translating bilhana 2A Second Look at Translations

To illustrate the difficulties with E. Powys Mathers' translation of Caurapañcáziká, {1} we look at a stanza common to the northern and western recensions, to be sure of having the source material. The first such in the collection is stanza 3 in the northern version and 6 in the western. The published translations:

Barbara Stolles Miller: her stanza 3 {2}

Even now,
if I see her again,
a lotus-eyed girl
weary from bearing her own heavy breasts—
I'd crush her in my arms
and drink her mouth like a madman,
a bee insatiably drinking a lotus!

E. Powys Mathers: his stanza 2 {1}

Even now
If my girl with lotus eyes came to me again
Weary with the dear weight of young love,
Again I would give her to these starved twins of arms
And from her mouth drink down the heavy wine,
As a reeling pirate bee in fluttered ease
Steals up the honey from the nenuphar.

Gertrude Clorius Schwebell: her stanza 5 {3}

And still today my arms are clasping
her curving body marked by the love god's arrow,
voluptuous burden of full blooming breasts.
They tremble as I'm drinking of her mouth intoxicated
as the bee drinks of a lotus.

The original Sanskrit {2} with the usual word-for-word translation: {4}

adyApi     tAM yadi punaH kamalAyatAkSIM
even now her  if     again  lotus with long eyes

pazAmi pIvarapayodharabhArakhinnAm
I see    plump breast burden suffering

saMpIDya bAhuyugalena    pibAmi vaktram
by force   greatly by brace I drink  mouth

unmattavan madhukaraH    kamalaM yatheSTam
intoxicated   bee causing     lotus       desired

As before, the Miller version is the most faithful, the Schwebell is the most pleasing as verse, and the Mathers is largely a fabrication.

Points and Problems

In prose meaning, Miller's is accurate and other two are not (of which more later). Making poetry from the words is another matter, however. We can say that Bilhana is using the stock phrases of Sanskrit poetry — mouths like honey, eyes as beautiful as lotuses — and that the poem's celebrity derives from its frankly erotic nature. But much Sanskrit love poetry is erotic by European standards, and Caurapañcáziká is good literature.

There are some poeticisms: madhu means 'honey-licker', i.e. bee, but also the juice or nectar of flowers, any sweet intoxicating drink, wine or spirituous liquor. Otherwise Bilhana uses blunt words: no arms but yugalena, a couple or to be yoked together (with yuga meaning not only a pair but all planets in two houses, i.e. something unusual, onerous and inevitable). No gentleness: saMpIDya means to compress, squeeze together, torment or reckon up (and astrologically to obscure or eclipse). No Platonic refinements: pIvara means fat, stout, large, plump, thick, dense, full of or abounding with.

But the poet also has compassion for the woman, throughout in the poem, and in this stanza, where her breasts are described as bhAra, a burden, labour or imposed task, and as khinna, distressed, wearied, suffering pain. Bilhana half-apologizes for his passion, which may also be the reason for the double appearance of the lotus. He uses kamala (lotus flower) and AyatAkSa (having longish eyes) in line one, and we should perhaps go beyond the stock epithet for beauty to imagine both flower and eye in their stainless beauty, and the splay of stamens that recalls the eyelashes. Into that beauty the bee thrusts itself, an image that Proust uses in À la Recherche du Temps Perdue. Mathers, also a homosexual writer, has grasped the significance with 'pirate', though his line continues with 'fluttered ease'.

If we want a free-verse rending or loose hexameter similar to that overleaf, we can write:

Even now I see her, and that heavy body,
with breasts that hurt her, and her long lotus eyes.
I squeeze her hard as bees that drink from lotus flowers
and am intoxicated with that honeyed mouth.

That dragging rhythm can also be rendered in iambic pentameters:

Even now I see her: heavy, breasts
That weary her, her longish lotus eyes.
But mad as bees in lotuses I press
Once more to drink that mouth where honey lies.

A pararhymed version:

Even now I see that that heavy body with
Its breasts that hurt her, longish lotus eyes.
Eclipsed as in the lotus flower the bees
Are mad with drinking honey at that mouth.

This fully-rhymed version transposes 'mad' to the bees, which the Sanskrit allows (but not compels):

Even now I see her in her weariness
Of breasts that hurt her, longish lotus eyes.
But mad as bees in lotuses, I press,
Intoxicated, where that honey lies.

And so on. Which is best?

