Nobody Says Who Is Speaking: The Rigveda's Dialogue Hymns and a Theatre Nobody Can Prove

A Quarrel With No Name Tags

RV 1.170 runs to five verses, at least three people talk in it, and the text never says who any of them are.

The opening verse is a piece of gnomic obscurity that translators have been fighting over for a century and a half: it is not now, it is not tomorrow, who knows what is wondrous? Then, with no transition and no introduction, two more voices:

“The Maruts are your brothers. Why, Indra, do you want to take our lives? Settle with them properly; do not strike us down in the encounter.”

“Agastya, brother, why do you slight us, you who are our friend? We know perfectly well what is in your mind: you do not intend to give us anything.”

Rigveda 1.170, verses 2 and 3, consecutive and by different speakers. After Griffith (1896).

By verse four somebody is ordering an altar prepared and a fire kindled, and by verse five the quarrel has been dissolved into diplomatic language about speaking kindly and tasting the oblations in their season.

So: a god, a war band of storm gods, and a human poet, in what looks like the tail end of a fight over who gets a share of a sacrifice. The hymn tells you none of that. Every speaker label you see in a modern translation, every bracketed [Agastya:] or [The Maruts:], was put there by an editor who reconstructed it from the verb endings, the pronouns, and the traditional index that assigns the hymn to a seer. The Sanskrit has no punctuation, no quotation marks, no dramatis personae, and in this hymn not one verb meaning “he said”.

There are roughly fifteen hymns like this in the Rigveda, and they are the strangest thing in the collection. Nearly all of the rest is a voice addressed upward, a poet speaking to a god who does not answer inside the poem. Then the mode breaks, and you get a dog bargaining with a cartel of hoarders, a woman arguing a point of cosmic law with her twin brother, a fire god negotiating the terms of his employment, and a wife opening a hymn with a complaint about an absent relative. Several of these hymns carry the only female speech the collection preserves, which is the subject of Hidden Women of the Rigveda. From 1883 onward, scholars began to suspect that these hymns were not poems at all but the surviving fragments of performances, and in 1908 one of them published a book arguing that the Rigveda contains the oldest plays in the world. The argument failed. What it turned up on the way is far more interesting than the thesis it was trying to prove.

The Text Has Almost No Way of Saying “He Said”

Classical Sanskrit marks quoted speech with the particle iti (इति), placed after the quotation like a closing bracket: gacchāmi iti abravīt, “I am going, he said”. Open any page of the Mahābhārata, any Purāṇa, any commentary, and iti is everywhere. It is the workhorse of Indian narrative prose.

In the Rigveda it is almost absent. Counting through the romanized Saṃhitā text that this site is built on, iti appears as a separate word twenty-one times in 10,528 verses, with roughly twenty more instances fused by sandhi onto the word in front of it. Call it about forty, or one quotative marker for every two hundred and sixty verses. The verbs are rarer still. The perfect āha, “he says, he said”, occurs ten times; uvāca, “said”, four times; the aorist abravīt and its relatives three times. Seventeen narrative speech-framing verbs in the whole collection.

Methods note. These counts come from the romanized Rigveda text in this repository, derived from the Sanskrit edition at sacred-texts.com, tokenised and normalised before matching. Two cautions. First, Vedic sandhi regularly fuses iti to the preceding word, so any word-level count undershoots; the fused forms were identified by inspection and the figure is approximate. Second, editions differ slightly in verse division, which is why the verse total here is 10,528 rather than the commonly cited 10,552. The point is the order of magnitude, not the decimal: the Rigveda’s poets had the particle and hardly used it.

The exception is worth looking at, because it shows what the Rigveda sounds like when it does bother to attribute. RV 1.161 is a hymn about the Ṛbhus, three artisan brothers who were set a test: take one sacrificial cup and make four. The hymn recalls the argument they had while working.

āpo bhūyiṣṭhā ity eko abravīd agnir bhūyiṣṭha ity anyo abravīt

“‘The waters are most abundant,’ said one; ‘Agni is most abundant,’ said another.”

Rigveda 1.161, verse 9. After Griffith (1896), with diacritics standardised.

