‘Could a rule be given from without, poetry would cease to be poetry, and sink into a mechanical art. It would be μóρφωσις, not ποίησις. The rules of the IMAGINATION are themselves the very powers of growth and production. The words to which they are reducible, present only the outlines and external appearance of the fruit. A deceptive counterfeit of the superficial form and colours may be elaborated; but the marble peach feels cold and heavy, and children only put it to their mouths.’ [Coleridge, Biographia ch. 18]

‘ποίησις’ (poiēsis) means ‘a making, a creation, a production’ and is used of poetry in Aristotle and Plato. ‘μóρφωσις’ (morphōsis) in essence means the same thing: ‘a shaping, a bringing into shape.’ But Coleridge has in mind the New Testament use of the word as ‘semblance’ or ‘outward appearance’, which the KJV translates as ‘form’: ‘An instructor of the foolish, a teacher of babes, which hast the form [μóρφωσις] of knowledge and of the truth in the law’ [Romans 2:20]; ‘Having a form [μóρφωσις] of godliness, but denying the power thereof: from such turn away’ [2 Timothy 3:5]. I trust that's clear.

There is much more on Coleridge at my other, Coleridgean blog.

Monday 11 February 2019

Pervigilium Finneganis

So I translated Finnegans Wake into Latin. You can buy the resulting volume as an e-book from Amazon for £1.54: all four parts of the Wake, 453 pages, plus an introduction by me. I'm not expecting you to spend money on it, of course. You're not insane. But it's a thing I have done.

Why have I done this thing? I explore the whys and hows in my preface and don't want to duplicate that material here. Obviously it's a kind of joke, and the joke depends both upon its pointlessness and its egregiousness. I might have just floated the idea, and maybe translated a few pages; but if a thing's worth doing, it's worth doing well, and that counts double if a thing was never worth doing in the first place. I feel Joyce would agree with me here.

Translation is a kind of reading. Indeed, it is a kind of very close reading. I wanted to give myself a reason actually to read the Wake, and I have now done that. But I accept that there's an aspect of this, taken to the lengths I have (manifestly) taken it, that looks a little ... weird. The sort of person who would sit with a copy of the Wake open in front of them, Lewis and Short at their elbow, and a pencil in their hand, who would go, sentence by sentence, through the whole of this immense, immensely complex novel turning it into carefully considered Latin is not the sort of person with whom you'd want to spend your free time. That's not what I did. You probably wouldn't want to spend your free time with me anyway, I'm well aware, but: that's not what I did.

What did I do? I collaborated with the machine. I fed text through Google Translate, English → Latin. Of course, that didn't get me very far: (a) because that programme is famously unreliable in itself when it comes to the accuracy and idiom of translated text, and (b) because most of Finnegans Wake registers as gibberish to Google Translate (obviously). So I had to fix-up both text and translation, and it was this process, rather than any fetishisation of a Latin Wake as such, that interested me; like one of those avant garde composers who works with prepared pianos. It's text and machine and the intriguing stuff that comes out the far side. That's the really compelling thing, I think.

Somebody needs to engage critically/theoretically with Google Translate. It's a fascinating programme in many ways. It doesn't ‘translate’. It compares source text with a triaged selection of already existing online translations and spits out the result. It's the world's most voracious and least discriminating ‘reader’ of texts. Puts professors to shame.

Anyway: at odd moments over a couple of weeks, and depending on my time and inclination, I would prep chapters of the Wake and feed them to the machine. Then I would work through the raw quasi-Latin that came out the other side. This involved sieving the text through various filters: sometimes search-and-replacing specific things throughout the whole text, sometimes going through line by line and altering bits of the finished project. Quite a lot of Latin is included in Joyce's original Wake actually, and in those cases I had to translate into English, obviously (there's some Greek too, and those bits I translated, arbitrarily enough, in French). And there were lots of other fiddly things that proved needful, or so it seemed to me. Indeed, I could easily have spent many months going through the text, titivating it in various ways, Latinising the myriad unrefined (or uncontaminated, depending in your point of view) nuggets that had passed through the filter of Google Translate unLatinised. But I had neither the free time, nor anything like the requisite motivation, to do that. Instead I mucked about with the text in various ways and then just walked away, leaving some parts of the Pervigilium Finneganis worked by me, some Latinspattered by unrevised Google Translate, and a few spots of the original Wake-y canvas visible between the application of the paint, like a Pollock painting. I worked quickly, until I reached a point where the book looked texturally interesting when I, as it were, stepped back and squeezed up my eyes. The judgment becomes whether its scribbled and scratchy texture approximates to the scribbled and scratchy texture of the Joycean original.

