आगमे विश्वासटिप्पन्यः

The oral and the written

Mantra and mantraśāstra**1

In the theoretical approach as well as in the practice of mantras in Hinduism, the premium has always been (and still is) placed on the power of the oral word: Revelation, there, is not Scripture, but śruti, ‘what has been heard’ (even though the first mantras were not heard, but ‘seen’ by the Vedic r̥śis), a premium overemphasised in Tantra. Writing, however, is also sometimes used, and the literature on the subject, the so-called mantraśāstra, the ‘teaching of the mantras’, is made up of a vast quantity of written texts. It thus appears that one writes on what is oral – which itself is sometimes transposed in writing. This written as well as oral ensemble is, of course, in Sanskrit. The oral literature we shall consider here is a part of the learned, lettered, not of the popular, domain, even though the learned (oral or written) sometimes ‘debases’ into the popular. The question of the relationship of the oral with the written – secondarily of the learned with the popular – insofar as it appears in the case of mantras, is what we shall try to see here.

A purely oral, spoken, ‘speaking’, word …

By their very nature, which is the nature of the Word, of vāc, mantras are word, utterance, voice. Vāc, however, since Vedic times, is also and foremost the Absolute, the all-powerful, the brahman: brahma vai vāc, says the Aitareya Brahmana, 6.21.1. Hence their power, their efficacy. Hence, too, their immateriality. Oral, uttered or unuttered, the word is ‘speaking’ or ‘spoken’ (vāc vadantī). Being made up in essence of the very highest plane of the word, mantras are therefore on the side of the Absolute: they are both word and the Absolute. There is an often-quoted saying in Tantric Śaiva texts: all mantras are made up of phonemes (mantrāḥ varṇātmakāḥ sarve), but the phonemes are of the nature of Śiva (śivātmakāḥ). The mantras are thus divine, and are therefore nothing to do with writing, which is purely human.

Being the vācakas of deities (or of other supernatural entities), that is, what ‘expresses’, or ‘says’ them, bringing them thus to existence, being the efficacious essence of these entities, mantras are powers.2 They have their own power or energy (mantravīrya or mantraśakti). They are efficacious utterances, able to effect various actions, to produce results, provided they are used according to prescription and by authorized persons: this is what they do when used in ritual.

All this is always done by word of mouth, by the spoken word, audibly or mentally uttered. The uttering of a mantra can be so low as to be inaudible, or else purely mental – the mental utterance being in fact the highest one, as was already stated in the Laws of Manu, 2.85. This utterance is a necessity: differing in this from the written word, a mantra always implies the existence of a speaker. It is never cut off from its production, from its origin, as is the case of the written word. A mantra may, of course, be deemed to act by itself without any human intervention, but this it does only insofar as it has been first uttered, thrust towards its goal by somebody being present. It cannot, like a piece of writing, become active after having remained dormant between the pages of a book, or the folios of a manuscript. It is necessarily associated not only with an enunciation but also (in spite of Frits Staal!) with an intention to enunciate. True, there are cases, as we shall see, when the mantra, cut off from its utterer, is supposed to act in writing; but these cases are either exceptional or are part of popular, ‘magical’ practices.

Written words, in fact, are elements of language. But mantras, as is everywhere theoretically admitted, are not part of language. They are made up of phonemes or syllables (varṇa, akṣara), sometimes of words – of the phonetic material of speech, that is; but they are not meant for communication. They do not depend from, nor are they conditioned by, any outside factor (they are nirākaṇkṣa, ‘autonomous’). Above all, they are not ruled by the conventions (saṃketa) which govern the world of speech as a means of communication. Being asāṃketika, ‘non-conventional’, the mantra is a form of word existing on a primary, divine plane, prior to and above language. It is not turned towards the outside world, but towards the divine to which it may lead the user, whilst being able also (and mainly) to act in the world. This non-linguistic character of mantras is not due to their being often made up of meaningless syllables for, even when they are made up of ‘normal’ Sanskrit words or sentences, mantras are not speech. This of course is a metaphysical or theological principle. It is, however, justified if one considers how mantras work: their efficacy is not due to what they ‘say’, but to how they say – to their actual wording. It rests on their phonetic tenor (hence the necessity of an absolutely perfect enunciation). This efficacy rests on the substance of the expression (to speak like Hjelmslev), on the phonetic structure, on the sounds being somehow emitted, but not on any literal meaning (supposing there is one).