Two speakers, two quotations, two verbs of saying, all inside a single line. The poet could do it. The verse before this one quotes an offer made to the brothers directly: drink this water, or if you will not, drink the rinsings of the muñja grass. When a Rigvedic poet is telling a story about other people, the machinery for marking speech is there and he uses it.

Which makes the dialogue hymns genuinely peculiar. In those, the machinery is switched off. The speakers are not introduced, not named, not quoted; they simply talk, one after another, and the only clue to a change of mouth is a verb that has shifted person or a pronoun that has changed its referent. A reader coming to RV 10.108 cold has to work out from the grammar alone that the voice saying “I am the messenger” belongs to a dog, and that her interlocutors are a group of cattle thieves, a story told at length in our piece on Saramā and the Paṇis.

The Hymns That Need More Than One Mouth

Tradition has a term for these texts: saṃvāda (संवाद), “conversation, colloquy”. It is not a Rigvedic word in that sense; it comes from the later ritual and exegetical literature looking back at the hymns and noticing that some of them behave differently. Lists differ on the margins, because a hymn with one quoted line is not obviously the same kind of object as a hymn that is nothing but quoted lines. Strict counts run to about fifteen; generous ones reach twenty.

Two features of that list matter. The first is where the hymns sit. Thirteen of the sixteen entries above fall in Maṇḍalas 1 and 10, the two books that linguistic and metrical evidence place latest in the collection; the exceptions are RV 3.33 in the old Viśvāmitra family book, RV 4.18, and RV 8.100. Whatever the dialogue form is, the Rigvedic tradition seems to have done most of it late, or to have admitted it into the canon late, which are different claims and hard to separate. On how the collection was assembled and what its ordering gives away, see Someone Sorted the Rigveda.

The second is that most of these hymns have no ritual job. The Rigveda survived because it was useful: its verses are the raw material of the śrauta liturgy, which is why the collection was memorised to the syllable for three thousand years, as described in the account of its transmission machinery. But a hymn in which an abandoned king snarls at a departing nymph that women’s hearts are the hearts of hyenas, a line whose speaker some readers would reassign to a chorus, does not have an obvious slot in a soma pressing. These texts were preserved despite being liturgically awkward. Something made them worth keeping.

And the stakes of guessing wrong are not merely literary. RV 10.124 is nine verses in which somebody defects. A god announces that he is abandoning his old allegiance and crossing to the other side, and in the course of the hymn the Asuras lose the form-making power called māyā, discussed in Māyā Before It Meant Illusion, while Indra is chosen as sovereign in place of the old order.

“I stayed many a year inside this altar. I leave the Father; my choice is Indra. Away pass Agni, Varuṇa and Soma. Lordship changes hands: I come over to the side that is rising.”

Rigveda 10.124, verse 4. After Griffith (1896), lightly modernised.

Who says that? Agni, on most readings, speaking to Varuṇa about Varuṇa’s own displacement. But verse 5 either continues in the same mouth or switches to Varuṇa’s, and the two options give you different theologies: in one, a deposed sovereign is invited to accept the new arrangement, in the other the defector is bidding for the throne himself. The hymn is one of the few places in the Rigveda that seems to register the great Indo-Iranian realignment in which asura slid from a title of honour to a word for demons. Our best evidence for how the poets understood that shift depends on a speaker attribution that the text declines to supply.

Oldenberg’s Missing Prose

Hermann Oldenberg proposed an answer in two papers, in 1883 and 1885. [1][2] His idea was that the dialogue hymns are not whole works at all. They are the verse skeletons of a mixed form, a narrative told in prose with the characters’ speeches fixed in metre, of a type attested all over later Indian literature, from the Jātakas to the ākhyāna sections of the Brāhmaṇas. In that form the prose is the connective tissue, improvised or loosely retold at each performance, while the verses carry the emotional weight and are transmitted word for word. Fix the verses in a canon, drop the prose, and you get exactly what the Rigveda preserves: speeches without situations.

