translating kalidasa 4Working Practices

Overleaf I recommended looking up every word in the Monier Williams dictionary, and then referencing elementary grammar as necessary. Unfortunately, as should have been obvious from the beginning, the approach suffers from two drawbacks. First, it doesn't always work. What does hArayiSyan mean? No, it's not a samasa involving hara (bearing/ charming/garland/girl of bad reputation, etc.) but the future participle of the causative of the verb hRi to carry: he caused to be conveyed. Kalidasa is relatively straightforward, but there are many things not given in Bucknell or elementary grammars. Second is the hermeneutic circle. Many words have alternative meanings, and all have shades of meanings, but we cannot know which to choose until we have understood the whole stanza, which of course depends on choosing the correct meanings in the first place. It's not an insuperable difficulty — stanzas do settle into sense eventually — but it does require us to note all the alternatives, and to draw hints and explanations from earlier translations (though these rest on the same conundrums of course).

Eventually — for those interested in doing something similar — I found this to be the best strategy:

1. Read Nathan, {1} Taylor {2} and Devadhar.{3}
2. Read the Devanagari aloud: {1} or {4}
3. Read the detailed exposition of M.R. Kale. {5}
4. Undertake a preliminary translation based simply on these four sources. This alerts us to where problems will arise in a. meaning and b. rhyme
5. Construct a table where each word of the original can be given its meanings, parts of speech and interpretations by previous workers.
6. Copy into the table the Meghaduta text from the Frankurt University site {10}
7. Check this text against other sources. {1} {2} {3} {4} {5} {10}. (There are a few slips.)
8. Add to the table the vocabulary provided by Hultzsche (two stanzas at a time is easiest).
9. Look up all words not given by Hultzsche in the Monier-Williams dictionary. {9} Also words where there are discrepancies in interpretation.
10. Write out a word-for-word, line by line, literal translation.
11. Rework the translation of 4. from the English words pregnant with meaning, aiming for sonorous beauty of phrase.
12. Now start thinking about the original carefully: what it means, what emotional effect was intended, how that was achieved. Aim to convey that, simply and clearly, eliminating unnecessary words or explanation.
13. Create a sonorous verse on the abcba pattern with the words given. Do not 'versify' the prose translations, but get the properties of the English words to create useful rhythm and sound patterns.
14. Rearrange to better meet the abcba rhyme pattern, changing the odd word as necessary.
15. Rework to improve its quality as verse per se.
16. Consult 5 and move the verse closer to 10, trying a. eliminate words or meanings not in the original, and b. work in words that the translation has not so far succeeded in including.
17. Again look up key words in Monier-Williams to a. check that pervious translations are essentially correct, b. understand better what the samasa mean and c. see if the alternatives offer a deeper meaning or words more useful to the verse.
18. Repeat 11 to 17 until the verse 'shapes up'.
19. Put away for days or weeks, polishing up the stanza(s) as they appear fresh to the eye. Aim to make them beautiful, moving and memorable.

Of course we have to avoid writing 'a poem on the theme of Meghaduta', or 'a response to reading Meghaduta', by sticking to Kalidasa's words, and not introducing many of our own. But we don't have to reproduce every word, and for these reasons:

1. We are not producing a student's edition. Anyone who studies Sanskrit at college will read Kalidasa, and tutors naturally want literal texts that give all the words and display an awareness of the grammar involved. We are seeking to convey the appeal in English that Kalidasa offers to a Sanskrit reader.

2. A poet uses a particular word for many things: its meanings, connotations, rhythmic properties, colour and vividness of imagery, melodic echo, consonantal texture and so forth. Most of these disappear in the exact, literal equivalent. Had he been writing in another language, the poet would have used words with quite different meanings to secure his effects, inevitably so: it is from this play of effects on many levels that the poetry emerges.

3. Word use is governed by conventions that go far beyond prose meanings: appropriate to genre, period, situation, class structure, tone, etc.

4. There is a some looseness and word repetition in the original, probably to meet the demanding Mandakrata metre.

5. The very free word order of Meghaduta requires us to transpose words and hazard guesses at meaning: a literal translation is a long way from acceptable English.