The inseparability of the mantra and of its utterer is underlined in many theoretical works and by a number of mantric practices. It appears also in the oftenproclaimed link between mantra and prāṇa, the vital breath linked to the Self, which is also one of the aspects of the cosmic energy (according to L. Renou, prāṇa is ‘the vitalist aspect of the ātman’). In the tantras and other Tantric texts, the utterance of a mantra is usually named uccāra/uccaraṇa: an enunciation which is an upward movement (ut-√CAR, to go upwards, to ascend) of the vital breath together with the mantra. Uccāra is in fact the ascending thrust of the phonic energy of the mantra. It is evidently not a respiratory breath. The term notes especially the ascent of the kuṇḋalinī caused by the upward moving conjunction, in the suṣumnā, of the prāna and apāna vital breaths: what happens there is, above all, the movement of the energy which goes upwards in the yogin’s ‘body’, but even in such a case a mantra is deemed to be uttered. Even when there is no vocal utterance, the mantra is nevertheless present and active in the consciousness and in the body (or more exactly in the image of the body, in the ‘lived’ or ‘experienced’ body) of the adept. One might describe this by saying that the mantra is ‘uttering itself’ in the consciousness of the adept whilst spreading out like a fluid in his body.3 The uccāra of a mantra, Abhinavagupta says when describing this process, is prāṇacidātma (TĀ 5.128): it is both ‘breath’ and consciousness. There is a conjunction of the vital and spiritual energy of the yogin and of the energy of the mantra. All this, evidently, cannot in any way be connected with writing.

We may remark here that all this is in complete conformity with the Indian traditional view according to which things of importance – especially when connected with religion – must not be dealt with in writing. The rule is as old as the Veda, which, as is well known, was transmitted orally during two millennia and is still being taught and used in ritual orally and with the most minute phonetic precision. Ṭhe superiority of the silent over the audible utterance of mantras, also, goes back to Vedic origins.

In the same way, mantras – those at least received in initiation to be used for spiritual or magical practices – can only be transmitted orally, by word of mouth, by the initiating master, the instruction going, according to the Sanskrit expression, ‘from ear to ear’ (karṇāt karṇopadeśena), for this only guarantees the secrecy of the transmission. Mantras are of divine origin, they were originally revealed to a first master (ācārya, guru) by the deity and then transmitted, always by word of mouth, through a succession of initiated masters down to the present times. Thus, only those who possess the needed qualification and authority (adhikāra) may transmit or receive, orally and according to precise rules, these secret and powerful sacred formulas. Many texts say that mantras that have not been received regularly (or, still worse, found in a written text!) have no value. They are ‘fruitless’, ineffective. They may even be dangerous.

It is interesting to note that these strict mandatory rules on the orality of mantras have survived for centuries in a social milieu in which a literature in Sanskrit was being produced (and there are many other languages in India) which is one of the largest in the world.

of which writing can only be a reflection or a comment …

Writing, in principle, appears as always devalued. It is merely a practical tool, an artefact without any value of its own. For Indians, as for Rousseau, ‘writing is a mere supplement of the spoken word (‘l’écriture ne sert que de supplément à la parole’), a ‘dangerous supplement’ (un dangereux supplément) even, since by its availability, by the fact that it extends the field of communication by spoken word or gesture, it wrecks the circle of direct interpersonal communication. By writing, one moves (down) from the qualitative, where the spoken word goes from the deity to the spiritual master, then to the disciple, to eventually turn back towards the divine through yogic practices or ritual action using mantras, to the quantitative where the word circulates freely in a homogeneous milieu where everybody can receive it, but where it has by this very fact lost all its power: when uttered it is mere noise, flatus vocis. Worse: once written, the spoken word becomes merely a stilted transfer, the visible, concrete, inferior form of a living, immaterial, supreme reality (as unauthorised copy, it may even get the piratical user into trouble4).