Reviewing Mrs. Miller's similar renderings in her earlier Bhartrihari (1967), Poetry said 'Aware that English cannot do this rhyme and meter, the translator has concentrated on "meaning, tone and closely knit texture of the original" and tried to find analogous syntactic forms and rhetorical devices in English", and succeeded very well.' I don't need to labour the point made so often in these pages, that rhyme and metre will serve perfectly well, given a little effort, but we should look at the texture of the original before crafting a final version. As before, the stanza is cast in the 14-syllable Vasantatilaka metre (where red syllables are short):

ad yA pi tAM ya di pu naH ka ma lA ya tAk SIM
pa zA mi pI va ra pa yod ha ra bhA ra khin nAm

saM pID ya bA hu yu ga le na pi bA mi vak tram
un mat ta van ma dhu ka raH ka ma laM ya theS Tam

in which we see certain patternings:

ad yA pi tAM ya di pu naH ka ma lA ya tAk SIM
pa zA
mi pI va ra pa yod ha ra bhA ra khin nAm

saM pID
ya bA hu yu ga le na pi bA mi vak tram
un mat
ta van ma dhu ka raH ka ma laM ya theS Tam

Little of that appears in the Miller version here, where the continuum is replaced by not-very-pleasing rhythmic units that can only loosely be termed free verse:

Even now |
if I see her again |
a lotus-eyed girl |
weary from bearing her own heavy breasts ||
I'd crush her in my arms ||
and drink her mouth like a madman |
a bee | insatiably drinking a lotus ||

That lotus image also causes problems. As we've noted, lotus is a stock image in Sanskrit poetry, invariably used for eyes, and evoking delicate, fragrant, lovely, fresh and unfading {5} — and anything of that nature: it doesn't really mean much. But if we repeat the word in English, we somehow tangle the bee in the woman's eye, which becomes a gross and repellent image. I'd suggest we change lotus to 'flower', and attach 'heavy' to it, so as to conjure up the picture of a woman over-burdened by her looks, just as the poet is over-burdened with his passion. Since Bilhana uses both intoxicated and desired (unmattavan . . . yatheSTam) we could also emphasize the headiness of passion by adding 'adrift', which doesn't appear as such in the original.

We also need to convey the rhythmic integrity of the original, and something of its intonation. The content is not so extended to need an extra line, and there is no compelling reason for rhyme if we can make the lines hang together by other means. One answer may be the rhythmically floating (i.e. somewhat indeterminate) line of E. Powys Mathers, but made a. more faithful to the meaning, b. into four line stanzas and c. consistent in stress or syllable count. Perhaps something like this, (syllables/stresses):

If even now I see her in that weariness:                            12/6
full breasts that hurt her, with her flower-heavy eyes.          12/6
As bees adrift in lotuses, I clasp her hard                           12/6
and fall intoxicated of that honeyed mouth.                        12/6

Or, if that's too much licence, then:

If even now I see her in that weariness                            12/6
of breasts too heavy for her and long lotus eyes.               12/6
As bees cling in the lotuses, I brace her hard                    12/6
and drink, intoxicated, of that honeyed mouth.                  12/6

That's possibly as far as free verse will take us, and we might do better to use traditional forms:

Even now I see her, in a loveliness
of full breasts hurting and of flowered eyes,
but mad as bees in lotuses, I press,
for that sweet mouth, and hard, where honey lies.

Approaches to Translation

What can we say about E. Powys Mathers's fidelity to the text? Ezra Pound could be cavalier with originals but didn't make up entire verses. Robert Lowell also had the goodness to call his free renderings 'Imitations'. But in the three Mathers stanzas we looked at overleaf, none of the phrases in italics is to be found in the original text, anywhere, in either rescension or its variants.

Even now
I know my princess was happy. I see her stand
Touching her breasts with all her flower-soft fingers,
Looking askance at me with smiling eyes.
There is a god that arms him with a flower
And she was stricken deep. Here, oh die here.
Kiss me and I shall be purer than quick rivers.

Even now
They chatter her weakness through the two bazaars
Who was so strong to love. And small men
That buy and sell for silver being slaves
Crinkle the fat about their eyes, and yet
No Prince of the Cities of the Sea has taken her,
Leading to his grim bed. Little lonely one,

You clung to me as a garment clings, my girl.

Even now
Only one dawn shall rise for me. The stars
Revolve tomorrow's night and I not heed.
One brief cold watch beside an empty heart
And that is all. This night she rests not well;
Oh, sleep, for there is heaviness for all the world
Except for the death-lighted heart of me.

Why was Mathers popular, indeed still is? {6} Because he packs his lines with what many consider poetry, much though the imagery belongs to 1920's work, the romance of travel to distant places or times: Valéry Larbaud {7} or James Elroy Flecker. {8}

But if the practice is not properly translation, it is one aspect of making something acceptable to a reader. Today the taste is for free verse, and translations are accordingly rendered in this style, often in spite of original forms or cultural contexts. Blurbs that promise 'a plain rendering for our time', or 'a fresh and idiomatic translation that brings us close to the original' should ring warning bells, but generally the cultural overprinting goes unnoticed.

