Is Sound Eternal?

Companion piece: “Whose Voice Is the Veda? Apauruṣeyatva Across the Darśanas” Part 1

I. The Question Part 1 Set Aside

Part 1 of this article pair asked whether the Veda has an author, and found three schools answering “no one”, “Iśvara”, and “no one, though it originates anew each cycle” — three verdicts on the pauruṣeyatva/apauruṣeyatva question. At almost every turn, those arguments leaned on a second, independent claim: is śabda — sound, specifically the phonemes (varṇas) that make up spoken language — a permanent kind of thing, or a fresh event that arises and perishes with each utterance? This is the nityatva/anityatva question, and it is this essay's sole subject.

The two questions are easy to collapse into one — an eternal Veda sounds automatically authorless, a produced Veda automatically someone's composition — and Part 1 showed why that collapse does not hold even for the schools that most look like they are making it. Mīmāṃsā needs eternity to secure authorlessness, but the two remain logically distinct claims requiring separate defence. Advaita Vedānta needs both eternity and origination, at different levels of description, to hold its position together at all. This essay works through what each school actually argues about sound itself, independent of what that argument is later used to prove about authorship.

II. Nyāya–Vaiśeṣika: A Sound That Is Made

A quality that comes and goes

Nyāya-Vaiśeṣika's starting point is not theology but physics. In its analysis of substances and qualities, śabda — sound — is classified as a guṇa (quality) residing in ākāśa (ether/space), perceptible by the ear. Ākāśa itself is eternal and all-pervading, but that tells us nothing yet about whether the qualities that arise in it are eternal too — heat is a quality of fire, and no one supposes each flash of heat has existed since beginningless time. The question the Nyāya-Siddhānta-Muktāvalī's Śabda-khaṇḍa takes up, in careful pūrvapakṣa-siddhānta form, is exactly this: is the guṇa śabda a standing feature of ākāśa, or a new product each time something strikes, splits, or is spoken?

The Mīmāṃsaka case, as Nyāya states it

Before refuting the Mīmāṃsaka view, the Muktāvalī states it fairly, and it is worth pausing on each argument in ordinary terms before seeing how Nyāya answers it.

• Recognition (pratyabhijñā): when we hear “ga” spoken by two different people at two different times, we say “that is the same sound,” not “that is a similar sound.” This feels like recognizing one continuing thing, not noticing a resemblance between two different things — much as recognizing an old friend feels different from noticing that a stranger looks like someone you know.

• No one perceives a sound dying: a pot's breaking is something you can watch happen. Nobody watches a sound “die.” What we call a sound ceasing is, in this view, just the sound becoming imperceptible again — the condition that was manifesting it has ended, not the sound itself.

• The manifestation (abhivyakti) model: speech-effort does not produce sound; it reveals an already-existing eternal sound, the way a lamp does not create the pot it lights up in a dark room — it only makes a pre-existing pot visible.

A fourth argument the Mīmāṃsaka offers here is theological — that a momentary word could not ground an eternal, authorless Veda. That argument belongs to the authorship question and is examined in Part 1.

Nyāya's refutation, argument by argument

Viśvanātha, following the line back through the Nyāya-sūtra, the Vaiśeṣika-sūtra and Udayana, answers each point and reaches the opposite conclusion — anityatva.

• Direct perception of arising and ceasing (utpatti-vināśa-pratyakṣa): we do hear a sound begin, and we do hear it fade to nothing — this is treated as a plain observational fact. What is directly seen to arise and cease cannot, by the ordinary rules of evidence, be called eternal. This is usually given as the primary reason (hetu) in the whole argument.

• Sound as a product of contact and separation (saṃyoga-vibhāga-janya): a drum sounds because stick meets skin; bamboo cracks because fibres separate; speech happens through the contact and release of the vocal organs. Effects that arise from a specific, locatable causal operation are, on Nyāya's general theory of causation, products (kārya) — and every product is non-eternal by definition.

• Loudness adds up: strike two drums together and the sound is perceptibly louder — an aggregation, an increase. A single eternal entity cannot become “more” of itself just because more instances happen to reveal it; growing louder only makes sense if each drum is contributing a genuinely new, distinct sound-token, not merely unveiling one old one.