In spite of the, always valid, principle of the strictly phonetic, oral nature of mantras, their users or inventors (in the two ancient meanings of the word: invent as well as discover) did not entirely escape the rule of writing since they lived in a milieu where writing was being used. We will indeed see several mantric practices where writing is not simply a practical device, but is sometimes linked to the very nature of mantras: there are therefore legitimate, ‘respectable’ uses of writing not as a substitute, but as an additive to the oral.5 To this can be added cases where it can be said that the oral debases into writing, cases we will merely allude to later on in spite of their important place in practice, the frontier between these two uses being in fact not always very visible, the ‘inferior’ uses being not merely modern or contemporary ones (nor purely written ones).

As we noted earlier, there is first the mantraśāstra, that is, all the texts, of all sorts, which describe, prescribe, or deal in some way with mantras. Thousands of pages have thus been written during some two millenaries on the purely oral mantras. What is more, these texts, while ceaselessly proclaiming that mantras are secret formulas, sometimes give explicitly, in writing, their phonic tenor – which is perhaps to be explained by the conviction that a mantra discovered in a book has no power or efficacy.

but which it can nevertheless help to reveal

Very often, nevertheless, such texts reveal only in veiled terms the phonemes that make up the mantra: using periphrases or conventional terms which only qualified masters or initiates can understand. They give, not the mantra, but its uddhāra:6 how to produce it or ‘extract’ it, how to make it arise out of the Word. This can safeguard its secrecy very effectively. In effect, if some conventional terms are common ones (as mayā for Ī, indu for bindu, etc.), and if there are manuals which give such meanings – bījakośa, bījanighaṇṭu, uddhārakośa, etc. – and also compendiums, listings of ‘secret’ terms, these riddles are sometimes very difficult to solve. Every tradition or initiatory lineage has its own system, carefully kept secret. The written, in such cases, assists the oral: it guards its secret against the ignorant, whilst disclosing it to those permitted to know it. Functioning as an aide-mémoire, it helps to keep up the oral tradition – a role it certainly had from ancient times.

The uddhāras do not merely use abstract notions or mythico-theological elements so as to indicate/refer to the letters of a mantra. They sometimes refer to their written form. Writing has in such cases a directly practical role. The uddhāra also often implies a spreading out (prastāra) of the phonemes of the Sanskrit alphabet so as to ‘extract’ (ud-DHR̥) from among them those which are to constitute the mantra. This prastāra consists in spreading out (pra-STR̥) the forty-nine to fifty letters of the alphabet, writing them on an appropriate surface, either grouping them down by phonetic categories (varga), or distributing them according to specific rules on a diagram. The letters once spread out, one ‘extracts,’ chooses that is, those which are to make up the mantra, these letters being often designated by conventional (secret) names. This use of prastaras underlines the fact that mantras, as aspects of the word, are born from the mātr̥kā, the phonic ‘mother’ of the universe of which the Sanskrit alphabet is part, from which they are extracted. It shows both the subordination of the mantras – made up of letters – to the totality of the Word, the Sanskrit alphabet (or the supreme deity insofar as it is made up of Word, alphabetical), and the fact that mantras draw their power and efficacy from that alphabetical totality.

The spreading out of the mātr̥kā is also made use of differently, by using secret, initiatory prastāras, named gahvaras, a term which means cavern, hiding place, secret. These are theoretically secret diagrams, of different shapes, where the letters of the Sanskrit alphabet are placed in a conventional order known only by initiates. They are thus ‘caverns’ where the secret formulas, the mantras are hidden.7 The user of a tantra that prescribes the use of a certain gahvara, knowing how to draw it and the places where the letters are to be written, it suffices that the places where they are written and the order in which the letters are mentioned to enable him to construct the mantra. The design of such diagrams, the placing of the letters and their ‘extraction’, are all codified, sometimes complex ritual operations. This is not surprising since what is being done there is not merely drawing a diagram and writing letters on it: it is invoking and manipulating the divine power of the word so as to gather it up in a highly effective formula, a mantra, the power of which the adept will then master through the ensemble of rites and conducts of the puraścaraṇa or mantrasādhanā,8 and will then use – in principle orally, perhaps also by placing it on a diagram, a maṇḋala – to some spiritual or mundane end. The mantra thus produced can also be the mantra given to the disciple in the initiation (dīkṣā) ritual. It is interesting to see here written letters being used to make appear ritually a formula which in principle must remain purely oral. There are other ritual procedures of that type used to verify if the mantra is well adapted to its user and to the result he aims at. This checking, named mantravicāra or mantraparīkṣ ā, is done using diagrams of different sorts.9

or even sometimes to show

One may perhaps be tempted to consider such procedures as rather external to mantras: they are preliminary to their use, with no influence on their nature or functioning. There are, however, also cases where it is in their very use, even in some measure in their essential nature, that mantras appear in a written form. Or, more, they may appear in a form that the user must imagine as being in writing. There are finally cases when mantras can be made use of only when not simply placed by nyāsa, but inscribed in a diagram.

