You are on page 1of 33

The Sound of Religion Author(s): Frits Staal Source: Numen, Vol. 33, Fasc. 1 (Jun., 1986), pp.

33-64 Published by: BRILL Stable URL: . Accessed: 01/12/2013 06:22
Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at .

JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range of content in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new forms of scholarship. For more information about JSTOR, please contact


BRILL is collaborating with JSTOR to digitize, preserve and extend access to Numen.

This content downloaded from on Sun, 1 Dec 2013 06:22:46 AM All use subject to JSTOR Terms and Conditions

Vol.XXXIII, Fasc. 1 Numen,




I Methodology of theScience of Ritual This paper* discusses concepts and methodology that are required forthe analysis of a ritualsong. In the past, ritual songs have been analysed by scholars of religion (e.g., van der Leeuw on Bach's HoheMesse) and by anthropologists (e.g., Raymond Firthin I In as far as can neither see, type of analysis has been Tikopia). successful; they have neither explained basic questions (e.g., why are such songs sung?) nor thrown any light on specificdetails. I found that different concepts and methodshave to be used in order to reach more satisfactory and promising results. These concepts and methods are familiar in other domains of inquiry, are fairly clear and intelligible,and will undoubtedlybe adopted by scholars of religion, anthropologists, and others,soner or later. In the pages I that follow shall tryto show that their time has come. At the outset, my approach to the study of ritual may appear to be closer to anthropological approaches than to those that have been adopted by scholars of religion; but it is differentfrom behavioristic, functionalistand structuralistanthropologies and fromthe "symbolic anthropology" thatis inspiredby hermeneutics and that is now America's favorite.Since thisarea of researchtends to get entangled in conceptual confusions, I shall startwith some remarks on methodology, specifically on the methodology of science. Few anthropologists and veryfewstudentsof religionhave recognized the relevance of the methodologyof science forthe conceptual clarificationsthat are so badly needed. An exception is Milton Singer, and his analysis will provide us with a good point of departure. The methodologyof science is generallyregarded as a shady discipline, on a par with all those apparentlyendless discussions on

This content downloaded from on Sun, 1 Dec 2013 06:22:46 AM All use subject to JSTOR Terms and Conditions



method, theoryand foundationsthat are so often-and with good reason-frowned upon by fieldworkers, and further suspect because of its original links with positivism,forthis discipline was brought to Chicago by philosophers who had been closely related to the Vienna Circle. Rudolf Carnap, for example, came from Vienna and Prague to Chicago, where he taughtfrom1935 till 1952 and where he edited, togetherwith Otto Neurath and Charles W. Science.At Chicago, Morris, the International Encyclopedia of Unified semantics was accordinglydiscussed and developed a quarter of a centurybeforethe emergence of what Edmund Leach has called the "Chicago dogma," namely, that "cultures are systemsof symbols and meanings" (Leach 1985, 156). Singer's methodologyof science is not the same as that of the Chicago philosopherswith whom he studied, nor is it of the same variety as the one to which I myself was exposed by my teacher Evert Beth in Amsterdam; but the similaritiesare sufficient forme to feel at home when I am reading or listening to him. Moreover, they remind me not only of Amsterdam but also of Madras-the cityin which both Singer and I did fieldwork (albeit of a different type) and where we both derived benefitfromthe guidance and vast erudition of the same Indian scholar, the late ProfessorV. Raghavan. The XVth International Congress of the InternationalAssociation forthe Historyof Religions (IAHR) takes me back to the same two cities. The IAHR was founded when a series of international congresses forthe historyof religions,which had been interrupted by the Second World War, was resumed in 1950 at Amsterdam. The 1950 congress was devoted primarily to a discussion of the "mythical-ritualpattern in civilization" (cf. Bleeker, Drewes and Hidding 1951). It was there that, as a young student of mathematicsand logic, I remained unimpressedby most of the selfstyled phenomologists (e.g., Mircea Eliade, E. O. James, Karl Kereflyi,H. W. Schneider) but was inspired by Gerardus van der Leeuw (then Congress President), A. D. Nock, Raffaele Pettazzoni, T. M. P. Mahadevan (whom I later followedto Madras), and especially Henri-Charles Puech and Louis Massignon. Subsequent decades witnessed the gradual replacement of that galaxy of scholars of religion by methodologistsof a kind different fromthe Vienna or Chicago variety,who emerged not fromdepart-

This content downloaded from on Sun, 1 Dec 2013 06:22:46 AM All use subject to JSTOR Terms and Conditions

mentsof science. it becomes relevant to turn once more to anthropologyand to Milton Singer. 1 Dec 2013 06:22:46 AM All use subject to JSTOR Terms and Conditions . G6del and Tarski. chants. philosophy. and semiotics. foralthough this fieldof studies This content downloaded from 134." "semantics.the sound that of the science of religion. that denied to the study of religion the status of a science on the basis of the belief that religion cannot be studied objectively-which implies that it cannot by studied at all. 155). Singer also uses the distinctionbetween and semantics. with its stock-in-tradeof persuasions and organizations. But is also requires a theory and an adequate methodologyforits analysis. a science of meaning (including sense and reference) traditionallystudied and developed by logicans such as Aristotle. I shall therefore begin with a fewremarkson pragmatics. but I do pay attentionto what I shall referto by the general termsemantics." In the above-mentionedreview. 48-52).30 on Sun.philology.58. began to pervade and even drown out the science of religion(an apand as a rough appellation I shall continue to use optimistically or the French proximation of the German Religionswissenschaft This in hand with went hand sciences development easily religieuses).I have not been able to make much use of these two disciplines insofar as I understand them. a pseudo-methodology inspired by continental philosophy. Leach statesthathe regardsthis splitinto three components as "unhelpful to say the least" (1985.the sound one one of the objects constitutes hears in the call of clarions and gongs.The Soundof Religion 35 or Oriental studies.253. or rather." and "pragmatics. especially as Singer has described it (1984. And so I decided to address myselfto what I regard as the realsound of religion. and since the science of religion has provided neither. This sound is akin to the music of the spheres about which Porphyrysaid that it cannot be heard by those whose minds are small (Life ofPythagoras. a strong sound of religion itself. Chapter 30). recitationsand ritual songs. disciplines Singer distinguishesbetween semiology which he associates withthe linguistFerdinand de Saussure and the philosopher Charles Peirce. he splits semiotics into three comsyntax ponents: "syntactics. respectively. Perhaps this is partlyor mainly due to the addition of pragmatics. a component that plays an importantpart in semiotics.In place of the search for objective truthand the concomitant emphasis on methodology.history but fromdepartmentsof religion oftenaffiliated with religious institutions. Frege. 5-6.

however.logic and linguistics. This content downloaded from 134.30 on Sun.253. and their users or contexts of use). D. pragmatics died in Jerusalem (Staal 1971). the status of pragmatics as a science has so far remained undetermined. 1 Dec 2013 06:22:46 AM All use subject to JSTOR Terms and Conditions .at least to anthropology. 19 ff." or "yesterday. given that most scholars of religion simply ignore it and that a respected anthropologistsuch as Sir Edmund Leach appears to regard it as unhelpful. Stalnaker and especially Richard Montague (1968. Pragmatic notions were currentin classical antiquity as well as in ancient India. M. is certainlynot true of syntax and semantics.). how to determine the truthvalue of sentences that contain "indexical terms. (Pragmatics is concerned withthe relationsbetween expressions. These two disciplines representbasically different approaches in mathematics. I myselfrelated pragmatics to J. Kamp. and semantics with the relations between such expressions and theirmeanings. Martin. Following Morris and Montague. that ritual songs arelanguageand I shall indeed argue that they are not." or "token-reflexives" (in the words of Peirce. it may be argued that. their meanings. in fact. At the same time.We shall see that the distinctionbetween syntax and semantics is as fundamentalto the analysis of ritual song as it is to the analysis of language. and Hans Reichenbach. D. we may distinguish theirprimaryconcerns: syntax syntaxand semanticsby contrasting is concerned with relations between (logical or linguistic) expressions." "egocentric particulars.36 FritsStaal has never turned into a broadly recognized discipline. followingwhich the fieldwas developed as a formalsystemby logicians such as H.however." Carnap initially considered pragmatics an empirical field of investigation(Carnap 1936-1937) but subsequently dealt with its theoretical foundations (Carnap 1947). Scott. Lewis. it addresses real problems: in particular. Bertrand Russell. from which it does not follow. they are undoubtedly here to stay. R.58. may not be superfluousto say a few words about it (for a fuller explanation see Staal 1984a. The distinction between syntax and semantics may seem so elementary as to require no furtherexplanation. respectively) such as "you. What may be true of pragmatics. Austin's "performatives" and attempted a linguistic analysis of one of its characteristicfeatures(Staal 1970a)." "here. L. Yet. 1970: reprinted in Thomason 1974). outside the domain of formallogic.