An example may make things clearer: the first line of the third stanza. We could write:

In tears withheld new fell that fall from grace,
From wealth attending on the King of Kings.
The otherworld that simmered in the clouded air
But felt discomfort when such longing brings
A breath that holds him to that neck's embrace.

Very neat of course, with its play on things outward and inward. But much too clever, almost as over-alliterated as the Act V players' performance in Midsummer's Night Dream. What about:

He held his tears, but felt that fall from grace,
From wealth attending on the King of Kings.
The otherworld that simmered in the clouded air
Is but discomfort when such longing brings
A breath that holds him to that neck's embrace.

But then we have held / holds. Is that the comparison Kalidasa intends? In fact it may be: the literal translation does indeed suggest a comparison of things near and far, of things inward and outward:

of_her having_stayed with_difficulty before desire impatience destitute_of_wealth cause
[inward tear] for_a_ long_ time attendant of_king_ of_ kings holding
cloud in[not_of_the_world] is_being comfortable although [other_than conceals] grief
head_and_neck embrace of_vital_breaths in_person how_much again distant

Nonetheless, it may be wiser — at least till we assess the balance in other stanzas — to leave these parallels in the background and write:

What tears he stifled on that fall from of grace,
From wealth attending on the King of Kings.
The otherworld which simmered in that clouded air
But felt discomfort now his longing brings
A breath that holds him to that neck's embrace.

The finished article, or that completed so far, is given here. The revised stanzas are below.

1. A year from amorousness: it passes slowly.
So thought a Yaksha by his master sent
For scanting duty to the Ramagiry:
To mope in penance groves as banishment
By rivers Sita's bathing there made holy.

2. Asadha's ending on the mountain found
Him weakened, gold chain slipping from his wrist,
And mixed his pleasure as a cloud came down
So playfully to hug the summit mist,
As elephants in heat will butt the ground.

3. In tears withheld he took that fall from grace,
From wealth attending on the King of Kings.
The otherworld that brimmed in cloudy air
Was still discomfort when far longing brings
A breath to hold him to that neck's embrace.

99. Excessive sorrows, and her sighs are his
As he in waiting sighs for her. Distress
That wastes her body reams out his. To block
Once lawful union with a hopelessness
Lies adverse fate athwart their path to this.

Texture of Sanskrit Poetry

Sanskrit is not an easy language, and many will prefer formal tuition at college or adult learning centre. Those intending home study will probably need an introductory course, and then find MacDonell {12} much easier than Whitney. {7} The online dictionaries are admirable, {13} {14} {15} {16} {17} particularly as they allow existing translations to be checked with English to Sanskrit searches, but they do need some basic grammar. Personally, I have found the Monier-Williams dictionary in book form {9} to be well worth the initial expense and inconvenience.

Being an ancient language, Sanskrit can be somewhat baffling: no relative or subordinate clauses, or even finite verbs very often, but long compounds and gerunds with the passive tense. Prepositions are widely used, and take various tenses, but it is generally the cases of the nouns and compounds that convey the sense.

For poetry that brings benefits and difficulties. The 'precise meaning' — in the way expected of European prose — is not always clear, and interpretations naturally differ. Compounds may also be long and involved — monstrously so in later poetry — but can create compact and evocative similes. A celebrated example is vIcikSobhastanitavihagazreNikAJcIguNA, the first line of stanza 28 in Meghaduta, which MacDonell {18} construes as "an appositional descriptive consisting of two main parts. The second, kAJcI-guNA, m. girdle-string, is a Tatpurusa [samasa]. The first is an adjectival descriptive in which the Tatpurusa vihaga-zreNi, row of birds, is described by vIcikSobhastanita, loquacious through wave agitation. The latter is a compound Tatpurusa, in which stanita is qualified by the simple Tatpurusa vIci-kSobha, agitation of the waves." Involved, yes, but through its use Kalidasa can draw a parallel between the river and a woman making her overtures of love. Compound similes operate throughout the Megaduta, where the cloud's life-giving passage across the parched Indian landscape is an extended metaphor for the sexual congress of all nature, one difficult to render in the European tradition and foreign to the Chinese.