In this domain, the oldest proven instances seem to be mainly Buddhist ones. They are numerous in Mahāyāna Buddhism, especially in Tibet where Sanskrit mantras are generally used. We may wonder whether this is not due in part to the influence of China, where the written sign, not the word, is essential. In effect, the best known and most characteristic Buddhist use of bījas, Sanskrit syllables, in a written form, is the Sino-Japanese (mostly Japanese, in fact, in the Shingon sect) of the siddham writing. This lies outside the Indian domain; the case is nevertheless worth mentioning here since, in some of the Tibetan traditions, the recitation and visualisation of the mantra are inseparable: there is a coalescence of the seen, meditated form and of the sound uttered and perceived inside, not merely the use of a phonic element in an outer visual frame.

Of all this, Hinduism also shows many instances, some being very ancient. One of the most evident is the visualization of bījas or of phonemes prescribed in the kuṇḋalinīyoga, where the yogin is to see with his mind’s eye, as present in the cakras of his yogic body,10 Sanskrit syllables or bījamantras written on the leaves or on the pericarp of the ‘lotuses’ which constitute the cakras. The sound, the oral utterance, of these akṣaras or bījas, is to be imagined, felt even, by the yogin as present in these centres. The kuṇḋalinī, too, as we have seen, when going upwards, may be identified with the upward thrust (uccāra) of a mantra. In some cases (for instance, in the japa of the śrīvidyā described in the third chapter of the Yoginīhr̥daya), the adept imagines the subtle phonic vibration (the nāda) of the mantra as ascending from the mūlādhāra to the dvādaśānta, the centre twelve finger-breadths above the head. In such a case, what is experienced by the adept is the phonic, oral/aural aspect of the process. It seems, however, that in some texts, the bījas or akṣaras/varṇas imagined as present in the visual structure of the kunḋalinī are to be visualised in their written form, on which it is probably easier to concentrate one’s attention than on a sound, by nature more evanescent.

There is, on the other hand, another mantric practice, where mantras or syllables are necessarily to be written: that of nyāsa. This consists in placing ritually (with a mudrā) on the body or on a support, a mantra, bīja or phoneme which is simultaneously to be mentally visualised (dhyayet, say the texts, dhyāna being here used in the Tantric sense of visualisation11), such visualisations being usually of their written form. (Admittedly it can also be the form of the deity of the mantra, but this is not suggested by the texts I have seen.) Some ritual treatises, sometimes explicitly, say that one is to visualise clearly the intensely shining letters of the mantra. Chapter 293 of the Agni-Purāṇa prescribes the lipinyāsa – the placing of written letters (lipi) – of the fifty letters of the Sanskrit alphabet, followed by the nyāsa of eight Lipīśvaras, ‘Lords of letters’, which are to be written (likhyate). The Purāṇa does not say how these deities are to be drawn (written?).

One has also speculated on the written aspect of these oral letters

Having said that in principle mantras cannot be but oral since their nature is that of the word, of the primal vāc, identical with the formless Absolute, the brahmansound (śabdabrahman) as it is sometimes called, we must now look at cases where writing comes in, not merely in the ritual uses or practices of mantras, but at their highest level, on the level of the deity. In such cases writing appears as linked to the very nature of mantras.

