

Predication of specific properties, such as existence or nonexistence, makes sense only with reference to particulars, but denotation, as direct and immediate signification, can convey only abstract and repeatable universal properties. These will be instantiated in certain particulars, and the particulars may hence be

358

Lawrence McCrea

indirectly designated through the universals that qualify them. And this indirect designation of one referent through its connection to a literal referent associated with it is precisely what secondary signification (lakṣaṇā) is understood to consist in—as when one says “A village on the Ganges” when one really means “A village on the bank of the Ganges”: the term “on the Ganges” designates, by secondary signification, the bank adjacent to the river, as the literal referent of the expression would make no sense in this context. If we accept the basic principle that a denoted meaning is that directly associated with and immediately conveyed by its signifier, then individuals can never be denoted. The fact that language can produce awareness of a particular only subsequent to and by means of the awareness it produces of a universal means that particulars can only be designated indirectly, via the universals that qualify them. This is summed up in an often quoted line of Kumārila’s (again, presumably from his lost Bṛhaṭṭīkā): “Denotation does not reach the qualificand (viśeṣya), as its power is exhausted [in denoting] the qualifier.”9 The primary referent of any term—what it denotes—can only be the qualifier, never the thing qualified by it, and the latter can hence only be designated indirectly and non-denotatively. Denotation of individuals is in fact impossible in principle— individuals are only cognitively accessible, designable, and memorable through their repeatable features and it is only these repeatable features that can be learned in association with individual words, and that are therefore what is first called to mind when these words are heard.

4 Conclusion If language is to refer at all to particular individuals, as everyday language typically does, or even to enjoin particular ritual performances, as the eternal Vedic text must be seen to do if it is to be of any use at all, it can only do so indirectly, through a process of secondary indication (lakṣaṇā). Whatever our intuitions about the atypicality of such secondary usage, it can be shown to be necessary in all, or nearly all, cases of sentential usage. This position was clearly Kumārila’s, and, in its essentials, seems to be a systematization and rational reconstruction of a set of positions already well established in the pre–Kumārila Mīmāṃsā tradition. Any position that argued for the direct denotation of sentence meaning would need to take a radically different approach, and, if formulated as a continuation of the Mīmāṃsā position, would need to rethink and reinterpret what appears to have been the

Kumārila on the Role of Implicature in Sentence-Signification

359

basic Mīmāṃsā approach to the question of sentence meaning and reference to universals and particulars, and it is just such a rethinking that we find in Prabhākara’s apparently unprecedented elaboration of his “construed denotation” (anvitābhidhāna) theory of sentence signification.

Notes 1 Mainly in the Vākyādhikaraṇa of his Ślokavārttika (Ślokavārttika, 845–950), but important discussions also occur in the Bhāvārthādhikaraṇa (PMS, 2.1.1–4), Vākyalakṣaṇādhikaraṇa (PMS, 2.1.46) and Kartradhikaraṇa (PMS, 3.4.12–13) of his Tantravārttika. 2 Śābarabhāṣya, on PMS 1.1.24, vol. 1, pp. 91–92: yady apy autpattiko nityaḥ śabdaḥ saṃbandhaś ca tathāpi na codanālakṣaṇārtho dharmaḥ. codanā hi vākyam. 3 vākyārtho lakṣyamāṇo hi sarvatraiveti naḥ sthitiḥ, quoted in Prakaraṇapañcikā, 396 and in TB, 153. 4 Ślokavārttika, Vākyādhikaraṇa vv. 342–43, p. 943: sākṣād yady api kurvanti padārthapratipādanam / varṇās tathāpi naitasmin paryavasyanti niṣphale //   vākyārthamitaye teṣāṃ pravṛttau nāntarīyakam / pāke jvāleva kāṣṭhānāṃ padārthapratipādanam // 5 Śābarabhāṣya, on PMS 1.1.25, vol. 1, p. 96: padāni hi svaṃ svaṃ padārtham abhidhāya nivṛttavyāpārāṇi. athedānīṃ padārthā avagatāḥ santo vākyārthaṃ gamayanti. 6 Śābarabhāṣya, on PMS 2.2.17: śṛṅgiṇi pucchavati lomaśe catuṣpadi dravyaviśeṣe paśuśabdam ākṛtivacanam upacaranti. 7 Śābarabhāṣya, on PMS 3.1.12, vol. 2, p. 689: sarvatra tu bādhite* padārthe vākyārtha upapadyate, nānyathā, sāmānyavṛtti hi padam, viśeṣavṛtti vākyam. sāmānyenābhipravṛttānāṃ padārthānāṃ yad viśeṣe ’vasthānam, sa vākyārthaḥ. tad etad uktam—“tadbhūtānāṃ kriyārthena samāmnāyo ’rthasya tannimittatvād” (PMS 1.1.25) iti. tatra pratyakṣataḥ padārthaḥ, vākyārthaḥ punar ānumānikaḥ. [*Most printed editions read “bodhite,” but the variant “bādhite” is necessary for the sense, and is confirmed by Kumārila’s comment on the passage (PMS 3.1.12, vol. 2, p. 689)]. 8 Ślokavārttika, Vākyādhikaraṇa vv. 310–311: astitvādyanapekṣaṃ hi sāmānyaṃ tena gamyate / astiśabdaprayogo ’pi tenaivātropapadyate // jāter astitvanāsitve na ca kaś cid vivakṣati / nityatvāl lakṣyamāṇāyā vyaktes te hi viśeṣaṇe // 9 viśeṣyaṃ nābhidhā gacchet kṣīṇaśaktir viśeṣaṇe. The verse is first quoted in Sucaritamiśra’s commentary on the Ślokavārttika (Ślokavārttika, Codanāsūtra v. 192, vol. 1), and becomes a staple of discussions of the scope of denotation in later Bhāṭṭa works, as well as in the field of poetics.

360

Lawrence McCrea

Primary Sources Mīmāṃsādarśana

Subbāśāstrī, ed. 1929–1934. Śrīmajjaiminipraṇītaṃ Mīmāṃsādarśanam. Vol. 6. Anandashrama Sanskrit Series 97. Pune: Anandasramamudranalayah.

PMS

Pūrvamīmāṃsāsūtra. In Mīmāṃsādarśanam.

Prakaraṇapañcikā

Subrahmanya Sastri, A., ed. 1961. Prakaraṇa Pañcikā of Śālikanātha Miśra with the Nyāya-Siddhi of Jaipuri Nārāyaṇa Bhaṭṭa. Banaras Hindu University Darśana Series 4. Benares: Benares Hindu University.

Śābarabhāṣya

Śābarabhāṣya of Śabara. In Mīmāṃsādarśana.

Ślokavārttika

Tailanga, Rama Shastri, ed. 1898. Ślokavārttika of Kumārila Bhatta. Benares: Vidya Vilas Press.

Ślokavārttika

Kāśikā of Sucarita Miśra. In Mīmāṃsāślokavārttika of Kumārila Bhaṭṭa.

Tantravārttika

Tantravārttikam of Kumārila. 1929–34. In Mīmāṃsādarśana.

19

The Intentionality of Words: Jayanta’s Syncretism of Nyāya and Mīmāṃsā Alessandro Graheli

1 Introduction Within the broader picture of the epistemological role of speech, that is, of verbal testimony, Bhaṭṭa Jayanta (ninth century) also discussed from various angles the process of denotation in sentences. Jayanta maintained that, in epistemologically relevant language, words possess two forces (śakti): referentiality (abhidhātrī śakti) and their tension toward the sentence meaning, intentionality (tātparyaśakti). Although “intentionality” is already a quite compromised word, in its Latin etymology from in-tendo, “to aim at,” or “to strive for,” it is an efficient translation for Jayanta’s tātparya, a noun derived from tat, “that,” and para, “aiming at” or “tending to.” This intentionality should certainly not be confused with a mentalist property, such as the intentionality of Franz Brentano and others, because tātparyaśakti is a causal capacity of words, where the speaker’s intention (vivakṣā) plays just an accessory role. Knowledge of the sentence meaning, therefore, is conveyed by the words themselves through their second force. Jayanta’s claim is that the assumption of this latter force explains any type of meaning, including implicatures, on both the semantic and the pragmatic level.

2  Jayanta’s Synthesis of Nyāya and Mīmāṃsā In Nyāya, the authority of the speaker is considered the ground for the validity of both common and Vedic language. Unlike in Mīmāṃsā, indeed, even the

362

Alessandro Graheli

authority of the Veda is founded on a trustworthy source. Such a trustworthy instructor is defined as follows: An authoritative instructor has directly experienced the essential property (dharma) of things, and he is moved by the desire to describe it as it is or as it is not.1

Since they are grounded on the utterance of a trustworthy instructor, episte­ mically valid expressions by definition cannot be permanent (nitya), because any utterance must necessarily occur at a given point in time. Moreover, the relation between a word and its referent is considered conventional, rather than natural, because it is based on stipulations established either by God or by human beings.2 These principles are all endorsed by Jayanta. Any speech, however, does not work by isolated words but, rather, by complex sentences. And yet, before Jayanta, in Nyāya there had hardly been any elaboration on the process of sentence signification. Linguistic discussions had focused on the signification of words, rather than sentences (see Chapter 7 in this volume). Well aware of this predicament, Jayanta approaches the issue as follows (NM, 5, II 135,15–136,10): [The problem:] The authors of the Sūtra and of the Bhāṣya did not provide any indication to the referent of the sentence. From where shall we learn about its nature, in order to explain it here? [Opponent:] Why did they not provide any indication? [Proponent:] Because this discipline of reasoning (ānvīkṣikī, i.e., Nyāya) is about epistemology, not about the reference of the sentence. [Opponent:] If this is the case, why were word-referents taught at all in Nyāyasūtra 2.2.66, “Referent of words are the particular, the form, the universal”? [Proponent:] This is a good point. There, however, the author of the Nyāyasūtra meant to establish the epistemic validity of śabda, and to put at rest the Buddhist (Dignāga) who argued that words and referents are not in contact. [Opponent:] But if this is the case, if the referent of the sentence is not established as an external object, the epistemic foundation of the science is undermined. An effort must be done in this area as well. [Proponent:] This is true. But by teaching word-referents the author of the Sūtra meant to embrace the sentence too, and this is why he did not separately teach the reference of sentences. His purport is that the word-referent itself is the sentence meaning. Not a single word-referent, however, but multiple wordreferents.3

Even while justifying the teaching of the sūtra, the Mīmāṃsā-savvy Jayanta must have felt the limits of the account of sentence meanings in Nyāya, which is

The Intentionality of Words

363

most likely why he went into considerable detail in his survey of the Mīmāṃsaka views on sentence signification. Bhāṭṭa authors appealed to a secondary process to explain how wordreferents—namely, the universals—jointly convey the sentence meaning— namely, the particular. Building on the Mīmāṃsā notion that words directly denote only universals, Bhāṭṭas believed that words designate particulars only indirectly and only in multi-referential combinations, because only the juxtaposition with other referents can satisfy the necessary criteria for indirect interpretation. Even in seemingly literal sentences, words invariably convey particulars indirectly, as each universal interacts with the others to yield knowledge of the particular through a process of mutual specification (see Chapters 9 and 18 in this volume). Prābhākara authors objected that one may resort to such a secondary process only when the denotative function fails, that actual language works by sentences and never by isolated words, that by overstressing the causal role of referents the Bhāṭṭa view demeans the epistemological primacy of language cherished by Mīmāṃsakas, and ultimately that words invariably denote connected wordreferents, that is, the sentence meaning (see Chapter 13 in this volume). Jayanta’s synthesis contains many elements of the Bhāṭṭa atomist theory, some correctives derived from the Prābhākaras’ criticism, and other adaptations to suit the Naiyāyikas’ needs.

3  Whose Intentionality? As already noticed by Kunjunni Raja and others, Jayanta was the first to postulate the force of intentionality (tātparyaśakti) as a specific characteristic of words: “Jayantabhaṭṭa […] was the first to bring forward the theory about tātparya being a separate vṛtti” (Kunjunni Raja 1963, 219). This much is well known. But Jayanta did not create his theory ex nihilo. The term tātparya, to begin with, is frequently found in Mīmāṃsā, Nyāya and Alaṅkāra works, though with different nuances, before Jayanta’s time.4 In Jayanta’s acceptation, the anaphora of the pronoun “that” (tat) from which the term tātparya is derived must refer to the sentence meaning (vākyārtha).

3.1  “That” Is Not the Speaker’s Intention The term is later commonly used to denote quite simply the speaker’s intention,5 but Jayanta is certainly not using tātparya in this sense, because the speaker’s

364

Alessandro Graheli

intention is a psychological state of the speaker that can be known only through inferences, not directly through speech, unless the speaker explicitly states something in the line of “I desire that […],” “I mean to say […],” “With this I intend to convey […],” and so on. There is another reason why for Jayanta this tātparya cannot be the speaker’s desire. He consistently maintains that intentionality is a distinctive force of words, and not of a sentient agent. He also unambiguously declares that the knowledge of a speaker’s desire is a product of an inference, and not of verbal testimony.6 In short, a sentient speaker’s desire to convey knowledge of a given thing can be known through an inference, but even this inference is necessarily based on the knowledge of this thing, which in turn is “that” (tat), the very sentence meaning known on the strength of the intentionality of words.7

3.1.1  “That” Is Not a Deontic Object but, rather, an Epistemic One In the ritual context of Mīmāṃsākas, “that” (tat) cannot possibly refer to the speaker’s desire. It rather denotes a deontic referent “that which ought to be done” (kārya), that is, the purpose of a Vedic ritual in the case of Vedic language.8 According to the Prābhākaras, on the other hand, the tātparya makes the primary significatory force itself capable of conveying not only the individual wordreferents, but their mutual connection as well. Thus even when tātparya is not taken as a separate vṛtti, it could be referred to as the motive force conveying the syntactic relation. (Kunjunni Raja 1963, 223)9

There certainly cannot be any speaker’s intention in a Vedic injunction, since the Veda is by definition authorless (apauruṣeya). In this context, Jayanta argues that Vedic sentences do not differ from mundane linguistic usages, insofar as both are uttered by a speaker, be it God or a human being. Consequently, even in Vedic sentences the anaphora of our tat is not the deontic “ought to be done,” which was devised by the Mīmāṃsākas as the referent of the sentence exactly to support their idea that the Vedas have no author. If the referent of the sentence is not deontic (kārya), it must then be an description of a state of affairs (prasādhita) (NM, NM 4, I 695,10–14).10 Jayanta here wants to abide by the Nyāya tenet that even the Veda has an author, namely God (īśvara). The Prābhākaras idea that Vedic injunctions are deontic in nature, and consequently their usage of the term tātparya with a tat as an anaphora of “what ought to be done,” is certainly not applicable to Jayanta’s position, particularly so because Jayanta’s introduction of the expression

The Intentionality of Words

365

tātparyaśakti occurs first in NM, NM 4, 696,1, in the context of the debate about the authorlessness of the Veda.

4  The Irrelevance of Indication or Suggestion If referentiality and intentionality are sufficient conditions for the apprehension of the sentence meaning, then there is no need of postulating other forces such as indication (lakṣaṇā) and suggestion (vyañjanā), as done by the Bhāṭṭa authors and the poeticians. Jayanta, indeed, never uses the term “indication” (lakṣaṇā) to describe his theory of sentence meaning, but he does use it elsewhere, and just like Kumārila he clearly distinguishes indication (lakṣaṇā) from secondary signification (gauṇatā). The following discussion occurs in Jayanta’s discussion on postulation (arthāpatti): [Opponent:] If there is just an all-powerful language with two distinct forces, referentiality and intentionality, how are primary (mukhya) usages to be distinguished from secondary and indicative ones?11 If there is no primary meaning the exegesis of the Veda becomes impossible. For instance, how could one apply principles such as “the previous is stronger” to solve conflicting evidences such as literal interpretation, inference, etc.?12 [Jayanta:] Even if the force of language is all-encompassing, there is no flaw, because these differences can easily be explained. It is not that once the causal force [i.e., Denotation] of words disappears the resulting cognition must disappear. That cognition [of word-referents, i.e., the primary meaning] is never completely abandoned; what follows is the variation of the original form.13 For instance, the cognition generated by the word “lion” in relation to the cub of a lioness is different from the cognition [of lion] in relation to Devadatta. And in the sentence “he bathes in the Ganges” the word “Ganges” acquires a different causality in respect to the sentence “the hamlet on the Ganges.”14

What emerges from this discussion is that Jayanta has little use for metaphors and other figurative tropes in his epistemological narrative. This could be a reason why he preferred to posit this new idea of a causal force of words called “intentionality,” rather than adopting the Bhāṭṭas’ principle of indication. From these examples, however, one can glean his awareness of the necessity to account for metaphorical and other oblique usages as well, perhaps in poetry or other situations where truth-values are clearly not in question.

366

Alessandro Graheli

Jayanta briefly evokes and dismisses Ānandavardhana’s theory of suggestion (dhvani). He points to the superfluity of postulating such a specific linguistic function and reduces poetic suggestions to the larger domain of sentence signification—namely, to the power of intentionality. The occasion of evoking dhvani is again the discussion around postulation.15 Another self-fancied scholar resorted to a so-called dhvani, which is also kept at bay by the mighty capacity (sāmārthya) of language. Knowledge of a prohibition comes from a request, and knowledge of a request from a prohibition, as in “Go freely, gentle monk” and “Do not enter the house, traveler.” According to the situation, there is just the power of the very words that teach the reality of things as it is known [by the speaker] through the other means of knowledge. Actually such a discussion with poets does not even look good. Even scholars are perplexed in the impervious path of sentence signification.16

In the light of his detailed treatment of compositionality, this “mighty capacity of language” encompassing poetical meanings must refer to the two forces of referentiality and intentionality, which are sufficient to explain the knowledge of sentence meanings.

5 Conclusion In sum, the term tātparya, in Jayanta’s expression tātparyaśakti, is intended by him as deriving from tatpara, where tat is quite simply the sentence meaning. The two forces of words are the causal capacities to bring forth knowledge of the individual referents and a unitary sentence meaning, respectively. The difference between the referents and the sentence meaning, reflected in the words and the sentence, is that while referents are individual entities, the sentence meaning is the combination (saṃsarga) of these. The individual referents are appraised in a hierarchical order (guṇapradhānabhāva), and are thus said to have a hierarchical relation (anvaya). The sentence meaning is nothing but the hierarchical order of the word-referents, which is known through the words occurring in that very hierarchical order. By extension, although Jayanta does not mention this aspect explicitly while discussing intentionality, one may assume an intentionality of sentences at supra-sentential level, in the hierarchical order of multiple sentences toward a composition of sentences. At the level of words, this intentionality produces knowledge of the hierarchical order of the referent of verb, the action (kriyāpadārtha), the factors of action (kārakapadārtha), and so on. This force, which presupposes the cognition of

The Intentionality of Words

367

the syntactic and semantic aspects at the level of word-referents, is in general imparted to words by God at the time of creation, but it can be stipulated at any time by human beings as well, for instance, when a new technical word is stipulated by a grammarian such as Pāṇini or a new verse is composed by a poet. Just like in the case of the referent of plain descriptive sentences, even that of injunctions, prohibitions, or other indirect expressions such as the poetical ones can be explained just by the postulation of intentionality, a necessary condition for the knowledge of the sentence meaning. When a speaker intends to express something that is not designated in what he actually says, this intention or desire cannot be known through the words themselves. In other words, for Jayanta this is a nonverbal aspect that cannot be known through words and their forces of referentiality and intentionality. According to Jayanta, verbal testimony functions through referentiality and intentionality only, and a message that cannot be conveyed and understood without knowing the speaker’s intention is not the epistemic object of verbal testimony. But Jayanta could not simply import Mīmāṃsā theories without modifications, because such views were rooted in assumptions that were incompatible with some basic Nyāya tenets: ●●

●●

●●

●●

●●

The idea of the Veda as authorless conflicts with the Nyāya foundation of testimony, which is based on the reliability of the source. A natural relation among words and referents is in contrast with the conventional relation of Nyāya. The notion that the referents of individual words are universals is opposed to the Nyāya concept of “property-possessor” (tadvat). The explanation of sentence signification through an inference of the correlation of word meanings, which would undermine the epistemological autonomy of śabda. The reduction of every statement to injunctions, particularly in the Prābhākara blend of Mīmāṃsā where both Vedic and common language are intrinsically injunctive, is not endorsed in Nyāya.

Notes 1 āptaḥ khalu sākṣātkṛtadharmā yathādṛṣṭasyādṛṣṭasya cikhyāpayiṣayā prayukta upadeṣṭā (NBh, ad 1.1.7). 2 See NBh, ad 2.1.55–56 for further details.

368

Alessandro Graheli

3 vākyārthas tu na kvacid api sūtrakārabhāṣyakārābhyāṃ sūcita iti kutaḥ śikṣitvā vākyārthasvarūpaṃ vayam ācakṣmahe. kim iti tābhyām asau na sūcita (sūcita] sūtrita MDUC 2606) iti cet pṛtahkprasthānā imā (-prasthānā imā] -prasthānā hīmā -prasthānāgamā MDUC 2606) vidyāḥ. pramāṇavidyā ceyam ānvīkṣikī na vākyārthavidyeti. yady evaṃ padārho ’pi kasmād iha darśitaḥ (darśitaḥ] pradarśito MDUC 2606) “vyaktyākṛtijātayas tu padārthaḥ” iti sthāne praśnaḥ. sa tu śabdānām arthāsaṃsparśitāṃ vadantaṃ bhadantaṃ (bhadantaṃ] rudantaṃ ca NM) śamayituṃ śabdaprāmāṇyasiddhaye sūtrakṛtā (sūtrakṛtā] om. MDUC 2606) yatnaḥ kṛtaḥ. yady evaṃ vākyārtham api bāhyaṃ (bāhyaṃ] sāhyāṃ MDUC 2606) vāstavam antareṇa śabda (śabda-] śāstrasya NM) pramāṇatā na pratiṣṭhāṃ labhata iti tatrāpi prayatnaḥ kartavya eva. satyam. padārthapratipādanayatnenaiva tu kṛtena tatra yatnaṃ kṛtaṃ manyate sūtrakāraḥ yad ayaṃ pṛthak padārthebhyo na vākyārtham upadiśati sma. tasmād ayam asyāśayaḥ padārtha eva vākyārthaḥ iti […] kiṃ tu naikaḥ padārtho vākyārthaḥ. anekas tu padārtho vākyārthaḥ. See also NM, 1, I.10,4, pṛthak prasthānā hi imā vidyāḥ. sā ca vākyārthavidyā, na pramāṇavidyā iti. 4 An historical account of the development of the concept was sketched by Kunjunni Raja 1963, 213–24. Prasad (1994) added valuable considerations, though focusing for the most on the later reception and implementation of the term. 5 For example, see Staal 1966, 308, “The term tātparya, lastly, refers to the speaker’s intention.” Cf. Tattvacintāmaṇi, vol. 4 (1), p. 327, NK, s.v. pratipādakecchaviṣayatvaṃ   tatparatvam / yaḥ śabdo vaktrā yadicchayā prayuktaḥ sa tatparaḥ //, and NSM, 84cd, vaktur icchā tu tātparyam. 6 E.g., see NM, I.696,4–5: kathaṃ tarhi puruṣavacanād uccāritāt vivakṣāvagama iti cet, anumānād iti brūmaḥ. 7 See NM, I.696,9–11: ayam artho ’syavivakṣita ity arthoparajyamānā tu vivakṣā na śakyā arthe ’navagate ’vagantum. arthaś cet prathamam avagato vākyān na tarhi tad vivakṣāparam. arthaparam eva bhavitum arhati. 8 For example, PrP, 378,9, tena tatraiva tātparyam, PrP, 400,12, tātparyāvagamād iti. 9 Prasad 1994, 321, perhaps in an attempt to find a synthesis of the two different applications, wrote that it is also used to mean “an intention, in the form of a demand, of a word in order to complete the sentence in which it occurs. Anything which is intended, such as explicit or inexplicit contexts, background assumptions, demands for incomplete expressions to be completed, and the volitional attitude or state of the speaker, is covered by intention (tātparya).” Moreover, “a theory of meaning is a theory of understanding, which takes ‘intention’ in its broadest sense. Therefore, the speaker’s intention is one of the meaning contents. But if he intends his utterance ‘to be understood’ by the hearer with all the meaning contents, then his intention must be taken in the broadest sense” (Prasad 1994, 341, note 22).

The Intentionality of Words

369

10 tatraitat syāl laukikavākyānāṃ vivakṣāparatvāt na kāryārthatvam. yathoktam. api ca pauruṣeyād vacanād evam ayaṃ puruṣo vedeti bhavati pratyayaḥ na tv evam artha iti. (na tv evam artha iti] naivam arthateti NM) vaidikāni punar apauruṣeyatayā kāryaparāṇy eva vākyānīti. etad api na peśalam, apauruṣeyasya vacasaḥ pratikṣiptatvāt, vede ’pi kartur īśvarasya prasādhitatvāt (prasādhitatvāt] sādhitatvāt NM). 11 Fujii (2001, 14) argued that in the NM Jayanta, after scrutinizing the two theories of abhihitānvaya and anvitābhidhāna, “advocates his own view, i.e. tātparyaśaktitheory. But here we must note that ANV [anvitābhidhāna] and ABH [abhihitānvaya] described in the NM are to some extent different from the ‘standard’ ones described in other texts such as Prakaraṇapañcikā […] the standard ABH adopts a metaphor (lakṣaṇā) as intermediate function through which the sentence meaning is indicated. Nevertheless, the ABH in NM lacks this sort of function” (that lakṣaṇā as used in the abhihitānvaya theory means “metaphor” is questionable). 12 nanv evaṃ sati sarvatra śabdavyāpārasambhavāt / mukhyasyāpi bhavet sāmyaṃ gauṇalākṣaṇikādibhiḥ // śrutiliṅgādimānānāṃ virodho yaś ca varṇyate / pūrvapūrvabalīyastvaṃ tat kathaṃ vā bhaviṣyati // 13 ucyate. saty api sarvatra śabdavyāpāre tatprakārabhedopapatter eṣa na doṣaḥ. na hi padānāṃ sarvātmanā nimittabhāvam apahāya eva naimittikapratītir upaplavate. tadaparityāgāc ca tatsvarūpavaicitryam anuvarttata eva. 14 anyathā siṃhaśabdena matiḥ kesariṇīsute / anyathā devadattādau pratītir upajanyate // gaṇgāyāṃ majjatīty atra gaṅgāśabdo nimittatām / upayāti yathā naivaṃ ghoṣādivasatau tathā // 15 On arthāpatti in relation to elliptic sentences, see Kunjunni Raja (1963, 169–74). 16 NM, I.129,7–130,2: etena śabdasāmarthyamahimnā so ’pi vāritaḥ / yam anyaḥ paṇḍitaṃmanyaḥ prapede kaṃcana dhvanim // vidher niṣedhāvagatir vidhibuddhir nisedhataḥ / yathā / bhava dhammiya vīsattho mā sma pāntha gṛhaṃ viśa (vl. viśaḥ MDUC 2606, 20r,5) // mānāntaraparicchedyavasturūpopadeśinām (vl. vasturūpāpadeśinām BORI 390, 37r,11) / śabdānām eva sāmarthyaṃ tatra tatra tathā tathā // athavā nedṛśī carcā kavibhiḥ saha śobhate / vidvāṃso ’pi vimuhyanti vākyārthagahane ’dhvani //

Primary Sources BORI 390

Pune, Bhandarkar Oriental Research Institute, Manuscript no. 390/1875–76. 432 foll. birch-bark, bound, Śāradā script.

Nyāyadarśana

Thakur, Anantalal, ed. 1997. Gautamīyanyāyadarśana with Bhāṣya of Vātsyāyana. Nyāyacaturgranthikā 1. New Delhi: Indian Council of Philosophical Research.

370

Alessandro Graheli

MDUC 2606

Thenjipalam (Malappuram District), Malayalam Department of the University of Calicut, Manuscript no. 2606. 188 foll. palm-leaf, Grantha-Malāyālam script.

NBh

Nyāyabhāṣya of Vātsyāyana. In Nyāyadarśana.

PrP

Subrahmanya Sastri, A., ed. 1961. Prakaraṇa Pañcikā of Śālikanātha Miśra with the Nyāya-Siddhi of Jaipuri Nārāyaṇa Bhaṭṭa. Banaras Hindu University Darśana Series 4. Benares: Benares Hindu University.

Tattvacintāmaṇi

Kamakhyanath Sharma, Tarkavagish, ed. 1990. The Tattvacintāmaṇi of Gaṅgeśa Upādhyāya. 4 vols. Delhi: Chaukhamba Sanskrit Pratishthan. Original edition Calcutta: Asiatic Society, 1888–1901.

Secondary Sources Fujii, Takamichi. 2001. “Jayantabhaṭṭa on Tātparyaśakti-Theory.” Journal of Indian and Buddhist Studies (Indogaku Bukkyogaku Kenkyu) 49 (2): 14–16. Jhalakīkara, Bhīmācārya. 1996. Nyāyakośa or Dictionary of Technical Terms of Indian Philosophy. Bombay Sanskrit Series 49. Or. ed. 1875. Pune: Bhandarkar Oriental Research Institute. Kunjunni Raja, K. 1963. Indian Theories of Meaning. Madras: Adyar Library and Research Centre. Prasad, Hari Shankar. 1994. “The Context Principle of Meaning in Prābhākara Mīmāṃsā.” Philosophy East and West 44 (2): 317–46. Staal, Jan Frederik. 1966. “Indian Semantics, I.” Journal of the American Oriental Society 86 (3): 304–11.

20

Rasa as Sentence Meaning1 Andrew Ollett

Rasa is a technical term which refers to a particular kind of emotional experience. It is generally reserved for emotions associated with works of art—that is, the emotions that are either represented in a work of art or according to the theory that would later come to dominate South Asian aesthetic thought, the emotions that a work of art engenders in the sensitive reader, spectator, or listener. The major discussions of rasa among intellectuals who wrote in Sanskrit have recently been presented in Sheldon Pollock’s Rasa Reader, which also frames those discussions in an intellectual-historical narrative (Pollock 2016). Many of the authors who have written about rasa have described it as a “meaning,” an arthaḥ, of a literary text. A few of them, namely Bhaṭṭa Nāyaka (late ninth century) and Dhanika (late tenth century), have described it more specifically as a kind of “sentence meaning,” or vākyārthaḥ. It is this equation, rasa as sentence meaning, that is the focus of this chapter: Precisely what does it mean to call rasa a sentence meaning, and what conceptual resources were drawn upon to make this equation? In the most basic sense, what we call the “meaning” of a linguistic expression is the content of the awareness that is produced by that linguistic expression. Hence, if I show you the word “cow,” you are likely to think of a cow. Rasa is, for its part, a kind of emotional experience, or perhaps more generally an affective state. As Pollock has demonstrated in his Reader, there were two competing positions, not always clearly articulated or distinguished from each other, regarding the locus of this affective state: Does rasa exist in the character, or in the reader? Thus we might consider rasa to be the meaning of a text either if it is an affective state in the character that the reader understands on the basis of the text or, alternatively, if it is an affective state in the reader herself that, once again, is produced on the basis of the text. Either way, however, there are certain problems associated with understanding rasa as a meaning. At the very

372

Andrew Ollett

least, if the equation is to hold, we are going to need a more sophisticated theory of meaning than a semiotic one which merely pairs linguistic expressions with mental states. The main arguments against thinking of rasa as a meaning in the narrowly semiotic sense were already clearly enunciated by Ānandavardhana, around 875 CE, in his revolutionary monograph, the Light on Resonance (Dhvanyālokaḥ) (see Pollock 2016, 90). Ānandavardhana himself did consider rasa—or more precisely rasādiḥ, a large set of elements that form part of the overall experience of a rasa, including intermediate affective states, psychophysical responses, and so on— to be a meaning, but it was precisely the inexpressibility of such a meaning through the standard “language functions” (śabdavyāpāraḥ) that led him to posit manifestation (vyañjanā) as an additional function. Ānandavardhana’s argument can be paraphrased as follows. Whereas a linguistic expression expresses its meaning, it does not seem to be the case that a literary text expresses a rasa. This is for two reasons, one pertaining to the supposed signifier, and one to the supposed signified. To begin with, we don’t seem to owe our awareness of a rasa in the character—and, as Pollock (2012) has argued, Ānandavardhana thinks of rasa primarily as located in the character rather than in the reader—to linguistic expressions as such. Ānandavardhana claimed, somewhat controversially, that one cannot successfully convey a rasa simply by using the words for it, for instance, “amazing,” or “heroic,” or “erotic,” and so on.2 Showing is more effective than telling. Authors instead speak of an aggregate of “aesthetic elements,” comprising foundational factors (ālambanavibhāvaḥ), stimulant factors (uddīpanavibhāvaḥ), reactions (anubhāvaḥ), psychophysical responses (sāttvikānubhāvaḥ), and transitory emotions (vyabhicāribhāvaḥ), which jointly raise a particular stable emotion (sthāyibhāvaḥ) to the level of a rasa. But if the cause of rasa is an aggregate, then it is no longer similar in this respect to a word meaning, which is expressed by a single word. It may, however, be similar to a sentence meaning, given that the awareness of sentence meanings arises, according to one view, from an aggregate of word meanings, just as the awareness (or experience) of rasa arises from the aesthetic elements in aggregate. The analogy of sentence meaning might also help to solve another problem: if rasas really are produced by an aggregate of aesthetic elements, given the fact that the aesthetic elements are not linguistic expressions, either individually or in aggregate, is it really possible to say that the aesthetic elements “express” a rasa? Surely, however, those aesthetic elements are themselves expressed by linguistic expressions, just as word meanings, out of which sentence meaning

Rasa as Sentence Meaning

373

arises, are expressed by linguistic expressions. Hence we may want to say that the literary text expresses a rasa in the same way that the words of a sentence express a sentence meaning. On the side of the supposed signified, there appears to be an important qualitative difference between rasa and other kinds of meanings. What, after all, are the kinds of things that can be signified by a signifier, and especially by a linguistic expression? To begin with word meanings, we might think of them as intentional objects, and it is clear that we don’t experience intentional objects and affective states in the same way.3 Similarly, one might think of sentence meanings as either structured configurations of such intentional objects or something along the lines of propositions, which can be evaluated in truth-conditional terms.4 In either case, rasa is evidently something quite different, neither analyzable into intentional objects nor capable of evaluation as true or false. These two concepts of sentence meaning, however, might be unduly restrictive. If we adopt a view of sentence meaning as more directly related to action and experience, as several South Asian theorists did, then the gap between rasa and sentence meaning will shrink. Thus, on the one hand, it does not seem possible to understand rasa as a meaning, if meaning is defined solely as what is literally expressed by a signifier. And for precisely this reason, theorists have favored verbs other than “express” to describe the relationship between the text and a rasa. Ānandavardhana led the way by claiming that a rasa could only ever be manifested (vyakta-). In his wake, Bhaṭṭa Nāyaka would claim that rasa is actualized (bhāvita-), and Mahima Bhaṭṭa would claim that it is inferred (anumita-). On the other hand, none of these thinkers denied that rasa is a meaning, and it might be possible to defend this equation if we entertain a different theory of meaning. As we have seen, the notion of sentence meaning offers a more promising parallel than that of word meaning. In the debate that followed the Light on Resonance, the defenders of Ānandavardhana argued that rasa was a kind of meaning that could not be expressed, and therefore had to be manifested, while his opponents argued rasa could be expressed (see McCrea 2008). This debate unfolded over the course of about a century and a half, from 875 to 1025 CE. During this time, there were a number of interventions in the theory of rasa itself. The most critical of these interventions, as Pollock has argued, was Bhaṭṭa Nāyaka’s insistence that rasa was an experience that the reader had in response to a compelling work of literary art (see Pollock 2010, 2016, 144–80). There were also, however, a number of interventions in the theory of sentence meaning. Between the