The theory explains a great deal at once. It explains the absence of attribution, because a narrator supplied it out loud. It explains the hymns’ allusiveness, their habit of referring to events they do not describe. It explains why the tradition kept them even though the liturgy had little use for them, and it explains what happened next: the Śatapatha Brāhmaṇa supplies the Purūravas story in prose a few centuries later, in a version that bends the hymn toward a ritual it wants to establish, a process examined in the piece on RV 10.95. The prose did not vanish; it was never fixed in the first place, so each generation made it again.

Paul Horsch’s 1966 study of the Vedic gāthā and śloka literature carried the argument forward into the whole stratum of semi-metrical narrative, and the clearest Vedic specimen is not a Rigvedic hymn at all but the story of Śunaḥśepa, the boy bought for sacrifice who talks his way out of the stake by composing hymns. [3] It survives in the Aitareya Brāhmaṇa as alternating prose and verse, and the tradition prescribes its recitation inside a royal consecration rite. [4] A story performed aloud, inside a ritual, by a priest with a speaking part: Oldenberg’s missing genre is not hypothetical in the Brāhmaṇa period. The question is whether it existed four or five centuries earlier.

A. Berriedale Keith thought not, and said so with some asperity in 1911. [5] His central objection was philological: the narrative Oldenberg leaned on, the Pauṣyākhyāna, turns out to be prose throughout, dialogue included, with only a few verses embedded in it, so it is not evidence for the mixed form at all. Keith went on repeating the objection for two decades, and his 1924 history of Sanskrit drama treats the whole family of Vedic-origin theories as a monument to wishful reading. [6] Karl Geldner, who had been sympathetic, later read at least one supposed ākhyāna hymn as a single connected poem instead.

1908: The Year the Rigveda Got a Cast List

Johannes Hertel pushed the case further in 1904, arguing that the dialogue poems were composed for delivery by several performers. [7] Leopold von Schroeder then went the whole distance. Mysterium und Mimus im Rigveda, published in Leipzig in 1908, translated a selection of the dialogue hymns, assigned every speech to an actor, and presented the results as the scripts of a mystery cult older than the fire ritual itself. [8] Schroeder did not stop at India. He connected the Maruts and their leader to the armed dances of the Corybantes, to Cretan and Italian ritual dance, to the sword dances of Germanic Europe, and read the whole complex as a single inherited religious drama.

The book is a period piece, and the period explains it. European scholarship around 1900 was saturated with the idea that drama grows out of ritual. Frazer’s comparative anthropology was at its height; Jane Harrison and the Cambridge ritualists were deriving Greek tragedy from seasonal rites; William Ridgeway was deriving it from cults of the dead. Keith returned to the subject in 1912 with a paper on the origin of tragedy and the ākhyāna written expressly as a reply to Ridgeway, which tells you how closely the Vedic and the Greek arguments were being run in parallel. [20] If tragedy came out of ritual in Athens, and the oldest ritual poetry anywhere happened to contain dialogues, the conclusion looked irresistible. Schroeder’s contemporaries had already rejected his separate claim that Pythagoreanism came from India, and the mystery-play thesis went the same way.

Aside. It is easy to be condescending about this, and worth resisting. Schroeder was wrong about the plays, but he was responding to something real in the text that nobody had accounted for, and his method of assigning speeches is still the first thing any editor of these hymns has to do. The modern translations that print [Indra:] and [The Maruts:] in brackets are doing a sober, minimal version of what he did floridly. The difference is that they call it a reconstruction.

graph TD
    A[Dialogue hymns noticed] --> B[Oldenberg 1883-85: lost prose]
    B --> C[Hertel 1904: several performers]
    C --> D[Schroeder 1908: mystery plays]
    D --> E[Keith 1911-24: no akhyana, no drama]
    B --> E
    E --> F[Horsch 1966: Vedic gatha literature]
    E --> G[Lidova 1994: drama grew with puja]
    F --> H[Jamison 2007: poetry as performance]
    G --> H
    H --> I["D'Intino 2018: the speaking persona"]

What Vedic Ritual Actually Staged

Strip away the plays and a residue of hard evidence remains, and it is not small. Vedic ritual, in the layer of texts just after the Rigveda, unmistakably contained scripted speech for more than one speaker.