e.The Soundof Religion Let us consider a general symbolic expression. and This content downloaded from 134. it needs phonology tax and semantics.e. Whether linguisticsalso stands in need of annotionsis open to question. If we analyze the same expression in syntactic terms. Thus L6vi-Strauss has been inspiredby the phonology ofJakobson. pragmatics have these two sciences.Linguistics is also concerned withthe propertiesof the symbolsthemselves. a. For anthropology. movements(three stepsup a staircase and thendown again). For of the letter the example. formingpart consisting regard (1) MNABCBAMN or of: MNABCBANM (3) (2) or as mirrorimage of itself. semantics and.we interpret thing or referringto something. but also to logic and linguistics. otherdisciplinethatdeals withpragmatic as we have already seen..58. on the and study other hand. perhaps. etc. (1) 37 If we analyze this in semantic it as meaning someterms. we disregard meanings and interpretations. the status of pragmatics as a discipline is thereforenot on a par with phonology. i.etc. words ("found sleepy on sleepy found"). The distinction between syntactic and semantic methods of analysis is basic to all systemsthat are formalor that use symbols in a systematic fashion. syntax and semantics. d. only configurations of five of: as units. chiefly to mathematics and the mathematical sciences. syntax. 1 Dec 2013 06:22:46 AM All use subject to JSTOR Terms and Conditions . we symbols.. numbers ("3 8 7 8 3").and in addition to synas these are primarilysounds.the general question naturally arises whether subdisciplines of logic and linguistics such as phonology.the answer anythingto offer is clear: it has been repeatedly inspired by developments in linguistics. If it is true thatsyntaxand semanticsare required forthe analysis of ritual songs that have so far been primarily studied in anthropologyand the science of religion. for example: ABCBA.a.30 on Sun. tones ("b flat.g.253.and these waves of inspirationhave flowedalong channels cut by at least some of these subdisciplines. b flat"). Singer by Peirce's semantics (a term I shall continue to use in a general sense.

Mauss had been introduced to these texts by Sylvain Levi who taught during de France. Oddly enough. at least according to most linguists. Just is closely dian mythology. Staal 1984a." Durkheim (1915." Hubert and Mauss offered a "scheme abstrait du sacrifice" which is primarilysyntacticin nature: forexample.38 FritsStaal without making the distinctionbetween semiotics and semiology) and Stanley Tambiah by the pragmatics of Austin and others (in his performative analysis of ritual: Tambiah 1979).but in the present context it should sufficeto state that sutras the syntactic supplementsand is analysis of ritual in the frauta in the to some extent supplemented by its semantic interpretation as the stand source of much of Inat the Brdhmanas Brdhmanas. 538) a course on Vedic ritual which was subsequently revised and published in 1898 as "La doctrinedu sacrificedans les Brihmanas. 386) concluded from Hubert and Mauss. that a rite can serve different ends.58.253. logicians and mathematicians. Why are we justifiedin characterizingthe analysis of Vedic ritual as syntactic? sutras To provide an adequate answer given in the i'rauta to this question would necessitate a lengthy excursus into Vedic literature.C. 3-8).specificallyfor Mauss (Mauss 1896-1897 at the Collhge 1969. what alone appears to be missing in thislist of examples is syntaxthe most basic of these disciplines. theymade. the elementary but fundamental syntacticobservation that rites have a beginning. a middle and an end.which is a pityforanthropology (cf. but he did not draw the further conclusion that ritual is thereforeto some extent independent of the ends it is supposed to serve. such as Schwab's 1886 monograph on the Vedic animal sacrificeentitled "Das altindische Thieropfer. Accordingly.30 on Sun.the syntacticanalysis of the drauta sutras This content downloaded from 134. III.a body of Sanskrit texts composed from approximatelythe eighthto the fifth centuryB. as well as from other anthropological data. Their analysis is to a large extent derived from the syntacticanalysis of Vedic ritual found in the irautasutras. 1 Dec 2013 06:22:46 AM All use subject to JSTOR Terms and Conditions .he did not pay much attentionto syntax. but in order to discern it we must look more closely. Syntax is not really lacking in anthropology. In theirfamous articleof 1899 entitled "Essai sur la nature et la fonctiondu sacrifice." Hubert and Mauss also referredin their article to published sources based upon the Vedic srauta sitras. forthe first time.

58.. In the present context I must referto what I have argued elsewhere. What.and in termsof the four subdisciplinesto which we have referred. inspiredby the Indian European phonologyhas been profoundly of features" "distinctive and Jakobson's concept grammarians. it is ratherthe anthropological study of ritual. with Noam Chomsky. Tambiah Singer In most of these relationships. Peirce.30 on Sun. Pa-nini. it is not really anthropology as a whole that has undergone the influence of syntax and pragmatics. the case is different. Jakobson phonology Pratisakhya. as follows: Anthropology Le'vi-Strauss Panini. Firstof all.The Soundof Religion 39 related to the origins of grammar. Chomsky syntax Linguistics pragmatics semantics Frege. least in India. and the consequences have a form been baffling: forLevi-Strauss introducedinto anthropology of analysis thatstressedbinaryoppositions and thatwas based upon This content downloaded from 134.253. the only exception is the ritual science of the drauta which I have listed in the table under "anthropology" for stitras. that uniquely Indian contribution that also incorporated the phonology and morphologyof the Vedic prdtiidkhya literature and that culminated in the Sanskrit grammar of Panini. lack of a betterheading and which influencedPaininiand the other Indian grammarians (Renou 1941-1942.we should properly include these ancient Indian forerunners. that (1) the ancient Indians possessed a science of ritualwhich used primarilysyntacticmethods. is the relation between these various disciplinesand the science of religion? Before we address this question we must make several observations. still bears the stamp of their influence. with Zellig Harris and more definitively exerted we the influence if had to Thus. In the case of phonology. and that (2) linguisticsoriginated. in close association with this syntacticanalysis of ritual (Staal 1982).however. Saussure Grice Austin. then. Staal 1982). seems to have entered Western linguistics much later. The study of syntax. & Mauss Hubert Srauta-suitras. 1 Dec 2013 06:22:46 AM All use subject to JSTOR Terms and Conditions .the influencehas been in the direction from linguistics (including logic and philosophy) to anthropology. by linguistics(and picture by logic and philosophyvia linguistics)on anthropologyin tabular form.

It will be the same with the Language are now firmly Science of Religion" (in: Waardenburg 1973. one might expect that the latter science should have been inspired by linguistics or logic as well. I. 86). why should twosuffice for man when so many other natural numbers exist and are found elsewhere in nature? If L6vi-Strauss had studied linguistics later. as a "systime oiCtout se tient.Moreover. by the combined and well directed efforts of many scholars.58. and the principles that must guide the student of the Science of established. instead of paying attentionto binary opposition. great resultshave been obtained. Levi-Strauss.40 FritsStaal distinctive features. 1 Dec 2013 06:22:46 AM All use subject to JSTOR Terms and Conditions . and have no place in syntaxor semantics. or Chomsky developed syntax earlier. Max Milller predicted: "It was supposed at one time that a comparative analysis of the languages of mankind must transcend the powers of man: and yet. What happened. mighthave introducedthe methods thathad been adumbrated by Hubert and Mauss syntactic in their study of ritual. in fact. Since ritual is one of the main areas of research common to anthropologyand the science of religion. though all linguists know that distinctive features had been postulated successfullyonly in phonology. in otherwords.hermeneuticsand other warring factions. but when the "Science of Language" made a methodological transition from comparative and diachronic philology to a synchronicscience of linguistics. In fact." Thus the science of religion failed to make the transitionwhich would have turned This content downloaded from 134.the "Science of Religion" lagged behind. It was leftto anthropologists such as Mauss to studyreligious like in the in which de Saussure had ritual manner phenomena studied language. introduced into anthropologya linguisticmethod which he regarded as a universal panacea. but that were not fully developed in linguisticsuntil Chomsky.30 on Sun. was that the comparative study of religions developed as Milller had predicted. L6vi-Strauss. it fell prey to phenomology.253. while in fact it is a technique that has been found applicable only in a subdiscipline of linguistics. existentialism. One and reason that it is a prioriunlikely that a might even go further method that can be used in phonology but not in syntaxor semantics could be usefulin anthropology. As early as 1867. Instead of developing a syntactic method of analysis appropriate to its object.the scene forsuch an influencehad long been set.