374

Andrew Ollett

late ninth and early eleventh century, the traditional theories of meaning were radically overhauled, revised, and elaborated. In this debate, then, there were two sets of moving parts. The answer to whether rasa could be thought of as expressed, either in the same way that a sentence meaning is expressed or, indeed, as a sentence meaning itself, thus depended on the models of sentence meaning that were available. The three authors who argue most explicitly for the equation of rasa and sentence meaning are Bhaṭṭa Nāyaka, Dhanañjaya, and Dhanika. Bhaṭṭa Nāyaka probably composed his Mirror of the Heart (Hṛdayadarpaṇam), which only survives in fragments, in Kashmir around 900 CE. Dhanañjaya wrote a work of dramaturgy called the Ten Dramatic Forms (Daśarūpakam), probably while at the court of Vākpati Muñja in Malwa in the last quarter of the tenth century. Dhanika was likely Dhanañjaya’s younger brother. As far as we can tell, his Observations (Avalokaḥ) on the Ten Dramatic Forms were composed in the second decade of the eleventh century.5 Pollock has argued convincingly that Dhanañjaya and Dhanika were deeply influenced by Bhaṭṭa Nāyaka, and in his Reader, he has treated their works as an “elaboration” of Bhaṭṭa Nāyaka’s ideas (Pollock 2016, 154; see also Pollock 2010). Pollock did not, however, discuss another one of Dhanika’s sources, who was as important in shaping Dhanika’s views on sentence meaning as Bhaṭṭa Nāyaka was in shaping his views on rasa. This was Bhaṭṭa Jayanta. Bhaṭṭa Jayanta was an exact contemporary, and countryman, of Bhaṭṭa Nāyaka. He, too, worked in Kashmir in the later part of the ninth century. His magnum opus was Racemose Reasoning (Nyāyamañjarī), a wide-ranging discussion of various aspects of the Nyāya philosophical system, and notable for its inclusion of many other philosophical perspectives, including those of Mīmāṃsā. In the book’s sixth chapter, Jayanta discusses two competing theories at length, one called “relation of expressed meanings” (abhihitānvayaḥ), and the other “expression of relational meanings” (anvitābhidhānam). These two positions were associated with two towering figures in the history of Mīmāṃsā: Kumārila Bhaṭṭa was held to be a partisan of the “relation of expressed meanings,” although he himself never used the phrase, whereas Prabhākara self-consciously defended the “expression of relational meanings.” If we look beyond Jayanta’s doxographic presentation, and toward the intellectual-historical landscape, we can see that Kumārila represented the older and common-sense view, while Prabhākara’s insistence that words could only express their meanings in the context of the sentence, and thus in relation to other word meanings, was a radical challenge to that view. It was in response

Rasa as Sentence Meaning

375

to that challenge that philosophers began to address the question of sentence meaning directly. One of the first to do so was Śālikanātha Miśra, Prabhākara’s commentator and interpreter, who wrote a seminal essay on sentence meaning in the eighth or ninth century (see Chapter 13 in this volume). Another was Jayanta himself. His discussion of sentence meaning in Racemose Reasoning is clearly responding, in the first instance, to the radical contextualism of Prabhākara, and his final position is quite close to Prabhākara’s. “I don’t accept the ‘expression of relational meanings,’” he writes toward the end of his discussion, “but if you want to call it ‘the conveyance of relational meanings,’ go ahead.”6 He covers much of the same ground that Śālikanātha had, although there is no clear evidence that he had read Śālikanātha’s writings. At the time that Jayanta was writing, say around 900 CE, the contextualism of Prabhākara was just beginning to make an impact. Earlier theorists of literature, such as Udbhaṭa (late eighth century), had gotten their Mīmāṃsā, and therefore their theories of sentence meaning, from Śabara and Kumārila. This is true, as far as we know, of Ānandavardhana as well, and also of Jayanta’s contemporary, Bhaṭṭa Nāyaka. The situation changes completely after Jayanta, who quickly became a point of reference for many intellectuals regarding the theory of sentence meaning. Jayanta had not only provided a clear presentation of the two competing theories of sentence meaning. He also provided a theory of his own as a kind of compromise. The most important and recognizable aspect of this theory is his introduction of a “power of purport” (tātparyaśaktiḥ), alongside the well-established “power of expression” (abhidhātrī śaktiḥ). The “power of purport” is a term that Jayanta invented, and his trademark contribution to the discussion, although scholarship has not always appreciated this point.7 Earlier authors had used “purport” (tātparyam), but in a slightly different sense. Literally, the word means “the fact of having that as its main purpose.” Mīmāṃsakas had always held that the main purpose of language, at least the language of the Vedas, is for people to perform the rituals described therein. Prabhākara and Śālikanātha, especially, maintained that an awareness of the practical purpose served by a particular sentence is a prerequisite for understanding that sentence’s meaning, which proceeds in a top-down rather than bottom-up manner. They argued that the meaning of a word is a relational meaning, that which a word contributes to a hierarchically organized structure whose main purpose (tātparyam) is some specific action, rather than its nonrelational meaning, which is the mere association that the word has when uttered independently of any particular sentence. For Jayanta, the purport was

376

Andrew Ollett

the overall meaning of a sentence, that which the speaker ultimately intended to communicate. Yet, because the expressive power of words, he maintained, was exhausted in communicating their non-relational meanings, he thought that there must be an additional power that is responsible for conveying the relation of the word meanings to each other, and hence the speaker’s purport. This he called “the power of purport.” He claimed that two powers, that of expression and that of purport, cooperated in conveying the meaning of a sentence in a two-stage process. In the first stage, the “power of expression” (abhidhātrī śaktiḥ), is responsible for expression as traditionally conceived, namely, calling a signified to mind in the presence of a signifier: To the domain of the “power of expression” belong not relational meanings, but the proper meanings of words that have been isolated through positive and negative concomitance.8

The “proper meanings” he mentions here are clearly non-relational. They are similar to dictionary definitions. By positive and negative concomitance, that is, by looking at the contribution that a particular word makes to a sentence meaning across all of its occurrences, we can determine a stable semantic value for that word. At the end of the first stage, we have word meanings such as these, and perhaps some idea as to how they can be fit together—since some grammatical relations are explicitly expressed, for example, by affixes—but no sense of an overarching structure of meaning into which they can be integrated. In the second stage, the “power of purport” (tātparyaśaktiḥ) supplies precisely this overarching structure: The “power of purport,” by contrast, extends to relational meanings, since it operates together with the power of expression, and its operation extends to the production of an awareness that has no unresolved dependencies.9

Jayanta is here referring to the traditional definition of the sentence in Mīmāṃsā, which is defined by the unity and independence of its meaning: within a complete sentence, any dependency that one word has for another, such as a transitive verb for a direct object, is resolved in that very sentence, whereas if the sentence is incomplete, at least some of those dependencies will remain unresolved.10 Independence, in terms of having no unresolved dependencies, is a characteristic of sentence meaning. Since the power of expression is blind to the dependencies of the meanings that are expressed—as Jayanta says in the following summary verse, the meanings it expresses are “either complete or

Rasa as Sentence Meaning

377

incomplete”—an additional power is required, in Jayanta’s view, to account for this characteristic: Cognitions that arise from other sources of knowledge, and that present a stable meaning that is either complete or incomplete, operate in one way; completely different is the verbal cognition that extends over its objects, terminating in the production of an awareness that has no further dependencies on the part of the listener.11

Having very briefly examined Jayanta’s theory of sentence meaning, we can now return to literary theory. We will compare those fragments of Bhaṭṭa Nāyaka’s Mirror of the Heart that speak of rasa as a sentence meaning with the celebrated presentation of rasa as a sentence meaning in Dhanañjaya’s Ten Dramatic Forms and especially Dhanika’s Observations. As noted above, there is a clear line of influence from Nāyaka to the sons of Viṣṇu. But they are nevertheless separated by a century. Jayanta’s influence is very clear on Dhanañjaya and Dhanika, whereas Bhaṭṭa Nāyaka’s fragments bear no trace of the influence of Jayanta, or indeed of Prabhākara, who began the contextualist revolution. The question of what conceptual resources Bhaṭṭa Nāyaka drew upon, or even what resources he was in a position to draw upon, is a difficult one, and even after Sheldon Pollock’s painstaking reconstruction of the major themes of Bhaṭṭa Nāyaka’s thought, only provisional answers are possible. It is, nevertheless, important to appreciate Jayanta’s influence on Dhanañjaya and Dhanika, because it allowed them to conceive of sentence meaning, and thus of rasa, in a way that was evidently not available to Bhaṭṭa Nāyaka.12 It may be that this is a distinction without a difference. Bhaṭṭa Nāyaka certainly did have a concept of sentence meaning, even if he didn’t call it “purport.” When Dhanika invokes “the power of purport” in support of his claim that rasa is a sentence meaning, is he really saying something all that different from Bhaṭṭa Nāyaka? I have to admit that I myself see Jayanta’s interventions in the theory of sentence meaning as largely terminological rather than conceptual. Nevertheless, there are two reasons for focusing on these seemingly terminological differences. The first is the philosophical significance of the theory that is at stake. Bhaṭṭa Nāyaka, using whatever resources were available to him, made one attempt to develop of a theory of meaning that could accommodate affective states. A century later, and using a new set of resources, Dhanañjaya and Dhanika took another pass. Were either of them successful? The second is my suspicion that what really distinguishes Dhanañjaya and Dhanika from Bhaṭṭa Nāyaka is not necessarily Jayanta’s calling card, the

378

Andrew Ollett

power of purport, but a more general insight that Jayanta himself had taken over from Prabhākara: language is only language in use, and in order to be used, there must always be something for language to do (kāryam). Whereas Bhaṭṭa Nāyaka invoked the special power of the Vedas to get its listeners to actualize its meanings, Dhanañjaya and Dhanika merely need to remind us that in every single use of language—whether in the Vedas, or in literature, or in everyday life—there is always something to be done. Bhaṭṭa Nāyaka believed that rasa was an experience that the reader had in response to the artful presentation of aesthetic elements in the literary text. He could think of rasa as similar to sentence meaning for two reasons. First, he borrowed from Mīmāṃsā—and probably more specifically from Kumārila Bhaṭṭa—the idea that a sentence meaning is fundamentally oriented toward action. Suppose someone hears the sentence “One who desires heaven should sacrifice with the full- and new-moon sacrifices.” If he recognizes himself as one who has both the desire for heaven and the entitlement to perform the ritual, he will be motivated to undertake it. And when he performs the ritual, he “actualizes” the result that is described in the sentence. In the same way, when one hears a work of literature, provided some additional conditions are met, he “actualizes” the rasa by experiencing it. Ritual and literary texts therefore issue, in similar ways, in the “actualization” of something that is encoded textually. And in both cases, the thing that is actualized is the key to the meaning of the text—the purpose to which all of the other elements are subordinated. If one conceives of meaning as a blueprint for actualization of this sort, rather than merely the content of a cognition, then it seems possible for rasa to be the meaning of a text.13 Two verses of Bhaṭṭa Nāyaka’s appear to make this point. The first is fragment #13, according to Pollock’s numeration: Rasa, manifested by the configuration of aesthetic elements, the object of a supreme awareness, and an experience consisting in savoring, is said to be the “meaning of the literary text.”14

It is not entirely certain that this verse is Bhaṭṭa Nāyaka’s. I will not go into its textual and contextual problems, which Pollock have been discussed at length in Pollock 2012. My reading and translation differs from his in a number of details.15 Most notably, he translates artha- as “purpose.” Against the background of Mīmāṃsā, this word may well have the sense of “purpose” as well as “meaning.” We can find two clues in this verse, however, that suggest that Nāyaka actually did claim that rasa is the meaning, and not just the purpose, of a literary text.

Rasa as Sentence Meaning

379

First, at least in the version transmitted by Mahima Bhaṭṭa, there is the claim that rasa is manifested by a particular configuration of aesthetic elements. Bhaṭṭa Nāyaka argued that what Ānandavardhana called “manifestation” was nothing more than what he called “expression,” but he was evidently happy to continue using the term. To say that rasa is manifested by the aesthetic elements is to say that we, as readers, owe our awareness of rasa to these elements. There is a parallel here with a traditional account of sentence meaning, according to which it consists in a conjuncture (saṃsargaḥ) of word meanings. Although only implicitly referenced by the word “configuration” (saṃyojanā) here, it will be explicitly referenced in the next fragment. Hence both rasa and sentence meaning share a kind of compositionality, in the weak sense that the character of the whole is determined by the character of its parts. Second, it appears to contain the claim that rasa, like meaning, is the object of an awareness. I read the end of the first line, somewhat counterintuitively I admit, as an endocentric compound (“an object of awareness”) rather than an exocentric compound (“its object is awareness”). That is first of all because the first half of the line refers to the manifestation of rasa, and what is manifested can thereby become an object of awareness. Secondly, however, I understand the three qualifications of rasa in this verse to correspond to the three stages of aesthetic experience in Bhaṭṭa Nāyaka’s theory. First, it is “expressed” by the literary text; second, it is “actualized” when the reader apprehends it by means of a special kind of awareness in which the differences between the apprehending subject and represented object are neutralized; and finally, it is experienced by the reader as a particular kind of affective state. When read this way, the verse addresses what we had earlier noted was a major problem with conceiving of rasa as a meaning: rasa is an experience, whereas meanings are cognitive objects. Rasa is a cognitive object at the beginning of the aesthetic process, and an experience at the end. Finally, to say that rasa is the “meaning” of an entire text may seem somewhat strained, in comparison to the claim that rasa is the “purpose” of the text, in the sense of the goal which the poet strives to achieve throughout the text. But the difference between the meaning of a sentence and the meaning of an entire text is quantitative rather than quantitative. Mīmāṃsā is clear that any textual unit, insofar as it is really a “unit,” has the same hierarchical organization that characterizes the sentence. Mīmāṃsā licenses us, in determining the meaning of a particular linguistic expression, to look beyond it, in ever-expanding concentric circles, until all of the dependencies are resolved, and in particular,

380

Andrew Ollett

its requirement for an overarching purpose is met.16 The purpose of a text is indeed its meaning. In another verse of Bhaṭṭa Nāyaka’s, numbered #12 by Pollock, the parallelism between rasa and sentence meaning is made crystal clear: Just as a sentence meaning in the Veda is thought to be the conjuncture, or whatever one wishes to call it, of word meanings—since, in view of its connection with a result, there must only be a single meaning for a given sentence—so too are the rasas, such as the erotic, here in literature.17

My translation reflects my interpretation of this verse as an analogy between rasa and sentence meaning, rather than an identification of them. Rasa in literature is like sentence meaning in the Veda. In both cases, they are constituted out of a “conjuncture” of other elements. These are the aesthetic elements in the case of rasa and word meanings in the case of sentence meaning. The verse gives a reason why each of them is thought of in this way, which requires a bit of explanation. One popular definition of a sentence was a “group of words,” and sentence meaning could be defined along similar lines as a “group of word meanings.” Why, after all, should we refuse to identify a sentence meaning with its constituent word meanings, and instead opt for something above and beyond the word meanings, such as their “conjuncture”? The reason is that a sentence meaning has to be single. Thus the conjuncture of word meanings is a possible candidate, whereas the word meanings themselves, because of their plurality, are not. And why does sentence meaning have to be single? Mīmāṃsakas maintained that the actualization of the meaning of a sentence—in effect, the performance of the sacrifice that a Vedic sentence enjoins—resulted in the production of one and only one result. The result may be final, such as heaven, or rain, or a son, or cattle, or it may be intermediate, helping to achieve a final result through some sacrificial procedure or another. But the sentence provides a blueprint for action, and that action is characterized by the production of a single result. The case of rasa is similar: although rasa is communicated through multiple elements, it is nevertheless said to be single, because the result of “actualizing” the rasa is also single, namely, the aesthetic experience itself. We can make a few observations about this verse. First, it is intentionally vague about what we should actually identify with sentence meaning, no doubt reflecting Bhaṭṭa Nāyaka’s awareness that thinkers had advanced different candidates. One ancient grammarian, Vājapyāyana, had identified sentence meaning with the conjuncture (saṃsargaḥ) of word meanings, and another, Vyāḍi, had identified it with their mutual difference (bhedaḥ) (Kunjunni Raja

Rasa as Sentence Meaning

381

1963, 191–92). Second, despite this show of agnosticism, the reasons that Bhaṭṭa Nāyaka adduced make it clear that the theory of sentence meaning he has in mind comes from Mīmāṃsā. More specifically, it probably comes from Kumārila Bhaṭṭa, who had actually identified sentence meaning with the actualization (bhāvanā) of a particular result (phalam), with which both of the earlier theories, conjuncture and difference, are compatible.18 In Kumārila’s thought, the result is the most important element in the structure of sentence meaning. The same is not true in Prabhākara’s interpretation. He had held that the result was but a subsidiary element of sentence meaning, the principal element being an obligation (niyogaḥ) or “something to be done” (kāryam). This suggests that Bhaṭṭa Nāyaka either did not know of Prabhākara or ignored him on this topic. The fact that he refers to sentence meaning as “conjuncture and so on” is another piece of evidence, even more circumstantial, for the same conclusion: after Prabhākara, and especially after his commentator Śālikanātha, almost everyone spoke of the “relation” (anvayaḥ) between word meanings, and the earlier term, “conjuncture,” came to have an old-fashioned ring. Finally, Bhaṭṭa Nāyaka analogizes rasa to sentence meaning, not in general, but specifically in the Veda. This is probably because the model he invokes was designed specifically for Vedic texts. By contrast, the model developed by Bhaṭṭa Jayanta, building on the insights of Prabhākara, was intended to apply to all sentences, whether Vedic or not. Hence Bhaṭṭa Nāyaka understands rasa in a way that presupposes an acceptance of a hermeneutics especially tailored to the Veda. This maneuver put a Jain author like Hemacandra in an awkward position: he wanted to present and defend Bhaṭṭa Nāyaka’s theory without relying on the example of Vedic hermeneutics.19 Bhaṭṭa Nāyaka here shies away from the claim that rasa is actually a sentence meaning. Instead, he claims that rasa is like a sentence meaning in a number of respects. Although it is itself singular, the awareness of it is produced by a conjuncture of different elements. These elements—the aesthetic elements in the one case and the word meanings in the other—are themselves communicated through linguistic expressions. It is, moreover, organized around the actualization or production of something, which is the object for the sake of which the ritual is performed in the case of Vedic sentences, and the aesthetic experience itself in the case of literature. In spite of these similarities, and in spite of Bhaṭṭa Nāyaka’s identification of rasa as “the meaning of the literary text” in the previously discussed fragment, a number of conceptual obstacles remain for the equation of rasa and meaning. Some of them have to do with the fact that Nāyaka, in offering a reader-

382

Andrew Ollett

centered and phenomenological account of rasa, broke decisively from an earlier tradition that Pollock has usefully characterized as “text-centered” and “formalist.”20 Thus, Bhaṭṭa Nāyaka might have responded to Ānandavardhana’s argument that rasa cannot be expressed by saying that rasa unfolds in a process wherein it is first expressed and then experienced. In that case, however, the meaning that is expressed in the first stage is not really rasa, and the rasa that is experienced in the second stage is not really a meaning. Moreover, if what the text means is simply defined as that which the reader experiences, then is there any principled distinction at all between meaning and response? Is it really equivalent to say “this is what the text means” and “this is what it means to me”?21 Finally, it may be that some of the conceptual difficulties don’t have to do with the application of rasa to the model of sentence meaning, but with the model of sentence meaning itself. One of the reasons why Prabhākara’s theory received so much attention in the ninth and tenth centuries was because of certain weaknesses in Kumārila’s theory: if sentence meaning really were to arise from the conjuncture of word meanings, then it would not be a “meaning” at all, in the sense of being conveyed by a linguistic expression, but the result of a secondary cognitive process (see Chapter 18 in this volume). Similarly, if rasa arises from the conjuncture of aesthetic elements, it is not really expressed by the text, but rather inferred from or suggested by the aesthetic elements themselves, and thus not a “meaning” of the text. These are some of the difficulties that Bhaṭṭa Nāyaka’s followers might have attempted to address, and in doing so, they might have had recourse to alternative theories of sentence meaning. In his Observations on the fourth chapter of Dhanañjaya’s Ten Dramatic Forms, Dhanika provides a summary of Bhaṭṭa Nāyaka’s views, and in doing so, he enlists the help of Bhaṭṭa Jayanta, although he never uses either of their names. Jayanta’s characteristic idea is that sentence meaning arises at the end of a two-stage process, driven, in the final stage, by language’s “power of purport.” He had not failed to observe that the phenomena that Ānandavardhana ascribed to “manifestation” could be accounted for otherwise, either by his own “power of purport” or, in any case, by a more expansive view of what language in general could do: “This expansive capacity of language also rules out that particular ‘resonance’ which someone else, thinking himself a scholar, had proclaimed.”22 In this connection, he cited two examples.23 The first is the famous Prakrit verse that Ānandavardhana had quoted at the beginning of his presentation of “resonance,” which begins “Go your rounds freely, gentle monk” (bhama dhammia vīsaddho). Ānandavardhana found this example serviceable because

Rasa as Sentence Meaning

383

the literally expressed meaning is a command (“go your rounds freely”), whereas the meaning that the reader ultimately understands is a prohibition (“leave immediately”), and he thought that his newly proposed function of language, manifestation, could account for the difference. The second example was not previously quoted by Ānandavardhana. Bhaṭṭa Jayanta seems to have introduced it into the debate. “Don’t enter that house, son. Eat poison instead.” Here the literally expressed meaning is a command to eat poison, but the reader knows not to “take it literally,” as we say. Ānandavardhana might have claimed that this example involves the suggestion of a figure of speech, namely distinction (vyatirekaḥ): going into the house and eating poison are both stupid and self-destructive things to do, but the former is even more so than the latter. Jayanta would claim that the meaning that is “suggested” or “manifested” is nothing other than the overall sentence meaning that is produced, in the second stage, by the power of purport. In his Observations, Dhanika provides a synopsis of the theory of manifestation in the voice of a proponent of that theory. Ānandavardhana himself, living about a generation before Bhaṭṭa Jayanta and apparently unaware of the contextualist theories on which Jayanta’s own position was based, did not anticipate the objection that the overall sentence meaning could, in fact, be identified with what he called the “manifested” meaning instead of what he called the “expressed” meaning. More specifically, using the two-stage model that Bhaṭṭa Jayanta had pioneered, the overall sentence meaning—“leave immediately,” and “entering that house is worse than eating poison”—arises in a second stage, which we might call an “all things considered” stage, after the words had conveyed their proper meanings in a first stage. The proponent of manifestation, ventriloquized in Dhanika’s Observations, presents this new objection and attempts to respond to it. The issues here are slightly beyond the focus of this chapter, but they are, first, whether the second example (“eat poison”) should be considered a case of manifestation by Ānandavardhana’s followers, and second, what, if any, differences exist between the meaning that arises in Bhaṭṭa Jayanta’s second stage, in which the power of purport operates, and the meaning that Ānandavardhana’s followers, in the wake of Jayanta, ascribe to a third stage, in which manifestation operates. The proponent of manifestation claims that, in the first example (“wander freely”), the literal meaning on its own constitutes a coherent sentence meaning, and the power of purport “rests” at this meaning. The speaker may well be telling a monk to wander freely. An additional meaning nevertheless arises (“leave immediately”), which can only be ascribed to a third stage beyond that of sentence meaning. By contrast, in

384

Andrew Ollett

the second example (“eat poison”), the power of purport cannot rest on the literal meaning, because it does not constitute a coherent sentence meaning. Thus the nonliteral meaning (“entering that house is worse than eating poison”) is selected as the sentence meaning within the second stage, without the need for a third stage. This, according to the proponent of manifestation, accounts for why the first example, and not the second, exemplifies Ānandavardhana’s idea of “resonance.” Dhanika quotes two verses from this author, who is otherwise unknown (Observations, 206–07). While this discussion has little to do with rasa, the proponent of manifestation is nevertheless made to say that rasa could never be a sentence meaning. This is because in the second stage, where sentence meaning arises, the reader does not understand a rasa, but rather understands the various aesthetic elements that will, in turn, manifest the rasa.24 This is a powerful argument against the equation of rasa and sentence meaning: the same process cannot be responsible both for conveying an understanding of the sentence meaning and for conveying rasa, because the apprehension of rasa is dependent on certain features of the sentence meaning, namely, the factors (vibhāvāḥ) and the other aesthetic elements. Dhanika responds to these arguments in two places. First, in a lost work of his, the Analysis of Literature (Kāvyanirṇayaḥ), he attacked the distinction between sentence meaning and manifestation that the proponent of manifestation had tried to maintain. In that work, of which he quotes a few verses in his Observations, he followed Jayanta in claiming that the powers of expression and purport alone were sufficient to account for anything that we might like to call a meaning, whether in everyday language or in a literary text. It is not the case that the power of purport comes to rest in a literal meaning in the first example (“go your rounds freely”), and does not do so in the second example (“eat poison”). In both cases, we are carried beyond the literal meaning by a careful consideration of the context. Second, when commenting on an important verse of Dhanañjaya’s Ten Dramatic Forms, Dhanika makes a compelling argument for understanding rasa itself as a sentence meaning—the type of meaning that is conveyed by the power of purport. Dhanañjaya’s verse is as follows: Just as an action—whether directly expressed or understood as being present by virtue of context or some other factor—constitutes sentence meaning when construed with its factors, so a stable emotion constitutes a sentence meaning when construed with the other aesthetic elements.25

Rasa as Sentence Meaning

385

Pollock has noted that this verse is an elaboration of a verse of Bhaṭṭa Nāyaka’s that was discussed earlier (fragment #12). Whereas that verse, however, identified sentence meaning directly with the conjuncture of word meanings, or something that played a similar role, this verse focuses on the conjoined elements themselves, namely, the action and its factors (e.g., its agent and patient). The relation between an action and its factors has often been taken to be paradigmatic of the relation that holds between word meanings within sentence meaning, however and in whatever terms that relation was conceived (Vājapyāyana’s “conjuncture” [saṃsargaḥ], Vyāḍi’s “difference” [bhedaḥ], Prabhākara’s “cross-connection” [vyatiṣaṅgaḥ] or “relation” [anvayaḥ]). Some authors argued that every single sentence meaning must include a relation of this form.26 There was a debate about whether the relation between elements of meaning needs to be conveyed separately from those elements of meaning themselves, or whether it is sufficient for grasping the relation to grasp the relata, but I am not sure that Dhanañjaya meant to take a position in this debate.27 Dhanañjaya’s verse claims that the aesthetic elements are conveyed to the reader in exactly the same way as word meanings. In fact, they are precisely word meanings. They may not be literally expressed by words in the text, although Dhanika mentions one example where they are: “my beloved bride becomes even more beloved” from Harṣa’s Nāgānanda.28 That is to say, they are not necessarily cognized in the first stage. If they are present at all, however, they are inevitably cognized by the second stage, in which the power of purport produces a comprehensive sentence meaning. Even when it is not literally expressed, Rāma’s love for Sītā, for example, can be conveyed as a meaning of the text in precisely the same way that the action of closing is conveyed as a meaning of the sentence “the door, please.” Note how much the ground has shifted in the century between Ānandavardhana and Dhanañjaya. For Ānandavardhana, it probably would have sounded ridiculous to say that a meaning is expressed at all if it is not literally expressed by a word. That is why he had to invent an entirely new concept of “manifestation” in the first place. Yet, in the wake of Prabhākara, Śālikanātha, and Jayanta, contextualism had become much more prominent in debates around sentence meaning. The meaning conveyed by a sentence, on this view, incorporates smaller elements of meaning of all kinds, both those that are literally expressed, and those that are contextually understood, and hence in order to arrive at a coherent sentence meaning, we crucially depend on the context within which a certain sentence is spoken. Bhaṭṭa Nāyaka, too, might

386

Andrew Ollett

have had a contextualist account of sentence meaning in mind, but it is telling that he does not refer to context at all in the available fragments. So much for the aesthetic elements. Dhanika then argues that the rasa which emerges from the mutual relation between these elements is not merely like sentence meaning but is in fact a sentence meaning of a particular kind. His argument relies heavily on Jayanta’s concept of purport. Here is what he says: Don’t go saying that something which is not actually the meaning of any of the constituent words cannot be the sentence meaning. For the power of purport culminates in something to be done. Let me explain. Every single sentence, whether man-made or not, is oriented towards something to be done. If that were not the case, then we would simply disregard it, like the words of a madman.29

The notion that every sentence culminates in something to be done is particularly associated with Prabhākara and Śālikanātha. When we actually map out a sentence meaning, they argued, this practical element is inevitably at the top, even if it is not literally expressed by something in the sentence—for example, an imperative or optative verb. All of the other elements of meaning must take a subordinate position with respect to it. If we identify sentence meaning with purport, as Jayanta does, then it follows that, until and unless we have identified a structure of meaning with this practical element at the top, we are not in possession of a complete sentence meaning. Take, for instance, the sentence, “Isn’t it hot in here?” The sentence meaning here is not a proposition about the temperature of the room. In most contexts—and remember that Jayanta requires us to understand sentence meaning against a contextual background—the sentence meaning is a polite request to open a window. Until we have understood that practical element, we have only understood a subordinate element of the overall sentence meaning, which theorists have called an “intermediate sentence meaning” (avāntaravākyārthaḥ). If we take the orientation toward “something to be done” to be a universal feature of language, then we must ask: What is to be done in the case of literary language? Here is Dhanika again (Observations, 211): In the case of literary language, we determine that what is to be done is precisely the production of the bliss proper to it, and we do so on the basis of positive and negative concomitance, for we do not encounter any other motivation for its use, either for the speaker or for the addressee.30

We might be puzzled at how quickly Dhanika appears to have reached the conclusion that aesthetic experience is the only purpose for which literary language is employed. In fact, Dhanañjaya and Dhanika had sarcastically

Rasa as Sentence Meaning

387

dispatched an alternative view, according to which literature serves the purpose of moral instruction, at the very beginning of the Ten Dramatic Forms.31 He continues: It is the stable emotion, conjoined with the other aesthetic elements, that we understand as the cause of the production of this bliss. Hence a particular rasa will draw the power of expression of a sentence towards itself until, through conveying at an intermediate stage the other aesthetic elements upon which the various proper meanings expressed in the sentence depend, it leads it to culminate in that very rasa. In such a sentence, the other aesthetic elements take the place of word meanings, and the sentence meaning is a stable emotion, such as desire, that is conjoined with them. Hence a literary text, so described, is a work of words and sentences (vākyapadīyam), the word meanings being the aesthetic elements, the sentence meanings the rasas.32

Dhanika is serious about the rasa being itself the sentence meaning, in the sense of the final purport, of a work of literature. But you can tell from his comment that “the other aesthetic elements take the place of word meanings” that he is less committed to the idea that the aesthetic elements are literally word meanings. In fact, as Pollock notes here, the other aesthetic elements may actually be sentence meanings, especially when they are communicated by literary ornaments. In such cases, however, they are not the final sentence meaning, which is axiomatically “something to be done,” but only intermediate sentence meanings. Two features of Dhanika’s explanation here mark it, in my view, not merely as an elaboration of Bhaṭṭa Nāyaka’s views but as an advance on them. And both of those features derive from the contextualist paradigm championed by Śālikanātha and Bhaṭṭa Jayanta. The first is the mention of a “dependency” (apekṣā) between the word meanings that have actually been expressed and the aesthetic elements. Dependency is one of the conditions (upādhiḥ, as Śālikanātha calls them) in the mutual relation between elements of meaning that is constitutive of sentence meaning.33 It is true that Bhaṭṭa Nāyaka had already invoked the conjuncture of word meanings (saṃsargaḥ) itself in this connection. Dhanika is simply more explicit about its mechanics: what it means to say that two elements of meaning are conjoined is, at least in part, that one of them has a dependency that is resolved by the other and vice versa. The other is a very striking image of the top-down way in which sentence meaning is supposed, on the contextualist paradigm, to work. Rasa is the purpose, “what is to be done,” and hence all of the other elements of meaning, be they word meanings or intermediate sentence meanings, are teleologically subordinate to

388

Andrew Ollett

it. These elements can only enter the structure of the overall sentence meaning insofar as they contribute, in some way, to this goal, the production of rasa. The contribution of each linguistic expression to the overall sentence meaning is its “relational meaning,” that is, its “proper meaning” (svārthaḥ) in relation to the other elements that are present in the complex, hierarchical, action-oriented structure. Directing this entire process, and “drawing the power of expression towards itself,” is rasa. Thus, using the resources found in the theory of sentence meaning that Bhaṭṭa Jayanta offers in his Racemose Reasoning—which, for its part, was made possible by the contextualist turn of Prabhākara and his followers—Dhanañjaya and Dhanika were able to sharpen two related insights of Bhaṭṭa Nāyaka regarding the operation of literary language. The first was that “manifestation,” a modality of language particular to literature, was unnecessary, given that we can account for the apprehension of meaning through the standard modalities of language, even for meanings we might describe as nonliteral, suggested, or implicit. This insight required, however, that we provide a more robust account of those standard modalities. It also implied that, contrary to what Ānandavardhana had claimed, rasa itself, or at least the aesthetic elements that served as its precursors, could be expressed as the meaning of a text. Jayanta’s top-down model, where the principal element in the structure of meaning was the purpose for which the text was composed, provided a suitable justification for the equation of rasa with sentence meaning. There remained, however, the distinction between meaning as a cognitive object and rasa as an affective state. This is where Bhaṭṭa Nāyaka’s second insight comes in: rasa is actually an experience that the reader has, and although the process begins with understanding rasa from the text as one of its meanings, indeed as its principal meaning, it culminates in the “actualization” of that rasa in the reader. In using the language of actualization, Bhaṭṭa Nāyaka meant for us to think of rasa as a result for which the text offers a kind of blueprint. Dhanika, while maintaining this terminology, had reinforced Bhaṭṭa Nāyaka’s insight by speaking of rasa as the keystone in the hierarchical and action-oriented structure of meaning that was increasingly coming to be called the “purport” (tātparyam) after Bhaṭṭa Jayanta. Rasa, or more precisely the experience thereof, was “something to be done,” around which, and for the sake of which, all of the other elements of meaning in a text took their place. The story of the intersection of these two lines of thought, one focused on the question of aesthetic experience, and the other on the mechanics of sentence meaning, by no means ends with Dhanika. Bhoja, who wrote one generation after him, similarly aimed to synthesize rasa with the theories of sentence

Rasa as Sentence Meaning

389

meaning developed by Śālikanātha and Bhaṭṭa Jayanta, both of whom he quoted extensively. But it was Abhinavagupta, an exact contemporary of Dhanika, who was perhaps most concerned to reconcile rasa with a comprehensive theory of meaning. This concern is most clearly evident in the first chapter of his commentary, called the Eye, on Ānandavardhana’s Light on Resonance, which evinces a deep familiarity with Bhaṭṭa Jayanta’s work. There he reprises some of the arguments that Dhanika had put into the mouth of the proponent of manifestation in his Observations, such as the following: He who thinks that even here suggestion is nothing more than tātparyaśakti (the power of the sentence meaning) does not know the truth of the matter. For in a sentence that conveys the vibhāvas and anubhāvas, the tātparyaśakti exhausts itself in giving the syntax (saṃsarga) [of the sentence] or its difference [in meaning from that of other sentences]; it does not concern rasa, the essence of which consists in the process of relishing. Let us say no more.34

This criticism is perhaps fair when it is read against Bhaṭṭa Jayanta’s own comments on the power of purport, which he had advanced precisely in order to account for that feature of sentence meaning, whatever we want to call it, that is not reducible to the individual word meanings. Yet when this theory is enhanced with the features that Dhanika had emphasized—the fact that the sentence meaning is always “something to be done,” and the fact that the experience of a rasa proceeds from its being comprehended as a sentence meaning—it is not clear that the power of purport should be limited in the way that Abhinavagupta wants, nor that that the expression of rasa through the power of purport excludes its being experienced.