Perhaps it will always go unnoticed. Readers do not realize that detours into contemporary idiom create a vacuum, where what's been left behind has to be reintroduced, often by adding matter foreign to the original. Gertrude Clorius Schwebel's rendering is very pleasing, for example, but the italicized words here are her invention:

And still today her image is embedded in my heart
how she sits at the window,
her head supported by her hand.
Anxious eyes are searching

Scholars are well aware of the liberties taken, but have no need of translations to discuss the original. The poetry may even become an irrelevance to them, as in this case:

Will the hero know, when he hears her plaintiff offstage song, that the woman of his dreams resides in the now-familiar wife and mother of his children? Or will he look again to the horizon for the new epiphany of the elusive Sri, the 'new creation of the jewel known as woman' (strIratnasRSTir aparA, 2.10 [9])? But this is really a rhetorical question. The lure of the playworld paradigm of emotional release never lost its force in kavya, or religion for that matter. Nevertheless, this play of Kalidasa's (and it is emblematic rather than unique in this regard) gives us reasons for doubting the psychological viability of this idea. The hero as self-possessed connoisseur of emotion, sporting in the feminized garden of Eros, becomes a child pleading for understanding from hidden authorities who, he fears, may be indifferent to his affective needs. {9}

No one can master Sanskrit literature without a deep love of the subject, but here literary matters are sidestepped by moving to the narrow (and doubtful) ground of Freudian psychoanalysis. But even if we judge Sanskrit literature by Sanskrit poetics, the central issue remains: how can we make a translation faithful to the original when we have to employ words with different connotations and histories of usage, and combine them in conventions that apply specifically to English poetry?

Broadly, there are two ways of approaching the problem. One is to find some common bedrock, some ur-language through which individual examples can be compared and transposed. The search here is for well-defined and universal terms, relationships that are true across all possible worlds, as in science, some aspects of the New Criticism, Chomskian linguistics and machine translation. The second approach is to think more of overlapping horizons of understanding. There is no universal measure of poetry, but a complex interrelationship of texts and interpretations, which differ between cultures and individual readers, and shift as the public itself changes its reading habits.

Machine Translation and its Theories

The shortcomings of machine or computer translation are readily acknowledged, but (as we have seen on the Racine page), the renderings can be useful. Provided the machine is coded adequately, it will largely cope, obviously better with straightforward facts than imaginative language of metaphor and connotation. Some theorists do believe the brain acts as a super-computer, and better machine translation simply calls for better coding. Others argue that human beings use language for a purpose, and that meaning cannot be reduced to grammar. {10} Pleasure in creativity, emotion and intention all have their part to play — when we are back in the philosophic difficulties of meaning. Language and thought are somewhat interdependent, moreover. Herbert and Eva Clark put the matter succinctly in their 1977 text book: "Language does not exist in a vacuum. It serves and is molded by other systems in the human mind. Because it is used for conveying ideas, its structure and function must reflect these ideas... Because it is used for communication within a complex social and cultural system, its structure and function are molded by these forces as well. Yet once people have learned how to use language, it wields a power of its own. It aids them in thinking about some ideas and hinders them in thinking about others. It molds many aspects of their daily affairs." {11} The shortcomings of the Code Model are listed by David Weber: communication is not mechanical, and more is needed than grammar and a lexicon. {12} As the name suggests, Source-Meaning-Receptor theories of translation place emphasis on meaning, which includes context and intention. A key principle is relevance, which states " that the hearer is justified in expecting to get adequate interpretive benefit for minimal processing effort" — perhaps explaining the compact nature of poetry, the formalism that communicates through small changes in word or word order.

SMR theories are technical, and involve parsing, lexical lookup, and computing a semantic representation, where representation is via agreed standards (just as it is when we read Racine or eighteenth-century English verse: we have to respond through the conventions that apply.) Meaning is an interpretation, an inference based on what is called explicature and context. Both explication (making implicit information explicit in the text) and context require a good deal of background information to be communicated, which poetry supplies indirectly, through conventions and previous examples (especially so in classical Chinese and Persian poetry, which are conservative and backward-validating). But then our brain processing is modified by replication: what we continually think or do predisposes us to behave in the same way afterwards, so that these models — which are much more detailed and technical than this note suggests — replace a view of meaning as entity by activities common to communities, which Dewey stressed long ago.

To aim for the ideal translation, one which perfectly captures the "meaning", may therefore not simply be posing a difficult task, but strive for what doesn't exist. Meaning is always underdetermined, context-bound and partial. Current work on classical translation theories in China tacitly accepts that the results will be culture-bound, but does not see this as a shortcoming. {13}

Hermeneutic Approaches

Writers have generally preferred the second approach, which embraces the continental school of hermeneutics. We live in a world of our reading and understanding, a dialogue between foreign traditions and our native literature. How that dialogue will work in practice we cannot know except by trial and error, by living out the possibilities and following the new directions suggested.