These compounds, the inflected nature of Sanskrit, and its euphony also allow great beauty of expression. If we look, for example, at our earlier stanza 1/99:

aGgenAGgaM tanuca tanunAa gADhataptena taptaM
sAsreNAsrudravamaviratotkaNTham utkaNThitena
uSNocchvAM saMsamadhikarocchvAsenah dUravartI
saMkalpaiskalpaiste vizati vidhinA vairiNA ruddhamArgaH

for which the word-for-word translation is:

with body body emaciated and with emaciated deep with distressed distressed
with flowing tear weaping continually longing with longing
ardent sigh brimful with sigh distant
with proper them conjoins by fate hostile blocked path

we can see the emotional closeness of the Yaksha and his lover are emphasized by the repetitions of emaciated, distressed, longing and sigh, which set up a monotonous, dispiriting chant. Then comes the quickening with alliteration on v, to be rendered hopeless with ruddhamArgaH, the word heavy with its two long syllables ruddhamArgaH and picking up rudravam, weeping.

aGgenAGgaM tanuca tanunAa gADhataptena taptaM
sAsreNAsrudravamaviratotkaNTham utkaNThitena
uSNocchvAM saMsamadhikarocchvAsenah dUravartI
saMkalpaiskalpaiste vizati vidhinA vairiNA ruddhamArgaH

And so on. Has our translation conveyed that pattern? Somewhat:

Excessive sorrows, and her sighs are his
As he in waiting wastes for her. How more
He feels her flooding tears, her hopelessness
As fate rebukes the lawful union that before
Flung breath and her full body into his.

or (the section is a workshop: I will decide on a final version eventually):

Excessive sorrows, and her sighs are his
As he in waiting sighs for her. Distress
That racks her body torments his. To block
Once lawful union with a hopelessness
Is adverse fate athwart their path to this.

But not entirely. The monotonous repetition is rendered, but the check of block comes at the end of the third line and the two succeeding lines fall back into the dispiriting rhythms, their ending sibilants echoing the sighing lovers. We can't usually capture everything in the one rendering, but to follow the Sanskrit word order more closely (free though it is) traditional verse lets us write:

Consuming tears and sighs on his behalf
He also feels and adds to, and recalls
A wasting body that is one with his.
Across that lawful union still there falls
A fate to hurt them and obstruct the path.

Other Attempts

For Hank Heifetz, however, who has produced a widely-praised translation of Kalidasa's Kumarasambhava, {19} traditional verse is precisely the problem. He writes: 'Scholarly translations of Sanskrit poetry into English have generally been of poor literary quality. . . characteristics are stiff, archaicizing diction (full of words like "wanton" and "charming"); the use of emotionally impoverished, merely "educated' language; antiquated inversions of sentence structure; and iambic rhythms (used directly or present as underlying patterns) that are inappropriate to the quantitative effects of Sanskrit verse and alien to the far more rhythmic achievements of twentieth-century poetry, developments which open up far more interesting possibilities for the translation of rhythm. The history of translation from Far Eastern poetry stands in interesting contrast. In this area, a tradition of good writing was established earlier in the century by Ezra Pound and Arthur Waley; such contemporary poets as Kenneth Rexroth and Gary Snyder have furthered it."

Previously, still in his Introduction, {20} he writes "I have attempted to create a poem in modern American English that conveys some of the greatness of the original through means available in living speech. . . In very case, I have tried to convey what I believe Kalidasa intended. I have sought out equivalents (but not imitations) in English for the rich, penetrating and emotionally precise effects of Kalidasa's stanzas. I have paid a great deal of attention to the rhythmic effects of individual stanzas and continuous sequences by seeking means in English for conveying the rhythmic import of Kalidasa's poetry. . ."

Worthy aims, and Professor Heifetz's comments on previous translations are largely true. The Origin of the Young God displays a deep love of the subject, and the detailed commentaries touch on matters that have not received sufficient attention before. However:

1. Does The Origin of the Young God achieve its aims?

A. Verse Structure

As is usual with longer Sanskrit poems, the various chapters of Kumarasambhava are written in different metres. Do the translations convey any of this? Here is the opening stanza of the first chapter :

Here the Sanskrit is written in the Zloka metre of eight syllables. The syllable count for the translation is 7, 13, 9, 11. There is no requirement for the caesura position, but the fifth syllable must be short, the sixth long, and the seventh alternately long and short in successive lines. English is not quantitative, but to the extent that stress can take the place of quantity, the fifth, sixth and seventh syllables in the four lines of the translation are 1. stressed, unstressed, stressed, 2. stressed, unstressed, unstressed, 3. unstressed, unstressed, stressed, 4. unstressed, unstressed, stressed.

