Basho and His Translators

By John Carley

Imagine this: that flashing red led in the kitchen turns out not to be the toaster; you have just discovered that your ultra-modern eco-friendly domestic nuclear fusion plant from Ingushetti Industries is just about to go critical and wipe out not just you, but your whole county. Unless you run through the emergency shut-off procedure, that is. With only 35 seconds to go you locate your instruction manual (under the cat) only to find that it has been translated from the Ingush by a bricklayer.  Ok then, by a technician – but not by a domestic nuclear fusion plant technician, by a dental technician.  Errrm…. open wide…. Boom! On the plus side: at least you won’t have to read any more dodgy haiku translations.

*    *    *    *

Poems are more complex than instruction manuals: there are few facts; their content is a matter of interpretation. Even if we accept without question the less-than-certain premise that the poet always knows his own mind – that there is a single correct meaning which reflects the poet’s concerted purpose – we can agree that the reader will always bring their own experience into play when responding to a poem. This is a commonplace, and all well and good – the more so in the case of haiku which seek from the outset to suggest and evoke rather than describe and delimit.

But we forget at our peril that the understanding of literature in translation is heavily mediated by the same process. And this is particularly problematic if we are seeking paradigms for our own writing. Every translator is first a reader; and our reading is the product of their reading.  Or rather: of their reading, followed by their writing.

So which is more difficult: reading or writing? Developmental psychology, and parenthood, tend to suggest that humans acquire the ability to read before the ability to write. Anyone who has done Ingush at night school will attest to the fact that deciphering the stuff is hard enough, but typing it is a nightmare – even if one can locate a keyboard overlay. And of course while ploughing through other people’s poetry is always tedious when compared to luxuriating in one’s own oeuvre, most poets will agree that it is in fact easier to read verse than to write it.

How odd therefore is the widespread assumption that to be able to translate Inuit whale-charming chants one must be an excellent whale-charmer with good English. Surely it is far better to be a good whale-charmer with excellent English. After all, the Inuit part of the equation is already in existence (well it would be, if it wasn’t fictitious) – it is the translation which must now be accomplished. Ideally by someone whose excellence in English has been honed principally through the practice of poetry rather than pedagogy.  Or bricklaying, come to that.

Let’s face it, the man in the saloon bar who says, “Surely it’s impossible to translate poetry” has a point.  The difficulty isn’t so much in understanding what the source text says, as in conveying it in the language of destination. Compromise is inevitable, even between co-sanguineous tongues such as Dutch and German, or Spanish and Italian. In the case of wholly unrelated languages – Japanese and English come to mind – the sticking plasters are bound to show.

But enough theorising – let’s look at some examples. After, that is, a word from our attorney…  All the extracts that follow are from books in my own possession. I have them because I value them. The purpose of my comments is not to denigrate the work of others by exposing alleged deficiencies, but rather to illustrate some of the choices that a translator must inevitably make. Mi’lud.

Haiku, like the hokku before them, are written in Japanese as a single line (or column). Excepting some cases where the calligraphy itself is a central feature of the art, there are no spaces between characters of groups of characters; so words and phrases are distinguished by the reader from an otherwise undifferentiated text. There is no capitalisation. Such punctuation as there may be is in the form of verbalised (and therefore written) interjections which are considered as words in their own right.

The following poem is a hokku (haiku) by Basho. I give it in its generally accepted original, followed by the same text entirely in Japanese phonetic script (i.e., without ideograms) – this is separated out into individual words with the metrical phrase boundaries shown by a double line. Then comes an approximate phonetic transliteration in the Roman alphabet. And lastly there is a crude word-for-word rendering.


なつくさ | や || つわものども | が || ゆめ | の | あと

natsugusa | ya || tsuwamonodomo | ga || yume | no | ato

summer-grass  | !/:/? || warrior | ‘s || dream | ‘s | mark/remainder

A thicket of summer grass
Is all that remains
Of the dreams and ambitions
Of ancient warriors

Yuasa, 1966

To their eternal shame some commentators have sought to vilify Yuasa simply for using a quatrain to translate Basho’s hokku. As we have already noted, the source text is written as a single undifferentiated column, and the assumption that the 5-7-5 metrical structure of the original must necessarily equate to three lines in English is questionable at best. Unless one is colossally ignorant, in which case it is self-evident.