Experience, said Dilthey, involves immediacy and totality. Immediacy gives meaning without ratiocination. Totality requires the meanings have sufficient weight and significance to unify the myriad moments of a person's life. Dilthey was talking about historical experience, but both factors apply to literature. In place of Kant's appeal to the synthesizing role of individual judgement, Dilthey appealed through individual creations to concerns of the community at large. Gadamer urges a wider concept of verification, for which he turns to games. Games have autonomy: they absorb the players, and have rules and a structures of their own. Art similarly absorbs both artist and viewer. Also like games, art does not permit unlimited free expression. The "right" representation has to be respected — "right" for the medium, and also representing something lasting and true, self-verifying though not self-evident, lasting through the changing circumstances of a man's life, showing itself in continually being re-experienced. "Right" does not come about through pouring effort into a certain conception of art, nor in slavishly following certain rules, but something which emerges in the hermeneutic struggle of artistic creation, the continual adjustment and readjustment of concept with medium, and of individual view with the wider social truths.

Yes, but how exactly? If all things are relative, no one interpretation is to be preferred to another.

Roman Ingarden believes that texts preserve the acts of consciousness on the part of their writer, which are then reanimated in various ways by the reader. One can distinguish four levels in a text — word sounds, meaning units, perspectives controlling states of affairs, and represented objectivities. Particularly prevalent in the last two levels are gaps or indeterminacies, which the reader fills with his own creations. But such gaps are not filled in an uncontrolled fashion, argues Wolgang Iser, but through a process of retrospection and anticipation that can overturn the text's "prestructure", the coding of the reader's usual habits and expectations. Reading indeed is a variable, complex business, but we can still arrive at an appreciation that broadly respects an art's form, with its fidelity to experience, emotive expression and internal consistency.

Unfortunately, these snippets, lifted from the theory section (see individual page references to read further) skate over many difficulties. In detail the theories are diffuse and contested, difficult to test or apply. {14} {15} Professional translators often ignore them. {16} In fact, as work on the Wang Wei piece suggests, continued practical application is still far more useful than literary theory. {17} {18}

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


1. Black Marigolds: A free interpretation of the Caurapañcáziká. E. Powys Mathers, pp. 66-77 in Mark Van Doren (Ed.) An Anthology of World Poetry (Albert and Charles Boni, 1928). Also reissued as Black Marigolds and Coloured Stars. Edward Powys Mathers (Anvil Press Poetry, 2004) "A heady mixture of rapture, longing, sorrow and imminent death", an Amazon reviewer called the translation. Information and selected verses on
Complete text is at: NNA. .
2. Miller, Barbara Stoles. Phantasies of Love-thief: Caurapancasika Attributed to Bilhana (Columbia Univ. Press, 1971).
3. Gertrude Clorius Schwebel, The Secret Delights of Love by the pundit Bilhana (from the Sanskrit). (The Peter Pauper Press, 1966).
4. Readers may want to check my rendering with online dictionaries. Cologne Digital Sanskrit Lexicon: Capeller's Sanskrit Dictionary: Apte Sanskrit Dictionary Search: NNA.
5. Aggarwal, Vinod, The Imagery of Kalidasa (Eastern Book Linkers, 1985), 188-200.
6. Sarah L. Russell. The most exquisite poem I have ever read is... Nov. 2004.
7. Valéry Larbaud. Extract from "Poésies de A.O. Barnabooth". Franceweb. NNA. .
James Elroy Flecker. The Literary Encyclopedia.
9. Robert E. Goodwin, Aesthetic and Erotic Entrancement in the Sakuntala in The Playworld of Sanskrit Drama (Motilal Banarsidass Publishers, 1998), 56-7.
10. Translation Research Group. 2003.
11. Herbert H. Clark and Eva V. Clark, Psychology and Language: An Introduction to Psycholinguistics (Harcourt Brace Jovanovich, 1977), 515, quoted by Michael Marlowe, The Effect of Language upon Thinking. Apr. 2004.
12. David J. Weber, A Tale of Two Translation Theories. 2006. NNA
13. Wang, H.Y. Critique of Translation Theories in Chinese Tradition :in Chinese (Hubei Education Publishing House, 2003) Short review in PDF form (English).
Craig A. Hamilton and Ralf Schneider. From Iser to Turner and beyond: Reception theory meets cognitive criticism. Winter 2002.
15. Leona Toker. If Everything Else Fails, Read the Instructions: Further Echoes of the Reception-Theory Debate. Originally published in Connotations 4.1-2 (1994-95): 151-164.
. Doug Robinson. 22 Theses on Translation. Originally published in Journal of Translation Studies (Hong Kong) 2 (June 1998): 92-117.
17. Ma Mon Hoc. The Nature of Meaning. NNA.
18. Quoted by Li Youzi in On the Subjectivity of the Translator. Oct. 2004.


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