Notes 1 When I presented a very early version of this chapter at the SAPHALA workshop in Vienna in December 2017, I had not taken Bhaṭṭa Jayanta’s influence on Dhanika very seriously. I was encouraged to think harder about the (now embarrassingly clear) evidence for this influence and what it entailed for Dhanika’s aesthetic theory by a number of participants in that workshop, including especially Daniele Cuneo, Hugo David, and Alessandro Graheli. This chapter owes its present form largely to their suggestions. I, of course, take responsibility for all mistakes and defects in the argument.

390

Andrew Ollett

2 Controversially because another eminent theorist, Bhaṭṭa Udbhaṭa, had indeed argued that rasa can be communicated by its “proper linguistic expressions.” See Pollock 2016, 70–71. 3 Although nothing, of course, prevents an affective state itself from being an intentional object: one can talk about emotions without experiencing them. 4 These two ways of thinking about sentence meanings are modeled on the two ways of thinking about propositions adumbrated in Kaplan 1989, 494, one defined by truth-conditional semantics (in Kaplan’s case, possible-world semantics), and the other defined as “structured entities looking something like the sentences which express them.” 5 This is the view of Kane 1961, 244–46, with which Pollock 2016, 154–55 largely agrees. The colophon of the text identifies Dhanañjaya as a member of the court of Vākpati, who ruled from 972 until he was captured by Tailapa in the early 990s. One manuscripts of the Observations, and one inscription referred to by Kane, identifies Dhanika, too, as an administrator (mahāsādhyapāla-) of Vākpati, and Dhanika once quotes a poem of Vākpati in his Observations. Dhanika also quotes the Deeds of Navasāhasāṅka by Padmagupta, which relates the story of Vākpati’s successor, Sindhurāja or Navasāhasāṅka, and which was probably composed around 1010 CE. Dhanika and Dhanañjaya both call themselves “son of Viṣṇu,” which makes it likely they were brothers, or possibly the same person. Bhoja, the successor of Sindhurāja, quotes Dhanika’s poetry, but never refers to either the Ten Dramatic Forms or the Observations, which is somewhat mysterious. See the discussion by Pollock 2016, 155 and Cox 2016, 58. 6 Racemose Reasoning, vol. 2, 218: tenānvitābhidhānaṃ hi nāsmābhir iha mṛṣyate / anvitapratipattis tu bāḍham abhyupagamyate // 7 See Graheli 2016 for tātparyam. In the introduction to their translation of the Light on Resonance, Ingalls, Masson, and Patwardhan claimed that “the school of ritualists founded by Kumārila held that there existed a third power which furnished a ‘final meaning’ to the sentence as a whole. They called this the tātparyaśakti, and defended its reality against their opponents, the Prābhākara ritualists, who claimed that the denotative force in each word kept on operating until at the conclusion of the sentence it worked automatically in harmony with the other words” (Ingalls et al. 1990, 14). 8 Racemose Reasoning, vol. 2, 216: nābhidhātrī śaktir anvitaviṣayā, kiṃtu   anvay​avyat​irekā​nugat​aniṣk​ṛṣṭas​vārth​aviṣa​yaiva.​ 9 Racemose Reasoning, vol. 2, 216–17: tātparyaśaktis tu teṣām. anvitāvagamaparyantā sahavyāpārād vyāpārasyaitadīyasya   nirākāṅkṣapratyayotpādanaparyantatvāt. 10 Mīmāṃsāsūtra, 2.1.46: arthaikatvād ekaṃ vākyaṃ sākāṅkṣaṃ ced vibhāge syāt. “It is a single sentence, on account of its single meaning, if it would have unresolved dependencies if it were divided.”

Rasa as Sentence Meaning

391

11 Racemose Reasoning, vol. 2, 217: anyathaiva pravartante pratyakṣādyudbhavā dhiyaḥ / arthaṃ pūrṇaṃ apūrṇaṃ vā darśayantyaḥ puraḥ sthitam // anyathaiva matiḥ śābdī viṣayeṣu vijṛmbhate / pratipattur anākāṅkṣapratyayotpādanāvadhiḥ // 12 In several publications, Pollock speaks of the “older theory by which the final purport of a sentence is produced (tātparya), to which Bhaṭṭa Nāyaka as a Mīmāṃsaka was committed” Pollock 2012, 235; see also Pollock 2010, 178. Mīmāṃsakas no doubt made use of similar concepts, such as overall meaning (paryavasānam), but the use of the concept of tātparyam specifically is, as noted above, a hallmark of Bhaṭṭa Jayanta, and not found in Bhaṭṭa Nāyaka’s fragments. 13 See Pollock 2010; David 2016; Ollett 2016; Pollock 2016, 144–54. 14 Critical Analysis of Manifestation, 70; see also New Dramatic Art [2nd ed.], vol. 1, p. 277 or New Dramatic Art [4th ed.], 271. bhāvasaṃyojanāvyaṅgyaparasaṃvitti­-  gocaraḥ / āsvādanātmānubhavo rasaḥ kāvyārtha ucyate //. For reasons given below, the conjectural reading bhāvasaṃyojanāvyaṅgyaḥ may be preferable. Pollock (2012, 242) takes Abhinavagupta’s reading and translates it differently: “The purpose of literature is rasa, which is an experience consisting of savoring; it may be said to be ‘manifested’ only by way of a manifestation called awareness, and its domain is the highest consciousness.” The translation in Pollock 2016, 149 is identical. 15 There are two issues. First is the difference in the reading between Abhinavagupta (saṃvedanākhyayā) and Mahima Bhaṭṭa (bhāvasaṃyojanā). I think Mahima’s reading makes much more sense, and hews more closely to what we know of Bhaṭṭa Nāyaka’s thought, and is less likely to have suffered corruption than the corresponding passage in Abhinavagupta. Second, Hemacandra, in his paraphrase of Abhinavagupta’s discussion, oddly substitutes the verse discussed below (saṃsargādir) for this one. I do not see it as quite as much of a non sequitur as Pollock does. 16 See the verse quoted by Jayanta (Racemose Reasoning, vol. 2, 218): prakṛtipratyayau yadvad apekṣete parasparam / padaṃ padāntaraṃ tadvad vākyaṃ vākyāntaraṃ tathā //. “Just as there is mutual dependency between a derivational base and a suffix, so there is between a word and another word, and a sentence and another sentence.” Similarly see Kumārila, Explanation of the System, 453. 17 saṃsargādir yathā śāstra ekatvāt phalayogataḥ / vākyārthas tadvad evātra śṛṅgārādī raso mataḥ //. Pollock 2016, 149 translates: “Just as in the Veda, where sentence meaning arises through the syntactic construal of the constitutive words— since that meaning must be a unity, given that it bears a relation to a single result— so in literature we hold rasa to constitute a kind of sentence meaning.” See also Pollock 2010, 164. 18 See Explanation of the System, 445: bhāvanaiva ca vākyārthaḥ svakārakaviśeṣitā, “sentence meaning is simply the actualization qualified by its factors.”

392

Andrew Ollett

19 Whereas Abhinavagupta (possibly although not certainly echoing Bhaṭṭa Nāyaka) likened the aesthetic experience to the performance of a Vedic sacrifice on the basis of hearing Vedic texts that commended it, Hemacandra had to substitute the use of a particular mantra on the basis of hearing Jain scriptures wherein the mantra is said to have worked. See Ollett 2016. 20 See, for example, Pollock 2012, 234. 21 Mahima Bhaṭṭa has a very interesting way of reading suggestive verses, in which an inevitable suggestion, one which the reader is invariably led to in order to make sense of the meanings presented in the text, really does belong to the text itself, while a suggestion that depends on the reader’s training, suspicion, conventions, and so on cannot properly be said to belong to the text itself. See Critical Analysis of Manifestation, 469. 22 Racemose Reasoning, vol. 2, 218: etena śabdasāmarthyamahimā so ’pi vāritaḥ / yam anyaḥ paṇḍitaṃmanyaḥ prapede kaṃcana dhvanim // 23 Racemose Reasoning, vol.2, 218: bhama dhammia vīsattho mā sma pāntha gṛhaṃ viśa. 24 Observations, 206: rasavākyeṣu ca vibhāvapratipattilakṣaṇadvitīyakakṣāyāṃ rasānavagamāt. 25 Ten Dramatic Forms, v. 4.37, pp. 211–212: vācyā prakaraṇādibhyo buddhisthā vā yathā kriyā / vākyārthaḥ kārakair yuktā sthāyī bhāvas tathetaraiḥ //. Translation modified from Pollock 2016, 170. Besides removing Pollock’s quotation marks around “sentence-meaning,” I have changed the translation in order to make it clear that kriyā and kāraka refer to word meanings, and not to words. Words, such as nouns and verbs, are not expressed, but word meanings are. 26 Notably Śālikanātha, in his Straightforward and Lucid Commentary, 383, yatrāpy arthāntaraṃ nāsti, tatrāpy antato ’styarthena vyatiṣaktaḥ, “in the end, even where there is no further meaning, the meaning of the verb ‘exists’ is cross-connected with it.” 27 Jayanta took the first position, evidently in opposition to Śālikanātha, who took the second. 28 The translation (of prītyai navoḍhā priyā) is from Pollock 2016, 170. 29 Observations, 211: na cāpadārthasya vākyārthatvaṃ nāstīti vācyam, kāryaparyavasāyitvāt tātparyaśakteḥ. tathā hi—pauruṣeyam apauruṣeyaṃ vā sarvaṃ vākyaṃ kāryaparam. atatparatve ’nupādeyatvād unmattādivākyavat. 30 kāvyaśabdānāṃ cānvayavyatirekābhyāṃ niratiśayasukhāsvādavyatirekeṇa pratipādyapratipādakayoḥ pravṛttiviṣayaprayojanāntarānupalabdheḥ / kāvyaśabdānāṃ ca svānandodbhūtir eva kāryatvenāvadhāryate. I have accepted Pollock’s interpretation of niratiśayasukhāsvādavyatirekeṇa as an intrusive gloss. I agree with Bhaṭṭa Nṛsiṃha’s interpretation of pratipādyapratipādakayoḥ as “the addressee and the speaker,” the two parties among whom a purpose, or motivation, for the use of language must always be sought. Pollock’s translation (2016, 171)

Rasa as Sentence Meaning

31

32

33

34

393

is: “In the case of the language of literature, we must conclude that its ‘actionoutcome’ is nothing other than arousing the bliss proper to it, since positive and negative evidence reveals no other performance-oriented purpose with respect to its signified and signifier.” Ten Dramatic Forms, p. 5 (1.6): ānandaniṣyandiṣu rūpakeṣu vyutpattimātraṃ phalam alpabuddhiḥ / yo ’pītihāsādivad āha sādhus tasmai namaḥ svādaparāṅmukhāya //. “Reverence to that good man who could be so small-minded as to say that the only result we get from stage plays, which surge with bliss, is moral instruction, like the epics, and who turns his back on savoring this bliss.” tadbhūtinimittatvaṃ ca vibhāvādisaṃsṛṣṭasya sthāyina evāvagamyate. ato vākyasyābhidhāśaktis tena tena rasenākṛṣyamāṇā tatta​tsvār​thāpe​kṣitā​vānta​ ravib​hāvād​iprat​ipāda​na-dv​ārā svaparyavasāyitām ānīyate. tatra vibhāvādayaḥ padārthasthānīyās, tatsaṃsṛṣṭo ratyādir vākyārthaḥ. tad etat kāvyaṃ vākyapadīyam. tāv imau padārthavākyārthau. Pollock (2016, 171) translates: “As for this outcome, it arises, we come to understand, when the stable emotion is ‘syntactically construed’ with the aesthetic elements. Accordingly, the expressive capacity of such a ‘sentence’ is elicited by a given rasa, and eventuates at last in producing it through communicating the specific aesthetic elements appropriate to its particular character. In this process, they elements may be taken to stand for words, while the stable emotion syntactically construed with them forms a sentence meaning. Thus, literature as such is a Vākyapadīya, a work concerned at once with word and sentence, the ‘words’ and ‘sentences’ being those just indicated.” This translation makes it appear as if Dhanika was presenting an analogy, whereas it seems to me that he actually meant to identify rasa as a sentence meaning. These three conditions (dependency, proximity, and compatibility) feature prominently in the theory that would come to be associated with Prabhākara, the “expression of relational meanings,” although they are listed, as far as I know, for the first time in Kumārila’s Explanation of the System. See my other essay in this volume (Chapter 13). Eye on the Light on Resonance, 84: yas tv atrāpi tātparyaśaktim eva dhvananaṃ manyate sa na vastutattvavedī. vibhāvānubhāvapratipādake hi vākye tātparyaśaktir bhede saṃsarge vā paryavasyet. na tu rasyamānatāsāre rasa ity alaṃ bahunā. The translation is from Ingalls et al. 1990, 110.

Primary Sources Critical Analysis of Manifestation

Dwivedi, Rewaprasada, ed. 1964. The Vyaktiviveka of Rājānaka Śrī Mahimabhaṭṭa. Varanasi: Chaukhamba Sanskrit Series Office.

394

Andrew Ollett

Explanation of the System

Tantravārttikam of Kumārila. In Mīmāṃsādarśana.

Eye on the Light on Resonance

Paṭṭābhirāma Śāstrī, ed. 1940. The Dhvanyāloka of Śrī Ānandavardhanāchārya: With the Lochana and Bālapriyā Commentaries by Śrī Abhinavagupta and Paṇḍitrāja Sahṛdayatilaka Śrī Rāmaśāraka. Benares: Chowkhamba Sanskrit Series Office.

Long Commentary

Ramanatha Shastri, S.K., and S. Subrahmanya Shastri, eds. 1934–67. Bṛhatī of Prabhākara Miśra [on the Mīmāṃsāsūtrabhāṣya of Śabarasvāmin] with the Ṛjuvimalapañcikā of Śālikanātha. 5 vols. Madras: University of Madras.

Mīmāṃsādarśana

Subbāśāstrī, ed. 1929–34. Śrīmajjaimi nipraṇītaṃ Mīmāṃsādarśanam. Vol. 6. Anandashrama Sanskrit Series 97. Pune: Anandasramamudranalayah.

Mīmāṃsāsūtra

In Mīmāṃsādarśana.

New Dramatic Art [2nd ed.]

Ramaswami Shastri, K.S., ed. 1956. Nāṭyaśāstra of Bharatamuni with the Commentary Abhinavabhāratī by Abhinavaguptācārya. Vol. 1. Second edition. Baroda: Oriental Institute.

New Dramatic Art [4th ed.]

Krishnamoorthy, K., ed. 1992. Nāṭyaśāstra of Bharatamuni with the Commentary Abhinavabhāratī by Abhinavaguptācārya. Vol. 1. Fourth revised edition. Baroda: Oriental Institute.

Observations

Daśarūpakāvalokaḥ by Dhanika. In Ten Dramatic Forms.

Racemose Reasoning

Varadacharya, K.S., ed. 1969–83. Nyāyamañjarī of Jayantabhaṭṭa: With Ṭippaṇī Nyāyasaurabha. 2 vols. Mysore: Oriental Research Institute.

Straightforward and Lucid Commentary

Ṛjuvimalā Pañcikā by Śālikanātha. In Long Commentary.

Rasa as Sentence Meaning Ten Dramatic Forms

395

Venkatacharya, T., ed. 1969. The Daśarūpaka of Dhanaṃjaya with the Commentary Avaloka by Dhanika and the Sub-Commentary Laghuṭīkā by Bhaṭṭanṛsiṃha. Madras: Adyar Library and Research Centre.

Secondary Sources Cox, Whitney. 2016. Modes of Philology in Medieval South India. Leiden: Brill. David, Hugo. 2016. “Time, Action and Narration. On Some Exegetical Sources of Abhinavagupta’s Aesthetic Theory.” Journal of Indian Philosophy 44 (1): 125–54. doi:10.1007/s10781-014-9256-1. Graheli, Alessandro. 2016. “The Force of Tātparya: Bhaṭṭa Jayanta and Abhinavagupta.” In Around Abhinavagupta: Aspects of the Intellectual History of Kashmir from the 9th to the 11th Centuries, ed. by Eli Franco and Isabelle Ratié. Leipziger Studien zu Kultur und Geschichte Süd- und Zentralasiens 6. Proceedings of the Abhinavagupta conference, Leipzig 2013. Berlin: LIT Verlag. Ingalls, Daniel H.H., J.M. Masson, and M.V. Patwardhan, eds. 1990. The Dhvanyāloka of Ānandavardhana with the Locana of Abhinavagupta. Harvard Oriental Series 49. Cambridge, Mass.: Harvard University Press. Kane, Pandurang Vaman. 1961. History of Sanskrit Poetics. Third, revised edition. Delhi: Motilal Banarsidass. Kaplan, David. 1989. “Demonstratives.” In Themes from Kaplan, ed. by Joseph Almog, John Perry, and Howard Wettstein, 481–563. New York: Oxford University Press. Kunjunni Raja, K. 1963. Indian Theories of Meaning. Madras: Adyar Library and Research Centre. McCrea, Lawrence J. 2008. The Teleology of Poetics in Medieval Kashmir. Harvard Oriental Series 71. Cambridge, MA: Department of Sanskrit / Indian Studies, Harvard University. Ollett, Andrew. 2016. “Ritual Texts and Literary Texts in Abhinavagupta’s Aesthetics: Notes on the Beginning of the ‘Critical Reconstruction.’” Journal of Indian Philosophy 44 (3): 581–95. doi:10.1007/s10781-015-9277-4. Pollock, Sheldon. 2010. “What Was Bhaṭṭa Nāyaka Saying? The Hermeneutical Transformation of Indian Aesthetics.” In Epic and Argument in Sanskrit Literary History: Essays in Honor of Robert P. Goldman, ed. by Sheldon Pollock, 143–84. New Delhi: Manohar. Pollock, Sheldon. 2012. “Vyakti and the History of Rasa.” Saṃskṛtavimarśaḥ 6: 232–53. Pollock, Sheldon. 2016. A Rasa Reader: Classical Indian Aesthetics. New York: Columbia University Press.

21

Meaning beyond Words: The Implicature Wars1 Daniele Cuneo

1  The Issue of Implicature and the Revolutions of Kashimirian Poetics Implicature, loosely defined, is the capacity of human language to express a sense that is not accounted for by the lexical meanings of words. Although it is often considered the least paradigmatic and the most refined use of linguistic communication, implicature in this broad understanding is a widely common phenomenon in human interaction, most apparent in instances such as ironic statements, satirical remarks, allegorical language, or even frozen figurative usage, as in dead metaphors such as “falling in love,” “time is running out,” or “the hands of a clock.” However, it is the case of poetical language that presents the most evident and salient illustration of the capacity of human language to push signification far beyond the letter of the word and the limits of the allegedly strict and conventional relation between one single word and one single meaning. This semantic hypertrophy of poetical language became a heated topic of discussion in the history of Sanskrit poetics (Alaṅkāraśāstra). This discipline stands out among the śāstric traditions that focus on many topics of what one might currently call “philosophy of language,” because of its relative marginality in the larger philosophical scenario of the Sanskrit culture. In particular, it never was and never became a full-fledged darśana, a systematic philosophical viewpoint on reality, unlike Nyāya, Mīmāṃsā, or even Vyākaraṇa, and as such it does not feature in the Sanskrit doxographies or even in the English-language introductions to Indian philosophy that all too rarely populate the shelves of contemporary bookshops. Nevertheless, this very marginality is arguably one of the reasons that allowed and maybe prompted authors of Alaṃkāraśāstra to

Meaning beyond Words

397

freely borrow and reinterpret concepts from different philosophical traditions and to normally imply and sometimes openly assert the trans-sectarian attitude of their enterprise.2 Such transdisciplinarity is particularly manifest in the treatment of the indirect usages of language, which are often discussed by consciously crossing the lines drawn by the available theories of linguistic significations proposed by the other philosophical schools. Crucial in the historical development of Sanskrit poetics is the fertile Himalayan valley of Kaśmīr. Crucial in this Kashmirian history are the ninth and the tenth centuries, which witnessed a series of theoretical revolutions that have been mapped more than once in secondary literature, and even in primary sources from the Valley (see Cuneo 2016b, 151–57). The main traits of this history, however, might well be sketched here, as well, given the hermeneutical slant championed by the present author. In my reconstruction of the intellectual history of the discipline, three such revolutions can be identified and ascribed to three Kashmirian authors: Bhaṭṭa Udbhaṭa, Ānandavardhana, and Bhaṭṭa Nāyaka. To cut a long and complex story short, it all started at the beginning of the ninth century with Udbhaṭa and his semanticization of the process of aesthetic appreciation by way of an active borrowing of cognitive and hermeneutical models from Mīmāṃsā, the Vedic school of textual exegesis, as argued by Bronner (2016), thus opening up Alaṃkāraśāstra to the contemporary developments in the wider philosophical scenario. This can be considered as the first breakthrough. The second one concerns, more directly, the issue of indirect language, so it is easier to get the third revolution out of the way first. This third paradigm shift3 is the one attributed to Bhaṭṭa Nāyaka’s Hṛdayadarpaṇa, probably belonging to the beginning of the tenth century, of which we have evidence only through Abhinavagupta and some later authors. In my understanding—which slightly differs from the interpretation that Pollock gives to this course of cultural events, both in a seminal article 2010 and in his game-changing Rasa Reader—Bhaṭṭa Nāyaka’s novelty lies in the interpretation of the concept of rasa, aesthetic emotion, as completely different from worldly emotions (bhāva). The latter are enmeshed in the endless vagaries of pleasure and pain, while the former are essentially delectable insofar as they are not touched by the attachment to real-world events and by the limiting factors that are space, time, and personal subjectivity.4 The present contribution focuses on some responses and on some rejoinders to those responses to the second revolution, the one heralded by Ānandavardhana (855–883 ca.) in his Dhvanyāloka (“The Light on Suggestion” or “The Light

398

Daniele Cuneo

on Implicature”). In order to account for the great varieties of indirect poetical meanings and specifically for the affective side (rasa) of the experience of poetry and drama, Ānandavardhana postulated a third function of language, variously called as dhvani, vyañjanā, dyotana, prakāśana, avagamana, and so forth, a function over and above the two commonly accepted functions of linguistic communication: denotation and indication. Indication was variously named by different authors, and with differences in scope and meaning, as lakṣaṇā, bhakti, upacāra, and more, and usually included both metaphorical and metonymical varieties of indirect communication.5 Ānandavardhana’s new-fangled theory was initially met with a strenuous resistance on the part of the Kashmirian intelligentsia, and for more than two centuries the issue of dhvani was heatedly debated in a number of extremely sophisticated works. The aforementioned work by Bhaṭṭa Nāyaka, for instance, was most probably composed as a concerted rejoinder to Ānandavardhana’s stance on poetical meanings and the way they are conveyed to the literary or theatrical audience. As far as historical generalizations go, one might safely state that by the very end of the eleventh century the theory of dhvani, poetical suggestion, had practically won the battle over the phenomenology, ontology, and teleology of poetical meanings. This victory, however, is the result of a complex debate involving, among others, the contrasting ideas of Mukula Bhaṭṭa (ninth century), the enormous and innovative contribution of Abhinavagupta (tenth to eleventh century), and the systematic reorganization found in the famous Kāvyaprakāśa, the magnum opus of Rājānaka Mammaṭa (eleventh century).6 The present chapter centers on the figure of Mukula Bhaṭṭa and on the Wirkungsgeschichte of his work and thought, with a specific focus on his influence on Mammaṭa.7

2  Mukula Bhaṭṭa’s “Abhidhāvṛttamātṛka” and Its Wirkungsgeschichte Mukula was the first author we know of who seemingly challenged the novel theory of Ānandavardhana, although in the very act of reticently proposing a different, more conservative theory he also absorbed most of Ānanda’s theoretical subdivisions of poetical meanings such as the threefold classification of suggestively conveyed sense: vastu, “a state of fact”; alaṃkāra, “a figure of speech”; and rasa, “an emotional mood.”8 Such mixed attitude of partial absorption and partial refutation will be found also in the reception of Mukula’s own work.

Meaning beyond Words

399

His Abhidhāvṛttamātṛkā, a work that might be translated tongue-in-cheek as The ABC of Signification, aims at a complete description of the workings of language, from the most common everyday usages to the most refined poetical statements. In this purposefully exhaustive process of analysis, Mukula also tried to undermine and detonate the very idea of a third function of language as postulated in the Dhvanyāloka.9 His theoretical weapon is an original and philosophically refined theory of secondary signification or indication (lakṣaṇā). Gaṅgāyāṃ ghoṣaḥ, “a hamlet on the Ganges,” is Mukula’s most famous and time-honored example, found for the first time in Patañjali’s Mahābhāṣya, together with other stock examples of secondary signification such as mañcāḥ krośanti, “the cots are crying,” standing for “the babies who are on the cots are crying”10 and kuntāḥ praviśanti, “the lances are entering,” standing for “the lance-bearers are entering.” Mukula’s theory accounts for a staggering variety of such examples, along with classifications, subclassifications, and alternative classifications of the phenomenon of secondary signification. His criteria include, among other things, the nature of the relation between denoted and indicated meaning, the degree of preservation of the primary meaning, the degree of perceived distinction between the denoted and the indicated, as well as the degree of intentionality on the part of the speaker uttering each and every instance of indication. This titanic theoretical effort cannot but harbor the clear intent of encompassing within itself not only the totality of everyday conversational and scholarly usages of language, but also the whole array of the poetical and affective meanings (rasas) that poetry as conceived by Ānandavardhana’s Dhvanyāloka was meant to convey. To lay some of my cards on the table, my contention is that the crux of the whole theoretical issue is the concept of bādha, usually translated as “obstruction,” “blockage,” or “inapplicability,” that is, the inapplicability of the direct meaning (mukhyārtha) of the words to the contextual situation in which a given instance of linguistic communication is used. In particular, the problem hinges on how widely or narrowly the concept of bādha was understood in order to account for the move from primary, lexical word meanings to the finally accepted sentence meaning, which often (or, according to Mukula, always) goes well beyond any fixed semantic relation. As hinted at, what mostly remains to be written is the Wirkungsgeschichte of Mukula’s ideas and the impact they had on the development of the theory of dhvani itself. Probably because of its hybrid theoretical collocation and scholastic affiliation, somewhere in the midst of various disciplines such as Vyākaraṇa, Bhāṭta as well as Prābhākara Mīmāṃsā and obviously Alaṃkāraśāstra,

400

Daniele Cuneo

Mukula’s Abhidhāvṛttamātṛkā has not attracted much attention among both Sanskrit authors and contemporary scholars.11 However, his ideas have been quite influential, to an extent often disregarded in previous English scholarship. This happened to be the case when his theory of indication forced itself, so to say, on the upholders of dhvani, specifically on Mammaṭa, arguably the author of the most important treatise of Sanskrit poetics of the second millennium.12 In my understanding, Mukula’s theory of indication provided Mammaṭa with a very sophisticated account of secondary signification that could be more easily appropriated, adapted, and fine-tuned, than rejected en bloc. It was thus easier and probably wiser for Mammaṭa to accept Mukula’s multifaceted theory in its general terms and only reject some specifics aspects, and a crucial one, in particular, just enough to argue that the power of indication could not account for poetical and emotive meanings and to thereby defend the existence of a third faculty of language, his cherished concept of vyañjanā “poetical suggestion.”13 This was no obvious move, but rather a bold choice on Mammaṭa’s part, as it meant the abandonment of the seemingly straightforward subdivision of secondary signification in simply two categories the metaphorical (gauṇī vṛtti) and the metonymical (lākṣaṇikī vṛtti), a stance maintained by Abhinavagupta, the great tantric master who perfected Ānandavardhana’s theory in his monumental Locana commentary on the Dhvanyāloka.14

3  Groping for Mukula’s Influence before Mammaṭa’s Appropriation Before delving into the Mukula-Mammaṭa relation, it is appropriate to remark that the issue of the inclusion of all possible non-direct poetic meanings within lakṣaṇā is tackled and dismissed already in Ānandavardhana’s Dhvanyāloka, however briefly. Strangely enough, in his Locana commentary to the Dhvanyāloka, dedicated to the defense and the improvement of the theory of dhvani against all its detractors, Abhinavagupta does not mention Mukula or his work by name, not even in the long sections devoted to the rejection of the inclusion of dhvani within secondary signification (bhakti, “associated meaning” in Ānandavardhana’s and Abhinavagupta’s terminology).15 However, not only does it seem unlikely that Abhinavagupta did not know Mukula’s work or that he did not repute it worthy of mention, but it is possible that a passage from the Locana on Dhvanyāloka 1.4 is aimed against Mukula without referring to him by name—and that’s the reason why even the learned English translators of the

Meaning beyond Words

401

Locana, Ingalls, Masson, and Pathwardan, did not signal the silent reference to him. The passage criticizes the idea of lakṣita-lakṣaṇā, “indication arising out of something that has been expressed through indication,” which Mukula uses in order to account for some cases of secondary signification.16 The passage of the Locana does have some interesting features: (1) the slightly contemptuous tone that Abhinavagupta has toward the unnamed detractor and (2) the fact that the refutation of the theory is based on the argument about the purpose of indication (see below), a line of argumentation originally found in the Dhvanyālōka, 1.17, then expanded by the Locana and finally taken over and perfected by Mammaṭa. Just to give an idea of Abhinava’s slightly dismissive tone, here is the relevant quote: “Accordingly, the name lakṣitalakṣaṇā (secondary operation arising from a secondary operation), given by a certain author to this type of operation, is a piece of stubborn perversity (Ingalls et al. 1990, 88).”17 But, of course, the reference might be to some other crucial figure in the debate who has simply been lost to the historical record here. A further hint at the fact that Abhinavagupta might have known the work of Mukula is the former’s use of the term and the concept of kakṣyā or kakṣā, “stage” or “phase,” in the explication of his theory of signification. To my knowledge, Mukula is the first to use the term in the sense of the various phases of linguistic understanding. In Mukula’s text,18 according to various linguistic theories upheld by Mīmāṃsakas or Grammarians, the various linguistic functions operate in one phase or another, that is, in a different order for each of the described view. The details are not of present concern, but Abhinavagupta does seem to borrow the term from the Abhidhāvṛttamātṛkā and to employ it for the phases in which his theory of signification subdivides the process of linguistic comprehension.19 To wrap up this rather speculative section, the absence of the Kashmirian work of Mukula in the oeuvre of Abhinavagupta still waits for a satisfactory historical explanation.20 But, in my opinion, the clues are pointing to the fact that Abhinavagupta did indeed know and dislike the Abhidhāvṛttamātṛkā.