Let us note first that in some texts (the Agnipurāṇa, 293.51, or the Brahmāṇḋapurāṇa, for instance) there exists a Lipidevī or Lipibhairavī: a ‘writing-goddess’; in the Śāradātilaka, 7.8–14, the goddess’s body is made up of letters (lipitanu) and she is seated under a ‘tree of letters’ (lipitaru). This alphabet-deity is the Mistress of the Word, Vāgīśvarī (or Devī Vāgīśī). It is in fact an aspect of Sarasvatī whose attributes she bears (notably the rosary and the book). Her plane, however, is not the supreme one: she is the goddess on the level where she gives birth to the world. In the non-dualist system of the Trika, on the other hand, it is the supreme goddess, Parā, who is paravāk, the supreme Word, a variant of Sarasvatī-Vāgīśī, who bears these attributes, the alphabet being present here, most subtly, as mātr̥kā, śabdarāśī or mālinī, as the supreme Energy or totality of the Word, that is, and therefore on a plane very much above the plane of writing. There is, however, a continuity or a proximity – underlined by the presence in both cases of Vāgīśvarī holding a book (pustaka) – between the totality of the word and the written Sanskrit alphabet. One could thus consider that there is here, as it were, a primary presence of writing: a ‘primordial trace’, or a ‘primordial writing’ (une écriture première), to use Derrida’s words; it being, however, understood that, in India as elsewhere, one may postulate the precedence of the oral over the written. It is only insofar as, in the course of time, one has speculated on the word and that the mantraśāstra has succeeded the mantra, that graphical elements appeared to reinforce the empire and power of the oral, to sustain, explain and describe its supremacy.

If we turn now towards the mantraśāstra, we will find striking instances – in fact, very well-known ones – of the intervention of the graphical in mantras or, to say it differently, of the interaction on all planes of the visual and the phonetic.

The most obvious example is probably that essential element of all bījamantras, the anusvāra, the nasalisation and prolongation of a syllable which one names bindu (which is to say, drop, dot) because of its written form as a dot placed over the nasalised phoneme. (It is sometimes called indu, ‘moon’: it is a luminous circle.) In mantras, it ends such bījas as OṂ, HRĪṂ, LAṂ, etc. Now, this written form is the basis of all the speculations on the bindu conceived as a ‘drop’ of concentrated phonic energy, as the point where the energy of the word gathers upon itself before dividing itself and spreading out to manifest outwardly its power and to manifest the universe.

Phonetically, the anusvāra, as a resonance which progressively fades out or which diminishes before a consonant, does not offer much food to the imagination. It is clearly and evidently its visual aspect which plays a role. Only the visual perception based on the written form can explain that the bindu at the end of a mantra could be generally described as the part which concentrates its dynamism, or (to use Kṣemarāja’s words in his Śivasūtravimarśinī) which ‘flies like the arrow thrust forward by the tautened bow’. Among the elements usually enumerated that characterise a mantra (the r̥ṣi, etc.11), the bindu represents usually the kīlaka, the ‘point’ or wedge that will pierce the target aimed at by the mantra.

In the traditional order of the letters of the Sanskrit alphabet (the varṇa-samāmnāya), the anusvāra is followed by the visarga, the sixteenth ‘vowel’, which, in Indian grammar, designates a release, an escape of breath after a vowel at the end of a word, its written form being two dots (:). It is thus graphically a division in two of the bindu, whereas – owing to its name and its place – it emits or produces (vi-SR̥J) the thirty-four consonants which follow it and, by this very fact, gives birth to the totality of the cosmic manifestation. This written form, a division in two of the bindu, also gives rise to speculations on the birth of the word. We may note here that very few mantras end with a visarga. I need not expatiate here on bindu and visarga, but refer readers to studies on the subject (for instance, Padoux 1990, chap. 5, pp. 272–286).

We may, on the other hand, apropos of the oral/written interaction as seen in the mantraśāstra, mention two other instances of speculations based on the written form of letters in the devanāgarī script.

First, on the drawing of the letter A, the first vowel, which is, in the Tantric metaphysical speculations on the word, the Absolute, the Unsurpassed (anuttara), the original phonic energy underlying all the other phonemes – therefore above all limits and all form. But, in emanative systems (as are Tantric traditions), it is necessary for the supreme deity, the Absolute, to hold in itself the seed or paradigm of all that it will emit. The energies that will create and sustain the cosmos are therefore all inchoately in the Absolute. Abhinavagupta elaborates upon this idea in the third chapter of his Tantrāloka (where speculations on the written form of phonemes are in Jayaratha’s commentary on TĀ 3.67). Such ideas are also to be found in another, older, text, the Vātūlanāthasūtra. The lines of the written form of the letter A are described as representing the four basic energies of Śiva: Raudrī is the head (-), Jyeṣṭhā, the weapon (3), Ambikā, the arm (/), and Vāmā (who ‘vomits’, that is manifests, the world) is the mouth (|). It is indeed not in its written form that A is the Unsurpassed with all its energies. But we see here the symbolic value given to the written form of a letter being used in support of the metaphysical interpretation of a phoneme. Another Sanskrit phoneme whose written form has given rise to the same speculations is E, whose written form is more or less triangular.13

