In a number of places, Analayo supports the contention that prose portions of certain texts, such as the Udāna, are later than their corresponding verses, on the grounds that the verse uses a metaphor which the prose takes literally. For example in an essay on a text of the Udāna (Anālayo (2012). “Dabba’s Self-cremation in the Saṃyukta-āgama.” Buddhist Studies Review 29.2, 153–174) he says:
… at times a stanza in the Pāli Udāna that evidently has a symbolic sense is accompanied by a prose narration that reflects a literal understanding of this stanza. An example that illustrates this is a stanza that inquires:
What is the use of a well,
if water is there all the time?
Having cut craving at its root,
what would one go about searching for?
While the stanza obviously uses the image of the well as an illustration, the prose that purports to record the original occasion when the stanza was delivered reports that Brahmins of a particular village had blocked a well with chaff in order to prevent the Buddha and his monks from drinking. When the Buddha requested Ānanda to fetch some water for him, the well by itself threw up all chaff and was filled with clean water to the brim. Pande (1957, 75) comments that ‘the author of the prose … seems to have grossly misunderstood the … verse, which intends “water” in no more than a merely figurative sense’. The Chinese parallel to this stanza in fact does not mention any such incident and instead provides an interpretation of the well imagery.
But this doesn’t seem persuasive to me at all. Of course a metaphor draws on a concrete image to make a more abstract point. That’s literally what a metaphor is. How else are they supposed to work? It’s one of the most characteristic differences between prose and poetry: prose is more literal, poetry is more metaphorical. Why are we surprised to see this?
I think it’s obvious in a text like the Udāna that the verses are, in fact, generally older than the prose. And I agree that there is a tendency to over-literalize things. And maybe that’s what has happened in these cases. But to me there is nothing unusual here, nothing that demands a historical explanation. It’s just … how language works. I don’t see how we can infer anything from this about the relative ages of texts.