4  “You Don’t Understand My Feelings. You Don’t Get Me At All” My understanding of the theoretical dynamics between lakṣaṇā and vyañjanā pivots on the necessity to account for the linguistic conveyance of emotional meanings, which cannot be expressed directly by simply naming the words

402

Daniele Cuneo

“sadness” or “love” or by any directly expressed statement whatsoever, as almost all authors after Ānandavardhana agree upon.21 As hinted above, it is my contention that it is the differences in the conception of mukhyārthabādha “obstruction of the primary meaning” heralded by Mukula and Mammaṭa that explain how emotionally charged language is differently accounted for by the two authors. Mammaṭa famously deals with the functions of language (śabdavyāpāra) in the second chapter of his magnum opus, the renowned Kāvyaprakāśa, where he already tackles the problems raised by Mukula and basically appropriates his sixfold subdivision of lakṣaṇā. But, in addition to this, he also composed a later,22 shorter and less-known work, the Śabdavyapāravicāra, which is even more clearly based on Mukula’s work as it focuses only on the functions of languages—three for him,23 two for Mukula—although Mammaṭa’s work is clearly aimed at its confutation. The Śabdavyāpāravicāra is a richly composite and intertextual work, as it shares much of its material, that is, the very wording of many of its sentences, with the second and the fifth chapters of the Kāvyaprakāśā, which in their turn were a re-writing of several passages found in the Abhidhāvṛttamātṛkā, and also includes the literal citation of several poetical verses from Mukula’s work employed as exemplifications of the various kinds of implicature. The very incipit of the work is a philosophically interesting passage that is not shared with the Kāvyaprakāśa and that connects morality, epistemology, and philosophy of language by justifying the very enterprise of any serious grappling with linguistic communication. Without spelling out all the theoretical consequences, Mammaṭa bluntly outlines the crucial epistemological-cumethical importance of dedicating a work to the functions of language and thus implicitly praises his resolutely unnamed source and enemy, Mukula, for the invention of what might be considered a new subgenre of Alaṃkāraśāstra.24 The opening passage reads: In this world the acceptance and the refusal of the things worthy of these two attitudes depend on valid cognition. And this [valid cognition] enjoys its validity insofar as it is based on certainty, and it is insofar as it is connected with language that certainty takes a thing as its object. And since language has a causal capacity without which one would not be able to account for [its] effect consisting in the comprehension of objects, one must postulate a function that is conveyed by terms such as denotation and so forth and that is [thus] multiple. Hence, it is in order to investigate this [issue] that this treatise that is The Analysis of Linguistic Functions is here initiated.25

Meaning beyond Words

403

In other words, without correctly obtained linguistic knowledge, that is, linguistic knowledge firmly based on certainty and thus epistemic validity, no morally meaningful choice could be made in front of the plurality of paths to be taken in human life.26 No stronger background constellation of concepts could be given to justify the philosophical study of language. For my present purpose, there is no need to investigate the hairsplitting subdivisions of denotation and indication as conceived by Mukula and accepted by Mammaṭa with several omissions and emendations.27 My current focus is the device that allows Mammaṭa to consent to the useful core of Mukula’s scholarly discourse by nonetheless defusing the powerful theoretical mechanism aimed at including the poetical meanings within the purview of indication (lakṣaṇā). To achieve this aim, it is necessary to start by considering the customary requirements that are needed for a lakṣaṇā to be triggered in the reader or listener of a sentence and how Mukula subtly alters them in order to include “the poetical,” in all its varieties, within the purview of the indicative function. The three requirements are mukhyārthabādha (the inapplicability of the primary meaning), sambandha (the relation between the denoted and the indicated meaning, also called by Mukula as mukhyārthāsatti, “proximity to the primary meaning”) and prayojana (the motive or purpose for the use of secondary signification).28 In the cherished example of the “hamlet on the Ganges,” the inapplicability of the primary meaning consists in the fact that it is physically impossible for the hamlet to be found straight on the very stream of the Ganges, the connection is the close proximity between the stream of the Ganges and its bank, the purpose is the implicit transfer of the qualities of coolness, purity, and the like from the stream of the Ganges to the hamlet itself. This is the standard account of the issue. As hinted at above, I argue that it is by a theoretical broadening of the concept of bādha, “inapplicability” or “obstruction,” that Mukula manages to achieve the theoretical feat of including poetical meanings within lakṣaṇā, without having to resort to a different function of language. Here comes one of Mukula’s most crucial examples, the instance of an emotionally charged poem: Cupid’s arrows are inescapable; in all directions spring blossoms forth; The rays of the moon madden the heart; the cuckoos steal the mind; And on top of it, this tender age is made intolerable by the burden of lofty breasts. O, my friend, how can I bear these five unbearable fires?29

With regard to this verse, the upholder of dhvani would argue that the aesthetic emotion (rasa) of “love in separation” (vipralambha-śṛṅgāra), or “love

404

Daniele Cuneo

thwarted” as Pollock (2016) has recently proposed to translate, cannot but be conveyed through the power of vyañjanā, “suggestion” or “manifestation,” since there is no obstruction to the understanding of the primary meaning of the words. The metaphorical language,30 at most, would just explain the harming nature of the five listed elements but not the intimate connection with the innermost feelings and torments of the young lady uttering this verse and how these are communicated to the listener of the stanza. With a similarly implicature-centered and rasa-oriented analysis in mind,31 Mukula comments the poem as follows: In this [verse] the unbearable character of cupid’s arrows and the other [four elements], on which the nature of fire is superimposed, is the meaning of the sentence. Hence it is also what is directly expressed (vācya) by that [sentence]. And it is by force of a reflection over the import of that [which is being said] that “love thwarted” (vipralambhaśṛṅgāra) is implied. Hence we have an inclusive indication [i.e., an indication in which the primary meaning as well is preserved, “included”] based on the sense expressed (vācya) by the sentence. Here there is no use of the consideration of the speaker’s character independently from the words [of the poem, as it is the case in other verses where the nature of the speaker determines the understanding of the listener, as in the famous example of the traveller enjoined not to fall in the lady’s bed, but actually prompted to do so], nor could the words of the sentence be construed without the implication of “love thwarted.” Since in this case the “love thwarted” is implied by a reflection on the form of what is directly expressed, this is a case of inclusive indication based on the expressed meaning.32

In this critical passage, as well as in others Mukula is re-interpreting and refunctionalizing, the very idea of bādha and along with it its positive counterconcept, the idea of yogyatā, “semantic compatibility,” usually refers to the mere congruity of sense within the various word-meanings of a sentence. A traditional example of a break in yogyatā would be a sentence such as vahninā siñcati, “he sprinkles with fire,” which is considered as no actual sentence, since a viable sentence meaning cannot be arrived at by way of the two seemingly incompatible word-meanings. Accordingly, “Mukulabhaṭṭa’s innovation is to expand the concept of compatibility, or yogyatā, to include consistency between contextual information and speaker intention, from the notion of compatibility as strictly involving semantic features internal to the sentence” (Keating 2013a, 440). To put it otherwise, in a way that will show how Mammaṭa understood the issue and managed to directly refute it, in Mukula’s reading, the preceding

Meaning beyond Words

405

verse makes no sense—that is, it has no semantic compatibility—without the secondarily conveyed implication of the emotion of thwarted love that is consciously meant and felt by the young lady uttering the sentence. The very point of the utterance, the real sentence meaning, is missed and therefore communication does not succeed unless the emotional flavor as meaning is also conveyed by the linguistic act and thereby understood by the listener. The feeling is the meaning. The emotional import is what has to be understood. By the clever device of fully including every shade of the speaker’s intention in what has to be understood for a sentence to make proper sense, in other words by extending the concept of bādha up to the inclusion of any understanding that falls short of the totality of the speaker’s intention, which comprises her inner emotional world that triggered the speech act, Mukula gets rid of the necessity of postulating a third function of language to account for affective meanings. In Mukula’s understanding of how human beings communicate, an understanding for which I cannot withhold my sympathy, some sentences just make no sense without grasping by way of an indicative implication the consciously intended emotional backdrop of what is being overtly expressed.

5  The “Conservative” Refutation of Mammaṭa Mammaṭa in his Kāvyaprakāśā, and more obviously in his Śabdavyāpāravicāra, attacks Mukula on his very ground by often quoting the same verses and by debunking, through their analysis, the all-encompassing nature of lakṣaṇā that Mukula supports. In Mukula’s system, virtually any linguistic expression conveys some figurative meaning, dependent as it is on an interpretation of the Bhāṭṭa Mīmāṃsaka understanding of language signification.33 Thus, in Mammaṭa’s Śabdavyāpāravicāra, the very same verse on the five fires is taken up, along with two other verses used by Mukula to exemplify other kinds of his lakṣaṇā,34 with the clear intent of refuting Mukula’s stance on the nonexistence of poetical suggestion (dhvani). To start, Mammaṭa simply states that the aesthetic emotion of “love thwarted” cannot be arrived at by indication because there is no obstruction of the primary meaning.35 Even more than this, in one of the subdivisions of poetry within the theory of dhvani as laid out by Ānandavardhana, it is precisely an unobstructed and directly expressed primary sense that determines the cognition of an aesthetic emotion. Clearly Mammaṭa does not subscribe to Mukula’s double expansion of the concept of yogyatā, on one side, and mukhyārthabādha, on the other.

406

Daniele Cuneo

Additionally, following a hint present in Ānandavardhana’s Dhvanyāloka and already fully developed by Abhinavagupta’s Locana,36 Mammaṭa takes issue with another aspect of Mukula’s theory. He argues that a linguistic function other than indication has to be accepted, when indication operates, in order to account for the understanding of the purpose (prayojana) of that very secondary usage. Such understanding, argues Mammaṭa, cannot be arrived at by any other means of knowledge, and he goes as far as to list the various means of knowledge (pramāṇas) and give reasons why it is not the case with none of them.37 More importantly, he argues that the purpose cannot be known through lakṣaṇā itself, and he gives four reasons for this: 1. Indication (lakṣaṇā) can be resorted to only if there is a purpose for it, and so the purpose cannot be its very object. For instance, the bank cannot be both the object of lakṣaṇā and the purpose for its use, otherwise there would be no difference between the simple denotation of the object “bank” and its indication. 2. There is no obstruction with regard to the indicated meaning, for example, the bank of the Ganges, which would be necessary to trigger the second indication which would account for the purpose. 3. There is no connection between the indicated meaning and the purpose, for example, between the bank of the Ganges and the qualities of coolness and purity; these extraordinary qualities are just connected with the river stream itself. Also there would not be any purpose for such a connection. 4. This retrieval of the purpose by indication would lead to an infinite regress, because one would need again a lakṣaṇā to understand the purpose of the lakṣaṇā used to account for the purpose of the first one and so on and on.38 Therefore, the purpose of lakṣaṇā must be arrived at by way of a further linguistic function, the vyañjanā that Mammaṭa has set his mind to prove. In his words, “Therefore, indication cannot be [the function at work in these examples]. But there is indeed that [cognition to be accounted for]. Consequently, one must accept a function whose object is the purpose [of indication], and it is what is expressed by terms such as ‘suggesting,’ ‘conveying,’ ‘manifesting,’ ‘revealing,’ etcetera.”39 In the passage of the Locana on which Mammaṭa’s reasoning is clearly based,40 Abhinavagupta had concluded its refutation of lakṣaṇā by arguing: “For if this operation [of conveying the idea of the Ganges’s coolness to the village] ran a halting course (skhaladgati)41 too, it could only be because its

Meaning beyond Words

407

primary goal was blocked, which could only occur because of some further purpose or intended goal (prayojana), so that an infinite series of intended goals would ensue (Ingalls et al. 1990, 87–88).” Mammaṭa’s historical role in this intellectual exchange was the reorganization and the active fine-tuning of Abhinava’s refutation by its expansion into a full-fledged and now four-spiked attack on Mukula’s stance, a personalized assault, especially insofar as it is embedded in a text that borrows so much from the Abhidhāvṛttamātṛkā. Mammaṭa then continues in the Śabdavyāpāravicāra to refute an objection that was only briefly dealt with in the Kāvyaprakāśa, the possibility that the object of indication is an entity comprehended as always and already qualified by the further elements that the upholder of “suggestion” would take as the very purpose of the indication to be conveyed by suggestion. In the standard example, the indicated meaning would then be something like the “bank possessed of the qualities of purity and the like” attributed to the Ganges.42 The passage of the reply deserves to be quoted in full to show the exhaustive nature of Mammaṭa’s discourse and how it again pivots on the three requirements of lakṣaṇā and especially on the definition of purpose or fruit (phala). We reply: the object of indication is the bank and so forth, and there is no purity and the like there. Thus, how could there be an indication of a qualified whole? Moreover, how could indication take as its object a qualified [and hence particular] whole, if convention is based on universals and that [indication] takes place on the basis of that specific kind of convention that consists in the three [requirements of “obstruction,” “relation” and “purpose”]? Yet, the object of suggestion is beyond convention. Furthermore, the domain of indication is the indicated meaning and not the purpose. In fact, blue is the object of perception, while [its] fruit is the vividness or the awareness [of that blue].43 And without fail the awareness of the indicated meaning is never said to be the fruit,44 for the same applies for the fruit of denotation. Moreover, the fruit of a linguistic expression is nothing but [its] awareness; and that is impossible without a function, and as a function cannot belong to another function, one must say that it belongs to language. Furthermore, as the presence of a fruit brought about by some function is regularly present when indication is present and absent when indication is absent, this must be its effect, and an effect is what is called “fruit.” The fruit of indication is nothing but the purpose brought about by the power of suggestion, just like45 the purpose obtained by different actions is the fruit of a sacrifice.46

To focus more closely on the general attitude that Mammaṭa takes toward the refutation of Mukula, it is fruitful to try and analyze his theoretical move from

408

Daniele Cuneo

the very viewpoint of the inner workings of lakṣaṇā. My contention is that Mammaṭa is pushing back against Mukula, precisely by closing in onto the other two requirements for the deployment of secondary signification, so that Mukula’s extension of the first requirement, bādha, gets thwarted. Mammaṭa shows how the three concepts—obstruction, connection, and purpose—are theoretically imbricated and umbilically tied. In other words, he demonstrates how, in his opinion, there is no way to change one of them without the collapse of the whole system. “Obstruction” (bādha) can only occur, and hence elicit a secondary signification, if the “connection” (sambandha) is between the primary meaning and the secondary meaning, and if the “purpose” (prayojana) for the secondary usage remains outside of secondary signification and lies within the activity of suggestion. So, just like with Alice eating her magic mushrooms, the transmogrified concept of bādha is shrunk back to its original function, the merely semantic inconsistency among the various word-meanings, and is thus forced to abandon the grand inclusion of both contextual information and speaker’s intention, which ultimately remain the purview of dhvani alone.

6 Conclusion As it is customary to indicate at least one further direction of research, a good start is the admission that the focus on the concept of bādha and the other requirements for lakṣaṇā has necessarily left out much of the complex web of arguments and counterarguments that are weaved together by Mammaṭa in his refutation of Mukula’s general stance and several of his specific points, both in the Kāvyaprakāśa and in the Śabdavyāpāravicāra. In order to fully grasp the reorganization as well as the intentional capsizing and overturning of Mukula’s scholarly discourse on Mammaṭa’s part, it would be necessary to pursue this research further, possibly by way of a visual grid containing the parallel passages of the Kāvyaprakāśā and the Śabdavyāpāravicāra, by thus highlighting the respective differences in argumentation, and of course a sentence-by-sentence confrontation with the text of the Abhidhāvṛttamātṛkā, which constitutes both the theoretical source and the polemical target of Mammaṭa’s attentive reuse and studious misuse of Mukula’s text.47 In the light of my analysis, the Śabdavyāpāravicāra represents a sui generis case of textual reuse as it is a scholarly work mostly constructed by using the words of a previous text, also through the medium of another text by its author, but with the clear aim of refuting the original source of his borrowings, and all this without even ever

Meaning beyond Words

409

mentioning the name of the text or of the author, let alone acknowledging the scholarly enterprise that lies behind such cunning misuse of reuse.48 As made clear by Gerow (1977, 272, n. 213), Rājānaka Mammaṭa’s oeuvre has not been highly valued by “modern critics, who generally regard Mammaṭa as the beginning of the ‘decline’ of Indian poetic thought.” Even Gerow himself, who is generally appreciative of Mammaṭa’s role as systematizer and synthesizer of Alaṃkāraśāstra views, has to admit that “Mammaṭa’s particular contribution to doctrine was very limited.” My current analysis of Mammaṭa’s theories and arguments as almost always based on Abhinavagupta’s work or as wholesale appropriations from Mukula’s Abhidhāvṛttamātṛkā does indeed show the cogency of this interpretation. However, my further contention is that the issue of originality cannot be simply reduced to the creation of new doctrines or arguments, but must entail also an author’s capacity to combine and reorganize into a new and all-encompassing entity a whole array of disparate theories, as in our current example of Mukula’s ideas on lakṣaṇā and Abhinavagupta’s version of the dhvani theory. With the risk of committing a teleological fallacy in intellectual history, one might well affirm with Pollock (2016, 224) that Mammaṭa’s work had the privilege and the merit of meeting the cultural and historical “need for a more comprehensive assessment of literary art.” By contrast, Mukula’s work deserves a very different evaluation, its originality in the conceptualization of lakṣaṇā being the main characteristic feature of his intellectual feat. However, its doctrine fails the test of history and gets absorbed only in a highly sanitized version by Mammaṭa’s appropriation. I do not think that the cogency of rational arguments can be at all regarded as the litmus test for the survival of any doctrine in the history of philosophy, but I do see it probable that Mukula’s failure to argumentatively release his highly sophisticated theory from the straitjacket of the three requirements of lakṣaṇā did play a role in the lack of success that his Abhidhāvṛttamātṛkā seemed to encounter. The idea that the purpose (prayojana or phala) of an indication must be necessarily arrived at by a different function was already raised by Ānandavardhana before Mukula himself, and only perfected and rehashed by Abhinavagupta and Mammaṭa. Therefore, Mukula’s analysis of the prayojana in the process of indication as “certainly not as thorough or clear as one might wish” (McCrea 2008, 300) remains intellectually puzzling and at least a plausible cofactor in the historical determination of the cultural trajectory that the Abhidhāvṛttamātṛkā ended up taking. To really conclude, I’ll circle back to the wording of my “indicative” title, “implicature wars” or ākṣepayuddha in its Sanskrit incarnation.49 I doubt that anybody has ever been slaughtered in the discursive wars over the functions

410

Daniele Cuneo

of language. However, the use of this particular instance of lakṣaṇā hinging on the word “wars,” which is meant to secondarily convey the idea of “heated debates,” must indeed have a specific purpose, as by now repeated ad nauseam. The purpose I intended is the suggestion of the crucial importance of linguistic communication in human life and therefore of linguistic exegesis as the pathway to veridical understanding. As the incipit of the Śabdavyāpāravicāra made clear, the entanglement of comprehension, certainty, epistemic validity, and morality haunts human existence and cannot but force human beings organized in complex societies to investigate epistemological issues within the field of linguistic philosophy, particularly with regard to the exegesis of normative texts. The case of Mīmāṃsā looms large in the Sanskrit culture, but also the issue extends way beyond the limits of strictly normative language. This was clearly the case in medieval Kashmir, especially because the wars over implicature were waged within a theoretical framework that fully recognized the crucial power that art wields over the shaping of the human “moral competence.”50 The exegetical hegemony over the normative capacity that art has to ethically educate and emotionally mold humankind was the bountiful kingdom to be conquered and defended over and over again in the discursive battles of the implicature wars.

Notes 1 I am deeply grateful to Alessandro Graheli for his precious remarks and suggestions. All mistakes, of course, are mine alone. 2 On this ecumenical nature of Alaṃkāraśāstra, especially when it merges with the Sanskrit dramaturgical tradition of Nāṭyaśāstra, see Cuneo 2016a. On the peculiar nature of Alaṃkāraśāstra in comparison with other knowledge systems (śāstra) of the Sanskrit episteme, see McCrea 2011 and Cuneo 2017. 3 On the loaded concept of paradigm shift, its Kuhnian pedigree, and the meaningfulness of employing it as a theoretical instrument in drafting the intellectual history of Sanskrit aesthetics, see McCrea 2008, 19–26. 4 On Bhaṭṭa Nāyaka, see the mentioned Pollock 2010 and Pollock 2016, without forgetting David 2016 and its rejoinder Ollett 2016. On the transformation of aesthetic emotions as ontologically different from ordinary emotions as the crucial philosophical move in the history of Sanskrit aesthetics, see Cuneo 2009 and Cuneo 2013. 5 The crucial weight of the philosophical model of teleological exegesis heralded by Mīmāṃsa in the development of this radical reinterpretation of the poetic experience is the central contribution of McCrea 2008.

Meaning beyond Words

411

6 This specifically Kashmirian story has indeed been told many times in great details, most recently and most originally in McCrea 2008. Starting with Mammaṭa, and more crucially after him, the core subject of controversy and theoretical development changes back to the definitions, classifications, and multiplications of the figures of speech, the alaṃkāras, the topic that kick-started and christened the very knowledge system of Alaṃkāraśāstra and that was partly subsumed and merged within the issues of implicature and aesthetic emotions during the centuries of the Kashmirian intellectual wars around these concepts. On alaṃkāra as poetical ornamentation, see at least the foundational Gerow 1971 and the eagerly awaited Bronner (forthcoming). 7 For the little information we have on Mukula’s life and cultural pedigree, see McCrea 2008, 261–66, a treasure trove of updated discussion on the life and work of all Kashmirian poeticians before Mammaṭa. The Abhidhāvṛttamātṛkā of Mukula (also improperly known as Abhidhāvṛttimātṛkā) is translated in Venugopalan 1977. A rich analysis of Mukula’s stance is again in McCrea 2008, 261–310. Not many other articles are dedicated to Mukula’s work: noteworthy are Agrawal 2008, Keating 2013a, and Keating 2013b. The quoted portions of the text will be referenced from Abhidhāvṛttamātṛkā, the editio princeps of Telang, which contains also Mammaṭa’s Śabdavyāpāravicāra, the other focus of the present chapter, also referenced from this edition. Other editions of Mukula’s work (Dwivedi, Avasthī, Mahanty) are mentioned in the Bibliography for the sake of completeness. 8 On this renowned division and a plethora of examples thereof, see the translation of the Dhvanyāloka and its commentary by Ingallset al. (1990, 80–82, 94, n. 11, and passim). The complex relation of partial acceptation and partial refutation that Mukula’s work shows with regard to the theory of dhvani reaches its apex in a troubling a passage of the Abhidhāvṛttamātṛkā in which the term vyaṅgya, “suggested sense,” is used approvingly without any hint at its refutation. McCrea (2008, 304, n. 79) cannot but comment: “Mukulabhaṭṭa’s use of the term ‘suggested’ (vyaṅgya) here is odd and seemingly rather careless.” Moreover, the whole portion of the text in which Ānandavardhana’s theory is more directly dealt with is marked off, at the end, by way of the possibly customary remark alam atiprasaṅgena, “enough with this digression” (Abhidhāvṛttamātṛkā, p. 21).   9 For some caveat on the absolute cogency of the interpretation that regards this refutation as the very aim of Mukula’s work, see the previous note. 10 For a study of lakṣaṇā starting from this very example, see Gren-Eklund 1986. 11 The scarcity of studies on Mukula is already mentioned in note 6, to which one can add that no ancient Sanskrit commentary is known. As to Mukula’s syncretic philosophical stance, it could well be sketched as follows: a monistic conception of language and reality as its theoretical and metaphysical background—directly deriving from Bhartṛhari’s philosophy of grammar—is joined with a realistic

412

12

13

14

15

16

Daniele Cuneo conception of language and the world at the level of common-day experience— mostly borrowed from the school of Bhāṭṭa Mīmāṃsā. Its importance can be easily deduced by way of a cursory glance at the entry of the Kāvyaprakāśa and its commentaries in New Catalogus Catalogorum or at the many pages dedicated to it in Cahill’s An Annotated Bibliography of the Alaṃkāraśāstra (Cahill 2001, 23–37), which goes as far as calling the Kāvyaprakāśa “the most often commented upon śāstra in Sanskrit literature (Cahill 2001, 23).” Pollock’s words (2016, 224) are also revealing: “The Light [Mammaṭa’s Kāvyaprakāśa] has been preserved in thousands of manuscripts all across India, and has attracted scores of commentators beginning as early as the mid-twelfth century, one of whom [Bhīmasena Dīkṣita in his Sudhāsāgara] refers to Mammaṭa in all sincerity as an ‘incarnation of Sārasvatī, goddess of language.’” Quoted passages from the Kāvyaprakāśa will be referenced from Abhyankar’s edition with the commentary of Māṇikyacandra (Saṅketa). For a quick summary of the issues regarding Mammaṭa’s date and thought as well as a viable bibliography, see Gerow 1977, 271–74. The acceptance of Mukula’s subdivisions of lakṣaṇā had a considerable fortune, visible even up to the most renowned Sāhityadarpaṇa by Viśvanātha Kavirāja (see Sāhityadarpaṇa, verses 2.6 and following) and beyond. This twofold subdivision boasted a significant pedigree, as it had been borrowed from the Tantravārttika of the Mīmāṃsaka Kumārila Bhaṭṭa and had in a way become the standard explanation of secondary signification in the late secondmillennium Sanskrit episteme. See Kumārila, Tantravārttika, 354, ad 3.4.12: abhidheyāvinābhūtapratītir lakṣaṇocyate / lakṣyamāṇaguṇair yogād vṛtter iṣṭā tu gauṇatāṃ iti //. For a history of secondary signification up to Kumārila and beyond, see Kunjunni Raja 1963, which goes through crucial loci classici such as Mahābhāṣya, vol. 2, p. 218, Mīmāṃsāsūtra, I.4.23 and Nyāyasūtra, II.2.62. A most recent work on secondary signification in the Buddhist milieu, which also surveys the whole early history of the topic, is Tzohar 2018. See Ingalls et al. 1990, 64–66, 84–88, 178–97. However, Abhinavagupta does state that the view that dhvani can be reduced to indication can be found “in written texts (Ingalls et al. 1990, 64)” (Locana, 28: pustakeṣu), glossed by the Bālapriyā as “in works on alaṅkāras (alaṅkāragrantheṣu).” It is the concealed presence of Mukula that might indeed be lurking beneath the surface of these vague references. This tentative identification might well be mistaken, because the term lakṣitalakṣanā is not restricted to Mukula only, but used, for instance, by Śālikanātha in his Prakaraṇapañcikā. However, to my knowledge, Mukula is the only one who uses it before Abhinavagupta in a text that deals with poetical and emotional language. However, the term lakṣita-lakṣaṇā does not play a central role in the text of Mukula.

Meaning beyond Words

413

17 Locana, ad 1.4, p. 61: ata eva yat kenacil lakṣitalakṣaṇeti nāma kṛtaṃ tad vyasanamātram. The rendering of vyasana as “stubborn perversity” might just be a bit of an over-interpretation by the translators, or it is probably a rendering of the pertinent gloss given by the Bālapriyā commentary, p. 61, which understands vyasanam as dhvaninirākaraṇanirbandhamātram, “the mere obstinacy in refuting dhvani,” which does point toward the identification of the upholder of this version of lakṣita-lakṣaṇā with someone directly involved in the debate centered on dhvani. Moreover, while commenting the portion on lakṣaṇā in Mammaṭa’s Kāvyaprakāśa, Māṇikyacandra (p. 27) concludes a passage in which he clearly has Mukula’s theory in mind by quoting this very yat kenacit reference by Abhinavagupta. This suggests that at least Māṇikyacandra could have well identified Mukula behind that mysterious kenacit. A more daring proposal would be to emend the text of Locana, ad 1.4 according to a parallel reference found in Locana, ad 1.17. What is found there (p. 149) is lakṣaṇa-lakṣaṇā, that is, precisely the term Mukula used to account for the “standard instance” of indication, the gaṅgāyāṃ ghoṣaḥ (Abhidhāvṛttamātṛkā, pp. 6–7 and passim), which is also the constant reference of Abhinava’s discussion. Ingalls et al. 1990, 187, n. 2 have emended lakṣaṇa-lakṣaṇā in lakṣita-lakṣaṇā on the basis of the text edited by Kuppuswami Śāstri together with the support of a fifteenth-century Keralese commentary by Uttuṅgodaya called Kaumudī (p. 274), and the twentieth-century commentary called Balāpriyā. In my opinion, however, it is at least a reasonable possibility that the wording lakṣita-lakṣaṇā found in the Locana ad 1.4 might well be a lectio facilior for lakṣaṇa-lakṣaṇā, and that consequently, in both cases, what Abhinvagupta was arguing against is precisely Mukula’s central theoretical term. This hypothesis would put on much firmer ground my tentative identification of Mukula behind the indeterminate persona concealed by Abhinavagupta’s kenacit. It is thus not out of the realm of possibilities that part of the textual transmission, the Sanskrit commentators and the contemporary editors and translators missed the implicit reference to Mukula’s work and read into the text the more common term lakṣita-lakṣaṇā. Of course, no final word can be pronounced on the matter before assessing the plausibility of my proposition by a thorough survey of the manuscripts of the Locana and the perusal of the two other existing Keralese commentaries on Abhinavagupta’s passage (see edition by Neelakandhan), which obviously lies beyond the scope of the present chapter. 18 Abhidhāvṛttamātṛkā, pp. 15–16. 19 Abhinavagupta’s discussion occupies the crucial part of Locana, ad 1.4, translated in Ingalls et al. 1990, 83–98. For a bird’s eye view explanation of Abhinavagupta’s theory of signification and its four phases, see Cuneo 2016b, 157–58. 20 Another piece of the puzzle is the fact that both Mukula and Abhinavagupta quote the verse of the Mīmāṃsā author Bhrtṛmitra on the five types of lakṣaṇā, but in

414

Daniele Cuneo

three slightly different versions. Mukula (Abhidhāvṛttamātṛkā, p. 17) reads it as abhidheyena saṃbandhāt sādṛśyāt samavāyataḥ / vaiparītyāt kriyāyogāl lakṣaṇā pañcadhā matā. Abhinavagupta quotes the verse twice. The first time, while commenting on 1.1, he reads the first hemistich as abhidheyena sāmīpyāt sārūpyāt samavāyataḥ. Then, while commenting on 1.18, he reads it as abhidheyena saṃyogāt sāmīpyāt samavāyataḥ, “which would exclude the metaphorical variety of secondary usage from lakṣaṇā (Ingalls et al. 1990, 67).” Of course, the difference in the readings might just be a matter of textual transmission, possibly to be settled only by critically editing the two works in questions. Further evidence on the relation between Abhinavagupta and Mukula is given in the study and translation of the Abhidhāvṛttamātṛkā by Keating (2019, 134, 269, n. 36, 270, n. 40, 274, n. 72, 276, n. 98, 278, n. 127). 21 See, for instance, the lapidary statement by Mammaṭa, Kāvyaprakāśa, p. 106: rasādilakṣaṇas tv arthaḥ svapne ’pi na vācyaḥ. 22 The Śabdavyāparavicāra is certainly later than the Kāvyaprakāśa, as the former reads at the end: etac cānyatra vistareṇa vicāritam iti saṃkṣepeṇehoktam iti śivam //. “Furthermore, this [linguistic function of manifestation] has been treated in details elsewhere [i.e., in the Kāvyaprakāśa]. Thus, it has been stated here in brief. Good be with you (Śabdavyāpāravicāra, p. 10).” Another typical trait of a later work is the fact that Mammaṭa in the Śabdavyāpāravicāra tightens up his view on some topics that he had left open to various possibilities in his Kāvyaprakāśa. For instance, the possible kinds of direct meaning (mukhyārtha) for which in the Kāvyaprakāśa Mammaṭa presents without showing his preference two rival doctrines, the conception of the four different kinds of directly signifying words (as related to universals, qualities, actions, and proper names) as propounded by the grammarians, starting with Patañjali, and the conception of the only existence of universal-denoting words attributed to the Mīmāṃsakas. In the Śabdavyāpāravicāra, p. 2, Mammaṭa clearly sides with the former view and argues in its favor in a passage that has no direct parallel in the Kāvyaprakāśa. 23 On Mammaṭa’s ambiguous position regarding the possibility of a fourth faculty of language, tātparyaśakti, accepted by Abhinavagupta on Bhaṭṭa Jayanta’s influence, see Cuneo 2016b. On tātparyaśakti, see Graheli 2016. 24 I am suggesting that by way of his Abhidhāvṛttamātṛkā Mukula is the inventor of the subgenre of short essay—a scholarly article if you will—dealing solely with the functions of language. To this subgenre it is possible to ascribe both the Śabdavyāpāravicāra by Mammaṭa and the much later Vṛttivārttika by Appaya Dīkṣita (see Gerow 2001), an extremely common textbook in present-day South Indian pāṭhaśālās. 25 Śabdavyāpāravicāra, p. 1: iha heyopādeyānāṃ hānopādāne pramāṇād eva / tac ca niścayātmatayā prāmāṇyaṃ bhajate / niścayaś ca śabdasāhityenārthaṃ viṣayīkaroti / śabdasya cārth​aprat​ītipr​atipa​ttila​kṣaṇa​kāryā​nyath​ānupa​patty​ā  

Meaning beyond Words

415

kārakatvāt kalpyamāno vyāparo ’bhidhādiśabdapratipādyo nānāprakāra iti tatparīkṣārthaṃ śabdavyāpāravicārātmakaṃ prakaraṇam idam ārabhyate //. The word pratipatti in the compound cārth​aprat​ītipr​atipa​ttila​kṣaṇa​kāryā​nyath​ānupa­­ patty​ā might well be a marginal gloss that has crept its way into the text, and hence I would be tempted to expunge it and I am not translating it. This passage is a tight and powerful summary of the incipit of Mukula’s Abhidhāvṛttamātṛkā, p. 1, translated in Venugopalan 1977, 219. 26 On the mutual imbrication of morality, language and the emotions as choices, see Cuneo 2015. For more thoughts on the implicit centrality of the issue, see Conclusion as well. 27 As a way of a relevant example of disagreement, one might mention a passage that is shared by the Kāvyaprakāśa and the Śabdavyāpāravicāra that is aimed at refuting the necessity of lakṣaṇā in the move from word-meanings to sentencemeaning. Mukula, following an interpretation of Kumārila, argues that the singular words directly signify universals only and that an “inclusive indication” (upādānalakṣaṇā) is needed to arrive at the particulars actually meant in sentences such as gaur anubandhyaḥ, “the cow must be tied up [for the sacrifice].” Mammaṭa accepts the “inclusive indication” for time-honored cases such as kuntāḥ praviśanti, but argues for the invariable concomitance between universal and particular as a sufficient ground for accounting for the passage from one to the other in the common linguistic expression and thus resists the veritable tsunami of indications needed in practically every singular speech act according to Mukula’s Mīmāṃsā-influenced theory. The end of the passage (Kāvyaprakāśa, p. 18 = Śabdavyāpāravicāra, p. 3) reads: “On the contrary, the individual is implied by the universal because of its invariable concomitance with the individual, just like there is an agent [implied] in ‘Let it be done,’ an object in ‘Do,’ and ‘a house,’ an imperative ‘eat,’ etcetera in expressions such as ‘enter, cake,’ etcetera” (vyaktyavinābhāvāt tu jātyā vyaktir ākṣipyate yathā kriyatām ity atra kartā, kurv ity atra karma, praviśa piṇḍīm ityādau gṛhaṃ bhakṣayetyādi ca). 28 The historical origin of this threefold criterion for indirect signification remains to be thoroughly investigated. The idea is certainly already put forward by Ānandavardhana, and survives almost unquestioned throughout the history of Alaṃkāraśāstra. The third requirement, in particular, the motive (prayojana), seems to be absent from the earliest understanding of secondary signification such as in Śabara or Vātsyāyana. The possible origin of this concept and its change in meaning are sketched in McCrea 2008, 192–93, n. 43: “For the Mīmāṃsakas, the figuratively expressed meaning must have some useful application (prayojana) in the appropriate (ritual) context. It is this criterion of usefulness that allows us to determine which of the many possible figurative senses of a term should be understood in any particular case. Ānandavardhana and the Ālaṃkārika who come

416

29

30

31

32

33

34

35

Daniele Cuneo after him understand the need for a prayojana in cases of figurative usage not as the requirement that the particular meaning expressed figuratively should apply usefully in the relevant context (this seems to be taken for granted), but rather that the resort to figurative expression itself should serve some communicative purpose (i.e. that there should be some reason for describing something figuratively rather than by a term which literally denotes it). While I have not been able to trace its history in any detail, this different understanding of prayojana is not original to Ānandavardhana; Dharmakīrti (ca. 650 CE) uses the term in this sense in his own commentary on Pramāṇavārttika, 3.1 (p. 170).” Abhidhāvṛttamātṛkā, p. 14: durvārā madaneṣavo diśi diśi vyājṛmbhate mādhavo / hṛdy unmādakarāḥ śaśaṅkarucayaś cetoharāḥ kokilāḥ / uttuṅgastanabhāradurdharam idaṃ pratyagram anyad vayaḥ / soḍhavyāḥ sakhi sāṃprataṃ katham amī pañcāgnayo duḥsahāḥ // An interesting implicit reference is the hint at the extreme ascetic practices during which a practitioner is surrounded by five fire, four actual burning pyres and the scorching Indian sun high up in the sky. As mentioned before (see note 8), one of the many subdivisions proposed by Mukula in his Abhidhāvṛttamātṛkā is simply a borrowing of Ānandavardhana’s subdivision of the three types of dhvani: vastudhvani, alaṃkāradhvani and rasadhvani. Abhidhāvṛttamātṛkā, p. 14: iti. atra hi smaraśaraprabhṛtīnāṃ pañcānām adhyāropitavahnibhāvānām asahyatvaṃ vākyārthībhūtam, atas tasya vācyatā / tātparyālocanasāmarthyāc ca vipralambhaśṛṅgārasyākṣepa ity upādānātmikā lakṣaṇā vācyanibandhanā / na hy atra vaktṛsvabhāvapariśīlanasya śabdarahitasyopayogaḥ / nāpi ca vākye padānāṃ vipralambhaśṛṅgārākṣepam antareṇānvayopapattiḥ / vācyasvarūpavicāreṇa tatra vipralambhaśṛṅgārākṣepād upādānātmikā lakṣaṇā vācyanibandhanā /. Some new manuscript evidence currently under scrutiny seems to offer some important variants for this passage, which might well modify the theoretical picture I have drawn from it in this section of the article. In support to this theory, Mukula cites an often quoted verse that Kamalākarabhaṭṭa attributes to Kumārila (possibly from his lost Bṛhaṭṭīkā): “Denotation cannot touch the qualified [particular] as its power is exhausted in the qualifying element [that is the universal]. (viśeṣyaṃ nābhidhā gacchet kṣīṇaśaktir viśeṣane).” By contrast, Ingalls et al. (1990, 92, n. 1) report the opinion of the Bālabodhinī commentary of Jhalkikar on the Kāvyaprakāśa, p. 44 that this quote comes from Maṇḍana Miśra. The two other verses are dṛṣṭiṃ he prativeśini (Abhidhāvṛttamātṛkā, p. 12 = Śabdavyāpāravicāra, p. 5) and prāpraśrīr eṣa (Abhidhāvṛttamātṛkā, p. 13 = Śabdavyāpāravicāra, p. 5). Śabdavyāpāravicāra, p. 5: vipralambhaśṛṅgāraś ca raso lakṣyata iti nodāhāryam / mukhyārthasya bādhābhāvāt.