Then there are also diagrams and amulets

One finds nearly always, in the centre of the yantras or cakras of Tantric female deities, a triangle point downwards, or two inverted superposed triangles symbolising the (sexual) union of Śiva and Śakti. There are also sometimes, in these diagrams, several inverted superimposed triangles (as, for instance, in the srīcakra of goddess Tripurasundarī) replicating and reflecting the first couple, symbolising thus the creative dynamism of the goddess united with the god. In the centre of the central triangle of the śrīcakra, there is also a bindu which is the goddess herself, a spot where both the oral and the written focus since the symbolism of the bindu is both geometrical-visual and phonetic-oral.

Admittedly, such representations are, strictly speaking, outside the domain of the oral/written relationship in mantras. It is, however, a neighbouring domain, which intersects and oversteps this relationship, and is an important one; for what is in question here is not only the relation of the phonetic to the written, but also, more largely, the relation of the oral to the visual. For this, indeed, is the context within which is set the more general problem of the relationship of mantras to ritual diagrams (maṇḋala, yantra, cakra), which is the problem of the relationship between the phonetic, temporal (and fugacious, evanescent) and the spatial (and lasting). There are, of course, mantras, bījas, or phonemes in all ritual diagrams which often are no more than a means to display the phonetic elements for putting them into action: the word, there, becomes active as being spatially placed, or, if one prefers, space is ritually made use of only when qualified, infused with power by the word. In such cases, the functioning of the yantra and of the mantra appears as being totally linked. There are also cases when they function in full parallel, cakra and mantra appearing to the yogin as the two faces of the same divine reality, as two interchangeable modes of access to this reality. This is what takes place with the śrīvidyā and the śrīcakra in the Tripurā tradition. It also happens (for instance, for the śrīcakra) that one goes on directly from one regime to the other: from the spatial to the phonetic, this being all the more easy when, as sometimes happens, the maṇḋala is not concretely marked out, being only visualised in imagination. This is the case when the yogin is to visualise diagrams in his body, in the cakras, for instance, visualised with their design and the deities that abide there together with mantras or bījas and phonemes.

But there are also what we may call diagrammatic mantras whose utterance and functioning – their use, that is, in mantrayoga – is constituted partly of visual, spatial elements (drawings and written letters), these being either concrete or mentally visualised.

There exists therefore a common zone where correspondences, isomorphisms, between visual and audible are played out. This is in fact all the less surprising in a civilisation where the Word, the oral, is placed at the origin of everything that exists, its supremacy being always extolled; but a civilisation for which, at the same time, the direct visual (sakṣāt) perception is the best possible proof of the existence of something; where, too, in the common religious practice, the view, the vision of the image of the deity by the devotee – the darśan – together with the deity’s gaze going towards the devotee, are of the utmost importance. We would be tempted to say here that we have, in the Indian case, the (apparent) paradox of the omnipotence of the Word, which is invisible, being proclaimed, extolled by ‘visual seers’. This Word was ‘seen’, in the beginning, by the Vedic poet-seers, the r̥ṣis: ‘the sacred Formula (brahman) that was first born in the East, the seer discovered it from the shining edge [of the world]’ (Atharvaveda 4.1).

Many instances of the presence of visual elements in the structure and uses of mantras – in a, by nature, phonetic (oral/aural) process – could easily be adduced. Typical is thus the fact that when enouncing the subtle phonic levels (kalās) which prolong the utterance (uccāra) of OṂ and are above all actual physical utterances, since they are infinitely subtle and eventually disappear completely in utter silence, the practitioner is to ‘see’ these levels as symbolic geometrical figures. One could also quote the case of the kāmakalā, a symbolic representation of the sexual union of Śiva and Śakti, which is something both graphically visible and phonetically invisible since it is a diagram with an inscribed letter (ĪṂ but also a bīja to be uttered.14