58. scholars of religionhave expressed an awareness of the need forsuch a transition.but this does not mean that specificpieces of linguistic analysis have been of great use to anthropology. Penner statesthat "the studyof linguisticsis the necessaryfoundation for explanations in religion" and that "language as we all know is composed of signs. and b.Of are postulated and never correimportanceis that all such structures to actual facts such as rites. by my own admission. and all linguistic signs have phonological.just as the laws of physicsdo not correspondto specific events that take place in my garden. structuresfound in natural languages or with which linguists are familiar (Staal 1979a.In my viz. these pieces of analysis "do not correspond to any existing ritual" or "to any actual ritual. or otherspecificphonological or syntactic tures.froma respectable branch of scholarship into a contemporaryscientificdiscipline as well. Rather. to featuresthat are abstracted fromactually existingrites. the most basic issue is not whetherritual or anthropological structuresare similar to linguistic structures. like linguistics. Belatedly.253.It would be correct to say that anthropologyhas been influencedby linguistics. L6vi-Strauss may have been wrong with regard to the specific strucbinary oppositions. Yet." He supportsthis by emphasizing that. in a recentreaction to my article "The Meaninglessness of Ritual" (Staal 1979b). Penner asserts that my syntacticanalysis of ritual is "irrelevant." Disentangling some of the misconceptionsthat are at the root of these assertions should help us to understand more precisely the relationshipsbetween the methodologyof science and the science of ritual-or anyempirical science. in the same article. however. 1 Dec 2013 06:22:46 AM All use subject to JSTOR Terms and Conditions . that some study of Vedic ritual I found somethingdifferent. theycorrespond. Hans H.The Soundof Religion 41 it..30 on Sun.For example. spond exactly if theyare adequate. but he is rightwith respect to method. This content downloaded from 134. that he postulated. 1984b). For example. forthatmatter. syntacticand semantic components" (Penner 1985). From the methodological standpoint. but to featuresabstracted from such events. some of L6vi-Strauss' work serves to warn us that an uncritical acceptance of distinctive featuresis articifialand unproductive within anthropology. ritual structurescan be generated by rules that are similar to the while othersare unlike any rules thatlinguistscall transformations.

viz. ensemble.42 FritsStaal What is basic to his method.. in the final analysis. a French philosopher by training. the only persons who have failed to understand it are certain types of such as behaviorists. Chapter 4). and most scholars of religion. 1 Dec 2013 06:22:46 AM All use subject to JSTOR Terms and Conditions . underlying structures that are postulated in order to account for empirical facts and events are necessarily mathematical or logical in character. What anthropologyhas taken fromlinguisticsis therefore. These general considerationsat the same timehelp to explain the importanceof logic and mathematics. and to all scientificmethod. that reality is differentfromwhat it appears to be (L6vi-Strauss 1968.These latterdisciplinesstudy in general. a rule-governed it is an activitythat is governed by rules. 61-62). to account for what is empirically given and visible. L6vi-Strauss has in fact told us where he came upon this idea. philosophers bienitonn'sde se trouver of and ordinary hermeneutics phenomenologists. Chapter 9): he took from geology. invisible-but that enable us. in fact govern it. Tambiah and L6vi-Strauss (whatevertheirdifferences). fail to appreciate and understand. that empiricists such as Edmund Leach. This is the common insight and methodology shared by anthropologistssuch as but Singer. Staal 1975. provided they are adequate." meanings attempted to demonstrate in my earlier studies of ritual is not that ritual is a Since activity. language. I discovered that ritual structures This content downloaded from 134. practitioners language philosophers such as Wittgenstein (cf.and which remained the main insightthat separated him fromexistentialists such as Sartre (see Sartre 1960 and L6vi-Strauss 1962. from Marx and from Freud the idea that is basic to all science. which was novel and startlingto him. like indeed it is to most ordinarypeople. is that structuresare postulated entitiesthat are not empirically giventhat are. includingthe most general structures structures thatcan be imagined. but that it is. In pursuing such questions. in fact.58. it becomes importantto discover what actual rules.253. This is obvious to scientists.30 on Sun. and what kind of rules. We are now in a position to see that the influenceof linguistics on anthropologyand on the science of religion does not necessarily or even implythat the objects of the lattertwo sciences are languages I of and What have "systems symbols. The abstract. nothing peculiarly linguistic but rather an insight into the characteristicfeaturesof science.

30 on Sun." "the" and "of" in a I agree that "it would natural language such as English do not refer. does not appear to be such a system. 1983a..58. nor anthropology. 16." But although the referenceis the same. 187. be odd to assert that they are meaningless because they do not refer. especially II.but by mathematicaland logical methods thathave also influenced linguistics (e. take into account the Fregean distinction and also failed to appreciate that "and. 1 Dec 2013 06:22:46 AM All use subject to JSTOR Terms and Conditions . nor the science of religion have performed such a task or are presentlyequipped to do it. As forthe distinctionbetween sense and reference. in the realm of ritual that funcIt is not clear that there is anything terms such as tions like names. verbs.253. Penner claims that I failed to between sense and reference. 127-134. Our conclusion should be rather different: to study ritual adequately we need a science of ritual akin to the Indian science of the srautasutras. that rarelycorrespondwith each other and that are accordingly distinguishedfromeach other by the relevant experts (e." with the distinctionbetween "analytic" and "synthetic" to which Penner tried to relate it (contrast Staal 1966). The name "Agnistoma".at least insofar as Vedic mythologyis concerned.for example.The SoundofReligion 43 can be analysed in syntactic terms not by methods specific to linguistics.Vedic ritual offers examples that are as clear as Frege's own.In developing such a science. for example. or syncategorematic or This has little to do. but neither linguistics. I triedto show that a syntactic analysis of ritual is not only possible but also fruitful. Vedic ritual and Vedic mythology are in fact remarkably separate developments. "the." But what I did is different." "of. methodology may prove helpful. 1986a). incidentally. 29.g. refersto the same ritual as does the name "Jyotistoma. Renou 1953. and in subsequent studies. but inaccurate and misleading. Dandekar 1982. I cannot say whetherothertraditionalobject areas of anthropological or religious study are also rule-governed systems. 77). Tsuji 1952.g. to which I shall returnin the Conclusion. It may be helpfulat thispoint to clear up one more confusionthat is also met with in Penner's article. Whether they will in turn benefitfromsuch a science of ritual remains an open question. Staal 1979a. the This content downloaded from 134..In that 1979 article. but mythology.and that such an analysis demonstratesthatthe assumption thatritualsexpress meanings like language is not only unnecessary. very "and". nouns.

1. It is the first verse of a hymn (RV 9." have differentsenses but the same reference.2 II Most delicious and inebriating In order to test some of the ideas discussed in the previous section. 1 Dec 2013 06:22:46 AM All use subject to JSTOR Terms and Conditions . and vice versa.toma a Soma celebration one with ceremony. Or rather.1) that belongs to a group of hymns all written in the same meter. that may or may not be derived froman expression of languagelike the verse of the Rigveda in the cases we shall be studying.253. These ritual songs belong to the category of ritual sound that in "morning star" and "evening star. Mantras and the ritual acts themselvescorrespond to each other. The text of this gayatrz-verse is: svidisthaya midisthaya pavasva soma dhiraya indraya p~tave sutaih(RV 9.44 FritsStaal senses are different: agnis. since Vedic A mantra is a ritual sound expression times.58.1).30 on Sun. because it is always assumed that they are. we shall now turn to the studyof a ritual song. five during the afternoonpressing. Whether the mantras themselvesshould be regarded as expressions of language is a topic that has never been discussed." jyotis. called gdyatrz. or first is sufficient to state that Agnistoma and Jyotistoma. The translationis straightforward and uncontroversial: (4) This content downloaded from 134. and two during the third pressing. the only book of the Rigveda that deals in its entiretywith Soma and its ritual purification and preparation.toma means "Praise of Agni" and means "Praise of Both refer to the same Light. has been called mantra. pressing day during which there are twelve chants and recitations: five during the morning pressing. In the present contextit chant. To studya ritual systemis to study a systemof mantras. We shall returnin the next section to the eleventh chant of the thirdpressing.we shall be concerned witha familyof ritualsongs derived froma verse of the Rigveda which extols the flow of Soma. The verse that is our point of departure is the firstof the ninth book.

