Discover more from Horace & friends
Often when I’m reading contemporary poetry in English I feel impatient with what feels like a kind of complacency, a readiness to be satisfied with mere description. I want poetry that does more than just describe, even precisely and beautifully, because the best prose does that too, and arguably (often) better or at least in a context that is more engaging and sustained. I want music, some sort of structure aside from and in counterpoint with that imposed by the ordinary rules of English syntax, the pleasure of form or of narrative (or both), and some tonal interest. I want to have that feeling of irreducibility and completion.
Or perhaps — sometimes — I just want it to be more in a way that is not really possible in English and the feeling of fretful dissatisfaction is linked to linguistic constraints. Some of the most perfect poetry I know, after all, is largely ‘pure description’. Kalidasa’s Meghadūta, written probably about 1600 years ago, does have a “plot” of a kind, but it’s a pretty basic one: a certain yaksha (a kind of nature spirit) is living in exile on Rāmagiri in the Vindhya mountains. After eight months, he sees an enormous cloud on the peak of the mountain. Realising that the cloud will move north, towards his home and his beloved wife, he asks the cloud to be his messenger (Meghadūta means ‘cloud-messenger’) and convey a message of love to her. Essentially the entire poem — 121 four-line stanzas, all in the same metre1 — is taken up with his description of the route the cloud should take and how he should identify his home when he gets there.
Compared to most (all?) contemporary poetry, this is an extravagantly artificial poem, in which nothing actually happens. But it is both ravishingly beautiful and profoundly satisfying — satisfying in a way that makes 99% of contemporary work feel like the merest froth. As Goethe remarked, ‘The English translator of the Cloud Messenger, the Meghaduta, is deserving of all honour, since a first encounter with a work of this sort marks a date in our lives forever’.2
Kālidāsa’s verses have been described as like individual pearls in a chain, each perfect on its own, and indeed they are often quoted in isolation. So perhaps, despite all the linguistic difficulties, its worth attempting to look at just one in a bit of detail. Here is the image of a single verse from the Meghadūta, followed by a transliteration (don’t worry about the various diacritics):3
gatvā cordhvaṃ daśamukhabhujocchvāsitaprasthasaṃdheḥ
kailāsasya tridaśavanitādarpaṇasyātithiḥ syāḥ /
śṛṅgocchrāyaiḥ kumudaviśadair yo vitatya sthitaḥ khaṃrāśībhūtaḥ pratidiśam iva tryambakasyāṭṭahāsaḥ // (verse 61 in my edition)
Those dauntingly long runs of letters largely represent compound words, which are particularly characteristic of Sanskrit in general, and high-register courtly Sanskrit in particular. Kālidāsa is actually fairly restrained in his compounding compared to many subsequent poets, but all the same this is one of the strangest features of the literary language for native English speakers. (I wrote a little bit about Sanskrit compounds in this earlier piece about translating Sanskrit.) Here is an image of the four lines — represented in this case as two long lines — with the compounds broken down into their constituent parts:4
So here you can see, for instance, that the long word in the first line, daśamukhabhujocchvāsitaprasthasaṃdheḥ, is made up of five main elements — daśamukha (meaning ‘ten-headed’), bhuja (meaning an arm or hand), ucchvāsita (the past participle of a word meaning to gladden, lift or raise), prastha (meaning a plain or the table-land at the top of a mountain) and samdhi (meaning connection, junction or joint). So the whole thing means something like ‘the joint of the top of the mountain having been raised by the arm of the ten-headed one’. The whole thing is in the genitive — the -eh ending of the word is the masculine genitive singular, and it all agrees with Kailāsya, also in the genitive, which is the name of the mountain, traditionally identified as the home of Śiva.