The brahmodya is the clearest case: a formal exchange of riddling questions and answers between priests, with the questions fixed, the answers fixed, and the whole thing timed to a point in the rite. The verbal duel has a deeper history than the ritual texts, and F. B. J. Kuiper argued in 1960 that an inherited institution of competitive speech lies behind a great deal of Rigvedic diction. [9] The Rigveda’s own riddle hymns look like the literary residue of that practice, and so do its impossible statements and its deliberate absurdities, collected in The Hare Swallowed the Razor.

The mahāvrata, an annual rite at the turn of the year, went further. Among its components is an agonistic scene between two assigned roles, an abhigara whose job is to praise and an apagara whose job is to revile, the two trading set abuse at each other in front of an audience of participants. Paola Rossi’s recent study of the terms, which barely occur in Vedic literature outside this rite, reads the scene as a formal contest with the violence channelled into speech. [10] Elsewhere in the same ceremonial complex there is antiphonal obscenity between a celibate student and a courtesan, and in the horse sacrifice a round of ritual insult among the officiants and the royal women, material analysed at length in Stephanie Jamison’s study of women in Vedic ritual. [11]

So the Vedic world had assigned speaking roles, scripted abuse, question-and-answer set pieces, and an audience. What it does not have, anywhere in the evidence, is a stage, a costume, a character distinct from the performer, or a plot carried by impersonation. That last absence is the one that matters, and it is why the drama hypothesis fails even though the performance hypothesis does not. Ritual speech in two voices is not theatre. It is ritual speech in two voices, which is a remarkable thing to find in a Bronze Age liturgy and did not need to become drama to be worth noticing.

The Tradition’s Own Origin Story

Indian theatre, when it finally appears in documentary form, claims descent from exactly this material. The Nāṭyaśāstra, the great treatise on dramaturgy compiled somewhere in the range of the second century BCE to the second century CE, opens with an origin myth: the gods ask Brahmā for a diversion that can be both seen and heard, and Brahmā composes the Nāṭyaveda, a fifth Veda, drawing recitation from the Ṛgveda, song from the Sāmaveda, the art of enactment from the Yajurveda, and rasa, aesthetic flavour, from the Atharvaveda. [12] The new Veda, unlike the four older ones, is open to everyone.

That is a claim about authority, not a report about history. A new art form in a culture that grounds authority in Vedic descent will produce a Vedic genealogy for itself, and the specific division of labour in the myth (recitation here, song there, gesture elsewhere) reads as a tidy retrofit. But it is also a datum. The dramatists did not derive their art from court entertainment or from foreign models, even though Greek contact was available to them; they derived it from the hymns. Something in the tradition’s memory connected performed poetry to the Veda.

Modern scholarship has mostly declined the genealogy. Natalia Lidova’s study of the question argues that classical Sanskrit theatre grew up not out of the yajña but alongside pūjā, the image-centred worship that displaced the fire cult, and that dramatic representation of divine figures in that setting is a better ancestor for the stage than anything in the śrauta ritual. [13] On any account there is a gap of something like a thousand years between the last dialogue hymn and the oldest Sanskrit play we can actually read, which survives as manuscript fragments of a drama attributed to Aśvaghoṣa, recovered from Central Asia and first edited in 1911. [6] A thousand years is a long silence to bridge with a hypothesis.

What the Hymns Do Instead

Give up the theatre and look at what is left, because what is left is better than a lost genre.

These hymns do something the rest of the Rigveda does not attempt: they characterise people by how they talk. The Maruts in RV 1.170 whine; their grievance is petty and exactly the grievance of a subordinate war band that suspects it is being cheated of its share. Yamī in RV 10.10 argues like a lawyer, pressing precedent and necessity, while Yama answers in the register of someone retreating behind the rules. Urvaśī deflects. Saramā, offered a bribe, is contemptuous. And in RV 10.28, a hymn otherwise occupied with soma and bulls, the opening line is spoken by a woman with a domestic complaint:

viśvo hy anyo arir ājagāma mamed aha śvaśuro na jagāma

“Everyone else has come, even the stranger. Only my own father-in-law has not come.”