B. Verse Texture

Heifetz is concerned to translate, i.e. convey to some extent, the alliterative and melodic patterns of the original. He analyzes two examples in his Introduction.

The first (2.41) is:

The interest here was the "dancing rhythm [that] builds up in short steps to the long elegant turn of the compound which ends the third line. . " That third line "nRtyAbhinayakriyAcyutam (literally, "fallen from the movements of his gestures in the dance:) dances around its beats of "a" and "y", while the consonants of kriyAcyutam seem to echo the very shaking loose of the dust. This compound, I felt, required an entire line for its movement into translation."

Again the translation says something different. The syllable counts for the lines are 12, 16, 13, 14. The third line has been transposed to the fourth. The first three lines do not have a dancing rhythm, and the fourth line is a loose anapaestic.

C. Greatness of the Original

Does the translation convey "some of the greatness of the original"? Readers must make their own judgements, but I wouldn't have thought so. As one reviewer put it: "Although Heifetz's notes call attention to alliterations, metrical subtleties and sonic effects in the Sanskrit, his translation often remains earthbound."

2. Are the aims achievable?

Heifetz's arguments are detailed, and important for translation generally. If we start with that innocuous phrase "modern American English", we can ask two things.

A. Is modern American English the appropriate medium? Kalidasa wrote in a highly-crafted version of elevated court language.

B. Why should modern American English be the touchstone? Isn't this a little neocolonialist? Would we want Paradise Lost so translated, for example, rather learning Milton's own thought and language to appreciate him properly? Yes, we can say that if Renaissance literature explored concepts unfamiliar to us, and employed words that have different meanings today, {21} we can nonetheless read its poetry in a way we cannot read classical Sanskrit. Milton was writing in and with a verse technique that we have inherited, that we still "read".

But properly? Heifetz is a contemporary poet, and pays the usual tributes to Modernist leaders, whose influence is turning older verse into a lost language. Neither Pound nor Rexroth had sufficient Chinese to produce reliable translations, and what they wrote was evocative prose, often beautiful but nothing like the original. Arthur Waley most certainly did read Chinese, but his cadenced prose set an unfortunate trend for translation, as it was not able to convey into English equivalents the verse on which the poetry of the original Chinese depends. Gary Snyder has a first-hand knowledge of oriental cultures, but not always the verse skill to convey that insight. {22}

No doubt a lot of traditional verse was incompetent, but "free verse" often avoids the challenge altogether. Subtlety and variety are only possible against some standard or regularity, moreover, and this is what "free verse" generally rejects. Like blank verse, contemporary styles are easy to write correctly, but phenomenally difficult to write well. They require certain departures from normal speech — melodic patterning, line breaks, unexpected juxtapositions, unusual wordings — to work effectively, and such departures, besides being intrinsically weak, are foreign to the classical tradition.

Hank Heifetz is not suggesting that the Sanskrit marriage of sense and rhythm should be replicated in English. "My interest," he writes, {20} "is in translating rhythm, by producing suitable American rhythms at the level of the speaking voice. This is a translation for the ear, meant to be read aloud in the natural emotional tone suiting each stanza or sequence and with the poetic line as the basic unit, receiving its slight stresses at the beginning and end."

Technically, those aims are achieved. The first of our examples in B. above is not prose, as we can see by running the lines together: The trees of the Nandana Grove where the wives of the immortals by hand would gently pick blossoms have learned from him to be cut through and fall. Nor is the second — Once it has come to touch that body, I know dust from the very ashes of the dead will purify the living and so the gods rub their foreheads with it as it falls from the play of his limbs in the language of his dancing. — though the line breaks could be arranged differently.

As will be clear, Heiftez's is a very pleasing translation, but somewhat lacks the variety, pace and emotive range needed to keep us turning the pages. Indeed, poets have usually accepted that everyday language needed crafting and heightening to convey the effects proper to poetry, and the argument is really over how this is to be done. "American rhythms at the level of the speaking voice" would be an admirable place to start — which is what the better contemporary styles perhaps recognize — but it would need to develop an extraordinarily wide vocabulary of expression to convey classical Sanskrit poetry. Perhaps the result would not have American rhythms in any simple fashion, but its own set of techniques and devices — be formal verse, though starting from a different place and with different premises.