• The recognition argument answered: “this is that same sound” is explained as recognition of similarity (sādṛśya) loosely spoken of as identity — exactly as we call a river “the same Ganges” though its water is never the same twice, or call a fire “the same fire” though its flames are renewed instant to instant. A shared universal (jāti — “g-ness,” “k-ness”) inheriting in each fresh sound-token fully explains why we recognize a type without needing one single eternal individual behind it.

• Against manifestation itself: revealing something (like a lamp revealing a pot) requires a thing that already exists somewhere, merely waiting to be perceived. Nothing of that kind is happening when a drumstick strikes a drum — the operation is straightforwardly generative, of exactly the same kind that produces any other new quality. Positing an extra “manifestation” apparatus here, when ordinary production already explains everything observed, is an unparsimonious extra hypothesis (gaurava-doṣa) — Nyāya's version of Occam's razor.

• Sound travels, and takes time: sound is heard progressively later the farther away one stands — an early Nyāya account explains this as a chain of successive contact-and-separation events rippling outward through ākāśa, sound producing sound the way one ripple produces the next (a wave analogy, taraṅga-nyāya). A static eternal entity has no business propagating outward in stages like that.

Nyāya's own account of why an authored, non-eternal Veda remains fully authoritative — the āptatva argument — is given in Part 1.

III. Mīmāṃsā: Eternal Phonemes, Never Produced

The Mīmāṃsaka's thesis is stated in the Vedāntaparibhāṣā:

वेदानां नित्यत्वेन निरस्तसमस्तपुंदूषणतया प्रामाण्यम् इत्यध्वरमीमांसकाः

The Veda is authoritative because, being eternal, every defect traceable to a person has already been excluded from it — so hold the Ādhvarīṇa (ritualist) Mīmāṃsakas. The Veda, on this account, consists of kūṭastha-nitya varṇas — phonemes that are eternal in the strongest sense: static, unchanging, never produced and never destroyed. What that eternity is used to prove about authorship is the subject of Part 1; what concerns this essay is the claim about sound itself, and the one mark of apparent impermanence that bears directly on it.

The evidence of ordinary speech about sound.

Everyone, learned or not, says utpanno gakāraḥ, naṣṭo gakāraḥ — “the sound g has arisen,” “the sound g has perished.” If phonemes themselves come and go, then words (aggregates of phonemes) come and go, and the Veda (an aggregate of words) comes and goes with them — impermanence at the smallest level should propagate all the way up to the whole text.

The Mīmāṃsaka's answer is the same abhivyakti (manifestation) theory Nyāya reports and rejects above: nothing about the phoneme itself is described as arising here. What arises is the manifesting condition — the vocal effort that makes an already-present phoneme audible in a given place. “Gg arises” is loose but valid usage for “g becomes perceptible here,” exactly as one might say “the mountain appeared” when the fog lifted, without meaning the mountain was created by the fog's departure.

The remaining marks Mīmāṃsā must answer — the taddhita derivation of śākhā names, mortal names inside the text, and the absurdity of trees and serpents in sacrificial session — are arguments about authorship, not about the eternity of sound, and are treated in full in Part 1.

IV. Advaita Vedānta: Three Levels of Eternity

Śaṅkarācārya's treatment of this question, in the Devatādhikaraṇa of the Brahma-sūtra-bhāṣya at Brahma Sūtra 1.3.28, and its later systematization in Dharmarāja Adhvarīndra's Vedānta Paribhāṣā, produces the most technically surprising position in the whole debate: the Veda originates, and is in that sense not eternal — and yet, at another level of description, sound itself is eternal after all. Getting to that combination takes several careful moves.

The setup: a challenge to word-meaning eternity

Earlier Mīmāṃsā (Pūrva Mīmāṃsā Sūtra 1.1.5) had already established that the relationship between a word and its meaning is autpattika — natural rather than a matter of convention or assignment — which is what makes Vedic authority anapekṣa, independent of anyone's intention. Śaṅkarācārya's bhāṣya opens by taking this as settled ground (अतः प्रभवात्, प्रामाण्यं प्रत्यनपेक्षत्वात्).