But now I'm mixing my conceptual metaphors. Is the finished product more like a painted canvas (of sub-sub-Pollockian sort) or more like an avant-garde composer's prepared-piano composition? The latter has the advantage of emphasising the machine, my co-translator (or lead translator). As Ezra Pound’s ‘A Retrospect’ (in 1918's Pavannes and Divagations) says, there's really no reason why learning to write poetry should be any more difficult than learning to play a piano. Even a prepared-piano machine. That mismatch interests me, actually, because whatever else we might say about Joyce's novel we will surely agree that it was elaborately and laboriously hand-made, over many years. My ‘translation’, by contrast, was machine-generated, quickly.

Machines have a relevance here, nonetheless. We're not (does it tempt fate to add: ‘yet’?) in a place where machines can actually do our thinking for us, much less our dreaming; but nonetheless the night-book that the Wake represents is surely about the way specific processes of thought over which we do not have conscious control infiltrate, shape and direct our minds—it's how dreams work after all. And at least in some ways those processes are pretty well approximated by machines like Google Translate. A machine like Google Translate obviously can't ‘think’, much less ‘create’; it does not look at a line of English and actually translate it into Latin. But what it does do, by sifting through and sorting the slumbering assemblage of human thought and creativity represented by centuries of human translations, is to bring a sort of collective subconscious of Latinity to bear on the source text, rising up from a vast passivity of history and culture to manifest itself, in immanent but unpredictable and sometimes eccentric ways; and this is not, when you come to think about it, a million miles from what Joyce was doing with history, myth and language in his original novel. It's the 21st-century now. Conceivably it's time the machine took its place in the ways Finnegans Wake is read.

If there is any merit in my project here [narrator's voice-over: there was no merit in the project] it might be in the way it shifts attention from the semantic meaning of the Wake and towards its texture. I don't know the (expansive!) body of critical literature on the book at all well, but the little I have seen tends to default towards elucidation, for instance by unpacking the various meanings in the strange neologisms out of which the book is construed. That's fine, obviously. Indeed, it's a common-sense way of approaching this puzzling work, even if it runs the risk of reducing it to a sort of massive screed of cryptic-crossword-puzzle clues. But there are other ways in which we can approach the Wake that aren't to do with unpacking the meanings of its words. We might be interested in its music, in its textural balance of roughness and smoothness (there is a lot of alliteration in the novel, for instance, and overall I'd say there are many more consonants here than there are vowels to service them). We might want, instead of engaging in the process of making the Wake easier to read, to explore the ways it resists being comprehended, even to the extent of producing a midrash upon it that moves comprehension markedly further away from us. That seems like it might be an interesting thing to do, don't you think? At any rate, that's what I did. My partner G.T. and I rustled up the text. Then I sourced an out-of-copyright image for the cover art (Gustave Doré's ‘Tom Thumb and the Sleeping Giant’: reproduced at the head of this post. It works quite well, I think, as an image for this novel) and banged the title on it before posting the whole thing to amazon's Kindle e-book platform. Bish, bosh.

One final thing occurs to me, now that it's up. A Roman would struggle to pronounce ‘Joyce’ after the manner we say it nowadays, with that lovely digging-in /dzh/ at the beginning and its bouncy /oi/ dipthong. S/he would instead utter the ‘J’ as ‘I’, the ‘y’ as ‘u’, would say the ‘c’ hard (as ‘k’) and s/he would pronounce the terminal ‘e’. The result would sound, it occurs to me, rather like the phrase ‘are you OK?’ as sung by Michael Jackson (‘Annie are you OK?’) in his hit single ‘Smooth Criminal’:

That's how I propose to pronounce ‘Joyce’ going forward, at any rate.


  1. Hmmmm....This reminds me of that (perhaps) strangest form of digital humanities, deformative reading, in which one uses a computer to rearrange a text in more or less arbitrary ways in hopes of stimulating the critic to interesting flights of interpretive fancy. It also reminds of the (inspired?) weirdness that ensues when one feeds Google Translate with nonsense strings. Language Logs has a number of posts on this subject. You can find them collected under "Elephant semifics", which gathers in some other nonsense as well.

  2. I love the fact that the Amazon suggested reading age for this is 11-18.
    You're spot on about Google Translate, by the way.

  3. This comment has been removed by the author.

  4. Ha- really interesting idea and I enjoyed this blogpost a lot. Finnegans Wake seems to have a long and global history of translation despite being clearly untranslatable in any normal sense of the word. Or even written in any particular language, I suppose.

  5. Latin me that, my trinity scholard, out of eure sanscreed into crowdblast in cellelleneteutoslavzendlatinsoundscript
    my enthusiastic to you braintaking endeavour and the sense of sharing the intellectual pleasure that you give me who translated FW into french with that kind of attitude, if I may https://sites.google.com/site/finicoincequoique/home

  6. This has inspired me. I'm currently 'translating' the Wake into asemic writing.

    1. I approve. There's a strong asemic element to what I have done here. (That, of course, may have been your point).