In many cases."(re)sounding.The Soundof Religion "With most delicious and inebriatingflow. A few more rules have to be taken into account beforewe can properlypronounce this verse. I . and the higher. Pressed for Indra to drink." 45 meterconsistsof threeoctosyllabicverses.58.253. ." and is writtenin the manuscripts(that are all of later date) with a horizontal bar below the syllable. the distributionof longs and shorts is as follows: u-uu- --u- S-u. but we shall not troubleourselveswiththem. and the followingat a higherpitch. Texts are never recited in this manner in the ritual itself:to be fitforritual use.. purifyyourself." writtenwith a vertical bar above it. Soma." and it is possible that it was originally spoken with raised pitch (but see Gray 1959a and b for a dissentingview). in which long The gayatri (-) and short (u) syllables are generally distributedas follows: -u- u- (5) In the case of the above verse. pupil purpose son. The result is as follows: svadisthaya madisthaya . the traditionalrecitationshave to undergo certain modifications. verse have to be turned into songs which are then incorporated in the Veda of Songs. "not raised. I pavasva soma dharaya indraya patave sutah (7) This mode of recitation. 1 Dec 2013 06:22:46 AM All use subject to JSTOR Terms and Conditions . The lower accent is called anuddtta.which is called svddhydya. svarita.- (6) The accent ""' which has been printedover certain syllablesof the text in (4) is called udatta."raised. or This content downloaded from was later marked by recitingthe previous syllable at a lower pitch.30 on Sun. only serves the from fatherto or to from teacher of transmission. Whatever its original pronunciation.

thus provided with a patternof song that immediatelyderives fromthe accentuation of the original. Before we can sing (9).the first the highest." from rk "verse" from which the appellation rgveda.46 FritsStaal Samaveda. esoteric aranzyageyagdna (or aranzyegeyagdna). The basic chants of the Samaveda. the first step of that process of incorThe are is svaritaand anuddtta uddtta.58. we have to know at which pitch the syllables not marked with a numeral have to be sung. This is in accordance with the chronologyproposed by Kiparsky (1982. Lecture II) forthe development of the Vedic accent system. It consists of the more common "songs to be sung in the village."list of verses. The rule is again very simple: these are sung at the pitch of the preceding syllable. poration extremelysimple. are listed in the firstpart of the Samaveda. "songs to be sung in the 1 2 3 1 2 3 2 1 2 3 12 12 3 1 2 3 (8) (9) This content downloaded from 134." is also derived.30 on Sun.253. which shows that the Samaveda is earlier and closer to the original Rigveda than the Rigveda systemof horizontal and vertical strokesused in the manuscripts. and the svaritaat an intermediate pitch. The next portion of the Samaveda lists the melodies to which these verses are sung. an interval a chanted at threedifferent each at of about secpitches. These numerals are writtenabove the syllables. ond fromeach other. "Veda of Verse." and the more grdmageyagdna. as arcikd. 1 Dec 2013 06:22:46 AM All use subject to JSTOR Terms and Conditions . The manuscripts referto these tones with the help of numerals: "1" for uddtta "2" for svarita "3" for anuddtta. which is to as Samhita--as in the case of the otherVedas-and also. In the above case.the last (anuddtta) (uddtta) the lowest. referred more appropriately. as follows: svadisthaya madisthaya pavasva soma dharaya indraya patave sutah (9) is closer to (4) than it is to (7). the Rigveda as it was known provided we insertthe Samaveda after to Panini. but before the Rigveda recension with which we are familiar.

Karnataka and Maharashtra). the melody needs to be listed only once.30 on Sun.In these lattertexts. which is stillfound in many parts of India (chieflyTamilnad. its manuscripts are rare and have. again This content downloaded from 134. there are in the Rigveda no variant formsor "readings" (a term based upon the Western assumption that these compositions are "texts" which are "read"). 88) as against two in the case of the Rigveda and Yajurveda each.a simple notation forthe pitch of the tones of the songs has been adopted in the manuscripts. In that case. the musical renderings vary greatly. Yet.. When one hears contemporarytraditionsof the Samaveda.58.253. remained unpublished.The Soundof Religion 47 forest.7 which designate pitches in their descending order.but more oftenthisformundergoes further modificationforthe sake of "ritual application" (prayoga). In the case of the transmissionof melodies. and this has led to some confusion among studentsof the Samaveda. There are numerous melodies in these gana collections because one verse is generallysung to different melodies. with a few exceptions.. most of which are of recent date (i. they are sometimes ritually used in theirgdnaform. But we also meet with the opposite: different verses are sung to the same melody. is stillrelativelystrong. no more than a fewcenturiesold). The Jaiminlya is almost extinct." Both these gana-books correspond to the svddhyaya of the other Vedas: they mainly serve the purpose of transmitting the traditionto the followinggenerations. and the Jaiminlya which is confined to Kerala and a few villages in Tamilnad. The printedtextshave adopted thisnotation in which the notes are referred to withthe help of the numerals 1.. which I shall referto as K-R.and its texts have been published in theirentirety. despite the vicissitudesof Indian history during the last three thousand years. The Kauthuma-Rana-yaniya tradition of two closely connected schools.. 1 Dec 2013 06:22:46 AM All use subject to JSTOR Terms and Conditions . The Vedic traditionis celebrated forthe care with which it has been handed down. two traditionscan be clearly distinguished:the Kauthuma-Ranaiyanlya. greatervariation is expected. This applies to the place of the accents as well: it never varies. and this may well account for the belief that there were originally a thousand schools in the Samaveda (Renou 1947.. This apis the actual of these and of the Samaveda purpose plication songs itself.e. Andhra.

There are seven notes in the Samaveda according to these manuscriptnotations. Three ofthe four Samaveda priests participate in such a chant. The five portions. and Pratiharta.they also induce lengthening of the preceding syllable. (2) udgftha ("chant"). The lattermodificationis well known This content downloaded from 134. The assignmentof bhakti is as follows: 1) prastavaby Prastota. 2) udgftha by Pratiharta. a few more conventions must be explained: "r" denotes lengthening. viz. (3) pratihdra upadrava ("accessory") and (5) nidhana ("finale").58. 1 Dec 2013 06:22:46 AM All use subject to JSTOR Terms and Conditions . In the gramageyagana of the K-R school. is the following: svadaisthaya / madaistha-ya// pavasvaso / madha'1 ra'23 ya // Ir 2 1 2 1 2 r 1 2 2 ta'343h / o'2345 i // // indrayapa//tavaisui'23 dia (10) In order to know how this actually sounds. When we compare (10) with the original verse (4). The firstof these. many of the short a's are lengthened into long d's. are chanted by different directions.30 on Sun. (4) ("prelude"). portionsto them Udgata. but the formersubdivision into five is constant in an importantclass of rituals: these five portions. are called: (1) prastdva ("response"). referredto as "GG 468. For example. The number of theselatter. provided with nine melodies to which it is sung. The song consistsoffiveportionsthat are separated by double bars ("//").1".48 FritsStaal with intervals of roughly a second. and i and e have become di. furthersubdivided into smaller portionsseparated by singlebars ("/").called are of fundamental importance in the Soma rituals. each lasting one beat or time unit (mdtra). 608-609).sitpriestsfacingdifferent they a ting in a particular fashion and following pattern that is always the same (see AGNI I. smaller subdivisions varies. we observe that certain syllables have been expanded.253. which should each be sung with a single breath. where bhakti. and 5) nidhanaby all three.. that of the Naradasiksa).g. or otherwisemodified. by Udgata. 3) pratihara 4) upadravaby Udgata.which is consistent with some of the traditional accounts (e.the numerals in the line sound the same as those writtenabove the syllables.. Prastota. our verse is listed as item it is and 468.