The ‘ten-headed one’ refers to Rāvaṇa, the demon king who is the villain of the Rāmāyaṇa. This verse is alluding to the story of how Rāvaṇa attempted to move Mt. Kailāsa to Lankā, ripping it from its foundations and alarming Pārvatī and the other residents of the mountain. Śiva pressed down on the mountain, crushing Rāvaṇa, who only escaped by propitiating Śiva. This kind of allusive density, in which the briefest references, even individual words, draw in myths which are not related is typical of Kālidāsa’s style, as it is of course of many other ‘classic’ texts, from Pindar to T. S. Eliot.
Using the epithet daśamukha for Rāvaṇa, without naming him, has a kind of indirection that you see a lot in Pindar, for instance. But in this line it also sets up a kind of chime with tridaśa, the element that begins the second compound. Daśa means ten and tridaśa means thirty, a shorthand for the ‘thirty deities’, or devas, a category of being generally opposed to the rākśasas and asuras, which are more like demons (Rāvaṇa was a rākśasa). The second compound goes on to specify that the vanitā (woman or wife) of the devas are referred to here — the female gods, who, looking down from heaven, use the bright peak of the mountain as their darpaṇa, or mirror.
My old student edition, edited by M. R. Kale and first published in 1916 (the world of Sanskrit student editions is a slow-moving one), offers the following translation:
Soaring still higher, you should be the guest of (the mountain) Kailāsa, the joints of whose peaks were loosened by Rāvana, which serves as a mirror to the females of the gods, and which stands occupying the sky with its peaks white like lotuses, as if it were the loud laugh of Śiva accumulated day by day.
In this translation, the two long relative phrases ‘the joints of whose peaks were loosened by Rāvana’ and ‘which serves as a mirror to the females of the gods’ are both rendered in the original Sanskrit by single compound words agreeing with Kailāsa, the name of the mountain: daśamukhabhujocchvāsitaprasthasaṃdheḥ, which we’ve already looked at, and tridaśavanitādarpanasya. The syntax of this first long line (or couplet, in the earlier transliteration) is actually very simple — aside from the opening absolutive phrase (an indeclinable form of the verb used widely in Sanskrit), there’s a single main verb syāḥ, in the optative (‘you should be’); a single nominative, atithi (‘guest’); and three genitives indication what you should be a guest of.
Any clear English translation is pretty much compelled to translate at least one of these compounds as a separate clause. Kale’s use of two relative clauses (the joints of whose . . . which serves) is helpful for anyone picking their way through the Sanskrit (because that’s generally what he does with compounds), but it’s very stilted in English, and also completely loses the effect of the contrast with the next line (or couplet), which actually does have a relative clause in Sanskrit. Here the mountain becomes the subject — yo, [the mountain, which].
Here’s a quite stylish more recent translation, from an interesting book by E. H. Rick Jarow, on the Meghadūta as a kind of eco-poetry. This translation does a good job of avoiding the awkward pile-up of relative clauses: Jarow puts the shorter and simpler of the compounds as a phrase in apposition (the mirror of celestial nymphs), and offers the reader a little more explanation of the myth alluded to in the first compound, by using a relative clause starting where:
And rising higher, you could be the guest of Mount Kailāsa,
the mirror of celestial nymphs; where the ten-headed Rāvaṇa’s cracked
the joints of its lofty peaks. It spread open like gleaming pure-white lotuses
spanning the sky, as if the howling-snow laughter of the Three-Eyed God
is heaped upin all directions.43
I like the solution here for the part of the verse, but I think the second half is actually less effective than Kale’s version, which retains more of the vivid strangeness of the Snaskrit, and also more of its syntax. The Sanskrit doesn’t say that the mountain — now the subject — is spread across the sky like lotuses. It says it is stood (stitaḥ), covering the sky with or by the bright lotuses of its pointed peaks. Jarow’s translation then changes the subject to the ‘howling-snow laughter’ — a rather good phrase, and a decent explication of the verse, but the Sanskrit line never specifies the snow. The two key words here rāśībhūtaḥ meaning ‘piled up’ and tryambakasyāṭṭahāsaḥ meaning ‘the loud laughter of the one with three eyes [i.e. Śiva]’ are linked to the mountain by iva, like — the mountain, filling the sky with its bright peaks, is like the loud laughter of Śiva, piled up day by day.