Perhaps more reasonable is the objection that the quatrain tends to a more extended treatment, and that this might induce some padding of the text.  In this case one might question whence that thicket of grass, or whether yume needs two words to convey it. And are those warriors really ancient?

In the case of thicket we know from the prose context (the Narrow Road ‘travelogue’) that the poem is not set in some huge area of grassland. There are after all few steppes in Japan. However the word itself is not in the source text. As for yume: most translations do indeed use the single word dream, but it can also hold secondary associations of ambition or aspiration. And, while no direct word for ancient appears in the original, the word tsuwamonodomo is a high register and rather archaic term that implies a historic setting – a factor confirmed by the prose context.

So has Yuasa given us pure embellishment? Are his choices distortions? Or do they simply allow the English-language reader to experience something similar to the harmony and dignity of the source text?

A mound of summer grass:  
      Are warrior’s heroic deeds    
              Only dreams that pass?

Britton, 1974

Let’s all laugh at Lady Bouchier (aka Dorothy Britton) for her blatant versification. It is after all far more rational to render highly structured poetry as free verse. Isn’t it?

Unlike Yuasa, Britton phrases her translation as a question, albeit rhetorical.  This duality is possible because of the nature of the word ya which may convey both surprise and doubt – it indicates a response to a circumstance which ‘gives one pause’. Here we are in the presence of a fabled ‘cutting word’ – a kireji, and Britton duly end-stops line one in order to generate outright juxtaposition between two phrases which, were it not for the colon, would be unrelated.

The thicket has now become a mound – although any such word is still absent from the source text – and it would seem that those ambitions have been realised, for we now have deeds. While this latter transformation may be a little problematic, the seemingly extraneous heroic is in fact rather better grounded: the possessive particle ga which applies to the departed protagonists is a literary usage which functions as an honorific. These were men worthy of respect.

summer grasses
where stalwart soldiers
once dreamed a dream

Ueda, 1992

Ueda’s text looks like a haiku is supposed to: three lines; no indents; no capitalisation; no punctuation. In short – no frills.

Surprisingly, given the conventions which had begun to concretise around this time, there is no em dash or similar at the end of line one. Like Yuasa before him, Ueda chooses to voice his translation as a single phrase and the cutting word, which sets off one phrase from the other in the source text, is not directly emulated. Here Ueda supplies the conjunction where although it might be objected that this would more readily suggest that Basho had used the post-positional particle ni rather than the exclamatory ya.

The warriors have become mere soldiers, though fortified by a reasonable, if not directly attributable, adjective. Stalwart they might be, but this alone does little to explain why they should be involved in that most unprepossessing of contemporary activities – dreaming the dream.  There is no verb in the source text: past, present or future.

a trace of the dreams
of warriors past
ah, the summer grass

Carley, 2010

Carley give us warriors past to convey the compound effect of the archaic tsuwamonodomo and the honorific ga.

As with all the translations we have seen so far there is a strong sense of an epoch almost forgotten. In the source text this derives in part from the word ato – an elusive noun which combines elements of mark, remainder, and location-of-past-occurrence. Carley uses the word trace and places it in a chain of dependencies which directly mirrors that of the original. Mirrors and inverts. For while the source text delivers summer-grass/warrior/dream/trace, the translation offers us trace/dream/warrior/summer-grass.

This translation presents us with a ‘soft’ juxtaposition pivoting on the utterance ah which, along with the comma, seeks to emulate the location, form and effect of the cutting word ya.  As with Britton, the translation offers an element of rhyme, albeit, in this case, partial.

*    *    *    *

For all their variety of approach one thing which unites these four English texts is that they all give the word natsugusa as summer grass. This is a reasonable, but partial, translation. In fact kusa (here gusa, which forms the latter part of the compound with natsu (summer), describes grasses and wildflowers that flourish together. One might imagine the English word meadow stripped of its connotations of pasture.

Similarly, while three of the translations content themselves with dream, and the other adds ambition, the word yume carries strong connotations of illusion – a factor that surely adds to the bittersweet tone of the original poem, and which, perhaps, the English texts cannot quite so readily regenerate by word choice alone. The question is: how tightly does the English word dream connote illusion? Or rather: are these connotations identical in nature and degree to those intended by the poet?

In the case of angels and pin heads one might sooner join the angels than live long enough to count them. So with translations: it is easier to learn the language and read the poem in the original than it is to advance a watertight argument for the unassailability of a particular approach. The general reader must simply accept that a poem in translation is a second-hand coat: it has been worn before. But the poet forgets at his peril that it may have been substantially altered in order to fit its new owner. Or to suit prevailing fashion.