They are meaningless and are often similar to the later mantras and dharanis of Tantrism (whether "Hindu" or "Buddhist").N. which would be intelligibleonly the specialists. At the same time. In order to study the ritual significanceof our song we need to be familiarwith the structure of the entire systemof its derivatives in the corpus of the Samaveda. which makes use of grammaticaltechnicaltermssuch as vrddhi and a greatmany others. in particular the terminationof the upadrava and all of the nidhana: o'2345 i //da // Such syllablesare called stobha. We finallyobserve that in (10) new syllables have been added." This illustratesin passing that the derivation of the ritual chants of the Samaveda from the Rigveda (and not only the formationof the Padapatha fromthe Rigveda. by the formal nature of the syllables of the original.or partly determined. in particular by theirlength. 1 Dec 2013 06:22:46 AM All use subject to JSTOR Terms and Conditions . These relationships are dealt with in texts such as the (K-R) Puspastitra.253. It stands to reason that these syllables were originallyadded in order to complete or fillthe gaps in a preexistingmelody. the notion of vrddhi the cornerstonesof Indo-European comparative phonology. Jha has shown) has contributedto the origin of the Sanskrit grammatical tradition.58.30 on Sun. as V. as well as the basis of our notion of "sound law. I have made use of recordingsand of a copy of a manuscript in the Malayalam script prepared by my collaborator Itti Ravi Nam- This content downloaded from 134. Since these songs have never before been published. Instead of pursuing this by numerical referencesto the published textsof the K-R Samaveda. the structureof the song is related to the original meterwhich is also based upon the therefore distinction between short and long syllables. Song (10) is also called a gayatri song-a circumstance to which we shall return. it will be more interestingand worthwhileto publish songs from the corpus of the Jaiminiya. but thisassumption can only be testedby studying a larger number of melodies in association with the textual sources to which they are set. These phonetic or sound modificationsare oftendetermined.TheSound ofReligion 49 to linguists: it is the famous vrddhi ("lengthening") discovered by the Sanskrit grammarians that is referredto in the firstsuitraof is one of Panini's grammar.

50 Frits Staal budiri. C. The original manuscriptis in the possession of Asko Parpola (University of Helsinki). Burnell in the 1870's in the Tamil region and both writtenin the Grantha script. Bake.the method of chanis portions different ting has to be further specified.2. Bake at the School of Oriental and African Studies. Here followthe nine songs based upon the gayatriverse fromthe Gramageyagana: Jaiminlya Gramageyagana 49.30 on Sun. A photograph of Caland's manuscriptwas in the possession of Dr. If the number of bhakti fromfive. 1 Dec 2013 06:22:46 AM All use subject to JSTOR Terms and Conditions . A. and therefore must be determinedfromoral tradition. The original is in the UniversityLibrary of Utrecht. I have mentioned the most importantvariant "readings" (referred to as "C") that occur in a handwrittencopy completed by Willem Caland in 1906. Two other manuscriptsof the Jaiminiya songs are known: they are B 497 and B 61-62 of the India Office Library.253.1-9 (1) svadayisthaya madayisthaya / pavasva somadharaya / in/ oyila // drayapa / tavayi sfitaih (2) svadisthaya iya iya madisthaya / pavasva so iya iya madharaya / indra-yapaiya iya / tavayi suitah/ oyi!a // (3) svadisthayau ho va iya madisthaya / pavasva sau ho va iya madharaya / indraya pau ho va iya / tavayi stitah/ oyila // C: poho (4) oyi svadi / sthaya madisthaya vuvova / pavasva soma dharaya o indra / ya pa au ho va / tave suitah// (5) uhuvayi svadi / sthaya madisthaya au ho va / pavasva soma dharaya uhuva indrayapa / tava au ho va / sfitah// (6) svadayisthaya madisthaya pavasva somadharaya ayindraya pa ha bu / tava yi suta bu / vaI// This content downloaded from 134. which was based upon B 497 and collated with B 61 and B 62.58. As the reader will see. Caland did not note the stobhas or the musical syllable notation given in B 62). A. (For the Aranyageyagana.This is easy to do because of the rule that one bhakti has to be chanted with one breath. There are still instances where these remain unclear in the manuscript tradition. both discovered by A. and was given to me by Mrs. some of the variantshelp to establishthe boundaries between the bhakti portions.

30 on Sun. or that it originated in an area of the North at a time when a Dravidian language was still spoken there.253. The third observation was made in AGNI. Firstof all. Staal 1961." because these scripts. "o" where theJaiminiyahas "a". froma textual This content downloaded from 134.The Soundof Religion 51 C: madisthaya / somadharaya / (7) svadisthaya madisthaya pavasva somadharaya indrayapa / tava fi tava u tava vu va au ho va / sutah // C: somadharaya / (8) au ho hm bha e hiya ha hayi svadayisthaya madi o yi madi / au ho hm bha e hiya ha hayi sthayapavasva so o sva so / au ho hm bha e hiya ha hayi ma dharaya indra o indra / au ho hm bha e hiya ha hayi ya pa tavayisuta oyi suitah / au ho hm bha e hiya ha ha au ho va / i // (9) svadisthayamadayisthaya/ pavasva soma dharaya / a yindraya ha vu va / suitah// patava A fewobservationsmay be made. are due to the Vedic school. fromwhich I quote (I. theJaiminiya does not use a numerical notation to referto the pitch of the tonal pattern which characterizes the melodies. The occurrenceof this sound may indicate thatthe Jaiminlya traditionoriginatedin South India. ferencesof affiliation however. 99. and the well known K-R stobha "o id1a" appears in theJaiminlyatradition always as "o ila. Thus minor phonetic differences may be determined not by difbut by geography(Tamilnad versus Kerala). It is not marked in the Malayalam manuscript." This is not due to differences between the Malayalam and Grantha both of can express "ai" as well as "ayi. 1 Dec 2013 06:22:46 AM All use subject to JSTOR Terms and Conditions . Secondly we can observe that the similarities between the Jaiminiya and the K-R version (above (10)) are phonetic: K-R oftenhas "ai" where the Malayalam manuscriptof theJaiminiya version has "ayi". For Important differences. but further back (see Renou 1947.58. pronounced like r in American English." using the Dravidian sound "1". 278): "When dealing with these songs and chants it should be rememberedthatthe Samaveda is repletewithwhat. There exists a Jaiminlya musical notation by means of syllables fromthe Grantha script. the has often K-R example.of which we noted that Caland copied it fromB 497 but not fromB 61-62. although Caland's manuscript also has "di. 69).

One is tempted to believe that thereis no answer to the question of why thisparticular song was selectedforuse at thisparticularpoint of the ritual. and thereforelinguistic correspondence--or is it This content downloaded from 134. when it is regarded as ferocious(krura) and dreadful (ghora). and also because it can be easily approached fromdifferent sides (AGNI I. assisted by the Pratiprasthata. we find that the rules of this game often escape us. and with many the stobha "au ho va'.7.however.To treat the textas ifit were corruptwould be to miss its very raison d'etre.52 FritsStaal point of view. GG 49. Why was this song chosen to performthis function?It fitswith the othersonly to some extent:it shares withmost a triplerepetition (of "tava ii").2. It came fromthe collectionlike a seed that fallsfroma blossom and is carried throughthe wind until it settlesdown somewhere. These features are characteristic of the Samaveda. What about the stobha "tava Ii" itself?Again.5).the Udgata sings a sequence of 57 samans. popularity partly from (Taittiriya the factthat it was subsequently interpreted withinthe perspective of Saiva theism (Gonda 1980. 1 Dec 2013 06:22:46 AM All use subject to JSTOR Terms and Conditions . thereare new deviations thatbreak the pattern.30 on Sun. Many formsthat may seem to be in printingmistakes or mistakes of the manuscripts are therefore fact what they should be. But even if we accept its playfuldeviations. To this end the Adhvaryu.pours a continuous libation of goat milk over the westernbrickof the northern wing." Both these features. KauthumaRanayanlya as well as Jaiminiya. 509 ff. Afterthe bird altar has been fullyconsecrated. far away fromthe center of the altar which is centerof power.253. forthcoming)." The seventh of this sequence of nine songs is used in its Gramageyagana formin the Agnicayana ritual. The fourth of these is our togethercalled Flow of Milk (ksfradhdrd). it has to be brought under control and pacified. we observe a phonological. This brick is chosen because it is relativelysoftand tender. Whenever regular patternsseem to emerge. The more complex featuresthat distinguisha series of later songs in the Flow of Milk sequence (cf. Staal 1983b) are not foundin it. are unexpected variations and varieties. the chief priest of the Yajurveda.). the Adhvaryu recitesthe famousSatarudrjya or Rudram which derives its Samhita 4. are quite common in the Samaveda.58. Arnold. During thisoblation.During this oblation and recitation.