Meaning beyond Words

417

36 See Dhvanyāloka 1.17 and Locana on it (Ingalls et al. 1990, 185–88). 37 Śabdavyāpāravicāra, pp. 5–6: “Moreover, this [purpose], unlike the reason for the obstruction of the direct meaning, cannot be known from any other means of knowledge. It is for this very reason that one uses the term ‘indication.’ Indeed perception has no bearing on a linguistic meaning, and not even inference does, as it is based on a previous perception, and not even a second inference would, as this would imply a regressus ad infinitum. Memory is of no avail, because of the absence of a [previous] experience of the object. (tac ca na mukhyārthabādhanimittavat pramāṇāntarād boddhavyam / tadartham eva lakṣaṇāśabdaprayogāt / na khalu śābde ’rthe pratyakṣaṃ kramate / nāpi tatpūrvakam anumānam / nānumānāntaram anavasthāpatteḥ / na smṛtiḥ tadanubhavābhāvāt /).” Again, the passage follows Abhinavagupta’s treatment in the Locana ad Dhvanyāloka 1.4. See, in particular, Ingalls et al. 1990, 87. 38 Śabdavyāpāravicāra, p. 6: na lakṣaṇā tasmin sati hi sā na tu tadviṣayā / nāpy asyā lakṣye bādho ’sti / lakṣyaprayojanayoś ca sambandhasya prayojanasya cābhāvāt /   tasyāpi lakṣaṇe ’navasthāpattir iti na lakṣaṇā syāt. This passage is a summary and a paraphrase of what was argued in more general terms, and not specifically against a verse brandished by Mukula, in Mammaṭa’s Kāvyaprakāśa, vṛtti ad 16 and 17ab, p. 27: yathā gaṅgāśabdaḥ srotasi sabādha iti taṭaṃ lakṣayati tadvat yadi taṭe ’pi sabādhaḥ syāt tat prayojanaṃ lakṣayet / na ca taṭaṃ mukhyo ’rthaḥ / nāpy atra bādhaḥ / na ca gaṅgāśabdārthasya taṭasya pāvanatvādyair lakṣaṇīyaiḥ saṃbandhaḥ / nāpi prayojane lakṣye kiñcit prayojanaṃ / nāpi gaṅgāśabdas taṭam iva prayojanaṃ pratipādayitum asamarthaḥ, and evam api prayojanaṃ cel lakṣyate tat prayojanāntareṇeti tad api prayojanāntareṇeti prakṛtāpratītikṛd anavasthā bhavet / 39 Śabdavyāpāravicāra, p. 6: iti na lakṣaṇā syāt / asti ca sā / tataḥ prayojanaviṣayo vyāpāro ’bhyupagantavyaḥ / sa ca dhvan​anāva​gaman​aprak​āśana​dyota​nādiś​a­­­ bdavyavah​āryaḥ​ /. 40 As cursorily noted in Ingalls et al. 1990, 95, n. 19. 41 As an aside, skhaladgati is the term that Ānandavardhana uses for describing the, so to say, subjective experience of “cognitive stumbling” when bādha, the obstruction of the primary meaning, is realized; the term probably comes from the Buddhist Pramāṇavāda tradition, as before Ānandavardhana I could only find it attested in Dharmakīrti’s Pramāṇavārttika and in the commentaries on it, and Ānandavardhana was probably close to the Buddhist logico-epistemological school as he is supposed to have written a now lost commentary on Dharmottara. 42 Śabdavyāpāravicāra, p. 6: nanu gaṅgātaṭe ghoṣa ityādisvaśabdād yādṛgarthapratītir na bhavati tādṛśī gaṅgāyāṃ ghoṣa ityādeḥ / tad eva prayojanam / pāvanatvādidharmākrāntaṃ ca taṭādi lakṣyate iti viśiśṭe lakṣaṇā na tu lakṣite viśeṣāḥ.

418

Daniele Cuneo

43 According to the Saṅketa commentary of Māṇikyacandra on the shorter parallel passage in the Kāvyaprakāśa, p. 28, the idea of vividness (prakaṭatva) belongs to the theory of Bhāṭṭa Mīmāṃsā, while the idea of awareness (saṃvitti) pertains to the Prābhākara view. 44 The editions I have consulted all contain a mistake here and simply read phalatvenoktaḥ, but the manuscript BORI 263/1875–1876 reads the sentence with the negative that it certainly needs to make proper sense: phalatvena noktaḥ (folio 4 verso, line 3). 45 If I am not mistaken, the idea is that what is obtained by actions that are not at all part of the sacrifice such as tending the fields or copulating with one’s own wife do produce the fruit of the sacrifice as their effects such as a plentiful crop or the birth of a son. 46 Śabdavyāpāravicāra, p. 6: ucyate / lakṣaṇāyās taṭādir viṣayaḥ / na ca tatra pāvanatvādayaḥ santi / tat kathaṃ viśiṣṭe lakṣaṇā / sāmānyaniṣṭhaś ca saṅketa iti tritayātmanā saṅketabhedena lakṣaṇā sā kathaṃ viśiṣṭaṃ gocarīkuryāt / vyajyate tv asaṅketita eva / kiṃca lakṣaṇāyā gocaro lakṣyo na prayojanam / pratyakṣasya hi nīlaṃ viṣayaḥ / prakaṭatā saṃvittir vā phalam / avyabhicārāc ca lakṣyasaṃvit phalatvena noktā / evaṃ hy abhidhāyā api phalaṃ / śabdasya ca vācakasya saṃvid eva phalam / tac ca vinā vyāpārān na sambhavati / na ca vyāpārasya vyāpāra iti śabdasyaivāsau vācyaḥ / lakṣaṇānvayavyatirekānuvidhānaṃ ca savyāpārasya phalasyeti tasyās tat kāryaṃ kāryaṃ ca phalam ucyate / kriyāntaraprāpyasya yāgaphalatvam iva vyañjanaśaktinirvartyasya prayojanasya lakṣaṇāphalatvam. 47 Just to give a quick example, out of the only six verses of the Śabdavyāpāravicāra—it is a short text after all—three are either obvious paraphrases or word-by-word borrowings from Mukula’s Abhidhāvṛttamātṛkā. Other important sources that might be investigated for a proper assessment of Mukula’s influence on Mammaṭa are the commentaries on the Kāvyaprakāśa, especially the ancient ones that seem to know the text of the Abhidhāvṛttamātṛkā firsthand, such as the works of Someśvara Bhaṭṭa, Ruyyaka and Māṇikyacandra. 48 On textual reuse in South Asian śāstric texts in general, see Freschi and Maas 2017. 49 The term ākṣepa is employed differently by different authors. In my jocose Sanskrit title, I use it in a general sense, just in order to indicate indirect meanings of any kind. Mukula uses it as the specific operation at play in the kind of lakṣaṇā that he dubbed as “inclusive.” Mammaṭa uses the term in the sense of an “automatic” logical implication that does not entail any use of lakṣaṇā. See, for instance, Kāvyaprakāśa, p. 22 = Śabdavyāpāravicāra, p. 4: “Moreover, when there is an invariable concomitance [such as the one between universal and particular] there is no use of the indicative function as [the sense] is already established simply by implication (avinābhāve cākṣepeṇaiva siddher lakṣaṇāyā nopayoga ity uktam).” 50 The scholarship on the intricate and tight relations between art and morality is understandably immense. A good book written from a Western perspective is Gaut

Meaning beyond Words

419

2007. My contribution on the issue as an interpretation of Abhinavagupta’s thought is laid out in Cuneo 2015, which is a concerted defense of the power and centrality of art in the development of a unified cognitive, emotional, and moral organ as its proper purpose. The straightforward connection between normative/prescriptive and poetical/affective texts in their capacity to prompt action is played out in the most evident way in a renowned passage by Bhaṭṭa Nāyaka quoted in Abhinavagupta’s Abhinavabhāratī, which has recently been the object of much updated and crucial scholarship (see, again, Pollock 2010; David 2016; and Ollett 2016).

Primary Sources Abhidhāvṛttamātṛkā

Abhidhāvṛttimātṛkā of Mukula Bhaṭṭa. In AVM-ŚVV.

Abhidhāvṛttamātṛkā (Avasthī)

Avasthī, Brahmamitra, ed. 1977. Abhidhāvṛttimātṛkā Mukulabhaṭṭakṛtā. Delhi: Indu Prakashan.

Abhidhāvṛttamātṛkā (Dwivedi)

Dwivedi, Rewa Prasada, ed. 1973. Abhidhāvṛttamātṛkā by Rājānaka Mukulabhaṭṭa with Hindī Translation and Explanation. Vidyabhawan Sanskrit Granthamala 165. Varanasi: Chowkhamba Vidyabhwan.

Abhidhāvṛttamātṛkā (Mahanty)

Mahanty, Sugyan Kumar, ed. 2008. Abhidhāvṛttimātṛkā of Mukulabhaṭṭa: With Subodhinī First Sanskrit Commentary and Saṅgamanī Hindī Commentary. Krishnadas Sanskrit Series 216. Varanasi: Chaukhamba Krishnadas Academy.

AVM-ŚVV

Telang, Mangesh Ramkrishna, ed. 1916. Abhidhāvṛttimātṛkā of Mukula Bhaṭṭa and Śabdavyāpāravicāra of Rājānaka Mammaṭācārya. Bombay: Nirnaya Sagar Press.

Dhvanyālōka

Sastri, Pattabhirama, ed. 1940. The Dhvanyāloka of Śrī Ānandavardhanāchārya: With the Lochana & Bālapriyā Commentaries by Śrī Abhinavagupta & Paṇḍitrāja Sahṛdayatilaka Śrī Rāmaśāraka. Benares: Chowkhamba Sanskrit Series Office.

Gautamīyanyāyadarśana

Thakur, Anantalal, ed. 1997. Gautamīyanyāyadarśana with Bhāṣya of Vātsyāyana. Nyāyacaturgranthikā 1. New Delhi: Indian Council of Philosophical Research.

420

Daniele Cuneo

Kaumudī

Kaumudī by Uttuṅgodaya. In Kuppuswami Sastri 1944.

Kāvyaprakāśa

Abhyankar, Vasudeva Shastri, ed. 1921. Kāvyaprakāśa of Mammaṭa with the Saṃketa by Māṇikyacandra. Pune: Anandasrama.

Locana

Dhvanyālokalocana of Abhinavagupta. In The Dhvanyāloka of Śrī Ānandavardhanāchārya: With the Lochana & Bālapriyā Commentaries by Śrī Abhinavagupta & Paṇḍitrāja Sahṛdayatilaka Śrī Rāmaśāraka.

Mahābhāṣya

Kielhorn, Franz, ed. 1880–1885. The VyākaraṇaMahābhāṣya of Patañjali. 3 vols. Third edition, revised and furnished with additional readings, references and select critical notes by K.V. Abhyankar, Pune: Bhandarkar Oriental Research Institute, 1962–72. Bombay: Government Central Book Depôt.

Mīmāṃsādarśana

Abhyankar, Kashinath Vasudev, and Ganesh Shastri Ambadas Joshi, eds. 1970a–1974. Mīmāṃsādarśana. Anandasramasamskrtagranthavali 97. Pune: Anandasrama.

Mīmāṃsāsūtra

Pūrva Mīmāṃsā Sūtra of Jaimini. In Mīmāṃsādarśana.

Nyāyasūtra

Nyāyasūtra of Gautama. In Gautamīyanyāyadarśana.

Pramāṇavārttika

Pandeya, Ram Chandra, ed. Pramāṇavārttikam of Dharmakīrti. In The Pramāṇavārttikam of Ācārya Dharmakīrti: With the Commentaries Svopajñavṛtti of the author and Pramāṇavārttikavṛtti of Manorathanandin.

Śabdavyāpāravicāra

Śabdavyāpāravicāra of Rājānaka Mammaṭācārya. In AVM-ŚVV.

Sāhityadarpaṇa

Krsnamohan Sastri. 1947–1948. Sāhityadarpaṇa of Viśvanātha. Kashi Sanskrit Series 145. Benares: Chowkhamba Sanskrit Series Office.

Tantravārttika

Tantravārttika of Kumārila Bhaṭṭa. In Mīmāṃsādarśana.

Meaning beyond Words

421

Secondary Sources Agrawal, M.M. 2008. “Mukulabhaṭṭa and Vyañjanā.” In Linguistic Traditions of Kashmir. Essays in Memory of Paṇḍit Dinanath Yaksha, ed. by Mrinal Kaul and Ashok Aklujkar, 28–40. New Delhi/Jammu: D.K. Printworld/The Harabhatta Shastri Indological Research Institute. Bronner, Yigal. 2016. “Understanding Udbhaṭa: The Invention of Kashmiri Poetics in the Jayāpīḍa Moment.” In Franco and Ratié 2016, 81–147. Bronner, Yigal. Forthcoming. A Reader on Alaṃkāra: A Historical Sourcebook in Classical Indian Ornamental Theory. New York: Columbia University Press. Cahill, Timothy C. 2001. An Annotated Bibliography of the Alaṃkāraśāstra. Handbuch der Orientalistik/Handbook of Oriental Studies, Abt. 2: Indien 14. Leiden: Brill. Cuneo, Daniele. 2009. “Emotions without Desire. An Interpretive Appraisal of Abhinavagupta’s Rasa Theory. Annotated Translation of the First, Sixth and Seventh Chapters of Abhinavagupta’s Abhinavabhāratī.” PhD thesis, “Sapienza” University of Rome. Cuneo, Daniele. 2013. “Unfuzzying the Fuzzy. The Distinction between Rasas and Bhāvas in Abhinavagupta and Bharata.” In Puṣpikā: Tracing Ancient India through Texts and Traditions. Contributions to Current Research in Indology, ed. by Nina Mirnig, Péter-Dániel Szántó Szántó, and Michael Williams, 49–75. Vol. 1. Oxford: Oxbow Books. Cuneo, Daniele. 2015. “Rasa: Abhinavagupta on the Purpose(s) of Art.” In The Natyasastra and the Body in Performance: Essays on Indian Theories of Dance and Drama, ed. by Sreenath Nair, 72–88. New York: McFarland. Cuneo, Daniele. 2016a. “Detonating or Defusing Desire. From Utpaladeva’s Ecstatic Aesthetics to Abhinavagupta’s Ecumenical Art Theory.” In Utpaladeva, Philosopher of Recognition, ed. by Raffaele Torella and Bettina Bäumer, 31–76. New Delhi: D.K. Printworld. Cuneo, Daniele. 2016b. “The Culmination of Sanskrit Aesthetics in Kashmir: A Hypothesis on Ruyyaka’s ‘Alaṃkārasarvasva’ in the light of Jayaratha’s ‘Vimarśinī.’ In Franco and Ratié 2016, 149–70. Cuneo, Daniele. 2017. “‘This Is Not a Quote.’ Quotation Emplotment, Quotational Hoaxes and Other Unusual Cases of Textual Reuse in Sanskrit Poetics-cumDramaturgy.” In Freschi and Maas, 219–53. David, Hugo. 2016. “Time, Action and Narration. On Some Exegetical Sources of Abhinavagupta’s Aesthetic Theory.” Journal of Indian Philosophy 44 (1): 125–54. doi:10.1007/s10781-014-9256-1. Franco, Eli, and Isabelle Ratié, eds. 2016. Around Abhinavagupta: Aspects of the Intellectual History of Kashmir from the 9th to the 11th Centuries. Leipziger Studien zu Kultur und Geschichte Süd- und Zentralasiens 6. Berlin: LIT Verlag. Freschi, Elisa, and Philipp A. Maas, eds. 2017. Adaptive Reuse. Aspects of Creativity in South Asian Cultural History. Abhandlungen für die Kunde des Morgenlandes 101. Wiesbaden: Harrassowitz Verlag.

422

Daniele Cuneo

Gaut, Berys N. 2007. Art, Emotion and Ethics. Oxford: Clarendon Press. Gerow, Edwin. 1971. A Glossary of Indian Figures of Speech. The Hague: Mouton. Gerow, Edwin. 1977. Indian Poetics. A History of Indian Literature 5. Wiesbaden: Otto Harrassowitz. Gerow, Edwin. 2001. The Vṛttivārttika or Commmentary on the Functions of Words of Appaya Dīkṣita. New Haven, CT: American Oriental Society. Graheli, Alessandro. 2016. “The Force of Tātparya: Bhaṭṭa Jayanta and Abhinavagupta.” In Franco and Ratié 2016, 231–61. Gren-Eklund, Gunilla. 1986. “The Cots Are Crying.” In Kalyāṇamitrārāgaṇam: Essays in Honour of Nils Simonsson, ed. by Eivind Kahrs, 79–97. Oslo: Norwegian University Press. Ingalls, D.H.H., J.M. Masson, and M.V. Patwardhan, eds. 1990. The Dhvanyāloka of Ānandavardhana with the Locana of Abhinavagupta. Harvard Oriental Series 49. Cambridge, MA: Harvard University Press. Keating, Malcolm. 2013a. “Mukulabhaṭṭa’s Defense of Lakṣaṇā: How We Use Words to Mean Something Else, But Not Everything Else.” Journal of Indian Philosophy 41 (4): 439–61. Keating, Malcolm. 2013b. “The Cow Is to Be Tied Up: Sort-Shifting in Classical Indian Philosophy of Language.” History of Philosophy Quarterly 30 (4): 311–33. Keating, Malcolm. 2019. Language, Meaning, and Use in Indian Philosophy: An Introduction to Mukula’s “Fundamentals of the Communicative Function.” London: Bloomsbury Academic. Kunjunni Raja, K. 1963. Indian Theories of Meaning. Madras: Adyar Library and Research Centre. McCrea, Lawrence J. 2008. The Teleology of Poetics in Medieval Kashmir. Harvard Oriental Series 71. Cambridge, MA: Department of Sanskrit / Indian Studies, Harvard University. McCrea, Lawrence J. 2011. “Standards and Practices: Following, Making, and Breaking the Rules of Śāstra.” In South Asian Texts in History: Critical Engagements with Sheldon Pollock, ed. by Yigal Bronner, Whitney Cox, and Lawrence McCrea, 229–39. Arbor, MI: Association for Asian Studies. Ollett, Andrew. 2016. “Ritual Texts and Literary Texts in Abhinavagupta’s Aesthetics: Notes on the Beginning of the ‘Critical Reconstruction.’” Journal of Indian Philosophy 44 (3): 581–95. doi:10.1007/s10781-015-9277-4. Pollock, Sheldon. 2010. “What Was Bhaṭṭa Nāyaka Saying? The Hermeneutical Transformation of Indian Aesthetics.” In Epic and Argument in Sanskrit Literary History: Essays in Honor of Robert P. Goldman, ed. by Sheldon Pollock, 143–84. New Delhi: Manohar. Pollock, Sheldon. 2016. A Rasa Reader: Classical Indian Aesthetics. New York: Columbia University Press. Tzohar, Roy. 2018. A Yogācāra Buddhist Theory of Metaphor. New York, NY: Oxford University Press. Venugopalan, K. 1977. “Mukula Bhaṭṭa. Abhidhāvṛttimātṛkā.” Journal of Indian Philosophy 4 (3/4): 203–64.

Chronology

The following table is a synopsis of the chronological indications contributed in this volume, corroborated by some external references. Dating South Asian authors and works is a notoriously complex operation, often based exclusively on their mutual and relative assessment, and as such still open to future improvements. Age

Author

Date, Place

Source

−500 Yāska

Fifth-third century

Cardona 1976, 270–3

−400 Pāṇini

ca. fifth/fourth century 350 post quem, Northwest India

Chapter 1 (Candotti) Bronkhorst 2019, 314

−300 Yāska Kātyāyana

Third century Third century

Bronkhorst 2019, 317 Chapter 1 (Candotti)

−200 Patañjali

Second century

Jaimini

200? last centuries BCE

Chapter 1 (Candotti), Chapter 4 (Saito) Bronkhorst 2019, 315–316 Coward and Kunjunni Raja 1990, 22 Chapter 2 (Nowakowska) Chapters 9 and 15 (Freschi)

200

Akṣapāda (Gautama)

200 ante quem

Chapter 11 (Graheli)

400

Vātsyāyana (Pakṣilasvāmin) Upavarṣa (Vṛttikāra)

Fifth century

Chapter 3 (Graheli)

480–540

Kataoka 2011, 112

Śabara

500–560 Third to fifth century Fifth century

Bhartṛhari

Fifth century

Kataoka 2011, 112 Chapters 9 and 15 (Freschi) Chapter 2 (Nowakowska), Chapter 4 (Saito), Chapter 16 (David) Chapter 4 (Saito), Chapter 6 (Ferrante) Coward and Kunjunni Raja 1990, 22; Bronkhorst 2019, 321 Chapter 8 (Kataoka) Chapter 6 (Ferrante)

150 post quem, Kashmir 150?, North of Ayodhya

500

middle of fifth century Dignāga

ca. 470–530 Sixth century

424 Age

600

Chronology Author

Date, Place

Source

Bhavya (Bhāviveka) Uddyotakara

Sixth century Sixth to seventh century middle of sixth century 550–610, Punjab

Chapter 16 (David) Chapter 3 (Graheli) Bronkhorst 2019, 318 Potter 1977, 9, 303

Kumārila Bhaṭṭa

Sixth to seventh century 600–650 middle of sixth century Seventh century

Chapters 9 and 15 (Freschi) Kataoka 2011, 112 Chapter 10 (McAllister) Chapter 4 (Saito), Chapter 6 (Ferrante) Chapter 8 (Kataoka) Chapter 21 (Cuneo) Chapter 10 (McAllister) Chapters 9 and 15 (Freschi) Chapter 16 (David) Kataoka 2011, 112 Chapter 16 (David) Kataoka 2011, 112 Chapter 4 (Saito) Coward and Kunjunni Raja 1990, 22

Dharmakīrti

Śākyabuddhi Prabhākara Miśra Maṇḍana Miśra

700

Uṃveka Bhaṭṭa Dharmottara Śaṅkara Śāntarakṣita Kamalaśīla Karṇakagomin

800

Ānandavardhana Śālikanātha

Padmapāda Sureśvara Bhaṭṭa Udbhaṭa Mukula Bhaṭṭa Bhaṭṭa Jayanta

Bhaṭṭa Nāyaka

ca. 600–660 ca. 650 ca. 660–720 Sixth and seventh century Seventh century? 620–680 660–720? 660–720 Eighth century 690? Eighth century 730–790 ca. eighth century Eighth century second half of eighth century Eighth century ca. 725–788 Eighth century ca. 770–830 ca. 855–883, Kashmir ca. 875 Eighth century Eighth to ninth century Ninth to tenth century 800–900 end of eighth century? Eighth to ninth century end of eighth century beginning of ninth century, Kashmir Ninth century, Kashmir end of ninth century, Kashmir

Ninth to tenth century end of ninth century, Kashmir beginning of tenth century, Kashmir

Chapter 9 (Freschi) Kataoka 2011, 112 Chapter 21 (Cuneo) Chapter 4 (Saito), Chapter 15 (Freschi) Chapter 16 (David) Chapter 5 (McClintock) Chapter 10 (McAllister) Chapter 5 (McClintock) Chapter 10 (McAllister) Chapter 21 (Cuneo) Chapter 20 (Ollett) Chapter 9 (Freschi) Chapters 13 and 20 (Ollett) Chapter 16 (David) Kataoka 2011, 112 Chapter 16 (David) Chapter 16 (David) Chapter 20 (Ollett) Chapter 21 (Cuneo) Chapter 21 (Cuneo) Chapters 3 and 11 (Graheli), Chapter 9 (Freschi), Chapter 13 (Ollett), Chapter 20 (Ollett) Chapter 4 (Saito) Chapter 20 (Ollett) Chapter 21 (Cuneo)

425

Chronology Age

Author

Date, Place

Source

900

Vācaspati Miśra

Ninth to tenth century Tenth century 950–1000 Tenth century 930–980 end of tenth century Tenth to eleventh century Tenth century, Kashmir 900–950, Kashmir Tenth century, Kashmir 925–975, Kashmir 980, Kashmir 980?

Chapter 4 (Saito) Chapter 16 (David) Kataoka 2011, 112 Chapter 9 (Freschi) Kataoka 2011, 112 Chapter 20 (Ollett) Chapter 20 (Ollett) Chapter 4 (Saito) Chapter 6 (Ferrante) Chapter 4 (Saito) Chapter 6 (Ferrante) Chapter 4 (Saito) Coward and Kunjunni Raja 1990, 22 Chapter 16 (David)

Sucarita Miśra Dhanañjaya Dhanika Somānanda Utpaladeva Helārāja

1000

Prakāśātman

950–1000

Abhinavagupta

Tenth to eleventh century, Kashmir 975–1025, Kashmir 1000–1050 Eleventh century Eleventh century Eleventh century, Kashmir Eleventh century, Kashmir Eleventh century Eleventh century Eleventh century ca. eleventh century

Chapter 21 (Cuneo)

Pārthasārathi Sarvajñātman Kṣemarāja Rājānaka Mammaṭa Ānandabodha Rāmānuja Jñānaśrīmitra Kaiyata

Chapter 6 (Ferrante) Kataoka 2011, 112 Chapter 9 (Freschi) Chapter 16 (David) Chapter 4 (Saito) Chapter 21 (Cuneo) Chapter 16 (David) Chapter 15 (Freschi) part IV (McClintock) part IV (Candotti)

1100

Haradatta Citsukha

Twelfth century? Twelfth to thirteenth century?

Chapter 4 (Saito) Chapter 16 (David)

1200

Veṅkaṭanātha

Thirteenth to fourteenth century

Chapter 15 (Freschi)

1300

Gaṅgeśa Mādhava Ānandapūrṇa

Potter 1977, 3 Chapter 4 (Saito) Chapter 16 (David)

Pratyaksvarūpa

ca. 1350 Fourteenth century Fourteenth century, SouthWest India Fourteenth century

1400

Uttuṅgodaya

Fifteenth century, Kerala

Chapter 21 (Cuneo)

1500

Appayya Dīkṣita Nṛsiṃhāśrama Madhusūdana Sarasvatī Bhaṭṭoji Dīkṣita

Sixteenth century, South India Chapter 16 (David) Sixteenth century, South India Chapter 16 (David) Sixteenth century, Bengal Chapter 16 (David)

Nārāyaṇa Bhaṭṭa

Sixteenth to seventeenth century 1590, Andhradeśa Sixteenth to eighteenth century

Chapter 16 (David)

Chapter 4 (Saito) Coward and Kunjunni Raja 1990, 22 Chapter 1 (Candotti)

426

Chronology

Age

Author

Date, Place

Source

1600

Kauṇḍa Bhaṭṭa

Seventeenth century 1640, Varanasi

Chapter 4 (Saito) Coward and Kunjunni Raja 1990, 22

1700

Nāgeśa Bhaṭṭa

Eighteenth century 1740, Varanasi

Chapter 4 (Saito) Coward and Kunjunni Raja 1990, 22

References Bronkhorst, Johannes. 2019. A Śabda Reader. New York: Columbia University Press. Cardona, George. 1976. Pāṇini, A Survey of Research. The Hague, Paris: Mouton. Coward, Harold G., and K. Kunjunni Raja, eds. 1990. Encyclopedia of Indian Philosophies: The Philosophy of the Grammarians. General editor Karl H. Potter. Delhi: Varanasi: Patna: Motilal Banarsidass. Kataoka, Kei. 2011. Kumārila on Truth, Omniscience, and Killing. Part 2. An Annotated Translation of Mīmāṃsā-Ślokavārttika ad 1.1.2 (Codanāsūtra). Beiträge zur Kultur- und Geistesgeschichte Asiens 68. Wien: Verlag der Österreichischen Akademie der Wissenschaften. Potter, Karl H., ed. 1977. Indian Metaphysics and Epistemology: The Tradition of Nyāya-Vaiśeṣika up to Gaṅgeśa. Encyclopedia of Indian Philosophies, vol. 2. Delhi: Motilal Banarsidass.

Glossary

Lemmata are sorted in English alphabetical order. The glossary largely consists of terms related to language. Some less specific technical terms used in Sanskrit by the contributors of this volume have also been added for reference’s sake. The right-end column provides the names of Sanskrit authors and/or traditions endorsing the definitions of technical terms, when definitions are present. abhi-dhāabhidhā

abhidhāna

abhidhātrī śakti abhihitānvaya

adhikāra

To denote; to designate; to express Denotation; direct designation; direct signification; expression; referentiality; cf. lakṣaṇā, tātparya The first power (see śakti) of individual words to denote their respective referents (see padārtha), distinct from their second power to denote the sentence meaning; cf. tātparya The power (vyāpāra) of words to express both primary and secondary meanings (mukhyalākṣaṇikayor abhidhāvyāpārayor atra vivekaḥ kriyate, AVM, 3,6–7) The power (vyāpāra) of words to express their direct (mukhya) meaning (sa mukhyo’rthas tatra mukhyo vyāpāro’syābhidhocyate, KPr, 1.8cd) Denotation; process of denotation The process of knowing things through audible sounds (śrotragrāhyavastukaraṇikā tadarthapratītir abhidhānakriyā, NM, I.399,1) Power of expression; see abhidhā Correlation of denoted meanings; relation of expressed meanings (Ollett); construal of the denoted (McCrea); cf. anvitābhidhāna The Bhāṭṭa theory of compositionality: wordmeanings are individually denoted and then correlated, thus indirectly (see lakṣaṇā) conveying the sentence meaning. The phrase abhihitānvaya, however, is not found in Kumārila’s writings, while it is used by both Śālikanātha and Jayanta in the ninth century Eligibility; qualification The qualification to perform a sacrifice in relation to the result, as well as the link between the obligation and the one who fulfills it

Jayanta

Mukula

Mammaṭa

Jayanta Jayanta

Bhāṭṭa

Prābhākara

428 adhikaraṇa

adhyavasāya

adhyeṣaṇa advaita ākāṅkṣā

ākāra ākāśa akhaṇḍārtha

ākhyāta

ākṛti

Glossary Substrate; the semantic-syntactic relation of location in Pāṇinian grammar Of the six semantic-syntactic relations (kārakas) in grammar, the one denoting a substrate (ādharo’dhikaraṇam, Aṣṭādhyāyī, 1.4.45) Judgment The link between the two levels of dravyasat and prajñaptisat, used as a justification of the veracity of knowledge obtained from inferences and from words (Kataoka) Soliciting A specific type of obligation (kārya) Nondualism Dependency; expectancy The syntactical and semantic dependency of the elements in a sentence, required for the completeness of a sentence meaning because without it the isolated meanings of words do not serve their purpose (abhidhānāparyavasānam ākāṅkṣā. yasya yena vinā na svārthānvayānubhāvakatvaṃ tasya tadaparyavasānam, TC, vol. 4 (1), p. 208,1–2) Phenomenal content of a cognition Space or ether, one of the five primary material elements in Vaiśeṣika ontology along with earth, water, fire, and air Undivided object The thesis of akhaṇḍārtha finds its roots in Padmapāda’s analysis of the concept of “convergence” (samanvaya) in Brahmasūtra, 1.1.4. Despite superficial resemblance in designation, the Vedāntic theory of akhaṇḍārtha is totally unrelated to Bhartṛhari’s view of the sentence (vākya) and its object (artha) as “indivisible” (akhaṇḍa) entities (David) Verb One of the four types of words, along with nouns, preverbs, and indeclinables (catvāri padajātāni nāmākhyātopasarganipātāś ca, VMBh, vol. 1, p. 3,17); a verb mainly means “becoming” (bhāvapradhānam ākhyātam, Nirukta, I.1) Universal; form or configuration; form or shape The arrangement of characteristic parts shared by a class of individuals, such as legs, hump, dewlap, and so on, in cows (sāsnādyavayavasanniveśātmikā ākṛtiḥ, NM, II.5,8); in Nyāya it is distinct from jāti, while it is used as its synonym in Mīmāṃsā

Pāṇini

Dignāga

Mīmāṃsā

Gaṅgeśa

Padmapāda

Yāska, Patañjali

Jayanta

429

Glossary akṣara

ākṣepa

ālambana Alaṅkāraśāstra alaṅkāra āmantraṇa ananvitārtha aṅga

anitya anubhava

anubhāva

anupapatti

Speech unit; phoneme; syllable; cf. varṇa Lit. “imperishable” (na kṣīyate na kṣarati iti vā akṣaram, VMBh, vol. 1, 36,5); in Pratiśakhya literature it was used for vowels, but in the VMBh, and later on, it is applied to any varṇa, and even to morphemes; later it is frequently used, by extension, to denote syllabic units in prosody Entailment; implication; implication (logical or ontological); implicature An automatic logical and/or ontological entailment, as in Śālikanātha’s discussion about a particular entailing the universal, or in Jayanta’s discussion of the relation of the middle and major term in syllogism, the relation of co-presence is by ākṣepa entailed in co-absence; the term is used in the same sense also by Mammaṭa, among others A specific operation in some types of lakṣaṇā Foundational factor; see vibhāva One of the two types of vibhāva in the process of aesthetic experience Poetics Figure of speech; poetical ornamentation (see dhvani) Inviting A specific type of obligation (kārya) Non-connected object; see svārtha Pre-suffixal base The base, deriving from either a verbal or nominal stem, after which a suffix is prescribed (yasmāt pratyayavidhis tadādi pratyaye’ṅgam, Aṣṭādhyāyī, 1.4.13) Transient; impermanent Immediate experience Immediate experience, as opposed to recollection (smṛti), for example, “it appears neither as a recollection, nor like an experience” (nābhāti smṛtirūpeṇa na cāpy anubhavātmanā, NM, I. 55.8) Defined as “a cognition different from memory” (smṛtiḥ. tadbhinnaṃ jñānam anubhavaḥ, TarkaSaṃ, 13) Aesthetic reaction One of the necessary ingredients of an aesthetic experience; see rasa Incongruity; impossibility A necessary condition for a postulation (see arthāpatti) One of the possible conditions for a figurative meaning