But we cannot dilate here on this sort of learned speculations and practices, intermixing oral and written, which are found in the Sanskrit texts of mantraśāstra. It would be interesting to know their origins and when they began, but on this we cannot do more than suggest possibilities. The earliest documents on the subject are in fact Buddhist and Chinese.15 Speculations on the Word, as well as magic, go back to Vedic times. It is therefore not surprising that even the magical uses of the word should have developed in a literate milieu. We may be permitted to believe that as soon as Indians began to write – at any rate very early – they used mantras as, shall we say, magic formulas, using for this purpose writing or lines of some sort, or by inscribing them in ritually drawn diagrams. Therefore, rather than a passage from the oral to the written attesting to a modern evolution, or a passage from the learned Sanskrit to the popular domain, there would have been from an early time, beside a strictly oral domain, the highest and most respectable one, a lower domain, nearer to magic (by which I mean the manipulation of sacred power to notably mundane ends), where oral and written co-exist but where the written plays often an autonomous or quasi-autonomous role. One must, of course, especially in modern times, make a distinction among the written elements: a devotedly preserved manuscript has not the same status as a printed text. The intrinsic value of the book matters little, in fact; what is important is to find in it the efficient formula and the way to use it.

If in India, today, one obtains mantras or amulets from specialists: tāntrikas, mantiravatis,16 pūjāris sometimes, or other magicians, these make them (supposedly) ritually, using rites where mantras are necessarily uttered. The oral mantra is somehow embodied in the written formula, thus investing it with power. Even for amulets, charms or yantras described in books, it is often specified (see, Chawdri, 1984) that, to put these magical contrivances into action, rites including the utterance of mantras and sometimes long japas are necessary, which shows the continuous power of the oral utterance. Nothing, in fact, is able to replace the enunciation, with the mental and bodily participation it implies: the presence in the body, one might say, of the word one knows ‘by heart’, or and perhaps better, in Sanskrit, which is kaṇṭhasthā :‘staying in the throat’, ready to be vocally uttered.

And the circle of users

But who are those who use these oral or written mantras? In what context? Of what kind is the relationship between the user and the mantra he uses? Being in Sanskrit, mantras, in principle, may be used only by Brahmins or at most by other twice-born groups (provided they are not Vedic mantras). This circle of users, however, became larger as a consequence of the ‘mantrisation’ of Hindu traditions in the course of centuries: any initiated person, of whatever caste or sex, can, if the prescribed rules are respected, receive a mantra he/she will be able to make use of if he/she conforms to the rules. The ācārya, sole person permitted to transmit a mantra, is still usually a Brahmin; but the receiver who uses the mantra is not), necessarily a Brahmin. More, as bhakti developed at the expense of ritual rules, any devotee could worship his chosen deity using the deity’s mantra. It is not only Brahmins, especially nowadays, who confer mantras on adepts or devotees – and their mantras are not less effective for that! The circle of users permitted to utter mantras tends thus to extend indefinitely till the most extreme (the lowest?) level, reached with the uses of mantras written on amulets or found in books.

We may note, too, concerning the relationship of the user to the mantra, that Sanskrit having ceased for more than a millennium to be the mother tongue of any high-caste Indian, to use a mantra, which is in Sanskrit, is for the user both to manipulate the power (intensified in the mantra) proper to that strange divine tongue. and to participate in the particular, learned, world of the Sanskrit culture. It is a peculiar relationship: to an unusual, both learned and timeless, form of speech. Sanskrit is the liturgical language of Hinduism. It is not the common language of the Hindu religion, still less that of the believers, who are alien to the Sanskrit world. This relationship of the user of a (Sanskrit) mantra remains the same whether or not he understands the formula he utters. This does not matter, in fact, since a mantra acts by its mere utterance, by its phonetic content, its elocutionary value, not by its meaning – supposing it has one.17

Whether or not the use of a mantra resorts to writing, it remains an aspect of the word whose power can only be put into action by oral utterance. The power of the oral is always what underlies and shapes all uses of mantras, be it a concrete (audible or inaudible) utterance, the inner utterance in the form of the ascending movement of prāṇa, or a purely mental vision of the sounds: one of these forms of orality is always present (even if momentarily) in all mantric practices, even when apparently written. This, I believe, is what appears through any careful study of mantraśāstra.