30 on Sun. in linguisticterms. "this one.58.The Soundof Religion 53 pseudo-linguistic? "Tava u" comes from "-tave. 99.2.This is quite common in these songs. tava tyad)." But thismeaning correspondenceis undoubtedly adventitious and withoutsignificance. This song is rituallyused in the constructionof the Agnicayana altar. When the bricks of the firstlayer are laid down (which is This content downloaded from 134. "tava" also has a meaning: it means "yours. for example. therefore. ed. 416). phdt phatphatphatphatphat: AGNI I.7 (corresponding to K-R AG 16. viz. 770 s."tavai" or "tavayi. "Tava." Jaiminlya Aranyageyagana 15." without carrying any deeper significance.v.. It is a well known Vedic stobha (see van der Hoogt 99: ayamvayau." which should yield.253. is perhaps related to the equally well-knownTantric stobha "aim" (see." The stobha can be taken to mean "yours U.and "u" is quite famous in later Tantrism (see." which is almost as good as "yours truly. 119). cf. 91: ayamayayam.which does not imply. however. ayamayamayamaya mayamayamauhova Again. 203-206). there is triple repetition. Bharati 1970. and it supportsthe idea that the derivationof songs fromverses is closely connected with the originationof linguistics in India. What we find.1) includes the followingsong based upon our verse: ayamayam ayamayam ayamayam svadisthaya madayisthaya / pavasva somadharaya / indrayapa tavayi suitah / ayamayam ayamayam ayamaya au ho va / i // C: ayamayamayamaya (2x)." as indeed it does in many other songs of the sequence GG 49. 1909. 11: ayam yah. that it could not evoke semantic associations among some users.111: ayam)and 1929. occurs in otherstobhas (see Puspasfitra.. Padoux 1963. for example. Next we shall consider some of the more esoteric "songs to be sung in the forest.and "ayam" also happens to mean something. Historical connexions between Vedic stobhas and Tantric bijamantras are likelyto exist (see Staal 1985a): a good example is the famous Tantric phat which occurred already in the Samaveda (see van der Hoogt a variation of a linguisticrelationship." moreover. 1 Dec 2013 06:22:46 AM All use subject to JSTOR Terms and Conditions . Simon. In Sanskrit.

Why this song is chanted here at this time is not known. There is no specific phonetic. which is the eleventh sequence of the entire pressing day. 1 Dec 2013 06:22:46 AM All use subject to JSTOR Terms and Conditions . These songs are listed in Uhagana 3. may we offerwith oblation.9. arise with a thousand! To you. notably one remarked 646-648). phonological.30 on Sun. apart fromthe fact that the recitationis again in the gayatrimeter and the chant is a gayatri chant-both varieties that are exceedingly common. semantic or pragmatic similaritybetween recitationand chant.2 c-d (see AGNI I. the songs are ordered in the same sequence in which they are used in the Soma rituals.58. 423): "Rising up fromevery stem. You who extend with a hundred.2. The ritual context of the Soma ceremonies has already been brieflymentioned in the firstsection. and the Udgata intones his song. derive) the Udgata chants occasionally. In these two collections. syntactic. at the third pressing. occurs in the correspondingpassages of the Samhita). diirva.7.withoutmelody. goddess brick.a hundredfold. to alert him. of which the firstand the fourthare based upon our verse.2-6 (see AGNI I. called cryate. No specific semantic connexion appears to exist between the ritual applications of these two derivativesfromour Rigvedic verse thatoccur in the two basic gana books of the Samaveda. the firstof these.etc. apparently only when it is the turn of a ritually significantbrick to be deposited and consecrated. extend to us a thousandfold. So how should they be sung? This content downloaded from 134. In the Agnistoma. fromeveryjoint. which is called Duirva afterthe grass of that name. AG 15. Other ritual uses of chants thatare derived fromthese songs pertainto the Soma ritual and are listed in the Uha and Uhya (or Rahasya) GCna collections of the Samaveda. At the consecration of the seventh brick. there are two sequences of chants and recitations." At the beginning of this recitation the Adhvaryu looks at the Udgata. caitya.54 FritsStaal from the root ci from which cayana. there are three pressings of Soma. but there are several difficulties. Caland 1907.253. citi. consists of a series of seventeen songs. upon by (Caland-Henry 180): the firstthree songs are notlisted there (though theirtextual form. and they are given in the actual formsin which they are sung. the Adhvaryu recitesTaittiriya Samhita 4.

This content downloaded from 134. "Savitrl"). This is a typicalcomment of that Brahmana-vacuous and ad hoc. 45." This mantra is also called Savitri afterthe god Savitr. It is often quoted in Vedic literature. 104). which is explained somewhere else. who is the "impeller" (prasavitr) of the gods. Gonda 1980a.v. who shall impel our thoughts.10) that is not listed in the Samaveda either.The SoundofReligion 55 The answer is that these are to be sung in the Gayatra melody. and is recited during many rites and on many occasions (see Gonda 1980b. s.58.10) is: tdtsavitdr vdrenzyam dhimahi bhdrgo devdsya yd nahpracoddydt.62.4. forexample. dhzyo (11) The distributionof long and short syllables in this verse is as follows: V--- - - - The translationis straightforward: "May we receive this desirable light of the god Savitr. The same holds for other verses in the meter." during which he receives his sacred thread.and is taught to every brahmin boy at the time of his Upanayana. but not in the corpus of the Samaveda.253. once and forall. There is some which has never been exspecial reason forthis separate treatment. all wishes of the sacrificer will be fulfilled by this mantra because they are impelled by Savitr.39).in particularthe famous Gayatri verse of the Rigveda gayatrz (3.62. an initiation ceremony or "second birth. 1 Dec 2013 06:22:46 AM All use subject to JSTOR Terms and Conditions . The text of the Gayatri (Rigveda 3.30 on Sun. According to the Satapatha Brahmana (2. only be found if we return once more to our point of departure. This mantra is reciteddaily during the Sandhya ceremony.sometimes in the context of a wish for inspiration(see. and for which a solution can plained.3. One of the most famous Vedic mantras is generally referredto afterthe common meter in which it is composed: the Gayatri.

It occurs everywhere. The answer to such questions is always the same: there is no answer. whose sequences definea Soma ritual. 89).once the mantra had acto find: this mantra had quired its special function.253. supreme significanceis attached to not the kind of mantra thatprovides the auspicious but meaningless sounds that make up the melodies that accompany ritual activity. it is placed at the beginning in some collections (in some manuscripts and in most of the printed editions). which in factobliterates almost all the meaningfulcharacteristics of the much that concerns the var~na system. For some arbitrary reason. distinguishbetween shorto and long 6o. The Savitri.The Pratihartaalways sings This content downloaded from 134. And so it remains significantand isolated.with one exception: it does not occur in the Samaveda.The otherbhakti portions are always sung in the same manner. remaining outside the numbered sequences and all classifications. The remaining bhakti portions are collectivelytreatedas follows. albeit in a different form:Ksatriyas have a Savitri in the tristubh so importantto a brahmin that it has spilled over to othertwice-borncastes. The reason for this special treatment. The Gayatri thus has meaning as well as social significance-unlike most mantras and other ritual chants-although this significanceis partly theoretical. They a treatment all receive the same treatment. Since only the prastavaretains the original text. Or not difficult too much meaning. 1 Dec 2013 06:22:46 AM All use subject to JSTOR Terms and Conditions . particularlylong o's. with very long o's (note that South Indian scripts. are set to this melody. A prayer for inspiration that is daily recited. It fell fromthe heap like the windblown seed we have already encountered.30 on Sun. it is only that portionthatis explicitlynoted in the tradition.while in systematically other collections it is simply not found. thatdefinesa brahmin and distinguisheshim fromothers.unlike Nagari. and Vaiiyas have one in thejagati meter(see Malamoud 1977. replacing them by stobhas and other meaningless sounds.58. The udgftha is always: 6 va o va o va hm bha o va.56 FritsStaal Why was this mantra picked to play such an important and auspicious role? There are hundreds of mantras in the Rigveda that say somethingsimilar. in fact. The Ga-yatri melody occupies a similar position in the ceremonies of the Soma ritual. Many of the stutior stotra chants.