Patañjali

Śālikanātha

Mukula Alaṅkāraśāstra

Mīmāṃsā

Pāṇini

Nyāya

Navyanyāya

Alaṅkāraśāstra

430 anusandhāna anuvāda

anvaya1

anvaya2

anvayavyatireka

ānvīkṣikī

anvita anvitābhidhāna

anyāpoha

Glossary Dynamic unification of cognitions corresponding to a reflexive awareness (Ferrante); cf. vimarśa Repetition; re-statement “Supplementary reference, allusion to something that has been laid down elsewhere; contrasted with vidhi, injunction (of something not otherwise laid down)” (MNP, 279) A re-statement yields a second-hand cognition, but without the unreliability of other second-hand cognitions such as memory (smṛti), because it implies the same causal process (namely, śabda2) that caused the first cognition (David) Co-presence; concurrent existence; antonym vyatireka Concurrence of middle and major term in inferences. In its application to inductive inferences, akin to Mill’s relation of agreement Syntactic-semantic relation; relation between word meanings within a sentence The difference of the Bhāṭṭa abhihitānvaya and the Prābhākara anvitābhidhāna mainly hinges on the different account of this relation (Ollett) Co-presence and co-absence In Vyākaraṇa, “concurrent occurrence” (anvaya) of a certain meaning and a certain linguistic unit and the absence (vyatireka) of a meaning and a unit (Cardona 1967, 337); In logic, see anvaya1 Investigation; see anumāna The discipline that teaches inferential reasoning, from anvīkṣā, syn. of anumāna (pratyakṣāgamābhyām īkṣitasyānvīkṣaṇam anvīkṣā. anumānam ity arthaḥ. tad vyutpādakaṃ śāstram ānvīkṣikī, NM, I.8) Correlated; see anvaya2 Denotation of correlated words; expression of relational meanings (Ollett); expression of [objects] in connection (David); construed denotation (McCrea); The Prābhākara theory of compositionality: Correlated words in combination denote the sentence meaning. The phrase is originally found in Prabhākara’s writings Vedāntins propose to replace Śālikanātha’s “view of the expression [of objects] in connection with an obligation” (kāryānvitābhidhānavāda) by another view known as the “view of the expression [of objects] in connection with others [in general]” (anyetarānvitābhidhānavāda) (David) Exclusion of other things (Kataoka); otherexclusion (McAllister); see apoha

Mīmāṃsā

Vedānta

Nyāya

Mīmāṃsā

Patañjali

Jayanta

Prābhākara

Vedānta

431

Glossary aparokṣajñāna apauruṣeya apekṣā apoha

āpta

apūrva

artha

arthakriyā

arthāpatti

arthāpoha

arthapratyāyaka

Immediate knowledge Authorless; lit. “impersonal” Expectancy; dependency; see ākāṅkṣā Exclusion The Buddhist theory about the object of words and concepts; in Dignāga’s semantics, the referent of a word is “qualified by the exclusion of other referents” (arthāntarapohaviśiṣṭe ’rthe, PS, ad 5.38d) (Kataoka) Authority; reliable person A trustworthy speaker has three qualities: (1) direct experience of the object, (2) desire to describe it, and (3) will to do it in a truthful manner (āptaḥ khalu sākṣātkṛtadharmā yathādṛṣṭasyārthasya cikhyāpayiṣāprayukta upadeṣṭā ca, NBh, ad NS 1.1.7, p. 14,4–5) Previously unknown; new; unprecedented An energy produced by the sacrifice and lasting from the time of the sacrifice till its result is achieved (bhaṭṭapakṣe ’vāntarapūrvaṃ śaktirūpam, GBh, 163,16) The duty known anew from an injunction (prabhākaramate tv avāntarāgneyādiniyoga eva, GBh, 163,17); it can typically be known only from śabda,2 and not through other means of valid cognition (mānāntarāvedya) Referent; meaning; object A polysemic word used in a wide range of contexts: the referent or meaning of a word in grammar and semantics (see padārtha and vākyārtha), an object or content of knowledge in epistemology (see prameya), a final cause in teleology and soteriology (see prayojana) Causal efficacy; practical efficacy Practical efficacy of the knowledge of an object, as a test of veracity in epistemology Postulation Postulation of an antecedent to explain an otherwise inexplicable phenomenon, which is either seen or heard (dṛṣṭaḥ śruto vārtho ’nyathā nopapadyate ity arthakalpanā, ŚBh, ad 1.1.5, p. 30) Remainder of referents The remaining referent after the exclusion of all its counter-positives, for example, in the case of a cow, the remaining referent after the exclusion of all the non-cows; cf. śabdāpoha Vehicle of meaning; Conveyor of meaning Cause of knowledge of a linguistic referent or an epistemic object (see artha)

Dignāga

Vātsyāyana

Bhāṭṭa

Prābhākara

Śabara

432

Glossary

Not touching reality The basic principle in Dignāga’s nominalism, that a śabda2 does not touch its artha (śabdānām arthāsaṃsparśitāṃ vadantaṃ bhadantaṃ, NM, II.135,18) By contrast, realist schools such as Nyāya defend arthasaṃsparśitā (NM, I.209,2) Possessing meaning; meaningful; unit endowed arthavat with meaning From Pāṇini’s definition of nominal stem (see prātipadika) one gleans that also verbal roots and even suffixes are considered as “endowed with meaning” (arthavad adhātur apratyayaḥ prātipadikam, Aṣṭādhyāyī, 1.2.45) The self ātman One of the twelve objects of knowledge (see prameya) One of the nine substances (dravya) Intermediate sentence; simple sentence; clause; avāntaravākya see mahāvākya avāntaravākyārtha Intermediate sentence meaning; see vākyārtha Insertion and extraction; association and āvāpodvāpa dissociation; cf. anvayavyatireka “Just as other words are learned with respect to their own referents through different insertions (āvāpa) and extractions (udvāpa), so the prescriptive endings (i.e., optative, imperative, gerundive, etc.) are learned with respect to what is to be done” (PrP3, 430,10–11) (Kataoka) Not belying avisaṃvādin The property of correspondence between a cognition and its content; the reality of a thing corresponding to its cognition (avisaṃvādakatvaṃ ca prāpakatvam ucyate. jñānasya ca prāpakatvaṃ sukha​duḥkh​asādh​anasa​marth​ apadā​rthap​rāpti​parih​ārahe​tubhū​tāyāḥ​ pravṛtter nimittapradarśakatvam eva, NM, I.61) Invalidation; annulment; blockage, obstruction, bādha inapplicability The blockage of the denotation of word-meanings as a cause of indirect signification (lakṣaṇā) in the abhihitānvaya theory of compositionality Invalidating factor bādhaka A superseding pramāṇa bhakti Associated meaning; see lakṣaṇā Bhāṭṭa The Mīmāṃsā tradition, or a Mīmāṃsā author, following Kumārila Bhaṭṭa’s tenets bhāva Emotion; see rasa Difference; particular, individual bheda arthāsaṃsparśitā

Dignāga

Nyāya

Pāṇini

Nyāya Vaiśeṣika Vaiśeṣika

Śālikanātha

Buddhism Buddhism

Bhāṭṭa

Mīmāṃsā

433

Glossary bhāvanā

brahman camatkāra

darśana dharma1

dharma2

dharmin

dhātu

dhātvartha

Actualization; causation of an effect; cf. vyāpāra The performative aspect (vyāpāra) of an action, which is denoted by verbal endings. It is of two types, śabdabhāvanā and arthabhāvanā (bhāvanā ākhyātapratyayārtho bhāvayitur vyāpāraḥ. bhāvanā dvividhā śabdabhāvanā arthabhāvanā ceti. liṅleṭloṭtavyapratyayamātragatā śabdabhāvanā. sarvākhyātagatā arthabhāvanā, MK, s.v., p. 3022b) Undivided absolute; unchanging primordial essence; Absolute Savoring A term first coined in poetics to connote aesthetic experiences Subjective, private dimension of consciousness (Ferrante) Philosophical system; lit. “vision”; Systematic viewpoint on reality (Cuneo) Duty; ritual duty The thing indicated by commands (codanālakṣaṇo’rtho dharmaḥ, PMS, 1.1.2.2) The performance of sacrifices, which is taught by the Veda (yāgādir eva dharma. tallakṣaṇaṃ vedapratipādyaḥ proyojanavad artho dharma iti, ArthaSaṃ, p. 4) The source of human perfection (yato ’bhyudayaniḥśreyasasiddhiḥ, VaiSū, 1.1.2); one of the guṇa1s inhering in the self (tasya guṇāḥ […]  dharmādharmasaṃskāra-, PrBh, 70,11) The meaning of Vedic obligations (see kārya, apūrva) Property A distinctive property (tasmād yo yena dharmeṇa viśeṣaḥ saṃpratīyate, PV I, 42) Property-bearer; see tadvat Lit. “possessor of a dharma,”2 typically a substance (dravya) Verbal base; verbal root; cf. prātipadika A word expressing an activity, vyāpāra, (kriyāvacano dhātuḥ, VMBh, vol. 1, p. 254,13), or expressing the “becoming” of something by means of the kārakas (bhāvavacano dhātuḥ, VMBh, vol. 1, p. 256,18) Object of a verbal root; see padārtha, kārya For example, the finite verb yajeta (“one should sacrifice”) consists of the verbal root yaj and the optative ending (liṅ) which respectively denote the activity of sacrificing (yāga) and a command (niyoga)

Mīmāṃsā

Alaṅkāraśāstra Abhinavagupta

Mīmāṃsā Bhāṭṭa

Nyāya-Vaiṣ.

Prābhākara Dharmakīrti

Mīmāṃsā

434 dhvani

dravya dravyasat ekārthatā

gauṇa gauṇavṛtti

gauṇī vṛtti guṇa1

guṇa2

jahallakṣaṇā jāti

jātimanmātra jātimātra jātiśabda

Glossary Sound; resonance; physical and nonlinguistic sound Sound as cause of manifestation of words according to Bhartṛhari, momentary sound according to Maṇḍana (Saito) Jayanta discusses and refutes it as the element manifesting śabda2, that is, sphoṭa (NM, I.526,16, II.151,1) and defines it as a property of śabda,1 according to the Nyāya subdivision into linguistic and nonlinguistic sound (dvividhaś cāyaṃ śabdaḥ varṇātmā dhvanimātraś ca, NBh, ad NS 2.2.39, p. 129,3) According to Ānandavardhana, whereas a linguistic expression expresses its meaning, a literary text manifests a rasa; dhvani is here synonym of vyañjanā, poetical suggestion; it is subsumed under bhakti, “associated meaning,” by Mukula Substance The existent thing, causally constructed The characteristic of having a single meaning According to Jaimini, the typical characteristic of a sentence (ŚBh, ad PMS 2.1.46) (Freschi) Secondary Secondary signification; cf. lakṣaṇā A secondary signification in which a word signifies something sharing the same qualities of the primary meaning, for example, “lion” for brave (Freschi) Metaphorical signification; cf. lākṣaṇikī vṛtti Metaphor, based on a transfer of qualities (guṇa1) Quality One of the six (later seven) categories (padārthas) in Nyāya-Vaiśeṣika metaphysics Full apophonic grade of vowels in Pāṇini’s system The second of the three degrees of a vowel (a, e, o, adeṅ guṇaḥ, Aṣṭādhyāyī, 1.1.2) in Pāṇini’s terminology A type of figurative usage, operating by abandoning the primary meaning; see lakṣaṇā Genus; universal; class; antonym vyakti, viśeṣa A form shared by multiple individuals (śābaleyā disakalagopiṇḍādisādhāraṇaṃ rūpaṃ jātir NM, II.5,9); cf. ākṛti, sāmānya The bare possessor of a universal; only the universal-possessor; see tadvat Bare universal; simple universal; only the universal; see jāti Generic noun; general noun; generic term According to Dignāga, a generic term such as “cow” can express, in a general form, anything that has not been experienced as a non-cow

Vyākaraṇa Jayanta

Alaṅkāraśāstra

Dignāga Mīmāṃsā

Mīmāṃsā

Mīmāṃsā Nyāya-Vaiṣ. Pāṇini Alaṅkāraśāstra Jayanta

Dignāga

435

Glossary jātiyoga jñāpaka

kalpanā kāraka

kārya

kevalapadārtha kevalārtha kriyā kṛt kṛti kṛtisādhyatva kṛtya

lakṣaṇa

lakṣaṇā

Relation with the universal; see jāti, vyakti Indicator; sign As a cause of knowledge, an indicator or sign such as an inferential mark or a word operates differently in respect to factors of knowledge (kāraka) such as the sense organs (liṅgaṃ hi jñāpakaṃ na cakṣurādivat kārakam, NM, I.295,10) Assumption; presumption; supposition; hypothesis; see arthāpatti Factor of action; thematic role; case In the Pāṇinian’s terminology, a label for the semantic-syntactic roles of nouns in relation to the verbal activity (sādhakaṃ nirvartakaṃ kārakasaṃjñaṃ bhavati, VMBh, vol. 1, p. 323,17,8), which are necessary conditions for the accomplishment of given actions (nimittaparyāyaḥ, Kāśikā, ad 1.4.23); Pāṇini lists six such kārakas, (1) agent (kartā), (2) patient (karma), (3) instrument (karaṇa), (4) donee (sampradāna), (5) point of ablation (apādāna), (6) substrate (adhikaraṇa) An instrumental cause of knowledge such as the eye, opposed to an indicator or sign such as an inferential mark (kāraka), as in (liṅgaṃ hi jñāpakaṃ na cakṣurādivat kārakam, NM, I.295,10) Duty; obligation; what ought to be done The meaning of a sentence (vākyārtha), something to be accomplished by an effort, rather than an instrument to achieve a result (phalasādhana) Bare word-meaning; individual word-meaning; see svārtha Isolated object; see svārtha Action; activity; the meaning of a verbal root; see dhātvartha Primary suffixes Effort; see prayatna The aspect of a thing as something to be accomplished by an effort; see kārya Gerundive suffixes; see liṅādi Pāṇini does not define the term, but uses it as the governing aphorism for the various gerundive suffixes (tavya, anīya, etc., Aṣṭādhyāyī, 3.1.95) Distinctive character; definition; see uddeśa, parīkṣā The property fixing the essence of an enumerated category, uddeśa (uddiṣṭasya tattvavyavasthāpako dharmo lakṣaṇam, NBh, 8,8) Indication; indirect signification; figurative signification; secondary signification; secondary usage; indirect denotation; secondary or figurative implicature

Jayanta

Vyākaraṇa

Jayanta

Prābhākara

Vyākaraṇa

Vyākaraṇa

Vātsyāyana

436

lakṣaṇavṛtti lākṣaṇikī vṛtti lakṣyārtha laṭ

laukika liṅ liṅādi

loka loṭ madhyamā

mahāvākya

Glossary A secondary signification in which a word signifies something connected with its primary meaning, for example, “cradle” for baby In the Bhāṭṭa theory of compositionality, particularly, the sentence meaning is said to be lakṣyamāṇa, “indicated” Every linguistic expression is a subcategory of lakṣaṇā, which thus includes even poetical meanings (cf. dhvani); lakṣaṇā is elicited on the basis of three conditions: an obstruction (mukhyārthabādha), a connection between secondary and primary meaning (mukhyārthāsatti), and a specific purpose (prayojana) Indicative reference (David); see lakṣaṇā Metonymical signification (Cuneo) Indirect meaning (Iwasaki); cf. śakya; see lakṣaṇā Prototype of verbal endings In Pāṇinian terminology a label for verbal endings, based on the general case of the present tense (vartamāne laṭ, Aṣṭādhyāyī, 3.2.123) Worldly; vernacular, humanly made language; antonym vaidika Endings of the optative mode; see liṅādi Exhortative suffixes; prescriptive endings They include the optative (liṅ), imperative (loṭ), and gerundive (kṛtya) suffixes (Freschi); in denoting a kārya, they are used in three specific senses: impelling (praiṣa), inviting (āmantraṇa), and soliciting (adhyeṣaṇa) They denote a new apūrva obligation, not just any obligation, because only the former is fit to be connected with the person commanded in the Veda (vaidikaniyojyānvayayogyatayā liṅādīnām apūrvakāryābhidhāyakatvanirṇayo nānupapannaḥ, PrP3, Vṛtti, 446,9–10) (Kataoka) World; people; common life; ordinary experience; society Endings of the imperative mode; see liṅādi Intermediate speech Linguistic thought, midway between vaikharī and paśyantī, purely mental and still endowed with a sequence (kevalaṃ buddhyupādānakramarūpānupātinī / prāṇavṛttim atikramya madhyamā vāk pravartate // VP, 1.166) Textual unit composed of multiple sentences; megasentence (McClintock) Discussed in relation to the hermeneutic value of an intermediate sentence (avāntaravākya), na ca mahāvākye saty avāntaravākyaṃ pramāṇaṃ bhavati, ŚBh, ad PMS 6.4.23)

Mīmāṃsā Bhāṭṭa Mukula

Pāṇini

Mīmāṃsā

Śālikanātha

Bhartṛhari

Śabara

437

Glossary

Mīmāṃsā

mukhya mukhyārtha mukhyārthabādha mukhyārthāsatti

nāda nāmadheya nāman

nāmarūpa Navyanyāya

nimitta

Mentioned as an intermediate segment of speech, in between a sentence (see avāntaravākya) and a whole text (prakaraṇa) (yathā […] vākyeṣu padāni na santi, tathā mahāvākyeṣv avāntaravākyāny api na syuḥ. tataḥ kim. mahāvākyāny api prakaraṇāpekṣayā na tāttvikāni syuḥ, NM, I.209) The system of thought traced back to Jaimini (ca. second century BCE) with a focus on exegesis, deontic statements, and sentence semantics Widely known among brāhmaṇical intellectuals as the “vākya-śāstra,” or “science of sentences” (McCrea) Direct; primary; antonym gauṇa; cf. mukhyārtha Direct meaning; denoted meaning; antonym lakṣyārtha Inapplicability of the primary meaning One of the three requirements of lakṣaṇā Relation between the denoted and the indicated meanings One of the three requirements of lakṣaṇā; also referred to simply as “relation” (sambandha) Nonlinguistic sound; bodily resonance, gross sound; Uttering or voicing; sound; see dhvani and varṇa Appellation; name Noun; substantive; name (McAllister) One of the four types of words listed by Yāska, along with verbs, preverbs, and indeclinables (catvāri padajātāni nāmākhyāte copasarganipātāś ca, Nirukta, I.1) Name and form, the Vedic principle through which the indistinct primordial reality is divided The philosophical system gradually evolved from older Nyāya and often associated to Gaṅgeśa (fourteenth c. CE), with a focus on logic and epistemology Cause; condition; necessary cause; ground or reason Cause or condition in general, particularly as ascertained on the basis of its effect (naimittika). It is the most proximate among the necessary causes, defined as a cause which is neither inherent nor noninherent (samavāyyasamavāyivyatiriktaṃ tu kāryotpattau nirjñātasāmarthyaṃ yat kāraṇaṃ tat nimittakāraṇam ucyata iti, NM, II.536,12–13) The nimitta of a word is the external referent, distinct from the meaning (artha). For instance, a cow is the nimitta for the use of the word “cow”

Jayanta

Mukula Mukula

Yāska

Nyāya-Vaiṣ.

Śālikanātha

438 nirākāravāda Nirvacana

nirvikalpaka niṣedha nitya niyoga niyojya Nyāya

nyāya

pada

pāda

padapāṭha padārtha

Glossary Theory of non-phenomenal awareness; Theory that awareness is not endowed with phenomenal content (McClintock) Semantic analysis The system of thought ascribed to Yāska and claimed to be the fulfillment of grammar; see also Vyākaraṇa Non-conceptual; see vikalpa Prohibition; antonym vidhi Permanent; fixed Command; obligation; see vidhi, kārya The commanded person Reasoning; The philosophical system ascribed to Akṣapāda (2nd c. CE) with a focus on logic, epistemology, and dialectics The discipline (śāstra) teaching inference, also called ānvīkṣikī (pratyakṣāgamābhyām īkṣitasyānvīkṣaṇam anvīkṣā / a numānam ity arthaḥ / tad vyutpādakaṃ śāstram nyāya, NM, I,8) Hermeneutic rule; principles Rules applied to word-meanings for the interpretation of complex sentence meanings; principles constraining the input-output relation between the words and the sentence meaning, for instance, criteria such as novelty (apūrvatā), teleology (ekārthatā) and feasibility of an injunction guide one’s interpretation of the injunction, and need to be assessed before the final ascertainment of the sentence meaning Word Inflected word (suptiṅantaṃ padam, Aṣṭādhyāyī, 1.4.14); originally a metrical foot, a unit identified in terms of form and rhythm and detached from meaning (Candotti) Foot; metrical foot One quarter, as with the chapters of Pāṇini’s Aṣṭādhyāyī, each divided in four sections called pāda Word-for-word recitation, meant for the preservation of the form of the Veda Referent of a word; meaning of a word; wordmeaning; Word object; see artha In the realist perspective of Nyāya, an external entity, Frege’s Gegenstand (evaṃ siddhe bāhye’rthe, NM, II.47.5); In the original formulation of the sūtra, it can be the individual, the form, or the universal (vyaktyākṛtijātayas tu padārthaḥ, NSū, 2.2.66), later synthesized in the concept of the property-possessor (tadvat, gavādijātiśabdānāṃ gotvādijātyavacchinnaṃ vyaktimātram arthaḥ, yas tadvān iti naiyāyikagṛhe gīyate, NM, II.5.3–4)

Yāska

Jayanta

Mīmāṃsā

Vyākaraṇa

Nyāya

439

Glossary

pakṣa paramārtha paramārtha parā vāc parīkṣā

pariṇāma paśyantī

phala Prābhākara pradhānakārya

praiṣa prajñaptisat prakriyā

prākṛtadhvani pramāṇa

In the case of nouns the meaning of the word is the universal (ākṛti) The six or seven categories in the ontology of Vaiśeṣika (substance, quality, dynamism, universal, particularity, inherence, with the later addition of absence) and sixteen in Nyāya epistemology and dialectics (pramāṇa, etc.) A view evaluated in a philosophical debate; argumentative or theoretical stance; see pūrvapakṣa, uttarapakṣa, siddhānta Ultimate reality; ontological reality The existent thing, in its ultimate reality; antonym prajñaptisat, dravyasat Highest speech The highest level of speech, even beyond paśyantī Examination; test of a definition The critique of a lakṣaṇa, whether the definiendum is congruous or not with the definition (lakṣitasya yathālakṣaṇam upapadyate na veti vicāraḥ parīkṣā, NBh, 8,8–9) Transformation; transmutation Internal speech; intuitive speech; cf. madhyamā and vaikharī The partless, sequenceless, luminous, internal, subtle, and unchanging linguistic essence (avibhāgā tu paśyantī sarvataḥ saṃhṛtakramā / svarūpajyotir evāntaḥ sūkṣmā vāg anapāyinī // VP, 1.167) Result; fruit The Mīmāṃsā tradition, or a Mīmāṃsā author, following Prabhākara Miśra’s tenets (seventh century CE) The main element; the main rite or prescription The main prescription in a ritual The command (niyoga) concerns the principal duty (kārya), denoted by exhortative suffix (see liṅādi) Impelling A specific type of obligation (kārya) The existent thing, conceptually constructed; antonym of dravyasat and paramārtha Derivation of specific linguistic forms A term originating the sixteenth-century flourish of “grammars by derivations,” such as Bhaṭṭoji Dīkṣita’s Siddhāntakaumudī (Candotti) Primary sound; see dhvani Instrument of knowledge; means of valid knowledge;

Mīmāṃsā Nyāya-Vaiṣ.

Dignāga Bhartṛhari Vātsyāyana

Bhartṛhari

Mīmāṃsā Prābhākara

Mīmāṃsā Dignāga

440

pramātṛ

prameya

pramiti

prasthāna pratibhā pratibimba

prātipadika pratipakṣa Pratyabhijñā

pratyakṣa

Glossary Source of knowledge; means of knowledge; valid knowledge That by means of which one knows the object (sa yenārthaṃ pramiṇoti vijānāti tat pramāṇam, NBh, 1,14). A complex cause, or causal apparatus, which includes multiple necessary causes (tad ucyate. avyabhicāriṇīm asandigdhām arthopalabdhiṃ vidadhatī bodhābodhasvabhāvā sāmagrī pramāṇam, NM, I.31) Knowledge of a previously unknown object (anadhigatārthagantṛtvam eva pramāṇalakṣaṇam, MK, s.v., vol. 5, p. 2772a) Direct knowledge (see anubhava) that is independently ascertained, that is, without depending upon any other piece of knowledge (anubhūtiḥ pramāṇam. kā cānubhūtiḥ. svatantraparicchittiḥ. kim idaṃ svātantryaṃ nāma. paricchede pūrvabuddhyanapekṣatvam, MK, s.v., vol. 5, p. 2774b) Knowing agent, knower The agent who is moved by a desire to obtain or avoid something (yasyepsājihāsāprayuktasya pravṛttiḥ sa NBh, 1,13) An agent who knows (pramiṇoti iti NM, I.34,6– 7); an agent who reflects (vicārako hi pramātā, NM, I.247,12); an agent intent in determining an object (arthaparicchedāc ca pravartamānaḥ, NM, I.431,8) Knowable object The artha which is known (yo ’rthaḥ pramīyate jñāyate tat, NBh, 1,14) Knowledge True knowledge of the artha (yattadarthavijñānaṃ sā, NBh, 1,15) Method; line of interpretation of Śaṅkara’s Bhāṣya Intuition; mental flash of a padārtha (McClintock) Reflection Reflection as in a mirror. The result of exclusion (see apoha) in verbal knowledge Nominal stem; nominal base; see dhātu Antithesis Recognition, term used to refer to the doctrines of a number of scholars and spiritual teachers who lived and taught in Kashmir between the tenth and eleventh century CE (Ferrante) Sensory perception intended as act, result or even object of knowledge; perception as experience

Vātsyāyana Jayanta

Mīmāṃsā Prābhākara

Vātsyāyana Jayanta

Vātsyāyana Vātsyāyana

Buddhism

441

Glossary

pravṛttikāraṇa pravṛttinimitta

prayatna prayojana

Pūrva-Mīmāṃsā pūrvapakṣa rasa

śabda1

Accepted as a reliable means of knowledge (see pramāṇa) by every school. It consists as sensory perception intended as instrument, act, result or even object of knowledge, according to the various authors and traditions Perception as experience, always devoid of conceptualizations (kalpanā) (McClintock) Cause for an activity; see pravṛttinimitta Condition for undertaking an action; the external referent of a word In general, the cause for seeking or abandoning what leads to pleasure or pain (prāpakatvaṃ sukha​ duḥkh​asādh​anasa​marth​apadā​rthap​rāpti​parih​ ārahe​tuḥ, NM, I.61) In the use of language, the condition for the use of a given word, as the universal “pot-ness” for the use of the word “pot,” that is, the referent of a word (padaśakyatāvacchedakaḥ / yathā ghaṭatvaṃ ghaṭapadasya pravṛttinimittam, NK, s.v.) Articulatory effort; mental effort Purpose The object (artha) for which an effort is done (yam arthaṃ adhikṛtya pravartate tat prayojanam, NSū, 1.1.24) Motive for the use of the secondary signification. One of the three requirements for an instance of lakṣaṇā “Prior Mīmāṃsā”; see Mīmāṃsā Preliminary thesis, the view evaluated in the dialectic process; prima facie; see uttarapakṣa, siddhānta Taste; aesthetic emotion Taste, one of the twenty-four qualities (see guṇa1) in Vaiśeṣika metaphysics The aesthetic emotional experience, generated through the combined effect of a primary emotion and various assisting factors (tatra vibhāvānubhāva-  vyabhicārisaṃyogād rasaniṣpattiḥ, NŚ, 272) The term is generally reserved for emotions associated with works of art, but it acquires linguistic significance when it is the result of the comprehension of literary works; It is the purpose, and hence the “thing to be done” (see kārya), in poetical sentences Sound in general Subdivided into linguistic and nonlinguistic sound (dvividhaś cāyaṃ śabdaḥ varṇātmā dhvanimātraś ca, NBh, ad NS 2.2.39, p. 129,3)

Dignāga

Nyāya

Mīmāṃsā

Nyāya Mukula

Nyāya-Vaiṣ. Alaṅkāraśāstra

Nyāya-Vaiṣ.

Nyāya-Vaiṣ.

442 śabda2

śābda śābdabodha

śabdabrahman śabdādvaita

śabdana

śābdāparokṣajñ āna

śabdāpoha

śabdārtha śabdavyāpāra

Glossary Speech; Word; language; linguistic expression; linguistic unit; meaningful unit An utterance which is the cause of knowledge of something—e.g., the thing with dewlap, hump, hooves, horns, that is, a cow (yena uccāritena sā snālāṅgūlakakudakhuraviṣāṇinām sampratyayaḥ bhavati saḥ śabdaḥ, VMBh, vol. 1, p. 1,10–12) The term śabda is well known in the sense of something audible, from which knowledge of the object (artha) is derived (śrotragrahaṇe hy arthe loke śabdaśabdaḥ prasiddhaḥ […] yato’rthapratipattiḥ syāt, ŚBh, ad PMS 1.1.5, p. 54,8) That by means of which an object is named and known (śabdaḥ śabdyate ’nenārtha ity abhidhīyate vijñapyate iti, NBh, 9,1–2) An audible sound that causes knowledge of an object (śrautragrāhyatvaviśiṣṭam arthapratipattihetutvaṃ śabdalakṣaṇam abhidhīyate, NM, II.172,8) Related to language; see śabda2 Verbal knowledge It becomes a preoccupation for Advaitins by the fourteenth century (David); the term is not used in early Nyāya and Mīmāṃsā, and becomes the staple term for verbal knowledge in Navyanyāya Supreme and undivided speech; eternal verbum, the ultimate reality; see brahman and śabdādvaita Linguistic monism (Saito) The view, often ascribed to Bhartṛhari, according to which reality consists in one indivisible linguistic principle, śabda, while the phenomenal world of differences, including the difference between śabda and artha, occurs at a lower and illusory level of reality Verbalizing “Languageing,” as a syntactical arrangement of mental states, under the assumption that any thought is internal speech Immediate verbal knowledge The idea of an immediate verbal knowledge is ascribed to Prakāśātman and remained associated with his name in later times (David) Remainder of words The word excluded from all the words for noncows; see apoha, arthāpoha See artha Function of language (Cuneo); see vyāpāra

Patañjali

Śabara

Vātsyāyana Jayanta

Vyākaraṇa

Pratyabhijñā

Vedānta

Dignāga

443

Glossary To be done; what ought to be done; the thing to be accomplished; target; to be realized; see siddha1 Language conveys something to be realized. This applies both to the Veda and common language Only Vedic language conveys something to be realized, while common language conveys something already realized (siddha1) (Freschi) 2 The probandum in inferential reasoning, for sādhya example, fire in the inference from smoke Theory of phenomenal awareness; theory that sākāravāda awareness is endowed with phenomenal content Presentification sākṣātkāra Power; potency; capacity; energy; semantic śakti relationship (Iwasaki) According to Jayanta, it is the power of a cause. Such a causal power can be either inherent (e.g., the power of clay in the production of a vase) or contingent (e.g., the power of the potter’s stick, or wheel) (yogya​tāvac​chinn​asvar​ūpasa​ hakār​isann​idhān​am eva śaktiḥ / saiveyaṃ dvividhā śaktir ucyate avasthitā āgantukī ca / mṛttvādyavacchinnnaṃ svarūpaṃ avasthitā śaktiḥ /   āgantukī ca daṇḍacakrādisaṃgarūpā, NM, II 403,8–10) Mīmāṃsakas use the instrument of postulation (see arthāpatti) to prove the existence of a śakti śakya Direct meaning Iwasaki; cf. mukhyārtha sāmānādhikaraṇya Co-reference Scriptural convergence samanvaya For Śaṅkara, “convergence” of sentences of various Upaniṣads on a single object (namely, brahman) sāmānya Universal; general term; cf. jāti As the universal inhering in substances, qualities, and actions, synonym of ākṛti (dravyaguṇakarmaṇāṃ sāmānyamātram ākṛtiḥ, ŚBh, ad PMS 1.3.30); The referent of individual words (sāmānye hi padaṃ pravartate, viśeṣe vākyam, ŚBh, ad PMS 1.1.24) The universal, one of the six or seven categories in Nyāya-Vaiśeṣika ontology, defined as being of two types, all-extensive (i.e., existence, sattā) and less extensive (param aparam ceti dvividhaṃ sāmānyam and paraṃ sattā aparaṃ dravyatvādi, TarkaSaṃ, p. 2 and p. 46) In Pāṇini’s grammar, sāmānya denotes a general rule, opposed to a viśeṣa rule General characteristic sāmānyalakṣaṇa The nonspecific domain that remains, not having been excluded by apoha, and is expressed by a word sādhya1

Prābhākara Bhāṭṭa

Nyāya

Mīmāṃsā

Vedānta Vedānta

Śabara

Nyāya-Vaiṣ.