Rigveda 10.28, verse 1. Translation mine from the Sanskrit; compare Griffith (1896). On the kinship vocabulary behind this line, see the Rigveda’s in-laws.

Nobody introduces her. Nobody explains why her irritation opens a hymn to a god. The voice simply arrives, does its work in one line, and gives way.

The strongest modern readings have stopped asking what genre these hymns belong to and started asking what the poets were doing with voice. Stephanie Jamison’s lectures on the Rigveda treat the hymns as verbal performances embedded in a ritual world, texts whose structure encodes their own delivery. [14] Silvia D’Intino has argued that a group of Varuṇa hymns builds a speaking “I” that is neither the poet nor a simple lyric voice but something held at a distance from both, and that this distancing is the ancestor of the dramatic character, more than any dialogue is. [15] On that reading, the thing the Rigveda invented is not theatre but the detachable persona: a voice that can be worn.

Aside. There is a methodological trap here worth naming. Once you decide the hymns are scripts, every obscurity becomes a missing stage direction, and the theory starts explaining everything, which is the point at which it stops being testable. The honest position is narrower. The dialogue hymns presuppose an audience that knew a story the text does not tell. Whether somebody acted that story out, or a narrator simply told it between the sung verses, the Rigveda does not say, and there is no evidence left that would settle it.

The attribution gap, meanwhile, turns out not to be an accident. Vedic transmission preserved what was metrically fixed. The accents were preserved, the syllable counts were preserved, the word-by-word analysis was preserved, every feature that could be drilled into a student was preserved with a fidelity that has no parallel in the ancient world. What was not preserved is everything that lived outside the metre: the prose, the gestures, the identity of the mouth. A tradition that could keep a pitch accent for three millennia dropped the cast list, because the cast list was not part of the text. It was part of the occasion.

Read It With Three People in the Room

Here is the experiment that makes the whole argument tangible. Take RV 1.170 in any translation, hand verse 2 to one person, verse 3 to a second, verses 1, 4 and 5 to a third, and read it aloud. Five verses, maybe ninety seconds. The hymn that looks like a knot of obscurity on the page resolves into something a reader of plays recognises instantly: a confrontation, a complaint, a mediation, a settlement. Then try RV 10.10 with two voices and notice how much of its force comes from the fact that neither speaker concedes.

The casting that makes the experiment work is the standard reconstruction, inferred from the verb persons and from the traditional ascription of the hymn to Agastya. It is a reconstruction and not a reading: nothing in the Sanskrit assigns a single line.

graph LR
    V1[Verse 1: gnomic brush-off] --> V2[Verse 2: plea for the Maruts]
    V2 --> V3[Verse 3: the Maruts complain]
    V3 --> V4[Verse 4: prepare the altar]
    V4 --> V5[Verse 5: settlement formula]
    A[Indra speaks] --- V1
    B[Agastya speaks] --- V2
    C[Maruts speak] --- V3
    B --- V4
    B --- V5

Nothing in that experiment proves that anyone in the second millennium BCE did it. What it does prove is that the material supports it, which is a fact about the poems and not about the theory. The Rigvedic poets, working inside a tradition built to praise, found room in it for speech that argues, wheedles, refuses, jokes and walks out. They did this with no narrator, no attribution and no way of marking a quotation, and the result is still legible three thousand years later, which is a technical achievement of a high order.

The silence about who is speaking has one last implication, and it is the kind of thing the Rigveda does on purpose elsewhere. A text that does not name its speakers makes the reader do the identifying, and a reader who has to work out who is talking is drawn into the hymn’s world in a way that a labelled script never achieves. The collection is full of this sort of deliberate withholding, as the survey of its vocabulary of concealment shows in What the Rigveda Hides. Whether the missing labels were a choice or an accident of transmission, the effect is the same. You are handed a conversation and left to work out who is in the room.

Open RV 10.95 next, and read it as what it is: eighteen verses of two people failing to persuade each other, with no narrator, no scene-setting and no verdict. It is a scene built out of nothing but speech, composed in a culture that has left us no trace of a stage, and nobody in it is named until you name them yourself.