" why d6coupage des vieux hymnes en formulesou meme fragments devenus des corps inertes dans la trame liturgique" (1960. forthe first song of the thirdpressingsequence of chants.2. The Yajamana and some otherpriests." Up to the presenttime. They give people. This is fromthe original verse: to be precisely what distinguishesmantras made into a mantra. breathes in. It is listed in the Sarnhita as follows: svadisthaya madisthayom (aiminiya Arcika 64. and thus fitforritualconsumption. "a sense of identity. or because the formeruses "a" where the latteruses "o.The Gayatri mantra itself by Nambudiri brahmans belonging to the Yajurveda or Samaveda and Nambudiris belonging to the Rigveda: the formerpronounce This content downloaded from 134. and chants "vak" "hrnm" (which comes fromnowhere) while he holds his breath--therefore almost inaudibly. We can now understand why. The fourthchant is listed in Uha Gana 3.30 on accordance withcomplex rules. 1 Dec 2013 06:22:46 AM All use subject to JSTOR Terms and Conditions .1). is often due to distinctionsthat rest upon meaningless phonetic variations. In all the cases of ritual application of mantras we have so far studied. nothing remains of the rich literatureof the Vedas but a collection of sounds and syllables. Entire passages that originally were pregnant with meaning are reduced to long o's. should also chant "o" (see AGNI I.a verse has to be subjected to formaltransformations. the Vedic schools themselvesare distinguishedfromeach other by such variations of sound that can be more easily explained in grammatical than in is pronounced differently religiousterms." That identity. operations that apply to to mantras formbut not to meaning.253. It preserves more of the original text although the udgftha has also disappeared: 6 ma da sva / yisthaya svadisthaya so o /a yindra/ 6 pa tava havu va / sil tah //(see AGNI I. This is why Renou referred he "le and mentioned as "poussiere v6dique. 603).1. Thus the Jaiminiya and Kauthuma-Ranayaniya schools differfrom each other by such characteristicsas vowel length.58. 646).1. Rigveda 9. We are now in a position to returnto the ritual uses in the Soma ceremonies of our verse.however. Ritual traditions have obvious social significance in that they identifygroups and distinguishthem fromeach other. only the prastdva need to be known.The Soundof Religion 57 togetherwith the Udgata. in thathackneyedcontemporaryphrase. 33). 76-77 quoted by Malamoud 1983.

37: and t may be substitutedin the ca. That the Yajurveda and Samaveda have adopted the preferred form is in accordance with the central place they occupy in the ritual tradition. and " P p" of this rule depends to the bilabial spirant "f'.16.1.58.In some cases these formsappear to be purely arbitrary.In This content downloaded from 134. but in two phonetic treatises(Pratisakhya) of the Yajurveda. "optionally. Actually. In the Gayatri.h" is generally believed not to occur in Sanskrit. Throughout this section we have observed that considerationsof form and formal derivations and transformations are foremostin the minds of the ritualists.. namely. replace of the visarga follow.3. especially those that introduce options. The interpretation on conventions that have been established elsewhere in the grammar. In this symbolism.2. 396).1. That the rules are to be understood in this order follows from 8." from 8." from8. and that the optional va means "preferable" has been shown by Kiparsky (1979). the Taittiriya of the Black and the Va-jasaneyi of the White Yajurveda.253.since theydo not correspondto any typeof formal relationshipthat is known or seems to make sense.3.36. Special symbols are needed to ad these sounds to the alphabet.58 FritsStaal the visarga at the end of "nah" in (11) as an "f. as in the manuscripts or printed editions of Panini's rule 8." spectively. (It is lacking again in the Madhyandina. "in the place of a word. another school of the White Yajurveda: Renou 1942-1957." Thus the variation between Vedic schools and between the Vedas themselvesis reduced to the optionality of grammatical rules.108. are used and lead to further ritual developments and proliferation. Thus we have to supply padasya. samhitdyam.9pauwhen a voiceless velar or a voiceless labial.2. " a sound that "."%k" refersto the velar "kh". The bilabial spirant "f" occurs not only in the grammar of Panini. which is believed to be absent from Sanskrit." from 8. it goes back to the origin of the tradition. and va. the voiceless labial which followsthe visarga "1h" is the initial "p" of "pracodayat. the bilabial spirant "f" is an optional variant prescribed by a grammaticalrule which also introducesanother sound. "in continuous pronunciation.30 on Sun. " kupvoh k. 1 Dec 2013 06:22:46 AM All use subject to JSTOR Terms and Conditions . In other cases the rules of grammar. This is not a modern phenomenon. the velar "kh".

related to the second. comand interlarded with and vowels there. but much of its meaning has disappeared and is no longer recoverable. of the meaning. but is presThe same cannot be said ent to the mind of the ritual performers. consider the nonsemantic approaches. For the original meaning of the verse is unknown to the relatedto any of the riteswith chanters.. most existing scholarlywritingon ritual will have been in vain. and the original formcan not only be recovered. This should alert us to the possibilitythatmeaning does not occupy center stage in the analysis of the ritual. and formsthat are generated by rules. concerns the extent Three importantquestions remain. The second question relates to whetherVedic ritual is representativeof a class of rituals. III Nonsemantic approaches In the previous section we saw how a verse of the Rigveda was turned into mantras forvarious ritesby being stretched here. This content downloaded from 134. The semantic approach to the synchronic at least in the case of our study of ritual therefore appears fruitless. are helpfulin elucidating these Vedic data.58.and neitheris it specifically which it has been associated. They are constantlyconcentratingupon rules. or is unique. Yet all these transformationsconform to precise rules.The SoundofReligion 59 both cases the chiefconcern of the ritualistsis with rules. viz. nor do these rites share any common meaning or function. finally syllables. This is the topic of the present section. is whetherthe semantic approach has established anything at all-for if it hasn't. and the syntactic. The third question. pressed chopped up to such an extent that its original formwas no longer recognizable. All their preoccupations illustratethe nature of ritual as a rule-governedactivity. the original verse is not only rendered well-nigh unintelligible. I shall take up let us the second and thirdquestions in the next section. For in the course of these same processes. Vedic data.253. the pragmatic. the phonetic or phonological. The first to which the nonsemantic approaches discussed in the first section.30 on Sun. But first. 1 Dec 2013 06:22:46 AM All use subject to JSTOR Terms and Conditions . Furtheranalysis confirms this.

Brahmanacchamsin or Acchavaka priests in accordance with specificrules. In Soma ceremonies.g." the ritual context determines throughoutwhat is to be recited or chanted. as well as when and on what occasion. Udgata and Pratiharta. the "relationship between the expressions and their users or contexts of use" that is characteristicof pragmatics is certainlya valid field of inquiry. As for "contexts of use..60 A.may throwlight on theirposition in the ritual enclosure. analysed the offices of two Rigvedic priests who take part in the Soma ceremonies. The ritual songs are always chanted by the Udgata.253. It is. the information contained in the presentparagraph. the relationshipis straightforward. But such "pragmatic" information of the depends on the structure ritual itself. the sastra recitations are always executed by the Hota. i. the Subrahmanyahvana: AGNI I. suitablyformulated. unlike the synchronicanalysis that seeks to understandthe ritual as a system. One contemporary scholar who has studied these historical problems fruitfully is the German Sanskritist Klaus for Mylius.58. would therefore take care of pragmatic problems as far as the users are concerned. Pragmatics FritsStaal In the study of Vedic ritual and mantras.. 596). The knowledge of these contextsof use is an important part of what constitutes a knowledge of the ritual-the knowledge that distinguishesa ritual specialist froma nonritualist. Mylius has. although on specific occasions other priests recite particular pieces. example. A pragmatic analysis of the functionof the users of the mantras.30 on Sun.and cannot replace the latter. the Pota This content downloaded from 134. Like the ritual acts themselves. well-known from the tradition (e. 386. the same in almost all ritual contexts. But such an analysis pertains to societyoutside historical reconstructionand is thereforea diachronic enterprise. most ritual recitationsare executed by the Adhvaryu on behalf of the Yajamana. Were we to add a list of exceptions and special cases. however. 1 Dec 2013 06:22:46 AM All use subject to JSTOR Terms and Conditions .e. partlyconcerned with one of those rare questions that have a relativelysimple answer: for insofar as the users are consince theyare invariably cerned. Maitravaruina. while the stutior stotra chants of the Soma ritual are always executed in the same fashionby the trio of Prastota. 369. of the priestlyfunctions. with a few exceptions.