Vyākaraṇa Buddhism Dignāga

444 samāsa samavāya samaya

sambandha saṃcāribhāva saṃsarga saṃskāra

samuccayajñāna saṅketa

sannidhi santāna śāstra sattā sāttvikabhāva siddha1

Glossary Compound word Inherence Semantic stipulation; semantic relation; stipulation of a word-referent relation; see saṅketa Lit. “coming together.” “The injunction of a restrictive relation between name and named” (abhidhānābhidheyaniyamaniyogaḥ samaya ucyate NM, I.592,1–2, quoting from NBh, ad NS 2.1.55) Relation Transitory emotion; syn. vyabhicāribhāva Correlation; association; conjuncture; syntax (Ollett) Disposition; latent impression; unseen effect, lit. “arrangement”; conditioning; trace; preparation; trace The term denotes a property of substances, a disposition to perform given functions, subclassified into three types: the psychological bhāvanā or vāsanā, relevant in linguistic discussions, dynamism (vega), and elasticity (sthitisthāpaka). The first of these three types is considered the direct cause of smṛti, and it is in turn a product of anubhava (arthānubhavasamādito hi saṃskāro vāsanā kathyate, NM, I.459.7, saṃskāro hi nāma yad anubhavajanitaḥ, NM, II.146.4) Cumulative cognition; (in Buddhism) cumulative awareness Linguistic convention; see also samaya Relation between a word and its referent, ostensively learned or conventionally established. The term is not found in early Nyāya, where instead samaya is used, while it is used by Dharmakīrti (e.g., saṇketena vinā sā arthapra tyāsattinibandhanā, PV, 3.46cd) and later by Jayanta (e.g., kila saṃjñopadeśinā panaso ’yam iti vṛddhavacasā cakṣurindriyeṇa panasajñānam utpadyate saṅketakaraṇakā, NM, II. 253,12–15) Proximity Continuity of the mind; continuous succession of instantaneous mental states as the cause of what we mistake for “individual identity” Discipline; knowledge system of the Sanskrit episteme (Cuneo); see uddeśa, lakṣaṇa, parīkṣā Existence psychophysical response (Ollett) Established, already given, ready; proven; see sādhya1

Nyāya

Vaiśeṣika

Dharmakīrti, Jayanta

Buddhism

Alaṅkāraśāstra

445

Glossary

siddha2 siddhānta Śikṣā skhaladgati smaraṇa smṛti

sphoṭa

sphoṭavādin śrutārthāpatti

sthāyibhāva

svabhāva svabhāvahetu svalakṣaṇa

svārtha

Common (laukika) language conveys something already realized (siddha1) (Freschi) Proven, demonstrated; see sādhya2 Synthesis; conclusion; proven fact; in dialectics, the final view evolving from the evaluation of a thesis and an antithesis; see pakṣa Phonetics, one of the ancillary disciplines aimed at the preservation of the form and meaning of the Veda Cognitive stumbling, as a requirement in the process of lakṣaṇā; see bādha Recollection; see smṛti Memory; recollection A cognitive state whose immediate cause is a saṃskāra (saṃskāramātrajanyaṃ jñānaṃ smṛtiḥ, TarkaSaṃ, 13) Memory is not reliable knowledge (pramāṇa), because it is indirectly derived from a previous cognition (smṛtis tu paricchede pūrvabuddhyapekṣaiveti na pramāṇam, MK, s.v., vol. 5, p. 2774b) sphoṭa (Saito); bursting sound (Candotti); The single and indivisible speech principle (Graheli); A trans-phonetic cause of the cognition of meaning (McClintock); see śabdabrahman Holist; propounder of the sphoṭa Postulation of an audible antecedent; antonym dṛṣtārthāpatti; see arthāpatti The postulation of an unheard piece of śabda2, (pramāṇagrāhiṇītvena yasmāt pūrvavilakṣaṇā ŚV, arthāpatti 2cd), intended to be as authoritative as śabda itself: tām arthagocarāṃ kecid apare śabdagocarām / kalpayanty āgamāc cainām abhinnāṃ pratijānate (ŚV, arthāpatti 52) Stable emotion One of the necessary ingredients of an aesthetic experience; see rasa Nature; autonomy of meaning from form (Candotti) Essence of the probandum used as inferential reason; Reason of the thing that an inference is about (McAllister) Particular character; particular (McAllister); antonym sāmānyalakṣaṇa The self-defined true object of perception, a phenomenologically perceptual instant Independent meaning of a word; proper meaning of a word (Ollett); see artha

Bhāṭṭa

Nyāya Mīmāṃsā

Bhāṭṭa

Alaṅkāraśāstra

Buddhism

446

svataḥ prāmāṇya taddhita tadvat

tātparya

tavya uddeśa

uddīpana

upacāra upadeśa

Glossary The meaning of a word taken in isolation, disconnected from the other word-meanings in the sentence; see anvitābhidhāna Intrinsic validity of cognitions Secondary suffix Universal-qualified particular; qualified by the universal; Property-bearer, possessor-of-that; see dharmin The possessor of a property, such as a jāti. According to Jayanta, it is the referent of generic nouns (gavādijātiśabdānāṃ gotvādijātyavacchinnaṃ vyaktimātram arthaḥ, yas tadvān iti naiyāyikagṛhe gīyate, NM, II.5.3–4) Contextual meaning (Freschi); intention (Iwasaki); purport or purpose (Ollett); intentionality (Graheli) In earlier Nyāya, primarily a property of words. Derived from para, “aiming at” or “tending to,” and tat, “that.” According to Jayanta, it is one of the two śaktis of words, along with abhidhā or abhidhātrī śakti; Udayana understands para as the uddeśa and tat as śabda2 (yaduddeśena yaḥ śabdaḥ pravṛttaḥ sa tatparaḥ, NKu, 620) It is accepted as the fourth function of language, in addition to abhidhā, lakṣaṇā and vyañjanā In Navyanyāya it overlaps with vivakṣā, and as such it becomes also a property of the speaker (kin tu pratipādakecchāviṣayatvam. yaḥ śabdaḥ vaktrā yadicchayā prayuktaḥ sa tatparaḥ, TC, vol. 4 (1), p. 327; vaktur icchā tu tātparyam, NSM, 84cd) Short for gerundive suffixes, tavya being the first to be mentioned in the Aṣṭādhyāyī; see kṛtya Enumeration; listing of categories; see lakṣaṇa, parīkṣā The bare list of names of categories at the beginning of a treatise (nāmadheyena padārthamātrasyābhidhānam uddeśa, NBh, 8,7–8), which constitutes a śāstra along with lakṣaṇa and parīkṣā Stimulant factor; see vibhāva One of the two types of vibhāva in the process of aesthetic experience Figurative usage of words; figurative process; see lakṣaṇā Instruction Equated to speaking, in the definition of śabda2 as epistemic instrument (upadiśyate iti ko ’rthaḥ. abhidhānakriyā kriyate, NM, I.398,1) Direct teaching, later “first teaching” (Candotti)

Prābhākara

Jayanta

Nyāya

Abhinavagupta Navyanyāya

Vātsyāyana

Alaṅkāraśāstra

Jayanta Vyākaraṇa

447

Glossary upādhi

upalakṣaṇa Uttara-Mīmāṃsā uttarapakṣa vāc vācaka vācya vāda

vādin vaidika vaikharī

vaikṛtadhvani Vaiśeṣika vākya

vākyapadīya

vākyārtha

Contingent condition; contingency; nonessential condition; incidental condition; condition ākāṅkṣā, yogyatā and sannidhi are the three upādhis of a sentence meaning (Ollett) Secondary characteristics “Subsequent Mīmāṃsā”; see Vedānta Antithesis; rebuttal of a pūrvapakṣa Speech; voice (Candotti) Signifier; expressive element; see arthapratyāyaka Signified; directly expressed; see artha Debate Debate under the conditions of an unbiased evaluation of all the arguments in favor and against a thesis by means of epistemically valid evidence (pramāṇatarkasādhanopālambhaḥ siddhāntāviruddhaḥ pañcāvayavopapannaḥ pakṣapratipakṣaparigraho vādaḥ, vītarāgavastunirṇayaphalo vādaḥ, NM, I.18) Disputant in a debate Vedic, related to the Veda; Vedic impersonal language, antonym laukika External speech; concrete speech (Saito); see madhyamā, paśyantī Articulated speech, caused by the emission of air through the places and organs of articulation (sthāneṣu vivṛte vāyau kṛtavarṇaparigrahā / vaikharī vāk prayoktṝṇāṃ prāṇavṛttinibandhanā // VP, 1.165) Secondary sound (Saito); see dhvani Atomism, the system of thought ascribed to Kaṇāda (perhaps second c. BCE) with a focus on ontology and metaphysics Sentence; see kriyā, kāraka The combination of a verb (see ākhyāta, kriyā), its complements (see kāraka), and the complements of such complements (ākhyātam sāvyayaṃ sakārakaṃ sakārakaviśeṣaṇam vākyasaṃjñam, VMBh, ad 2.1.1, vol. 1, p. 367,10); a textual unit having a single verb (ekatiṅ vākyasaṃjñam, VMBh, ad 2.1.1, vol. 1, p. 367,12) Of words and sentences Any literary text, lit. “consisting of words and sentences” The name of Bhartṛhari’s magnum opus Sentence meaning; referent of a sentence Among ancient grammarians, Vājapyāyana identifies sentence meaning with the conjuncture (saṃsarga) of word meanings, and Vyāḍi with their mutual difference (bheda) (Ollett)

Śālikanātha

Nyāya

Bhartṛhari

Patañjali

Vyākaraṇa

448

varṇa

varṇasaṃvit varṇavādin vāsanā Vedānta

vibhāva

vicāra vidhi vikalpa vimarśa virodha

viṣaya viśeṣa

Glossary A complex state of affairs (viśiṣṭārtha in the Mīmāṃsā jargon): “The sentence conveys the specific” (viśeṣe vākyam, ŚBh, ad PMS 1.1.24); Kumārila and Prabhākara both endorse the principle of ekārthatā, but this artha is a prayojana for Kumārila, and a kārya for Prabhākara (Freschi) Śālikanātha’s view of sentence meaning involves all three shades of artha: it is not only what is signified, but also a knowable state of affairs (see prameya), and it possesses a teleological aspect (see prayojana); in a sentence the kārya is the main element in relation to its subservient elements, and as such it is considered the referent of a sentence (Kataoka) Multiple word-referents constitute the sentence meaning (padārtha eva vākyārthaḥ […] kiṃ tu naikaḥ padārtho vākyārthaḥ, anekas tu padārtho vākyārthaḥ, NM, II.136,12) Speech unit; single sound; phoneme Mīmāṃsā authors are extremely careful in distinguishing the unchangeable varṇas from their phonic representations through sounds (dhvani). Phonemic consciousness as a common ground of linguistic sounds (Ferrante) Propounder of phonemic speech units; proponent of the doctrine that phonemes are the conveyor of meaning; Upholder of phonemes Trace; see saṃskāra The philosophical school primarily focusing on the exegesis of the Upaniṣads; also known as Uttara-Mīmāṃsā (Freschi); Lit. “the end of the Veda” Factor One of the necessary ingredients of an aesthetic experience; see rasa critique, critical reflection Injunction; prescription; see kārya, niyoga It consists of a niyoga, a niyojya, and a viṣaya Conceptualization Reflective awareness Conflict; obstruction One of the possible causes in figurative meanings; see lakṣaṇā Object; content of a cognition Particular; individual; specific meaning According to Śālikanātha, a specific meaning achieved by ākṣepa from a sāmānya, that is, the sentence meaning (Ollett)

Bhāṭṭa

Prābhākara

Jayanta

Mīmāṃsā

Alaṅkāraśāstra

Śālikanātha

449

Glossary

viśeṣaṇa viśeṣya viśiṣṭādvaita Viśiṣṭādvaita vivakṣā

vṛddhi

vyabhicāribhāva Vyākaraṇa

vyakti

vyaktimātra vyañjanā

vyāpāra

vyāpti vyatireka

vyatiṣaṅga

Species or particulars, one of the six categories in Vaiśeṣika ontology In Pāṇini’s grammar, a particular rule, opposed to a sāmānya rule Qualifier; qualification Qualificand Qualified nondualism (Freschi); see advaita The Vedānta school established by Rāmānuja (eleventh century CE) (Freschi) Speaker’s intention; desire to speak; desire to utter In the Buddhist reduction of śabda2, that is, knowledge from words, to an inferential process, the speaker’s intention is what is inferred Elongated apophonic grade of vowels; see guṇa2 The third of the three degrees of a vowel (ā, ai, au, vṛddhir ādaic, Aṣṭādhyāyī, 1.1.1) in Pāṇini’s terminology Transitory emotion One of the necessary ingredients of an aesthetic experience; see rasa Grammar, lit. “act of analyzing”; The system of thought ascribed to Pāṇini (perhaps fifth century BCE), Kātyāyana (third century BCE), and Patañjali (second century BCE), which became a philosophy of grammar particularly through Bhartṛhari (fifth century CE) (Candotti) Individual; particular; manifestation (McClintock); see sāmānya, jāti The individual, a substance (dravya), the substrate of qualities and actions (śuklādiguṇādhikaraṇaṃ kriyāśrayaś ca dravyaṃ vyaktiḥ, NM, II.5,7) Bare individual (Dasti); see vyakti Manifestation Posited as a third vyāpāra of śabda2 by Ānandavardhana (Ollett), also called dhvani Energy; function; see śakti That which possesses this property is an efficient cause of something Invariable concomitance; see anvaya1, vyatireka Co-absence; concurrent absence; “what exceeds” (Candotti) Also difference, distinction, separation. In its application to induction, akin to Mill’s relation of difference, the absence of a term in absence of the other; antonym anvaya1 Interconnection; cross-connection [between word meanings] (Ollett); see anvaya2

Nyāya-Vaiṣ. Pāṇini

Buddhism

Pāṇini

Alaṅkāraśāstra

Nyāya

Alaṅkāraśāstra

Nyāya

Prābhākara

450 vyavahāra vyutpatti

yāga yogyatā

Glossary Common linguistic usage; day-to-day communication; Transactional agreement; discursive practice Language acquisition; language learning The process of teaching and learning a conventional signifier-signified relation (saṅketagrahaṇa, NM, II 202.8) Sacrifice Compatibility; semantic fitness; semantic compatibility According to Śālikanātha, “suitable to be Mīmāṃsā connected” (yat sambandhārham, VM, p. 390); for example, water is suitable to be sprinkled, while fire is not The counter-concept of bādha, but Mukula intends Mukula with it the consistency between the contextual information and the speaker’s intention (Cuneo) Absence of obstructing true beliefs Navyanyāya (bādhakapramābhāva, NK, s.v.)

Glossary’s References ArthaSaṃ

Ācārya, Nārāyaṇa Rāma, ed. 1950. Arthasaṅgraha of Laugākṣi Bhāskara, with the commentary Mīmāṃsārthasaṅgrahakaumudī by Paramahaṃsa Rāmeśvara Śivayogi Bhikṣuvara. Mumbai: Nirṇayasāgara. Aṣṭādhyāyī Sharma, Ram Nath, ed. 1987–2003. The Aṣṭādhyāyī of Pāṇini. Edited and translated, 6 vols. New Delhi: Munshiram Manoharlal. AVM Telang, Mangesh Ramkrishna, ed. 1916. Abhidhāvṛttimātṛkā of Mukula Bhaṭṭa and Śabdavyāpāravicāra of Rājānaka Mammaṭācārya. Bombay: Nirnaya Sagar Press. GBh Shah, Nagin J., ed. 1972. Cakradhara’s Nyāyamañjarī-granthibhaṅga. Lalbhai Dalpatbhai series, 35. Ahmedabad: L.D. Institute of Indology. Kāśikā Shobhitamishra, Pandit, ed. 1952. Kāśikāvṛtti. 2 vols. Varanasi: Chaukhamba Sanskrit Pustakalaya. KPr Abhyankar, Vasudeva Shastri, ed. 1921. Kāvyaprakāśa of Mammaṭa with the Saṃketa by Māṇikyacandra. Pune: Anandasrama. MK Kevalanandasaraswati. 1992. Mīmāṃsākoṣaḥ. Sri Garib Dass Oriental Series, 148-154. In seven volumes. Or. ed. 1952, Prajnapathasala Mandala, Wai. Delhi: Sri Satguru Publications. MNP Edgerton, Franklin. 1929. Mīmāṃsānyāyaprakāśa of Āpadeva. Introduction, Sanskrit Text, English Translation, Glossary, and Notes. New Heaven: Yale University Press.

Glossary NBh Nirukta NK NKu NM NŚ NSM

NSū PMS PrBh PrP3 PS PV PV I

ŚBh ŚV ŚV

451

Thakur, Anantalal, ed. 1997. Gautamīyanyāyadarśana with Bhāṣya of Vātsyāyana. Nyāyacaturgranthikā 1. New Delhi: Indian Council of Philosophical Research. Sarup, Lakshman, ed. 1984. The Nighaṇṭu and the Nirukta. Original edition 1927. Delhi: Motilal Banarsidass. Jhalakīkara, Bhīmācārya. 1996. Nyāyakośa or Dictionary of Technical Terms of Indian Philosophy. Bombay Sanskrit Series 49. Or. ed. 1875. Pune: Bhandarkar Oriental Research Institute. Goswami, Mahaprabhulal, ed. 1972. Nyāyakusumāñjali of Udayanācārya. Mithila Institute Series, Ancient Texts 23. Darbhanga: Mithila Research Institute. Varadacharya, K.S., ed. 1969–1983. Nyāyamañjarī of Jayantabhaṭṭa: With Ṭippaṇī Nyāyasaurabha. 2 vols. Mysore: Oriental Research Institute. Ramakrishna Kavi, Manavalli, ed. 1926. Nāṭyaśāstra of Bharata Muni. With the commentary of Abhinavagupta. Gaekwad’s Oriental Series 36. Baroda: Oriental Institute. Gangadhara Shastri Bakre, Mahadev, ed. 1934. The Kārikāvali of Vishwanātha Panchānana Bhatta. With the Commentary Siddhānta Muktāvali. 8th. Bombay: Pāndurang Jāwajī, Proprietor of the Nirnaya Sagar Press. Nyāyasūtra of Gautama. In NBh. Pūrvamīmāṃsāsūtra. In Mīmāṃsādarśanam. Dvivedin, Vindhyesvariprasada, ed. 1895. The Bhāṣya of Praśastapāda together with the Nyāyakandalī of Śrīdhara. Vizianagram Sanskrit Series 5. Benares: E.J. Lazarus & Co. Subrahmanya Sastri, A., ed. 1961. Prakaraṇa Pañcikā of Śālikanātha Miśra with the Nyāya-Siddhi of Jaipuri Nārāyaṇa Bhaṭṭa. Banaras Hindu University Darśana Series 4. Varanasi: Benares Hindu University. Pind, Ole Hoten. Dignāga’s Philosophy of Language: Pramāṇasamuccayavṛtti V on Anyāpoha. Part 1 and Part 2. In Pind 2015. Pandeya, Ram Chandra, ed. 1989. The Pramāṇavārttikam of Ācārya Dharmakīrti: With the Commentaries Svopajñavṛtti of the author and Pramāṇavārttikavṛtti of Manorathanandin. Delhi: Motilal Banarsidass. Dharmakīrti. 1960. “Pramāṇavārttike Prathamaḥ Paricchedaḥ.” In The Pramāṇavārttikam of Dharmakīrti: The First Chapter with the Autocommentary; Text and Critical Notes, ed. by Raniero Gnoli, 1–176. Serie Orientale Roma 23. Rome: Istituto Italiano per il Medio ed Estremo Oriente. Śābarabhāṣya of Śabara. In Mīmāṃsādarśana. Dvarikadasa Shastri, ed. 1978. Ślokavārttika of Śrī Kumārila Bhaṭṭa with the Commentary Nyāyaratnākara of Śrī Pārthasārathi Miśra. Prachyabharati Series 10. Varanasi: Tara Publications. Dvārikādāsa Śāstrī, ed. 1978. Ślokavārttika of Śrī Kumārila Bhaṭṭa with the Commentary Nyāyaratnākara of Śrī Pārthasārathi Miśra. Prāchyabhārati Series 10. Varanasi: Tara Publications.

452 TarkaSaṃ TC VaiSū VM VMBh

VP

Glossary Sharma, Mahadev, ed. 1919. Tarkasaṃgraha of Annambhaṭṭa with Nyāyabodhinī and Padakṛtya. Sixth edition. Bombay: Nirnaya Sagar. Kamakhyanath Sharma, Tarkavagish, ed. 1990. The Tattvacintāmaṇi of Gaṅgeśa Upādhyāya. 4 vols. Delhi: Chaukhamba Sanskrit Pratishthan. Original edition Calcutta: Asiatic Society, 1888–1901. Muni Jambuvijaya, ed. 1961. Vaiśeṣikasūtra of Kaṇāda with the Commentary of Candrānanda. Baroda: Oriental Institute of Baroda. Vākyārthamātṛkā of Śālikanātha. In Prakaraṇapañcikā. Kielhorn, Franz, ed. 1880–1885. The Vyākaraṇa-Mahābhāṣya of Patañjali. 3 vols. Third edition, revised and furnished with additional readings, references and select critical notes by K.V. Abhyankar, Pune: Bhandarkar Oriental Research Institute, 1962–72. Bombay: Government Central Book Depôt. Rau, Wilhelm, ed. 1977. Bhartṛharis Vākyapadīya, die mūlakārikās nach den Handschriften herausgegeben und mit einem pāda-Index versehen. Wiesbaden: Franz Steiner Verlag.

Annotated Bibliography of Indian Philosophy of Language

This annotated bibliography is presented in chronological order. The focus is on publications in English language that may engage in a dialogue with presentday philosophers. Publications that are of no use without Sanskrit competence have been neglected. The literature on the subject is vast and this selection is based on the subjective view of this volume’s contributors, so there is no claim of exhaustiveness. John Brough. 1951. “Theories of General Linguistics in the Sanskrit Grammarians.” Transactions of the Philological Society 50:27–46 One of the most influential articles on the Grammarians’ theory on the linguistic unit (śabda) and on the sphoṭa. On the basis of the arguments on the nature of śabda in the Mahābhāṣya, where Patañjali also uses the term sphoṭa, and then focusing on Bhartṛhari’s arguments in his Vākyapadīya, in this article Brough shows how the original concept of sphoṭa was purely linguistic, and that its doctrine is well-founded in actual linguistic usage. (Akane Saito)

John Brough. 1953. “Some Indian Theories of Meaning.” In A Reader in the Sanskrit Grammarians, ed. by J.F. Staal, 414–23. Studies in Linguistics 1. Cambridge, MA: MIT Press Sketches the Indian theories of compositionality, abhihitānvaya, and so on, discusses homophones and polysemy, and concludes in favor of Bhartṛhari’s account of the provisional nature of divisions within a sentence. Citing Ānandavardhana’s theory of dhvani, Brough notices how theories of meaning formed the foundation of theories of Sanskrit poetics. This is possibly the first English study engaging with both Sanskrit semantics and philosophy of language of the twentieth century, a precursor of Kunjunni Raja 1963 and others. (Alessandro Graheli)

K. Kunjunni Raja. 1963. Indian Theories of Meaning. Madras: Adyar Library and Research Centre

454

Annotated Bibliography

A systematic survey of South Asian theories of meaning, attempting to show their relevance to modern linguistic discussions. After (1) an introduction on the very concept of meaning, the book unfolds in six chapters: (2) primary meanings of words, (3) the sphoṭa, (4) the four conditions of sentence meaning (e.g., ākāṅkṣā), (5) the relation of words to the sentence (e.g., anvitābhidhāna), (6) metaphor, and (7) poetical meanings. Bhartṛhari’s theories on semantics and Ānandavardhana’s views on poetics stand out as prominent. The book remains one of the most quoted secondary sources in the field. (Alessandro Graheli)

Madeleine Biardeau. 1964. Théorie de la connaissance et philosophie de la parole dans le brahmanisme classique. Paris: Mouton An unsurpassed synthesis on the epistemology of language in Mīmāṃsā, Nyāya and grammar, with a particular stress on the work of Bhartṛhari. (Hugo David)

K. A. Subramanya Iyer, ed. 1965–71. The Vākyapadīya of Bhartṛhari with the Vṛtti: English Translation. Poona: Deccan College Offers an English translation of the first chapter of the Vākyapadīya, together with the oldest extant commentary on it, the Vṛtti. Bhartṛhari discusses here several topics relevant to philosophy of language, most notably his theory of sphoṭa. Iyer is an authority in the field and his translation is not only reliable but also fairly readable. Yet his renderings are not always literal, so for a deeper study the volume should be read together with the Sanskrit original. (Marco Ferrante)

K. A. Subramania Iyer. 1966. Sphoṭasiddhi of Maṇḍana Miśra. English translation. Poona: Deccan College Post-graduate and Research Institute An English translation of Maṇḍana Miśra’s monograph on the sphoṭa, accompanied by the edition and annotations. Iyer also provides a lengthy introduction on (1) how the theory of sphoṭa was discussed by Bhartṛhari and attacked by the Mīmāṃsakas, (2) how Maṇḍana justifies the existence of sphoṭa, and (3) some deeper investigations on some passages of the Sphoṭasiddhi, especially in relation to the relevant discussion found in the Vākyapadīya. (Akane Saito)

K. A. Subramanya Iyer, ed. 1969. Bhartṛhari: A study of the Vākyapadīya in the light of the Ancient Commentaries. Poona: Deccan College Though dated, the best comprehensive introduction to the work of Bhartṛhari, providing information on the author, his philosophical thought, and contributions to grammar. The volume is accessible to beginners, but it is also a useful tool

Annotated Bibliography

455

for more advanced students and scholars, who will particularly benefit from the huge selection of textual passages appended at the end of the book. (Marco Ferrante)

Pierre Filliozat, ed. 1975–1986. Le Mahābhāṣya de Patañjali avec le Pradīpa de Kaiyaṭa et l’Uddyota de Nāgeśa. Pondichéry: Institut Français d’Indologie. Four volumes A French translation of the first lessons of Patañjali’s Great Commentary, till Aṣṭādhyayī 1.3, where some very broad philosophical issues, along with some strictly grammatical debates, are tackled: the aim and method of linguistic description, the interpretation of the practice of definitions and of the device of substitution, the opposition between meaningless and meaningful linguistic units, and so on. Not an easy text to read, but extremely precise in all the technical details and enriched by the translation of two major later commentaries, namely the Pradīpa of Kaiyaṭa (around eleventh CE) and the Uddyota of Nāgeśa (late seventeenth CE). (Maria Piera Candotti)

George Cardona. 1976. Pāṇini, A Survey of Research. The Hague, Paris: Mouton The first of two volumes collecting the virtual totality of research devoted to Pāṇini and, less exhaustively, to its tradition. See Cardona 1999 below for further details. (Maria Piera Candotti)

S.D. Joshi and J. A. F. Roodbergen, eds. 1986. Patañjali’s Vyākaraṇa-Mahābhāṣya. Introduction, Text, Translation and Notes. 1968 Samarthāhnika (2.1.1); 1969 Avyāyībhāvatatpuruṣāhnika (2.1.2–49); 1971 Karmadhārayāhnika (2.1.51– 72); 1973 Tatpuruṣāhnika (2.2.2–23); 1974 Bahuvrīhidvandvāhnika (2.2.23– 38); 1975 Kārakāhnika (1.4.23–55); 1976 Anabhihitāhnika (2.3.1–7); 1980 Vibhaktyāhnika (2.3.18–45); 1981 Prātipadikārthāhnika (2.3.46–71); 1986 Paspaśāhnika; 1990 Sthānivadbhāvāhnika (1 1.56–58). Pune: University of Poona Eleven volumes that offer the edition and annotated translation of some crucial “daily lessons” (āhnika) of Patañjali’s Great Commentary to Pāṇini. The topics of the lessons target some crucial issues in the grammatical debate: the first lesson, or Paspaśāhnika, offers the first systematic reflections on the method and aims of teaching grammar; the other lessons concern some crucial morpho-semantic linguistic facts (compounding, the concept of nominal base, etc.) or some more syntactic ones (the concept of syntactic link, the syntacticsemantic role of agency-factors, and the principle of the unique syntactic marker, anabhita). One last lesson is devoted to the device of substitution and

456

Annotated Bibliography

its management (sthānivadbhāvāhnika). In each volume the edition and the translation are preceded by an English introduction with a useful summary of the main arguments developed in the lessons. (Maria Piera Candotti)

Arindam Chakrabarti. 1989. “Sentence-Holism, Context-Principle and Connected-Designation anvitābhidhāna: Three Doctrines or One?” Journal of Indian Philosophy 17:37–41 Delineates the differences and similarities of the three main theories of language in Brahmanical India, namely Kumārila’s theory of abhihitānvaya, Prabhākara’s theory of anvitābhidhāna, and Bhartṛhari’s theory of the sphoṭa, each attributing lesser and lesser weight to individual words, in respect to whole sentences. The three theories are respectively labeled by Chakrabarti “connected-designation,” “context-principle,” and “sentence-holism.” The article is highly recommended to readers newly approaching Indian theories of language, before getting into the subtleties of each theory. (Elisa Freschi)

Bimal K. Matilal. 1990. The Word and the World: India’s Contribution to the Study of Language. Delhi: Oxford University Press Partially inspired, in structure and contents, by Kunjunni Raja 1963, it discusses India’s philosophy of language, with references to classical Indian Sanskrit texts, from the perspective of the Western philosophy of the time. It consists of three parts: (1) General issues such as Indian grammar and linguistics traditions, relationship between words, meanings and things, theory of kārakas, and verbal knowledge. (2) A discussion of the theory of sphoṭa, of its proponents and critics, as well as an interesting problem of translation in the light of Bhartṛhari’s concept of language. (3) Three appendices devoted to the mystic dimensions of language, the semiotic conceptions in the Indian tradition of argumentation, and the idea of meaning in poetry and literary criticism. (Monika Nowakowska)

Harold G. Coward and K. Kunjunni Raja, eds. 1990. Encyclopedia of Indian Philosophies: The Philosophy of the Grammarians. General editor Karl H. Potter. Delhi: Varanasi: Patna: Motilal Banarsidass To date the most complete presentation of the school of grammar (vyākaraṇa). A first part presents a brief historical résumé of the principal authors—though neglecting the earliest authors and beginning instead with Bhartṛhari—and a brief presentation of the main philosophical topics discussed in the domain of metaphysics, epistemology, and linguistics. The second part consists of a very useful “Survey of the Literature of Grammarian Philosophy” with useful summaries of the most important works. The volume is closed by a vast, though not anymore up-to-date, bibliography. (Maria Piera Candotti)

Annotated Bibliography

457

Saroja Bhate and Johannes Bronkhorst. 1994. Bhartṛhari, Philosopher and Grammarian: Proceedings of the First International Conference on Bhartṛhari, University of Poona, January 6–8, 1992. Delhi: Motilal Banarsidass This volume collects the contributions presented at the First International Conference on Bhartṛhari, which revived the studies on this fundamental author and proposed a plan for the complete translation of his magnum opus that is nevertheless yet to be fulfilled. Many papers enclosed here have thus a broader aim than what is normally expected from such editorial products. Particularly interesting is Ashok Aklujkar’s “Introduction to the Study of Bhartṛhari” that identifies some crucial issues and trends of research. Several contributions focus on the relationship between Bhartṛhari and Buddhist philosophy. (Maria Piera Candotti)

Madhav. M. Deshpande. 1992. The Meaning of Nouns: Semantic Theory in Classical and Medieval India. Dordrecht: Kluwer Academic Publishers A wide-ranging study on a fundamental question of Indian semantics, considered from the point of view of the “new” school of grammar in South India. (Hugo David)

Bimal Krishna Matilal and Arindam Chakrabarti, eds. 1994. Knowing from Words. Western and Indian Philosophical Analysis of Understanding and Testimony. Dordrecht: Kluwer A collection of original articles on the epistemology of testimony, or wordgenerated knowledge. This is a monumental work in the field of comparative or border-less philosophy, to which foremost philosophers from both Indian and Anglo-American traditions contributed: P.  F. Strawson, J.  N. Mohanty, Keith Lehrer, Ernst Sosa, Sibajiban Bhattacharyya, Arindam Chakrabarti, Elizabeth Fricker, Julie Jack, John McDowell, C. A. J. Coady, Michael Dummet, Gordon Brittan, Michael Wel-bourne, Badrinath Shukla, Visvabandhu Bhattacharya, Bimal Matilal, and Sukharanjan Saha. The articles do not presuppose the reader’s knowledge of Sanskrit or technicalities of analytic philosophy. (Yoichi Iwasaki)

Jan E.  M. Houben. 1995. The Saṃbandha-samuddeśa (Chapter on Relation) and Bhartṛhari’s Philosophy of Language. Groningen: Egbert Forsten This translation of the chapter on the relation between the word form and its meaning—from the third book of the Vākyapadīya—is a re-appraisal of Bhartṛhari’s philosophical thought through the lenses of the specific and yet

458

Annotated Bibliography

enlightening topic of the relationship between a word form and its meaning. (Maria Piera Candotti)

Jonardon Ganeri. 1996. “Meaning and Reference in Classical India.” Journal of Indian Philosophy 24 (1): 1–19 Presents the debate on the nature of referents (padārthas) in relation to Kātyāyana’s fifth aphorism, whereby a word is able to refer to a particular because it possesses a certain quality. For the grammarians this quality is a context-sensitive linguistic rule, for Mīmāṃsā authors a universal that is context-invariant, and for Nyāya ones it has both demonstrative (particular) and qualifying (universal) components. The article is very ingenious but not easily accessible without some familiarity with the jargon of contemporary semantics. (Marco Ferrante)

Gerdi Gerschheimer. 1996. La théorie de la signification chez Gadādhara. Le Sāmāṇyakāṇḍa du Śaktivādavicāra. Publications de l’Institut de Civilisation Indienne 65. 2 vols. Paris: De Boccard To date, the most elaborate study of the views on language of the most famous Bengali author of the Navyanyāya school, with the best available translation of his major work on the topic. It contains a very useful and up-to-date introduction, clarifying essential concepts of Navya-Nyāya semantics, but more generally also relevant to philosophy of language. (Hugo David)

Peter M. Scharf. 1996. The Denotation of Generic Terms in Ancient Indian Philosophy: Grammar, Nyāya, and Mīmāṃsā. Transactions of the American Philosophical Society Held at Philadelphia for Promoting Useful Knowledge 86.3. Philadelphia: American Philosophical Society What is the meaning of a generic term, such as “cow”? Is it the universal cowness, particular cows, or/and something else? The book is dedicated to this time-honored question of Indian semantics. After a short introduction, the views of Pāṇinian grammar, Nyāya, and Mīmāṃsā on this issue are thoroughly presented with abundant citations of Sanskrit sources accompanied by their English translation. In particular, the book contains a translation of relevant sections from Patañjali’s Mahābhāṣya, Vātsyāyana’s Nyāyasūtrabhāṣya, and Śabara’s Bhāṣya. As such, it may serve as a source book for anyone interested in Indian semantics. (Yoichi Iwasaki)

Jan E. M. Houben. 1997. “Part II: The Sanskrit Tradition.” In The Emergence of Semantics in Four Linguistic Traditions: Hebrew, Sanskrit, Greek, Arabic, ed. by Wout van Bekkum et al., 51–145. Amsterdam Studies in the Theory

Annotated Bibliography

459

and History of Linguistic Science. Series III—Studies in the History of the Language Sciences 82. Amsterdam/Philadephia: John Benjamins Publishing Company A part of an editorial project on a corpus of revered texts from various traditions, it offers a clear and comprehensible overview, useful also to a nonspecialist, of the beginnings of the linguistic and semantic inquiries in the Sanskrit language, accompanied by some sociolinguistic context and terminological explanations. It discusses (1) semantic aspects in the Vedas and the Vedic ancillary disciplines, with a particular focus on the Nirukta, (2) the exegetical tradition of Mīmāṃsā, (3) Pāṇini’s grammar and its main commentators, (4) the philosophical schools of Nyāya and Vaiśeṣika, and (5) early Jaina and Buddhist contributions. The essay continues with a large portion on Bhartṛhari’s semantic ideas and is concluded by brief sketches of later philosophical developments, namely the theories of apoha, sphoṭa and śābdabodha. (Monika Nowakowska)

Eivind Kahrs. 1998. Indian Semantic Analysis. The “Nirvacana” tradition. Cambridge: Cambridge University Press The main work in the Nirvacana tradition of semantic analysis is the Nirukta of Yāska, more or less coeval with Pāṇini’s Aṣṭādhyāyī. Often misinterpreted as offering “(folk-)etymologies,” the Nirukta main goal is rather the identification of the real essence of Vedic words. Historical and etymological criteria as understood in the Euro-American world play little role in it. Kahrs explains how the Nirukta works on the basis of a substitution model. For instance, the Nirukta does not claim that a given word is historically derived from a given root, but rather that it replaces that root when a given meaning needs to be expressed. Kahrs further notices the permanence of this model throughout South Asian philosophy of language, and especially in the Śaiva philosophy of language of Kashmir. The book is recommended to scholars who work on philosophy of language in Sanskrit sources, as well as to historians of linguistics. (Elisa Freschi)

George Cardona. 1999. Recent Research in Pāṇinian Studies. Delhi: Motilal Banarsidass An update of Cardona 1976. The bibliographical list of references is preceded by a presentation and discussion of the main research topics and ongoing debates in the scholarly community. Besides some more philologically oriented issues (authorship and date of the relevant texts) some crucial broader topics are also tackled, such as the general structure of the Aṣṭādhyayī: the role of metalanguage and metarules, the role of semantics, and so on. On the most

460

Annotated Bibliography

important issues, the author discusses the different scholars’ views and advances his specific viewpoint on the matter. The two volumes are an invaluable tool to understand the state of the art (even though not up-to-date anymore) in Pāṇinian research. (Maria Piera Candotti)

Krishnacharya T. Pandurangi. 2004. Prakaraṇapañcikā of Śālikanātha, with an Exposition in English. New Delhi: Indian Council of Philosophical Research Śālikanātha is the best-known author of the Prābhākara sub-school of Mīmāṃsā. His so-called Prakaraṇapañcikā is a collection of essays on various topics. One of these essays, the Vākyārthamātṛkā, is one of the most influential philosophical texts on language ever written in South Asia. Pandurangi’s book contains a loose English translation of selected passages of the Prakaraṇapañcikā, linked with each other’s by brief summaries of the untranslated portions. It thus offers a bird-eye view of the whole Prakaraṇapañcikā and it is an excellent introduction to the Vākyārthamātṛkā. (Elisa Freschi)