This wonderful comparison rests upon a Sanskrit convention that associates laughter with brightness and gleaming (of the exposed teeth, I suppose). Laughter is bright, and of course this laughter is piled up day by day because the unexpressed term is that of the snow. Snow isn’t in this line, but it is the snow that makes the peaks of Kailāsa shiny enough to serve as a mirror to the gods and as bright as lotuses.
English is an enormously rich literary language, and every language has its particular resources. Many of the features that make kāvya (ornate classical Sanskrit poetry) so distinctive have equivalents in English: the English tradition has plenty of metrically inventive poetry, for instance, and plenty of allusively dense verse marked by a rich use of simile and metaphor. (We might think of something like Shakespeare’s Venus & Adonis, for example.) Other elements of Sanskrit poetry, like the great concision and flexibility of word order produced by a highly inflected language, are found also in Latin and Greek verse. But there is just no way, I think, in any of these languages to reproduce the syntactic counterpoint so common in Sanskrit poetry of this kind — even much less sublime examples than that of Kālidāsa — between a sentence structured, on the one hand, around complex inflection (in this example, the nominative, accusative, genitive and instrumental) and, on the other hand and within the same sentence, by the uninflected relationships between the elements of a compound. This feature, I think, helps to make kāvya unrivalled as the poetry of pure description.
For the metrical enthusiasts among you, the metre is Mandrākrāntā (‘slow-stepping’), a form perhaps invented by Kālidāsa for this poem. Each line has 17 syllables, in the following pattern: | – – – – | u u u u u – | – u – – u – x ||. Note that Sanskrit metre, like Latin and Greek, is structured by the quantity of syllables (whether they are long or short), not by stress. The Sanskrit technical vocabulary for discussing metre is rather more sophisticated than for Latin and Greek (and vastly more so than the pretty hopeless confusion that reigns in relation to English metrics). The eagle-eared may have noticed that mandrākrāntā is itself – – – –, giving you the beginning of the line. You can read a bit more about this metre and the useful Sanskrit metrical mnemonics here.
Translation from the German by Eric Ormsby, in Goethe, West-Eastern Divan: Complete annotated new translation including Goethe’s ‘Notes and Essays’ and unpublished poems (London: Gingko, 2019), p. 499.
This is verse 61 in my student edition; verse 60 in the online text from which the image is taken; verse 62 in the complete analysed text online from which an image below is taken. The numbering varies because there are a handful of contested verses included in some editions and not in others.
From a fantastically useful website, the Digital Corpus of Sanskrit, offering a complete analysis of many classic Sanskrit texts. I only discovered this while writing this piece but it’s a great boon if you’re stuck, and the tool that shows you the frequency of the word within the corpus is brilliant.
Subscribe to Horace & friends
Thoughts on poetry, translation and related matters.
‘Howling snow-laughter’ transformed my experience of ‘Sodding snow-soddenness’ into something much more magical after Paris made its first attempt at winter weather this afternoon. Perfect timing Victoria! In old Norse and English poems, and I think in Gaelic, kennings likewise (tho more modestly) enlarge the scope for description by means of
Such a brightly gleaming essay that it made me laugh! It strikes me that part of poetry’s role is to articulate the allatonceness of language itself, as it carries word meanings and cultural histories along, as lightly as a cloud does its water droplets. And this reminds me of Rody Gorman, who writes in Irish and Scottish Gaelic and English, and recently published a translation of the medieval Irish text Buile Shuibhne, The Madness of Sweeney, subtitled An Intertonguing, where the three languages intersect. A random example: My curse on Sweeney, he-it loveshareoffencedestroyedme the hourtime he javelindashattacked my little clusterclockbell in a javelintroopspray frenzy.