45. respectively. as was. B. 24-25. the original meaning has largely or entirelyvanished.. It is clear fromthese examples that a pragmatic analysis may be relevantto historicalreconstruction. viz. in theiroriginal context. pragmaticsdoes not have the same functionhere as it has in linguisticsor philosophy. phonetics and phonology are distinguishedby a consound cern." Yet his officecannot have been created long afterthe Rigveda. but that the Acchavaka occupied the lower status of the two. Phonetics and Phonology The sounds of the mantras formthe subject matterof various ancient treatisesattached to the Vedic corpus.with "mere" sound and with significant rather: "sound that makes a the distinction is less (or difference"). His analysis led him to conclude thatboth these priestshad a lower statusthan the majority of the other officiants.The SoundofReligion 61 and the Acchavaka (Mylius 1977. owing to the fact that they fail to distinguishbetween two rather different questions: the question of the meaning of mantras in their ritual This content downloaded from 134. This possibility. etc. the Acchavaka is of later origin and he is not "elected" but "called. However. In view of this situation it is not surprisingthat ancient Indian ritualistssuch as Kautsa defended the view that mantras have no howmeaning (see Staal 1967. seriouslyby Western scholars. The mantras are oftenderived fromverses that.we noted that this is true only in the sense of historicalreconstruction. But when we looked more carefully.30 on Sun. has not been taken ever. To maintain the distinction in this domain would beg one of the questions we have posed.58. 1982).the answer to this latter question seemed obvious: they do have meaning. but to priestly functions which are stereotypicalroles.had meaning. 47. where he is assigned menial jobs. 1 Dec 2013 06:22:46 AM All use subject to JSTOR Terms and Conditions . This was amply illustratedin the preceding section.253. At firstsight. But in the ritual contextin which these verses have been transformed into mantras. Whereas in the study of language. easily maintained withrespectto mantras. The Pota is referred to in the Rigveda. and an almost indefinitely large number of similar illustrationscould be added. whethermantras have meaning. because it does not apply to individuals who determine the truth value of a statement. he is in fact related to the vaisja class. the officeof the Gravastut.). for example.

we must now record a remarkable fact: the phonological study of mantras has been one of the great Indian pastimes fromthe Vedic period onward. but also ordinary speech is an interesting object of scientific inquiry.58. The situation is similar to that in etymology: the etymologyof a word may throwlight on its original meaning. Part of the genius of Panini in fact consists in his recognition that not only the sacred speech of the Vedas. but it is also borne out by the facts: for the Vedas themselves offeredspeculations about mantras and language riddles long before a more systematicstudy of sounds originated in the Pratisakhya and Siksa literaturesand in the grammatical works of Panini and his successors.and although it was ignored by some Western philosophers(e. to mantrasbut withUsing the term"phonology" withreference out making any assumption about their meaning.. it has been re-discoveredby others (e. but it may have no connexion with its actual meaning or use.g. In fact. This has always been known to Indian and Western linguists. it is likelythat the study of the strange and peculiar of these sounds-strange and peculiar because they characteristics are different from ordinary language and seem to deviate from some of its rules-paved the way for the phonological study of "regular" sounds. the sounds of language..62 FritsStaal context.which is the contextthat definesthem as mantras. viz. and the rules that can account forthem.253.This may be explained a priori.30 on Sun. This is different however. certainlyto Indian ritual. and the question of the meaning of the expressions of language fromwhich mantras are derived in those cases in which they are derived from language.because man is more puzzled and attracted by the extraordinarythan by the ordinary. 1 Dec 2013 06:22:46 AM All use subject to JSTOR Terms and Conditions . The phonological study of the sound of religion as it appears in Indian ritual is thereforedefinitelypromising. This content downloaded from 134.The conclusion we draw fromthese conceptual distinctionsis thatmantrascan verywell be meaningless even in cases where theyare derived from verses that possess meaning. Wittgenstein). Heidegger)..g. from Levi-Strauss' preoccupation with distinctive featuresand binary opposition as a general method forthe analysis of ritual and social structure. This early emform on the explains phasis linguistic grammatical flavor that atand taches to almost everything Indian. thus leading to the origin of linguistics.

He knows something like the followingstructure(I say "something like. he knows how The ritualist." for he This content downloaded from 134. however. structural relationships between formal expressions are foremostin the minds of Vedic ritualists.58.first of all. without such knowledge they could not carry out their ritual tasks. A practisingritualistwho is asked forthe meaning of a recitationor song invariably replies: go to a scholar of language or philosophy.Actually. depending on his expertise.But there is no traditionaltransmissionof the meaning the verse.From what we have seen in the previous section. A brahmin who belongs to the Samaveda knows more: he knows into songs.30 on Sun. with other Samavedins and it still does not qualify him as a ritualist.1. and if he is really good. how the verses of the Rigveda have been transformed He will probably know various chants derived from that verse. he will know where these occur in the collections of the Samavedic corpuswhich is large and complicated. thisverse. If he does. as we have seen. Only the ritual expert or vaidikaknows in addition to an "ordinary" Rigvedin or Sa-mavedin (or Yajurvedin. confining ourselves to the ritual applications of Rigveda 9. knows to reciteit.The Soundof Religion C. 1 Dec 2013 06:22:46 AM All use subject to JSTOR Terms and Conditions . In fact. So let us first try mary. It is a ritualexpert that identifies knowledge of these complex structures him fromoutsiders. for that matter) the association between mantras and acts. Syntax 63 With syntaxwe come to the heartof the matter:forit is only syntax that provides the tools with the help of which we are in a position to studyritual structureand account forthe relationshipsthat obtain between ritualfacts.1.253. and when and where in the ritual these have to be insertedand executed. and so he does not necessarilyknow its meaning. forthis is what he learned when he learned the Rigveda by heart as a boy-a knowledge he shares with many other brahmins who also belong (by birth) to the Rigveda but are not of ritualists. He shares such knowledge. he must be a Sanskrit scholar and it has become a matterof personal interestto him. That is.the substance of the and distinguishes that previous section can be summed up by outliningthe structures a sumsuch to formulate make up thisknowledge.

1 MANTRA 49. semantic as well as nonsemantic). An exhaustive synchronicanalysis of the ritual can be given in terms of the structural relationships between such forms. (To be continued) * at theXVth International form. I am grateful helpful of thisarticle.64 FritsStaal uses a system of referencerooted in oral tradition and in some fromthe systemused by Western scholars.1985). If and when the system changes (which may be due to a varietyof causes. The syntacticapproach can therefore completely account for the ritual facts. earlierdrafts to theclass ofSoma ceremonies 2 The term jyotistoma mayalso be used to refer of whichagnistoma is the prototype.7 Aranyageyagana Arcika64.58.1 Stuti I ACT Flow of Milk 4 Layer I. Vedic ritual makes sense. Paper read. and this can again be studied in syntacticterms. the ritual. which respectsdifferent the availability of printed texts): presupposes LANGUAGE Rigveda 9.2.2 Stuti IV Sequence 11 I Such a structure makes sense only withinthe larger patternof hunwhich togetherconstitutethe edifice of dreds of similar structures. and is therefore syntacticin nature. but it is structuralsense.30 on Sun. Such an analysis will referto the rules that relate structuresto each other.1. in abbreviated Congressofthe International Association for ofReligions(Sydney. 1 Dec 2013 06:22:46 AM All use subject to JSTOR Terms and Conditions . without referenceto meaning or external function. This content downloaded from 134.7 Gramageyagana 15. the rules or the relationsbetween them change. This article theHistory will also be published in the International Journalof Asian Studies(Milton Singer to Pamela MacFarland fornumerous comments on issue). Brick7 Sequence 11 I Uhagana 3.253.