Bimal Krishna Matilal. 2005. Epistemology, Logic and Grammar in Indian Philosophical Analysis. Oxford: Oxford University Press When first published in 1971, this book broke new ground in the study of Indian philosophy with its careful analysis of the connections among theories knowledge, reason, and language across treatises from the Naiyāyika, Buddhist, and Grammar traditions and with reference to Western analytic philosophy. While Matilal was to develop his ideas even further in subsequent works (especially Matilal 1990), this book remains foundational in its clear exposition of the centrality of language for understanding Indian philosophical thought. This revised edition, edited by Matilal’s student Jonardon Ganeri, incorporates Matilal’s own revisions. (Sara McClintock)

Lawrence J. McCrea. 2008. The Teleology of Poetics in Medieval Kashmir. Harvard Oriental Series 71. Cambridge, Mass.: Department of Sanskrit / Indian Studies, Harvard University An in-depth study of the major ideas developed in poetics and literary criticism, hinging around Ānandavardhana’s theory of poetic suggestion (dhvani) and in relation to the hermeneutic techniques of Mīmāṃsā scholarship. One of the many merits of the volume is bringing to the fore the previously understudied influence of Mīmāṃsā hermeneutics in the poetic discourse, far beyond the precincts of Vedic exegesis. A groundbreaking section discusses the panfigurative theory of language of Mukula, treated by McCrea as a refutation of dhvani. (Alessandro Graheli)

Annotated Bibliography

461

Mithilesh Chaturvedi, ed. 2009. Bhartṛhari: Language, Thought and Reality. Proceedings of the International Seminar, Delhi, December 12–14, 2003. Delhi: Motilal Banarsidass This volume is an ideal prosecution of Bhate and Bronkhorst 1994. It collects contributions of the most important scholars who have devoted many years of their life to the study of this author (in particular Aklujkar, Bronkhorst, Deshpande, and Ogawa). The different contributions highlight both the links of this author with the Vedic and Brahmanical philosophical tradition and its contribution to the development of the analysis of the different expressive powers of language, which paved the way for later reflections on pragmatic and metaphoric aspects of communication. (Maria Piera Candotti)

Lawrence J. McCrea and Parimal G. Patil. 2010. Buddhist Philosophy of Language in India: Jñānaśrīmiśra on Exclusion. New York: Columbia University Press This important publication consists in a study, translation, and printed edition of the Apohaprakaraṇa, or “Monograph on Exclusion,” by the eleventhcentury Buddhist Jñānaśrīmitra. The introduction details developments in the apoha theory from Dharmakīrti onward, with special attention to Dharmottara’s innovations later adapted by Jñānaśrīmitra. While continuing to address questions of linguistic reference, these later thinkers also further contribute to Buddhist epistemology through the application of apoha to their theories of concept formation and perceptual knowledge. Jñānaśrīmitra’s text illustrates how exclusion becomes a key Buddhist analytical tool not only for philosophy of language but also for philosophy of mind. (Sara McClintock)

Johannes Bronkhorst. 2011. Language and Reality. On an Episode of Indian Thought. Brill’s Indological Series 36. Leiden/Boston: Brill A cornerstone booklet identifying the broader philosophical implications of the “correspondence principle,” as labeled by the author, which assumes a one-to-one relationship between word-forms and word-meanings in a sentence. Bronkhorst identifies different traces of this implicit principle not only in the grammarians’ tradition (or in the broader domain of philosophy of language) but also in many debates developing in the centuries around the beginning of the Christian Era in many philosophical schools, such as the doctrine of sātkāryavāda (preexistence of effect in cause) or Nāgārjuna’s statement on the intrinsic inconsistency of world-representations. (Maria Piera Candotti)

462

Annotated Bibliography

Jonardon Ganeri. 2011. Artha. Meaning. Foundations of Philosophy in India. New Delhi: Oxford University Press An update and revision of Ganeri’s own Semantic Powers (1999). Addressing philosophers, the book touches upon most of the main topics of the Indian philosophy of language, in particular semantic ideas. It discusses such topics not from a historical or synchronic perspective, but as individual philosophical claims evaluated against each other and against Western and modern propositions, translating and analyzing them through symbolic notations. The author problematizes and discusses various theories of meaning and understanding offered by representative Indian thinkers, mostly focusing on the systems of Nyāya and Navya-Nyāya. The latter, especially, informs the two final chapters on the semantics of names and pronouns. (Monika Nowakowska)

Mark Siderits, Tom J.  F. Tillemans, and Arindam Chakrabarti, eds. 2011. Apoha: Buddhist Nominalism and Human Cognition. New York: Columbia University Press This anthology of investigations of the nominalist Buddhist theory of exclusion (apoha), by leading scholars from a range of disciplines, includes close readings of Indian texts alongside critical interpretations in light of contemporary philosophy, linguistics, and cognitive science. Excellent introductory essays early in the volume highlight the contributions of the Buddhist progenitors of the theory, Dignāga and Dharmakīrti. Later contributions include studies of the reception of apoha among Buddhists and non-Buddhists in both India and Tibet. The volume ends with a series of philosophical accounts of the potential contributions of apoha to contemporary semantics and philosophy of mind. (Sara McClintock)

Patrick McAllister, ed. 2017. Reading Bhaṭṭa Jayanta on Buddhist Nominalism. Wien: Verlag der Österreichischen Akademie der Wissenschaften Contains a critical edition and an annotated English translation of the section of Bhaṭṭa Jayanta’s Nyāyamañjarī on the Buddhist theory of exclusion (apoha), as well as a collection of essays, mostly on apoha and to a lesser degree on Jayanta’s treatment of language. Jayanta’s treatment of apoha is an excellent source-based introduction to the topic and confirms his ability to reconstruct his opponents’ positions with remarkable clarity. Though the Sanskrit edition and some essays are intended for a specialist audience, the book also contains an accessible and yet accurate English translation, as well as some contributions— particularly, Freschi and Keidan’s, and McAllister’s own—for the use of linguists and philosophers. (Elisa Freschi)

Annotated Bibliography

463

Vincent Eltschinger et al. 2018. Dharmakīrti’s Theory of Exclusion. Part I: On Concealing. Studia Philologica Buddhica 36. Tōkyō: The International Institute for Buddhist Studies The first part of a planned three-volume translation of the portion of Dharmakīrti’s Pramāṇavārttika on exclusion (apoha) with the auto-commentary. It is the latest stage of research on the topic of Buddhist nominalism. (Hugo David)

Johannes Bronkhorst. 2019. A Śabda Reader. New York: Columbia University Press An anthology of seminal Sanskrit passages translated into English, addressing even nonspecialists. The volume, published in the series of Readers directed by Sheldon Pollock, contributes to fill the gap of readable English translations of influential Sanskrit passages on linguistic themes. The volume mostly covers the discussion on semantics and metaphysics, with the bulk of its space allotted to the perspectives of Vaiyākaraṇa and Buddhist authors, yet only marginally touching upon the rich South Asian debate on figurative language and other implicatures. (Alessandro Graheli)

Index abhidhā  226 conventional according to Nyāya  227–8 denotation, discussed under pramāṇa in Nyāya  155 referentiality  226 restricted to individual word meaning according to Śabara  356 abhidhāna process of denotation  173 strong and weak  174 abhihitānvaya, see also compositionality construal of the denoted  353 relation of expressed meanings  253, 374 term coined by Śālikanātha  253 Abhinavagupta  12, 141, 351, 352, 389, 397 commentator of Utpaladeva’s work  135 reconciling rasa and theory of meaning  389 systematizer of Pratyabhijñā  135 adhikāra  256 adhyavasāya judgment linking dravyasat and prajñaptisat  166 adhyeṣaṇa, see prescription advaita  297, 314 affix  20, 23–5, 78 ākāṅkṣā, see dependency ākāśa, see ether akhaṇḍārtha according to Prakāśātman  327–30 first occurrences of the concept  332 n.10 undivided object, expressed by sentences  315, 325 Vedāntic theory unrelated to Bhartṛhari’s akhaṇḍavākya theory  331 n.9 akhaṇḍārthavādin  330

ākṛti  157–9, see also word meaning Akṣapāda, see Gautama akṣara, see syllable ākṣepa implication  256, 417 n.49 Alaṅkāraśāstra  396–7 ecumenical nature of  410 n.2 its status of śāstra  397 āmantraṇa, see prescription Ānandabodha  325 Ānandapūrṇa  330 Ānandavardhana  351, 372, 378, 397 author of Dhvanyāloka  372 his critics  398 influenced by Mīmāṃsā  375 his innovative theory  397–8 as innovator  397 rasa cannot be expressed  382 ānantya endless extension of referents of a common noun  230 infinite number of referents  170 ananvitārtha, see word meaning anubhava  60 anumāna, see inference anupapatti condition for indirect signification  323 impossibility  323 anusandhāna  138 anuvāda  322 not veridical (pramāṇa)  334 n.30 anvaya, see also syntax concurrent existence  176 hierarchical relation  366 insufficient to explain universal relations  176 in Prābhākara  258 relation between word meanings  253 anvayavyatireka as what remains and what exceed, in grammar  28

Index anvitābhidhāna, see also compositionality construed denotation  353 expression of relational meanings  253–4, 374 in Vedānta  320 apauruṣeya  364 apoha, see exclusion Appayya Dīkṣita  324, 330 on criteria for immediacy of cognition  318 substantiating the vitality of Prakāśātman’s ideas  318 āpta criteria for trustworthiness  362 trustworthy witness  362 apūrva  289–90 something previously unknown  251 apūrvakārya, see also prescriptive endings something unknown to be done  289–90 artha  239, 254–5, see also sentence meaning; word meaning in Early Mīmāṃsā  299–301 in Viśiṣṭādvaita Vedānta  307–9 arthāpatti  196–7 arthapratyaya, see knowledge from words arthāsaṃsparśitā  167 arthavat unit endowed with meaning  25 Aṣṭādhyāyī, see also Pāṇini one of the macrotexts in Pāṇini’s grammar  19 word-derivational device  20 ātman, see self bādha obstruction as condition for figurative meaning  399 bases nominal  23, 25–9, 77–8 verbal  22–4, 28–9 Bhartṛhari  25, 82 advocate of sentence unit  32 his subclassification of sound  80 Bhāṭṭa, see also Kumārila sub-school of Mīmāṃsā  182 Bhaṭṭa Jayanta, see Jayanta Bhaṭṭa Nāyaka author of Hṛdayadarpana  374, 397

465

his paradigm shift  397 on rasa as sentence meaning  378, 380 three stages of rasa experience  379 Bhaṭṭa Udbhaṭa  397 his paradigm shift  397 his semanticization of aesthetic experience  397 influenced by Mīmāṃsā  375 Bhaṭṭoji Dīkṣita  88 bhāvanā actualization, as sentence meaning  381 Brahman unqualified, as import of Upaniṣadic statement  316 Bronkhorst, Johannes  32 Brough, John  88 camatkāra  142 Cardona, George  41, 71 causation, see also śakti either natural or contingent according to Jayanta  227–8 Nyāya theory of  226–7 semantic relationship based on  169 Citsukha  325 compatibility  260, 263, 265–8, 404–5 compositionality, see also anvitābhidhāna Jayanta on  366 in Mīmāṃsā  430 principle of  253 Śālikanātha on  258–9, 262, 269 in tension with contextualism  253 in tension with particularism  258 concomitance  168–9, 176, 376 contextualism, see also compositionality Jayanta on  375, 385 in Prābhākara Mīmāṃsā  189 principle of  253, 256 Śālikanātha on  258 co-reference  170–2, 229 darśana  4, 396 denotation, see also knowledge from words Buddhists on  153, 155–7, 161–2 Dignāga on  167–6 indirect  322–4

466

Index

Kumārila on  353–9 Mukula on  402–7 in Vyākaraṇa  29 deontic element intrinsic in commands  296 logic and imperative logic  309 and linguistics  305 subject to truth tables  309 dependency (expectancy)  194–5, 260, 266, 376 ākāṅkṣā  232 expectancy among word meanings or objects  194, 326 expectancy, one of the three criteria for sentence signification  194 explained by Śālikanātha  264 one of the conditions for sentence meaning  387 syntactical in nature  195 derivation, in semantic analysis  20 descriptive sentences  295 Dhanañjaya  374 author of Daśarūpaka  374 Dhanika  374 commentator of Daśarūpaka  374 as innovator  387 younger brother of Dhanañjaya  374, 390 n.5 dharma constituent of reality  31 duty, the meaning of the Veda  39, 48, 186, 300–4, 307, 354 property  209–10 Dharmakīrti  84, 90, 172, 209 anti-sphoṭa philosopher  76 on conceptual cognition  209 critique of his theory of phonemes  84–5 emphasizing the fallacious aspect of language  166 his theory of meaning  115–17 introducing apoha theory  208 introducing the notion of judgment (adhyavasāya)  166 on properties  210 dharmin entity  227 property-possessor  232 dhātu, see bases

dhvani dismissed by Jayanta  366 sound, distinct from nāda  79 suggestion of meaning  352 three types of, according to Ānandavardhana  398 Dignāga  137, 165 his critique of Nyāya semantics  170–2 his theory of apoha  167–8 influenced by Bhartṛhari  165 on tadvat  170–2 on verbal cognition as inference  167 dravya, see substance dravyasat causally existent  171 elements  57–8 ether material element in Vaiśeṣika metaphysics  58 substrate of sound  45, 47, 57–8 exclusion (apoha) criticized by Kumārila  166 Dignāga’s semantic theory  165 in the form of reflection, as awareness a word, according to Kamalaśila  125 the real nature of denotation  176 existence  170, 173, 229, 357, see also non-existence of the causally existent (dravyasat)  165 of the conceptually existent (prajñaptisat)  165 mere existence (sattāmatra)  117 three levels of, in Yogācāra philosophy  165 of the ultimately existent (paramārtha)  165 figure of speech (alaṅkāra)  398 Fillmore, Charles J.  71 n.13 fitness (yogyatā), see also compatibility in Mīmāṃsā  191, 194–5 in Navya Nyāya  342–3 Frege, Gottlob  202, 204, 273 n.8 Ganeri, Jonardon  274 n.16 Gaṅgeśa  344

Index and God as the root of veridical testimony  344 and intention as necessary for knowledge from words  346 Gautama  56 God author of Veda  154, 364 semantic relation as his will  241, 342 source of semantic relation  154, 227, 239, 362 Helārāja  102 n.46 Hemacandra  381 imperative sentences, see prescriptive endings; prescriptive sentences implicature  3, 186, 240, 351, 357, 396 gauṇa gauṇavṛtti  193 secondary cognition  43 secondary signification  365 lakṣaṇā  161, 328, 400–2, 358, 404–9 bādha, sambandha and prayojana its three conditions  403 basis for sentence signification according to Bhāṭṭa Mīmāṃsā  351 distinct from gauṇa  193, 365 indirect reference or indirect denotation  322–3 the move from universal to particular according to Kumārila  357 not conducive to truth  365 poetic meaning a subcategory of, according to Mukula  352 reminder of non-related object (ananvitārtha)  323 requiring impossibility and semantic affinity as conditions  323 its role according to Jayanta  365 secondary indication  358 secondary reference, discussed under chala in Nyāya  155 indication, see implicature (lakṣaṇā) individuals  230, see also particulars bare individual not play role in language  156 denotation of  358

467

property-bearing (tadvat)  153 reference to  356 inference Buddhists on  166–70 Jayanta on  232–3 in Mīmāṃsā  304 Śālikanātha on  282–3 inherence  57–8, 280 intension  210 and extension  257 intention disambiguating of polysemy  340 hearer’s understanding of  340 speakerless  345 intentionality  361, 364–7, see also tātparya īśvara, see God jahallakṣaṇā, see also implicature type of lakṣaṇā  328 Jaimini  182, 299 jāti genus  60 universal  156–7 jātiśabda, see nouns Jayanta  2–3, 56–7, 226, 374 on causation  226–8 discussing abhihitānvaya and anvitābhidhāna  374 and his own theory of sentence meaning  375 on knowing referents  232 on ontology of referents  228–31 on poetic meanings  351 his position on sentence meaning quite close to Prabhākara’s  375 his syncretic theory of sentence signification  367 on tadvat  230–1 Kaiyaṭa  94 kalpanā assumption  44 conceptualization  112 Kamalaśīla  114 Kaplan, David  390 n.4 kāraka factor of action  64, 366 participant of action  71 n.13 syntactic relation  63

468

Index

Karṇakagomin  109, 218 n.22 Kārya  303 distinct from simple action  279, 284–6 its three aspects, impelling, invitation and solicitation  281–2, 289 known through perception and inference  282 obligation  278 something to be done  251, 280, 295, 303 kāryatā distinct from phalasādhanatā  280 including any kind of imperative  281 Kātyāyana  32 Kauṇḍa Bhaṭṭa  91 knowledge from words genuinely inferential according to Dignāga  166–7 immediate verbal knowledge (śābdāpa­­ rokṣajñāna) 317, 333 nn.19–21 known through inferior inference According to Dharmakīrti  167 major preoccupations for Advaitins  315 Prakāśātman’s new theory about  315 theory of, proposed by Dharmakīrti  208 verbal understanding  340 kṛtya, see prescriptive endings Kṣemarāja  88 Kumārila author of lost Bṛhaṭṭīkā  358 author of Ślokavārttika, Tantravarttika and Ṭupṭīkā  182 commentator of the Śābarabhāṣya  182 on denotation  353–9 founder of the Bhāṭṭa school  182 system builder  252 Kumārila Bhaṭṭa, see Kumārila lakṣaṇā, see implicature lakṣya figurative meaning  239 language acquisition (vyutpatti)  227 explained by Śālikanātha  261–3 of obligations  278–9 two-step approach for Vedic sentences  286–7

laukika everyday speech  43 vernacular language  39 worldly usage  183 liṅ, see prescriptive endings liṅādi, see prescriptive endings loṭ, see prescriptive endings McCrea, Lawrence  258, 409–11 Mādhava  108 Madhusūdana Sarasvatī  325 madhyamā, see speech mahāvākya, see sentence Mahima Bhaṭṭa  379, 392 n.21 Mammaṭa  398 author of Śabdavyāpāravicāra  408 his critique of Mammaṭa  405–7 his historical role  407 influenced by Mukula  398 the Kāvyaprakāśa his magnum opus  398, 402 Maṇḍana  78, 240, 316, 335 n.38 author of Brahmasiddhi  326 his theory of sphoṭa  85–6 Maṇḍana Miśra, see Maṇḍana meaning  240–1, see also, sentence meaning; signification; word meaning associated (bhakti)  400 contextual  188 direct  399, 414 n.22 lexical  188 proper and relational  258–64, 375, 388 prototypical  188 secondary  161, 239, 257, 408 (see also indication; lakṣaṇā) śabdārtha  153, 203 memory and fresh cognition  322 not qualifying as valid knowledge  322 prove of a permanent and unchanging self  138 recollection  226 reification of momentary cognitions in Buddhism  137 according to Prakāśātman  322 according to Śālikanātha  191–2, 267–8

Index re-statement through same causal complex  322Mīmāṃsā  182–4, 297 on the meaning of artha  299–301 science of sentences  353 morpheme  11, 91, 181 mukhyārtha, see meaning, direct mukhyārthabādha, see bādha Mukula Bhaṭṭa author of Abhidhāvṛttamātṛka  399 his impact on theory of dhvani  399 refuting dhvani  399 nāda, see also dhvani resonance, as distinct from dhvani  79 uttering or voicing  48 Nāgeśa Bhaṭṭa  90, 102 n.47 nāmadheya appellation, common notion, designation  31 nāman name, equated to property  210 Navya Nyāya  239, 340 nimitta necessary cause  43 pravṛttinimitta, ground or reason for the use of a word  192 nirākāravāda  131 n.45 nirvacana, see semantic analysis niṣedha prohibition  305 nitya fixed, as opposed to variable  40, 184 obligatory rite  286 permanent, as opposed to temporary  26, 82, 362 niyoga command  278 obligation  381 niyojya, commanded person  289 non-existence  167, 357 nouns, see also word; word meaning common, meaning of  161 general  240 Nṛsiṃhāśrama  330 Nyāya  56, 161, 226, 239, 295 nyāyas principles guiding in understanding sentences  269–70

469

obligation, see kārya; niyoga pada, see word padapāṭha functional segmentation  16 word-for-word recitation  16 padārtha, see word meaning Padmapāda  315 Pāṇini  19 author of Aṣṭādhyāyī  19 his grammar being a macro-text  19 types of rules in his grammar  20 parā, see speech Pārthasārathi  182 particulars  170, 221 n.33, see also individuals actionable  262 conveyed, denoted or designated by words  229, 363 not denoted  174 predication of  357 relational meanings as particulars  270 paśyantī, see speech Patañjali  22, 26, 82, 183, 405 perception  63–6 devoid of concepts (kalpanāpoḍha) according to Dignāga and Dharmakīrti  112 of sphota  86–7 phalasādhanatā character of being a means to attain a result  280 distinct from kāryatā  280 phoneme, see varṇa Pollock, Sheldon  371–4, 377–8, 382, 390, 397, 404 polysemous words  340 Prābhākara, see also Prabhākara Miśra sub-school of Mīmāṃsā  182 Prabhākara Miśra commentator of Śābarabhāṣya  182 founder of the Prābhākara school  182 prajñaptisat conceptually constructed existent  165 parallel to Bhartṛhari’s upacārasat  177 n.2 prakaraṇa, literary genre  316

470 Prakāśātman  250 author of Śābdanirṇaya  315 commentator of Padmapāda’s Pañcapādikā  315 endurance of his semantic hypotheses  334 n.26 familiar with Śālikanātha  334 n.27 follower of anvitābhidhānavāda according to other Vedāntins  324 his anvitābhidhāna in Vedāntic garb  320–1 initiator of Vivaraṇa-line of interpretation of Śaṅkara’s Bhāṣya  315 paraphrasing Śālikanātha  320 prakriyā derivation of specific linguistic forms in Vyākaraṇa  20 type of grammar such as Siddhāntakaumudī  21 prākṛtadhvani, see sphoṭa pramāṇa  62–6, 71 n.14, 72 n.17, 155, 192, 299 source of knowledge  345 prātipadika meaningful along with verbal bases  28 meaningful telic element according to Pāṇini  25 nominal base that cannot be meaningful  27 Pratyabhijñā doctrine of Kashmirian scholars  135 its metaphysics  137–8 pratyakṣa, see perception pravṛttikāraṇa, see nimitta pravṛttinimitta, see nimitta predicate predicate or attribute, translating Frege’s Bestimmung  210 predicate term  205 and subject  218 n.20, 353 and svabhāva as property  218 n.20 prescription (vidhi)  281 prescriptive endings apūrvakārya being their primary referent  290

Index kriyākārya being their secondary referent  290 prescriptive endings  278 prescriptive sentences  295 proximity (sannidhi), see also sentence signification explained by Śālikanātha  264 between mental images according to Śālikanātha  195 one of the three criteria for sentence signification  194–6, 260, 263, 268 between words  195 Pūrvamīmāṃsā, see Mīmāṃsā Quine, W.V.O.  206–8, 216, 217 n.9 Rājānaka Mammaṭa, see Mammaṭa Rāmānuja  297, 307–9 rasa aesthetic emotion  351 cannot be conveyed by using words for it  372 emotional experience  371 emotional mood  399 its distinction from other types of meanings  373 not expressed but manifested according to Ānandavardhana  373 semiotic theory not sufficient to explain it  371–2 as sentence meaning  352, 372 Recanati, François  188, 198 reference four theories of, discussed by Dignāga  170 referent as being produced and transient  175 higher and lower according to Dignāga  175 Śabara  39 on abhidhā  356 author of Śābarabhāṣya  182 commentator of Pūrvamīmāṃsāsūtra  182 on signification of individual words  356 on universals  356–7

Index śabda, see also dhvani; nāda; speech; testimony; word, sentence as constant, fixed, invariable (nitya)  40 constituted of phonemes  50 distinct from the object  41 its definition in Mīmāṃsā  41 its impermanence because it is a product, according to Jayanta  59 possessing a universal  59–60 syllogism in favor of  59 three reasons for, according to Vidyabhushana  58 its worldly (laukika) and Vedic usages  183 not translatable as “word”  181 quality of ether (ākāśa)  57–8 as source of knowledge as speech  47, 64, 77 as testimony, by definition impermanent  362 translated as “linguistic expression” or “linguistic unit”  182 śābdabodha, see knowledge from words śabdabrahman and sphoṭa  101 n.39, 127 n.11 supreme and undivided speech  90 śabdādvaita linguistic monism  72 n.20, 88 śābdāparokṣajñāna, see knowledge from words (immediate verbal knowledge) śabdāpoha, see exclusion śabdārtha, see meaning śabdavyāpāra  372 sādhya, see also siddha prescriptive meaning  307 sākāravāda  131 n.45 sākṣātkāra first-hand or direct experience  67–8 as foundation of authority  68 presentification  317 śakti  136, 239, 321, see also causation as causation  227–8 known by postulation  260 power to generate a cognition  259, 375–6 of words, according to Jayanta  19 word’s expressive power  321

471

śākya direct meaning opposed to figurative meaning  239 Śākyabuddhi  109, 218 n.22 Śālikanātha Miśra  251–2 author of Vākyārthamātṛkā  251 barely quoting his teacher Prabhākara  252 builder of Prābhākara Mīmāṃsā  252 commentator of Prabhākara’s works  182 held Kumārila in esteem  252 innovator  182 samānādhikaraṇa, see co-reference sāmānādhikaraṇya, see co-reference samanvaya, see also Vedānta and akhaṇḍārtha  326 convergence of Upaniṣad and Veda  314, 325 sāmānya as synonym of jāti  60 sāmānya and viśeṣa their use in grammatical treatises  27 their use in liturgic literature inspiring Pāṇini’s grammar  21 their use in theory of meaning  256 sāmānyalakṣaṇa conceptually constructed things called general characteristics  165 different from particular characteristics (svalakṣaṇa)  165 samaya, see semantic relation sambandha, see semantic relation saṃsarga, see also sentence signification; syntax conjuncture of word meanings  380 saṃskāra conditioning  116 disposition  60 preparation to Scripture  319 trace left by previous, first-hand cognition  322 unseen effect  43 Śaṅkara  87, 297, 314 saṅketa, see semantic relation sannidhi, see proximity Śāntarakṣita  109 Sarvajñātman  331 n.8 sattā, see existence segmentation  13–14, 16–18, 24–5

472 self  58–9, 63–4, 108, 137–9, 296, 319 semantic analysis  15 claimed to be the fulfillment of Vyākaraṇa  16 semantic relation based on God’s will  343 defined by God  243 lexical convention  243 negative and weak, proposed by Dignāga  176 social agreement  243 transient and not innate according to Dignāga  166 semantics  11, 60 Buddhists on  165–6, 168, 174 in Navya Nyāya  239–43 Prakāśātman on  315–16 Śālikanātha on  263–5 sentence  26 hierarchical organization within  258, 261, 379–80 hierarchical unity of  258, 379 mega-sentence (mahāvākya)  125 unity of  379 sentence meaning content of awareness produced by an expression  371 deontic object  364 determinate particular  256 epistemic object  364 in Mīmāṃsā  185–92, 194–6 qualitatively different from word meaning  255 secondary cognitive process according to Kumārila  382 something to be done  181 something to be done according to Dhanika  389 undivided object (akhaṇḍārtha)  315 sentence signification according to Bhāṭṭa Mīmāṃsā  186–9, 363 according to Prābhākara Mīmāṃsā  189–93, 363 ākāṅkṣā, yogyatā and sannidhi being the three criteria for  195 necessarily requiring figurative implicature  354–5

Index novelty, teleology and feasibility as guide for interpretation of exhortative sentences  307 siddha already given as synonym of nitya  27 state of affair  307 Siderits, Mark  203–5 Skinner, Quentin  256, 274 n.16 smaraṇa, see memory smṛti, see memory Somānanda  135 speech (vāc) classifications of  89–90 definition of  50 highest (parā)  89, 102 n.46 Jayanta on  65–8 parrot’s  341–6 three levels of, according to Bhartṛhari  89 concrete (vaikharī)  89 intermediate (madhyamā)  89 intuitive (paśyantī)  89, 127 n.11 sphoṭa according to Bhartṛhari the indivisible form of sound manifesting to both utterer and hearer  79 the process of its manifestation  80–1 according to Maṇḍana directly perceived  86 according to Patañjali distinct from sound  77–9 its etymological meaning as bursting sound  31 according to Premodern Grammarians equated to the highest reality  93 indivisibility of  9 manifested by a sequence of phonemes  92 phoneme-/sphoṭa/  91 requiring the concept of phoneme (varṇa)  90 sentence-/sphota/  91–2 universal sphoṭa  92–3 word-/sphoṭa/  91–2 according to Śaiva philosophers the highest reality  76, 88 and linguistic monism (śabdādvaita)  88

Index five historical variants of the concept  76–7 and the fourth level of speech theorized by Nāgeśa  90 Jayanta’s criticism  87 linguistic monism (śabdādvaita) wrongly correlated to  88 Śaṅkara’s criticism  87 Śāntarakṣita’s and Kamalaśīla’s criticis eternality and sequence incompatible  110–11 relying on Kumārila’s criticism  108–9 unity and permanence of sphoṭa impossible on mereological bases  108 unity of sphoṭa inconsistent with the segmentation of language  108 sentence-sphoṭa might be Jayanta’s invention  100 n.34 and the three level of speech theorized by Bhartṛhari  89 sphoṭavādin  113, 119, see varṇavādin śrutārthāpatti  196–7, 275 n.40 substance (dravya)  156, 181 Sucarita Miśra  182 Sureśvara  315 svabhāvahetu  208 svalakṣaṇa particular  206 particular characteristics  165 syllable  13, 16–17, 24 synonymy  206, 208–11 as cognition of things in the same class  210–11 as having the same form (ekapratibhāsa)  214 syntax, see also anvaya association of word-objects  323 combination of referents  366 conjuncture of word meanings  379 as opposed to semantics  194, 264 phonosyntax of Vedic hymns  33 saṃsarga  389 tadvat analogous to apohavat  170 property-possessor  367

473

referent of individual words and not of sentences  231, 233 synonym of jātimanmātra  170 that-possessor  228–31 theory of, criticized by Dignāga  170 tātparya  341, see also intention based on the hierarchy of individual referents  366 distinct from speaker’s intention  361 etymology of  361 the fact of having “that” as its main purpose  375 intentionality, as opposed to referentiality  226 Jayanta’s hallmark  391 n.12 property of words  226, 361, 363–5 purport  375 purpose  341 sentence meaning  363 speaker’s intention  341 speaker’s will  341 transferred from words to speaker in Navya Nyāya  341 tātparyaśakti power of purport  376 testimony (śabda)  48–9 defined as the cause of veridical understanding  155 by definition impermanent  362 Dignāga on  177 n.6 Jayanta on  62, 67, 69, 364, 367 Udayana  341 Uddyotakara  56, 155 universals, see also jāti; sāmānya according to Śabara  356 Buddhists on  124, 215 in Mīmāṃsā  154, 185–8, 261, 356–9 in Nyāya  59, 156–8, 160 Umbeka, see Uṃveka Bhaṭṭa Uṃveka Bhaṭṭa  182 upacāra  398 figurative meaning  357 upadeśa, see also testimony direct teaching in grammar  23, 35 instruction, as definition of testimony  65 upādhi  318, 329 condition, as opposed to upalakṣaṇa  263

474

Index

upalakṣaṇa secondary characteristic, as opposed to condition  263 undenotated indicator  235 n.20 Upavarṣa  42, 183, 199 n.5 Utpaladeva  95, 102, 135 Uttaramīmāṃsā, see Vedānta in relation to Pūrvamīmāṃsā  297, 314 Uttuṅgodaya  413 vāc, see also speech speech  89 voice  14 vācaka related to Saussure’s signifiant  17 signifier  13 Vācaspati Miśra  316 vācya related to Saussure’s signifié  17 signified  13, 17 Vājapyāyana  385 vākya, see sentence vākyārtha, see sentence meaning varṇa  11, 17, 60, 91, 103 n.51, 199 n.2 Abhinavagupta on  143 acoustic manifestation of consciousness  143 atomic unit of speech  40 Dharmakīrti on  115–25 impermanent according to Śāntarakṣita  122 Kumarila on  59–61 Maṇḍana on  83–6 in Mīmāṃsā  41–5 phoneme  11, 40 Śaṅkara on  86 single sound  17 specified meaning, according to Mīmāṃsā  185 and sphota  89–94 synonym of akṣara  40 truly existing according to Mīmāṃsā  183 in Vyākaraṇa  29–30, 78 varṇasaṃvit phonemic consciousness, in Pratyābhijñā  143 varṇavādin  113, 115, 118–19 Mīmāṃsā authors known as  183

vāsanā, see saṃskāra Vasubandhu  109 Vātsyāyana (Pakṣilasvāmin)  56, 156 Veda authorless (apauruṣeya)  39 its definition  296–7 the source of knowledge to know dharma  354 uttered and listened to, rather than written and read  50 Vedānta  297–8, 314 lit. “end of the Veda”, religiousphilosophical movement  313 in relation to Mīmāṃsā  314 Veṅkaṭanātha  298, 307 verbal knowledge, see knowledge from words viśeṣa, see particulars; sāmānya; viśeṣa viśeṣaṇa process of qualification  357 Viśiṣṭādvaita  297 its view on signification  308–9 vivakṣā, see also tātparya desire to utter  115 distinct from intentionality  362 speaker’s intention  68, 167, 362 Vṛttikāra  42 vyabhicāra ambiguity  230 deviation of a word referring to many referents  170 Vyāḍi  380, 385 Vyākaraṇa biplanar system  30 history of the term  13–15 twin discipline with Śikṣā and Nirvacana  15 vyakti, see particular, individual vyañjanā manifestation  372 suggestion  404 vyāpāra, see śabdavyāpāra vyatiṣaṅga, see also syntax cross-connection  385 vyavahāra discursive practice  262 human communication and interaction  12 transactional agreement  124 vyutpatti, see language acquisition

Index Wittgenstein, Ludwig and correspondence view of language  295 and ostensive definition  262 word (pada) definition of  27, 35 n.29 history of the term  16–17 inflected word  26 meaning-bearer according to Patañjali  25 word meaning form or shape  157, 160, 162 hierarchical relation of word meanings  366 individual, universal, and form  156, 362 mental flash (pratibhā)  125 non related (ananvitārtha)  323 ontological status of, in Mīmāṃsā  184–5 reflection (pratibimba)  125 relational meaning  375

475 as sentence meaning, according to Jayanta  362 as sentence meaning, according to Prakāśātman  327 that which we cognize when understanding aptly used words  156 what is intended by the speaker  341

Yāska author of Nirukta  16 proponent of Nirvacana tradition  16 on semantic analysis  16–19 on synonymy  18 yogyatā  260, 343, 404–5 explained by Śālikanātha  264 fitness, one of the three criteria for sentence signification  194, 342 linked to bādha  404 semantic compatibility  260, 404 semantic in nature  195, 342

476



477

478