The pūrvapakṣin then presses an objection. Many Vedic words denote deities — Vasu, Rudra, Āditya, Indra — described as embodied beings who act, fight, are born and die, much like human individuals. If word and referent are supposed to be eternally linked (नित्यस्य शब्दस्य नित्येनार्थेन नित्ये सम्बन्धे), how can that hold when the referents themselves are impermanent, produced and destroyed? “Vasu” and “Rudra” look exactly like human names such as “Yajñadatta,” assigned to a person only after his birth. If so, does not the eternity of the Vedic word collapse together with the mortality of what it names?

The resolution: individual versus universal

Śaṅkarācārya's answer turns on a distinction familiar from ordinary language, once it is pointed out: between the individual (vyakti) and the universal or form (ākṛti).

• The world of individual beings — including deity-individuals — arises from the eternal word (अतः प्रभवात्), not the other way round. Śaṅkarācārya marshals genesis-passages as evidence: Prajāpati creates by first uttering (“eta iti ha vai prajāpatir devān asṛjata...”), and the Manu Smṛti states that Maheśvara first formed names and forms from the very words of the Veda (वेदशब्देभ्य एवादौ निर्ममे स महेश्वरः).

• Just as the word “cow” denotes an eternal universal (ākṛti/jāti) though individual cows are perpetually born and die, “Indra” denotes a standing office or form, occupied successively by different individuals across cosmic cycles — exactly as “commander-in-chief” (senāpati) names a role rather than one particular mortal man. Whoever currently occupies the post is called by that name.

• Śaṅkarācārya is careful to distinguish this “arising from the word” from Brahman's causal relation to the world as material cause (upādāna-kāraṇatva, discussed at BS 1.1.2). Given the word as eternally fixed in its meaning-relation (नित्यार्थसम्बन्धिनि), what “arises” is only the manifestation of individuals suited to that pre-existing linguistic usage — a subtler, non-causal sense of “arising” than ordinary production.

With this move, the deity-mortality objection dissolves without yet touching the deeper question of what, exactly, the eternal śabda underlying all this actually is.

The sub-debate: sphoṭa or phoneme?

Having secured that some eternal śabda underlies creation, Śaṅkarācārya turns to that harder question, weighing two positions.

(a) The grammarians' sphoṭa.

Individual phonemes obviously perish the instant they are uttered — each utterance produces and destroys them afresh (उत्पन्नप्रध्वंसित्वात्). To preserve a real, unitary, eternal word-essence, the grammarians posit a separate, partless entity, the sphoṭa, which the sequence of transient sounds “manifests” (vyajyate). Meaning, on this account, is grasped from the sphoṭa itself, not from the string of individually meaningless phonemes.

(b) Upavarṣa's varṇavāda, adopted by Śaṅkarācārya.

Bhagavān Upavarṣa's older position needs no extra entity: the phonemes themselves are eternal. The evidence, once again, is pratyabhijñā — we recognize the very same “g” every time it is spoken (सोऽयं गकारः, “this is that same g”), and this recognition is never overturned by any stronger means of knowledge (न प्रत्यभिज्ञानस्य प्रमाणान्तरेण बाधानुपपत्तेः). Apparent differences of pitch, nasalization, loudness or individual voice belong only to the manifesting sound or effort (abhivyañjaka-dhvani), not to the phoneme's own nature — exactly as a lamp's light is not itself altered by the different objects it happens to illuminate.

Śaṅkarācārya explicitly prefers Upavarṣa's account for its economy (laghīyasī kalpanā) — it requires no unobserved (adṛṣṭa) entity — over the sphoṭa theory's heavier ontological commitment (gurutarī kalpanā, दृष्टहानिरदृष्टकल्पना च, “abandoning the observed and positing the unobserved”). The eternal śabda that grounds the eternal word-meaning relation is, on this account, simply the eternal phoneme-type, recognized afresh across every utterance — not a mystical extra unit, and not the momentary individual sound-event either.

Pravāha-nityatva: eternity as a recurring stream

The adhikaraṇa continues, at Brahma Sūtra 1.3.29–30, to generalize this account to the Veda as a whole and to explain how the same Veda recurs at the start of every cosmic cycle (kalpa) in an identical sequence (anupūrvī) to the Veda of the cycle before — proceeding from Īśvara's will, but patterned after, not independently invented from, what came before. This is later named pravāha-nityatva: eternity as a beginningless, unbroken stream of recurrence, as distinct from kūṭastha-nityatva, a single static entity that simply is, without process or repetition at all. This distinction is the hinge on which the entire apparent tension with later Vedāntic texts turns — and, as Part 1 shows, the same hinge on which Advaita's redefinition of authorship turns.