Naturally, we can't argue from negatives, and say that the absence of any effective translation of Sanskrit poetry into free verse proves that one cannot be done. But free verse does have its problems, and if its practitioner avoids a lumbering translation in formal verse he may turn out the merely unobjectionable. Every approach comes at a cost, and it would have been less contentious to have employed a four-line stanza for the Meghaduta translation, for example, and to have avoided rhyme altogether. But a fifth line I found necessary to capture the content of the long Mandakrata lines, and the acknowledged quality of the Sanskrit called for the larger resources of English verse.

That's probably all the average reader wants to hear of this tiresome subject, but I look again at quantitative verse through a second translation of Bartrihari, still experimenting with appropriate forms.

To round off matters, we undertake a verse translation of Heiftez's key example: verse 79 of Chapter 5. The transliterated devanagari showing the vaMSastha (x -x | - - x | x - x | x - x || metre in its long and short syllables:

Kumarasambhava Chapter 5 Verse 79

ta daG ga | saM sarg am | a vAp ya | kal pa te
dhru vaM ci | tA bhasm a | ra jo vi | zuddh a ye
ta thA hi | nRt yA bhi | na yak ri | yAc yu taM
vi lip ya | te mau li | bhi ramb a | rau ka sAm

The word-for word rendering:

or doing that body contact having obtained is able
certainly funeral pyre ashes dust completely purified
for so it has been said movements of the body performing fallen
is anointed with the head of heavenly beings because of who

Rearranging for a literal translation:

For that body contact the heavenly beings anoint their heads with the purifying ashes of the funeral pyre shaken down in the fervour of his dancing.

A simple verse translation that brings out the meaning:

The gods who crave a contact with that body
smear their foreheads with the funeral ash
shaken down by gestures of the dancer's
deeper fire in purifying dust.

Which we can then vary in tempo to echo the original: drum beats in the first line, maintained in the second, a rhythm to echo Shiva's dancing in the third, and a quiet ending in the fourth.

To touch the dust his limbs have shaken down
the gods incline their foreheads as forever
the fire-wreathed fervour of that dancing body
falls in purifying funeral ash.

A preliminary attempt. It's not particularly close to the literal version, though considerably more to a deeper rendering. {23} We'd want to devise forms for each Sanskrit measure before going further, and modify them as the translation took shape.

Looser styles are the norm today, and readers may prefer Heifetz's translation. But I don't think it can be said that free verse is intrinsically the better medium, either for conveying the original features, or for that amalgam of beauty and content that makes poetry worth reading.

Getting Back to the Original Words

Returning to Meghaduta, even this brief incursion into Sanskrit verse should have shown the dangers of simply reworking previous translations. Important elements of the structure can be missed, and each translation takes us further from the original.

Translations can also be somewhat quixotic, building on words that seem not to be in the text. An example is stanza 1.24, where most versions mention thunder in some form: your thundering on the border of its bank (Kale), at your soft thunder along her banks (Nathan) or which roars pleasantly at the edge of her banks (Taylor). But the original:

teSAM diSu prathitavidizAlakSaNAM rAjadAnIM
gatvA sadyaH phalam avikalaM kAmukatvasya labdhA ( phalam atimahat; tv avikalaphalaM. (Kale): phalam api mahat )
tIropAntastanitasubhagaM pAsyasi svAdu yat tat ( svadu yatra; svadu yuktaM; svadu yasmat; svadu yat tva?)
sabhrUubhaNgaM mukham iva payo vetravatyAz calormi

has a word-for-word rendering:

in that from direction celebrated Vidizá sign of capital
having gone suddenly fruit full of being a lover obtaining
bank side murmuring pleasing of water
frowning face like water from Vetravati moving wave

with no thunder, whatever the version. Devadhur in fact omits thunder The river Vetravati awaits your charming call in her earlike fringe — simply seeing the stanza as an extended play on the cloud's courtship of the river, but introduces words of his own. Why this happens I do not know — translators may be referring to earlier commentaries — but the safer approach, I think, is to return to the text as given and make such poetry as we can from its words.

