There are endless challenges for the literary translator as well as the certainty that the product will never be flawless.  Some versions strive for literal precision while others seek more freely to capture an effect analogous to that of the original. The extreme of the first sort is the old Loeb Library’s facing translations which, for all their Victorian fustian, serve best as a crib of the original. The second sort might be represented by Pound’s versions of Li Po or Robert Lowell’s “imitations.” Most literary translators situate themselves somewhere between.
Some works presented as translations, however, present an entirely different set of questions. For instance, poems in which the sounds of a foreign language are rendered in something close to the same sounds in the target language with no regard for the original meaning, sometimes called homophonic translations, have appeared for the last sixty years. The tradition is generally dated from the versions of Catullus rendered by Louis and Celia Zukofsky between 1958 and 1966 and published by Cape Goliard in 1969, though these in fact represent a compromise between a reliance on sound alone and a conventional translation.
A few examples will suffice to illustrate the Zukofskys’ method. Here is the Latin followed by Celia Zukofsky’s literal translation and then the semi-homophonic version in their collection.
Multus home es, Naso, neque tecum multus homost qui
descendit: Naso, multus es et pathicus.
Much a man you are, Naso, and that you much a man it is who
comes down: Naso, much you are and pathetic/lascivious.
Mool ’tis homos,’ Naso, ’n’ queer take ’im mool ’tis ho most he
descended: Naso, mool ’tis – is it pathic, cuss.
A Latin-derived word like “descend” passes between the languages with at least some semantic relevance (including hints of obscenity), but most words are not similarly accommodating . Even were it lacking the enigmatic “mool,” this translation clearly veers in the direction of gibberish. For instance, the concluding “cuss” seems wholly reliant on sound. Yet, for the reader familiar with the original and perhaps for others as well, the Zukofsky rendering can seem oddly effective. The slangy tone of the Zukofsky certainly mirrors the colloquialism of the original, and even its sniggering indecency seems to have a place.
A slightly longer piece may provide a better measure. Here the Latin is followed first by a plain prose translation by Leonard C. Smithers (1894) and then the joint Zukofsky rendering.
Nulli se dicit mulier mea nubere malle
quam mihi, non si se Iuppiter ipse petat.
dicit: sed mulier cupido quod dicit amanti,
in vento et rapida scribere oportet aqua.
No one, says my lady, would she rather wed than myself, not even if Jupiter himself sought her. Thus she says! but what a woman says to a desirous lover ought fitly to be written on the breezes and in running waters.
Newly say dickered my love air my own would marry me all
whom but me, none see say Jupiter if she petted.
Dickered: said my love air could be o could dickered a man too
in wind o wet rapid a scribble reported in water.
Here again the anchorage in meaning the Zukofskys retained affords their verse a weird aptness, with new elements added through the “mistranslation.” For instance the repeated use of “dickered” in the place of forms of dicere suggests the repeated if minor squabbles of a couple and the transformation of a form of petere (to ask, seek, pursue) as petted suggests intimacy. The garbled syntax could be thought to represent the addled lover’s mind.
Yet the choices seem sometimes almost arbitrary. In accordance with their mixed mode of work the Zukofskys were satisfied to begin with “newly” solely because it sounds like “nulli” and to end by simply translating Catullus’ final acqua as water, there neglecting the sound altogether.
Since the Zukofskys’ Catullus was published, it has received negative reviews from Classicists and markedly mixed reviews from poetry journals. Paul Mann undertakes to speak for the majority when he says, “For most translators, the name Zukofsky represents a scandal. It is a name better left unspoken, and when it is spoken, it signifies grotesque infidelity, gratuitous distortion, the deliberate abuse of a poem for the translator’s own aesthetic satisfaction.” According to his account the “only” readers who “respond sympathetically” to Zukofsky’s book are those “devoted” to his “overwhelmingly difficult” poetry in general. 
On the other hand, to Curtis Faville “Louis Zukofsky's Catullus stands as one of the major works of literature of the 20th Century, right alongside Ulysses, “The Waste Land,” Mrs. Dalloway, Spring & All, Harmonium, The Naked Lunch, Light in August; in other words, in the company of those works which propose revolutionary, new paradigmatic conceptions of form and method.”  With only a bit more modesty on behalf of his poet, Peter Quartermain calls the book “one of the most imaginative and resourceful texts produced in English in the last half of the twentieth century.” 
To my knowledge no one has followed the Zukofskys’ method since. Yet a more extreme if less demanding procedure of homophonic translation which wholly ignores the signification of the original text, following only its sound, has become nearly commonplace. In Charles Bernstein’s widely used The Practice of Poetry, he recommends so “translating” a work from a language of which one is ignorant. Bernstein has himself essayed homophonic translation in his “From the Basque” and “me Transform – O!” 
Not surprisingly, others among the so-called Language poets, whose goal seems at times to make poetry as boring as possible, have taken a fancy to this procedure. Among the more widely known examples of homophonic translation is David Melnick's treatment of the opening of the Iliad in which Homer’s menin aeide becomes “Men in Aida” (1983).  Ron Silliman produced a new Duino Elegy under the title “Do we know Ella Cheese?” that opens
when itch scree
hurt as much
Then how's their angle
or known gun?
Honky sets selves,
his name a eye nor much.
Plows lick answers . . . 