The Vedānta Paribhāṣā: a sharper vocabulary for the same doctrine

Centuries later, the Vedānta Paribhāṣā states the position with scholastic precision, and at first glance it looks like a reversal of Śaṅkarācārya's own words:

अस्माकं तु मते वेदो न नित्यः, उत्पत्तिमत्त्वात्

In our view, the Veda is not eternal, because it has an origin. This is supported by the same śruti evidence used elsewhere in the tradition — the Veda as the out-breath (niḥśvasita) of the great Being — and by the reasoning that the Veda, as an ordered collection of phonemes, words and sentences (वर्णपदवाक्यसमुदाय), arises at creation and dissolves at pralaya exactly like ākāśa and the other elements.

But in the very next breath, the Paribhāṣā denies that phonemes are momentary (kṣaṇika):

य एव वेदो देवदत्तेनाधीतः, स एव मयापि

The very Veda recited by Devadatta is the very same one I recite — proven by the identical pratyabhijñā argument (सोऽयं गकारः) that Śaṅkarācārya had already used at Brahma Sūtra 1.3.28. Against the alternative — that each recitation produces a brand-new, momentary phoneme (dvikṣaṇāvasthāyitva) — the Paribhāṣā points out that this would require positing endlessly many distinct “g”-sounds, an unparsimonious burden (gaurava) the theory should not have to carry. As for why an unuttered phoneme goes unperceived: what is missing is only the manifesting condition, since utterance manifests phonemes rather than producing them — andhakārasthaghaṭānupalambhavat, “like a pot in a dark room, unperceived but not destroyed.” The ordinary cognition utpanno gakāraḥ is accordingly read either as simply mistaken (contradicted by the stronger pratyabhijñā) or as valid only with respect to the arising of the manifesting sound, which relates to the phoneme only indirectly.

So the Paribhāṣā is not undoing Upavarṣa's thesis of phoneme-eternity — it is using it, at the level of the individual phoneme-type. What it denies eternity to is something at a different level of description entirely: the Veda taken as a structured whole across cosmic time.

Three levels, not one

The apparent contradiction between “nitya śabda” (Śaṅkarācārya) and “vedo na nityaḥ” (the Paribhāṣā) dissolves once the discussion is separated into three levels that both texts track implicitly but never lay out side by side.

Śaṅkarācārya's bhāṣya, focused narrowly on refuting sphoṭavāda, operates almost entirely at Level 2 and speaks of “nitya śabda” there without pausing to distinguish it from Level 3. The Paribhāṣā's blunt vedo na nityaḥ is a statement about Level 3 alone — and it immediately reaffirms Level 2 (denying kṣaṇikatva) in the very next line, precisely so as not to be misread as a reversion to naive impermanence.

It is this three-level structure that lets Advaita hold utpattimattva and apauruṣeyatva together without contradiction — the technical move, resting entirely on the distinction between Levels 2 and 3 set out here, that Part 1 examines in full.

V. A Fourth Voice, Briefly: Vyākaraṇa's Sphoṭa

The grammarians' own elaboration of sphoṭavāda — most fully worked out by Bhartṛhari in the Vākyapadīya, where the sphoṭa is extended beyond the single word to the sentence as an indivisible unit of meaning (vākyasphoṭa) — is only glimpsed above through the lens of Śaṅkarācārya's refutation of it. A proper treatment of the grammarians' own case, and of how later Vyākaraṇa authors answered Vedānta's economy objection, would need its own textual basis in the Vākyapadīya and its commentaries rather than being reconstructed secondhand through an opponent's summary; it is flagged here as a genuine gap rather than filled in from general recollection.

VI. Three Schools, Compared on Both Axes

Read together with Part 1, the three fully-treated positions differ on both axes examined across these two essays, and converge only on the shared verdict that the Veda is authoritative.