You'll come to Vidizá, the capital
Well known across the compass of these quarters,
When, like a lover, at the Vetravati,
Draw near her face to have the frowning waters
Turn to murmuring: and drink your fill.

The author's full (and free) translation of Kalidasa's Meghaduta is published in pdf format by the Ocaso Press.

A 568-page free pdf ebook on practical verse writing is available from Ocaso Press. Click here for the download page.


1. Leonard Nathan, The Transport of Love: Kalidasa's Megadhuta (Berkeley, 1976).
2. Kalidasa's Meghaduta or 'The Cloud Messenger' McComas Taylor. May 2001. NNA. Biligual text.
3. C. R. Devadhar, Works of Kalidasa: Vol. 2 (Motilal Banarsidass, 1984).
4. Kalidasa: Meghaduta Based on the edition by M.R. Kale. NNA. Simple transliteration, employing UTF-8 convention.
5. M.R. Kale, The Meghaduta of Kalidasa (Motilal Banarsidass, 1969).
6. Roderick S. Bucknell, Sanskrit Manual: A Quick-reference Guide to the Phonology and Grammar of Classical Sanskrit (Motilal Banarsidass, Delhi, 1991)
7. William Dwight Whitney, Sanskrit Grammar (D.K. Publishers, 1875/1924)
8. E. Hultzsch (Ed.), Kalidasa's Meghaduta: Manuscripts With the Commentary of Vallabhadeva and Provided With a Complete Sanskrit-English Vocabulary. Foreword By Albrecht Wezler. (Munshiram Manoharlal Publishers, 1911/1998).
9. Monier Monier-Williams, English-Sanskrit Dictionary (Munshiram Manoharlal Publishers, 2003).
10. 20. Kalidasa: Poems: Meghaduta. Transcripts and (free) Titus fonts.
11. Nathan 1976, op. cit, viii.
12. Arthur A. MacDonell, A Sanskrit Grammar for Students: 3rd Edition (OUP, 1971)
13. Cologne Digital Sanskrit Lexicon. Based on the Monier-Williams Sanskrit-English Dictionary, with 160,000 main entries.
14. Capeller's Sanskrit Dictionary. 50,000 entries, input governed by Harvard-Kyoto convention.
15. Apte Sanskrit Dictionary Search. NNA. Based on The Practical Sanskrit-English Dictionary of Vaman Shivaram Apte.
16. Gérard Huet's Sanskrit-French dictionary. NNA. Free to use online or download.
17. Online Sanskrit Dictionary. Cologne University's Sanskrit dictionary, plus a good listing of others.
18. MacDonell 1927, 176.
19. Hank Heifetz, The Origin of the Young God: Kalidasa's Kumarasambhava (Univ. California Press, 1985 / Motilal Banarsidass Publishers, 1990)
20. Heifetz 1990, op. cit, 12-15.
21. Isabel Rivers, Classical and Christian Ideas in English Renaissance Poetry (Routledge, 1992).
22. Eliot Weinberger, 19 Ways of Looking at Wang Wei: How a Chinese poem is translated (Asphodel Press, 1987), 42-43. Weinberger admires Snyder's translation, but it seems to me a clumsy rendering, with an idiotic concluding word: moss only forms on trees away from the sun, and more than a trailing proposition is needed to invoke a spiritual dimension.
23. Stanza 5.79 typifies many problems of Sanskrit translation. First there is an error in my copy of Kalidasa's Kumarasambhava. The opening word is tadaGgasaMsargam and not tandaGgasaMsargam. M.R. Kale (M.R. Kale, Kumarasambhava of Kalidasa, Motilal Banarsidass, 2004: 92) gives the correct text, and tand etc. will not fit the vaMSastha metre. Then there is the word order, so free that one cannot talk about a rhythm coincident with the sense. And finally there is the difficulty of tying down the meaning properly. ln the literal translation I have accepted the interpretations of Kalidasa's many commentators, as I think Frank Heifetz has, but in the verse translations have gone back more to the meanings of individual words (reading eternity for dhruvaM rather than certainty, etc.).


Translating Kalidasa: page one   |  page two   |   page three   |  page four


Material can be freely used for non-commercial purposes if cited in the usual way.