In these examples sound is not only foregrounded; it is given unlimited license. Thus the author may arrive at a text through a process little different from free association. The words of an original in a different language serve as a basis for generating random meaning amid a constantly changing vortex of half-meaning, mistaken meaning, and willful defiance of meaning.
Another sort of false translation is the poem presented as a translation for which no original exists. In the trans-European trading about of narratives during the Middle Ages, a text not infrequently claims to be a translation of an earlier poem. If that source is unknown, it can be impossible to determine what, if anything, the present work owes to prior models. At times the assertion that the poet is merely transmitting an older story rather than composing altogether afresh is designed to enhance the received value of the work (though for moderns a claim to originality is privileged).
In order to attach bardic significance to his work in the proto-Romantic moment James Macpherson published what purported to be the works of the legendary Ossian yet which was largely original. Since exoticism served the emerging Romantic sensibility as well as antiquity, William Thomas Beckford’s claimed his Vathek to be translated from an unpublished Arabic manuscript. Whether readers did or did not believe the source was other than English scarcely matters: the same semantic spin exists in either case.
Ezra Pound’s “Papyrus,” while shaped by scholarly publication of finds such as the Oxyrhynchus papyri, also valorizes the Modernist qualities of fragmentation.
Though evoking antiquity and Sappho in particular (who mentions a Gongula), Pound’s one syllable first line implies the entire reverdie tradition of the Middle Ages. The two-syllables that follow echo the complaint continuous in poetry from the Bible  to the “Hesitation Blues,” and the three-syllable final line (drawn out in languid longing) is a name liquid on the tongue. The pretense of translation justifies the elliptical syntax and places the object of desire impossibly distant in time and place.
Doubtless the grandest monument on the shelf of translations without originals is Armand Schwerner’s The Tablets which purports to be a translation with commentary by a “Scholar-Translator” from ancient Sumero-Akkadian cuneiform. Rather like Carlyle had done in Sartor Resartus or Nabokov in Pale Fire this strategy purchases ironic distance and indicts the authority of the single authorial subject, refracting a multiplicity of possible attitudes and interpretations of experience. This is hardly the place for an exposition of this ambitious poem which, according to those who saw him, remains far less marvelous on the page than when performed by the author.
The difference in literary value between Zukofsky’s Catullus or Pound’s “Papyrus” and the mass of utterly forgettable compositions by so-called experimental writers who keep pursuing the same, now traditional, avant-garde techniques is unmistakable. In spite of the mysteriously compelling appeal of dreams to the dreamer, they seem to be, alas, composed for an audience of one, for very little is as boring as another person’s dream. Elementary school students now compose exquisite corpses (after their unit on haiku, perhaps) which have precisely the appeal of those by celebrated poets who write in French. Tzara’s “Dada Manifesto on Feeble & Bitter Love," which prescribes the cut-up method for generating text, is itself a satisfactory poem. What results from following its instructions will, unfortunately, not be. Does anyone feel pleasure at the prospect of a half hour of Jackson MacLow? Does even Gary Sullivan read flarf?
The fundamental problem of all aleatory methods is that they remove intention from composition. Intention may be mistaken or twisted or self-deceiving or vicious, but it must be present to generate meaning. A sunset may be beautiful, but it is not a work of art because it lacks intention. Significance arises only in the interpretations of nature, not in its creation. I am not thinking here of authorial intention as the “correct” reading of a poem or story.  As Blake knew, the author may not know a new work’s potential. In fact the writer makes many revelations to readers without knowing what is going on. Yet there is some impulse of desire, some feeling of a moment, some shade of affect preserved in every work of art. The arts’ unique role arises from their ability, unshared with other artifacts, to render the evanescent gossamer of human consciousness in permanent form. The random generation of words can produce only an arid sham, what a cheeky child might call “the Avant-garde Emperor’s New Clothes.”
1. This is only a more pointed version of the imperfection of all texts. See my “Sweet Treason” in Dada Poetry: An Introduction or on this website for a fuller treatment of conventional translation.
2. Translation Review, Volume 21-22, Issue 1, 1986, “Translating Zukofsky's Catullus” pp. 3-9.
4. Stubborn Poetries: Poetic Facticity and the Avant-Garde, University of Alabama,p. 60.
5. The influential journal of translation Circumference features such homophonic translations regularly. See Horáček, J. "Pedantry and Play: The Zukofsky Catullus." Comparative Literature Studies 51.1 (2014): 106-131. Bernstein also composes in semi-gibberish, notably in “Johnny Cake Hollow.”
6. The opening scene is in a gay bathhouse. Melnick’s version of the first three books of the epic were published under the title of Men in Aida by Uitgeverij (The Hague and Tirana 2015).
7. First published in Roof V in 1978. The German is “Wer, wenn ich schriee, hörte mich denn aus der Engel/ Ordnungen? und gesetzt selbst, es nähme/ einer mich plötzlich ans Herz.”
8. See Habakkuk 1:2, Psalms 13 and 35, Revelations 6:10, etc.
9. With the publication of their influential essay “The Intentional Fallacy” in The Sewanee Review in 1946, authors W.K. Wimsatt and Monroe Beardsley questioned further the value of searching for authorial intention while others, while such a search is essential for others, such as E.D. Hirsch and M. A. Abrams.