Let us return briefly to the fabulous kireji. Almost all contemporary definitions of the haiku form in English refer to a bipartite structure which pivots or turns on a syntactic disjuncture at the end of line one or two, frequently reinforced by an em dash, or similar. Oddly though, as we have seen above, though Basho duly uses the classic cutting word ya in the expected place, only two of the translations seek to emulate it directly. And while Britton locates it at the end of line one, in the form of a colon, Carley gives it at the head of line three thanks to an exclamation and comma.

There is little here to support the standard model of English-language haiku. But such reinforcement as may be found elsewhere should not necessarily to be taken at face value either. Makoto Ueda’s Basho and His Interpreters is justly considered to be required reading for any haijin serious about their art. The following three poems all appear in that book.


さらばち | も || ほのか | に | やみ | の || よひ | すずみ

sarabachi | mo || honoka | ni | yami | no || yoi | suzumi

crockery | also/yet || indistinct | in | darkness | ’s || evening | coolth

plates and bowls
dim in the twilight –
the evening cool

Ueda, 1992

dishes and bowls
grown indistinct against
the cool of evening

Carley, 2010

Ueda’s translation adopts the familiar form of much contemporary English-language haiku complete with disjuncture and em dash. However there is no kireji in the source text which instead uses a possessive particle as a conjunction at the end of the second metrical phrase. This is, in fact, a single-image hokku  – a type which Basho considered to be the preserve of the master poet.


あき | の | よ | を || うち | くずしたる || はなし | かな

aki | no | yo | o || uchi | kuzushitaru || hanashi | kana

autumn | ’s | evening | object marker || strike | to crumble || chat | …/suchness

autumn night –
striking and making it crumble
this jovial chat

Ueda, 1992

the autumn night
brought to nothingness
by our chatter —

Carley, 2010

Again, the impression is of a standard construction in which the fragment autumn night is set off by a cutting word against the image of the jovial chatter. Unlike the previous example there is a cutting word in the source text, but rather than being a device which seperates one metrical and semantic unit from another, it is actually the terminal intensifier kana which combines the functions of demonstrative adjective, ellipsis points, and wistful exclamation. No wonder Ueda settled on the em dash!

Lastly let’s consider what is a rare, but far from unique, example:


めいげつ | の || はな | か | と | みえて || わたばたけ

meigetsu | no || hana | ka | to | miete || watabatake

harvest moon | ‘s || blossoming | ?/! | comparative marker | seemingly || cotton field

looking as though
the harvest moon has blossomed
a cotton field

Ueda, 1992

the harvest moon 
in bloom? – or so it seems
this cotton field

Carley, 2010

Were it submitted to many of our top haiku publications one can’t help feeling that this poem would go straight into the circular file (bin). It is indeed bi-partite, and it uses a regulation cutting word (in this case ka). But it does so slap bang in the middle of the central metrical phrase – a device which any dictionary definition of English-language haiku will tell us is incorrect. So how can the translator make this poem comprehensible, if not to push the break in syntax back to the end of the line? God forbid he, or Basho, should be assumed to have committed some facile blunder!


Translation is a dark art, not a science. It offers very little in the way of assurance for the poet seeking paradigms for his own writing. Just as readily the collective certainties of the audience can induce a translator to adopt those same somewhat dubious conventions, either unconsciously, or on the grounds that in this way the work will be familar and approachable … a circumlocution which the cynic would render as: might sell a few copies.

So, as Hanibal might have put it: caveat lector. The act of translation will inevitably involve interpretation, choice and compromise, while notions of comparative prosody are prone to the feedback loops of fashion.

Speaking of feedback loops –does anyone know why that red led keeps flashing in the kitchen? And what does annihilatik istanti mean?


Editor’s note: This article was written especially for Haiku NewZ.

John Carley (1955-2014) lived in Lancashire, England, was a former renku editor for Simply Haiku and the author of the Renku Reckoner (Darlington Richards, 2015). He invented the four-verse yotsumono, celebrating the form with a collection written with several authors, The Little Book of Yotsumonos (Darlington Richards, 2012). A nijuin led by John won the 2013 Haiku Society of America Einbond contest. Read Morning Heat.

Working in association with the poet and renku theorist Eiko Yachimoto John published numerous retranslations of classic Basho kasen (36-verse renku).