One of the conundrums of ancient epic poetry, both the Indian and the Greek varieties, is the question of how they came to assume their canonical form in the first millennium BCE. Not only literary criticism is at stake: a great deal of nationalist rhetoric depends on the origin of the great epics and the language used to compose them.

Both the Greek and Indian epic poems were both originally orally transmitted, since at the time of their composition there was no writing system in use1. The oral transmission of the epics creates a problem because we have no record of the development of the text. So instead, scholars turned to linguistic analysis and the archaeological record to try and separate the original core of the epics from later accretions. However neither of these methods proved effective.

While the language of Homer is (like classical Sanskrit) a literary one that was never used colloquially, under the “light patina of Attic forms” there is “an indissoluble mixture of two different dialects, Aeolic and Ionic. But the attempts of the linguists to use this criterion for early (Aeolic) and late (Ionic) ran into the dilemma that Aeolic and Ionic forms sometimes appear inextricably tangled in the same line or half line” (The Iliad, introduction by Bernard Knox, Penguin 1990 p13).

Analysis based on the archaeological record fared no better. Although the events of The Iliad would have occurred in the Bronze age and the majority of the epic describes weapons made of bronze, “in the fourth book the Trojan archer Pandarus has an iron arrowhead, mentioned quite casually as if that were normal… in this passage iron is obviously cheap… elsewhere we meet proverbial phrases like ‘heart hard as iron,’ which indicate complete familiarity with the metal. It certainly looks as if these are different historical layers, but once again, there is no way to extract them.” (ibid)

This problem was solved by an American scholar named Milman Parry. He noted the insight of German analytical scholars that there were certain ornamental epithets used to describe the characters in The Iliad which made it easy to meet the demands of the metre of Greek heroic poetry, the dactylic hexameter. For example, “Achilles is ‘brilliant,’ ‘godlike,’ or ‘swift-footed,’ Apollo is ‘one who shoots from afar,’ the Achaeans are ‘strong-greaved,’ or ‘bronze-cloaked,’ Hera is ‘white-armed’ and ships are ‘black,’ ’round,’ ‘hollow’ or ‘swift’” (ibid). There are also a set of stock formulaic phrases padding out the text.

Parry’s breakthrough was to demonstrate that “the system was more extensive and highly organized than anyone had dreamed, and he also realized what it meant. It meant that the system had been developed by and for the use of poets who improvised… The oral bard who uses such formulaic language is not… a poet reciting from memory a fixed text. He is improvising, along known lines, relying on a huge stock of formulaic phrases, lines and even whole scenes; but he is improvising. And every time he sings the poem, he does it differently. The outline remains the same but the text, the oral text, is flexible. The poem is new every time it is performed” (ibid, p17). Exactly the same process can be seen at work today in performances of Indian classical music.

This theory explains the lack of consistency in the archaeological record and the linguistic tone of the Greek epics. Epithets which include references to diverse material cultures and use vocabulary from different dialects are chosen because of their fit with the metre. Indeed it renders the epics virtually useless as a detailed guide to the culture and society of the people represented; it tells us rather more about that of the poet.

The same theory can also be applied to Indian epics such as the Rāmāyaṇa2, of which books two to six of the canonical form are almost certainly the work of a single bard (sūta) known as Vālmīki. It is widely accepted that the story of the Rāmāyaṇa existed long before Vālmīki composed it, and since Vālmīki there have been innumerable retellings of the story. Like the Greek epics, it contains “iteration, formulaic composition, simple metrical forms preferably subject to musical or quasi-musical recitation, copiousness, heavy use of epigrams and sententia, hyperbole and tales of wonder.”3 Like Homer, Vālmīki was drawing on a tradition of learned patterns to create his work.

In Indian oral tradition in particular, the use of stock phrases serves another important purpose: memorisation. This is especially important in works like the Rāmāyaṇa which have religious importance. Probably the most extreme example of this can be found in the Pali canon, the massive collection of the Buddha’s sermons preserved for more than four centuries by recitation. This was no doubt partly due to the Buddha’s pedagogical style. But the frequent, long repetitions of stock phrases found in the canon also served another purpose: they served as a mnemonic aid to memorisation. Although the intention was “to preserve the Master’s words as accurately as possible… it should also be remembered that it was not all a mere matter of mechanical repetition, though this undoubtedly occurred occasionally too.” (Dīgha Nikāya, trans. Maurice Walshe, Wisdom 1987)

Oral epics are still alive in India today. In “the Singer of Epics”, William Dalrymple tells of his experiences researching Rajasthani performers of local epics, known as bhopas. These illiterate villagers still perform epics thousands of stanzas long, each taking several nights to play out. Because these epics are sacred, they are performed rote. Dalrymple reports that during a performance

I asked another guest, who understood Mewari … if he could check Mohan Bhopa’s rendition against a transcription by John D. Smith, of Cambridge University, of a version performed in a different part of Rajasthan in the 1980s. Give or take a couple of turns of phrase, and the occasional omitted verse, the two versions were nearly identical, he said. And there was nothing homespun about Mohan Bhopa’s language, he added. It was delivered in incredibly fine if slightly archaic courtly Mewari diction. (Nine Lives, Bloomsbury 2009, p97)

Socrates, in his work Phaedrus, reports the King of Egypt saying of Thoth’s gift of writing: “it will set forgetfulness in the minds of learners for lack of practice in memory.” This is exactly what happened to the Serbian poets and is happening to their Indian cousins: “Just as the blind can develop a heightened sense of hearing, smell and touch to compensate for their loss of vision, so it seems that the illiterate have a capacity to remember in a way that the literate simply do not. It was not lack of interest, but literacy itself, that was killing the oral epic” (Dalrymple, idem, p95). However the techniques used to create them—the use of rote-learned patterns to enable memorisation and improvisation—can be seen throughout the creative arts, martial arts, and science (including computer programming).

This points to the hypothesis that this mechanism, the chunking up of memorized patterns in order to create larger scale works, is fundamental to human creativity. I’ll be exploring this hypothesis in future posts.


1In the case of the Greeks, the only native alphabet, Linear B, was developed by the Mycenaeans for administrative purposes, and fell out of use in the twelfth century BC with the destruction of the Mycenaean civilisation. The adaptation of the Phoenician alphabet by the Greeks only shows up in the archaeological record by the eighth century BC. Sanskrit was also originally a purely oral language. The oldest evidence of written Sanskrit turns up in about the first century BC using the Brahmi script, which at that time had already been used in India to write Asoka’s rock edicts. The most widely accepted hypothesis is that Brahmi is also descended from the Phoenician alphabet (as is almost every script currently in use, from Arabic to Thai), and indeed some early Brahmi inscriptions are written from right to left.

2The Mahābhārata is considered itihāsi, or history (although its historicity is not its most compelling attribute), rather than kāvya or poetry, and its structure is rather more complex and heterogeneous. It is relatively straightforward to decompose this work into strata (the Bhagavad Gitā, for example, in the form that exists today, is composed in a style of Sanskrit that can be dated to around the 3rd century CE), and doing so reveals fascinating insights into the development of Indian society.

3There is an additional constraint imposed on Indian epics due to their being subject to the rules of Nātyā, which can be roughly understood as the dramatic arts. As a result, the stock epithets, metaphors and figures of speech must evoke the correct rasa or mood, of which one of the most important in kāvyais viraha or love in separation—in this case Rāma’s feelings for his abducted wife, Sītā, which are the subject of some of the most evocative verses in the work.