What the table makes visible is that nityatva and apauruṣeyatva are not, despite appearances, two labels for one commitment. Nyāya shows that pauruṣeyatva and reliability can go together, provided the author is flawless. Mīmāṃsā shows that apauruṣeyatva and unqualified eternity can be made to entail each other, at the cost of a heavy metaphysics of static phonemes. Advaita shows the pair coming apart altogether: an authorless text that nonetheless originates, because what makes a text authored, on its technical account, is not that someone spoke it, but that the speaking was not itself a re-speaking of something prior.

VII. What the Argument Is Actually For

It is tempting to read two essays' worth of careful argument about phonemes, authors and deities as scholastic elaboration around a conclusion nobody was ever going to abandon — of course the Veda is authoritative, on every account; why does the machinery matter? That reading misses what actually drives the dispute. Dharma, and the connection between a ritual act (yāga) and its unseen fruit, are pramāṇāntarāgocara — reachable by no other means of knowledge at all. No perception, no inference, no ordinary testimony can independently confirm or disconfirm a claim about dharma; śabda, specifically Vedic śabda, is the sole access route into that entire domain. If that is so, then the question of whether that sound is eternal, and the question of who, if anyone, is speaking it, are not formalities appended to the text after the fact. Together, they are the entire question of whether anything in that domain can be known by anyone, ever.

Nyāya answers both questions by making the speaker maximally trustworthy: sound is a produced, changeable thing, but an omniscient author cannot be wrong, so his testimony transfers its reliability whole regardless. Mīmāṃsā answers both by removing the speaker and grounding the text instead in the static eternity of its own phonemes: with no person behind the words, no personal defect has anywhere to attach, and the text's reliability needs no external guarantor at all. Advaita answers by separating the two questions more finely than either rival does — sound is eternal at the level of the phoneme-type, produced anew at the level of structure, and the speaker himself never once speaks for the first time — and it is that separation, not the absence of an utterer, that carries the weight apauruṣeyatva is meant to carry elsewhere. Three answers, two entangled questions. The problem is real, which is why four traditions of serious philosophers thought the argument was worth having at all — and why the argument's texture, across both of these essays, is worth reading closely.

VIII. A Short Glossary, for Readers New to the Terms

IX. Sources and Editorial Notes

Primary sources drawn on

References follow APA 7th edition style, drawn from the specific print editions consulted (the same editions cited in Part 1). Details are given as printed on each edition's title page; where a detail could not be confirmed from that page, it is flagged rather than supplied from memory.

Thangaswami Sarma, R. (1996). Mīmāṃsāmañjarī (prathamaṃ kusumam). Indian Council of Philosophical Research. (Distributed by Munshiram Manoharlal Publishers)

Used for: §6 (pūrvamīmāṃsādarśane vedasyāpauruṣeyatvaṃ nityatvaṃ ca), pp. 22–26 — the Mīmāṃsaka case for eternal phonemes and its supporting evidence.

Dharmarājādhvarīndra. (n.d.). Vedāntaparibhāṣā [with ‘Prakāśa’ Hindi commentary] (G. S. Musalgaonkar, Comm.; S. S. Musalgaonkar, Ed.) (The Vidyabhawan Sanskrit Granthamala, No. 100). Chowkhamba Vidyabhawan.

Used for: the Āgama chapter, for the Advaita summary statements, the three-level analysis of nityatva, and its reconciliation with Śaṅkarācārya's bhāṣya.

Śaṅkarācārya. (1938). The Brahmasūtra Śāṅkara Bhāṣya with the commentaries Bhāmatī, Kalpataru and Parimala (Maha-M. A. Sāstrī, Ed.; re-ed. Bhārgava Sāstrī, 2nd ed.). Nirnaya Sagar Press.

Used for: Brahma Sūtra 1.3.28–30 (the Devatādhikaraṇa) — for the type/token resolution of the deity-mortality objection, the sphoṭa/varṇa sub-debate, and pravāha-nityatva.

Viśvanātha Nyāyapañcānana Bhaṭṭācārya. (1932). Kārikāvalī, saṃvalita with the [author's own] Nyāyamuktāvalī [commonly cited as Nyāya-siddhānta-muktāvalī], together with the Dinakarī and Rāmarudrī glosses (Ātmārāma Śarmā, Rev.; 2nd ed.). Nirnaya Sagar Press.

Used for: the Śabda-khaṇḍa, for the Nyāya pūrvapakṣa/siddhānta on the nityatva/anityatva of śabda as a guṇa of ākāśa.