This was originally a comment on one of Larry's posts, but I think it got lost in a spam filter somewhere.

Larry, of David R. McCraw's Du Fu's Laments from the South, sez:

The preface claims that one reason Du Fu doesn't get the respect he deserves as one of the world's great poets is that he hasn't been effectively translated, and correctly admits that this book will not change that -- these renderings aim for compressed and end up crabbed, with recondite vocabulary and rebarbative punctuation. (Some frequently used words, like "alcedine," are in no dictionary at my disposal — my best guess there is "kingfisherlike.")

This set me off on a trail that ended at "Cedules from a Berkeley Workshop in Asiatic Philology", in which author Peter A. Boodberg argues (in cedule 14, "On Chromatographic Effects in Chinese Poetry"):

The rich spectrum of Chinese chromatonymy, multilined and multibanded, has not received the attention it deserves. Most chromatonyms are not too well defined in our dictionaries, and translation equivalents are chosen haphazardly according to context, with little consideration paid to semantic nuances. Among the many Chinese color-terms crying for simple and effective rendering is the adjective TS'UI <ts'jwəd (C124, 'feathers', as semantic, +tsu<ts'jwət as phonetic) /A/, 'vivid green-blue-purple-black', originally descriptive of the glossy iridescent plumage of the kingfisher, TS'UI being the second hemiphthegm of the dissyllabic name of the Asiatic kingfisher (Halcyon), FEI-TS'UI /B/. 'Kingfisher-green' (-blue, -black, -brown) is an awkward polysyllabic way to translate TS'UI which may describe women's penciled eyebrows as well as foliage. With due regard to the fact that kingfishers in Chinese literature were probably both Halcyoninae and Alcedininae, is there any reason why we should not use the term ALCEDINE (from L. alcedo, 'kingfisher') to designate exactly what TS'UI connotated to the Chinese? ALCEDINE is a handsomely tailored word, sonorous and precise, yet broad enough to be safely applicable as a color-epithet to a variety of things.

I... I think this is where the word "alcedine" was invented. I don't have OED access, but a quick glance through the Google Books suggests that in English the word isn't used outside the fields of (a) ornithology, and (b) Chinese classics. You can see how this situation would arise, too: there's a concept in classical Chinese that's important enough to appear quite regularly, but difficult to translate into English. Someone invents a word for it. The next generation of Chinese scholars encounter the word as part of their studies, figure out from the context/source text what it means, and latch onto it as the ideal translation for TS'UI, never realizing that it is actually artificial Chinese-classics-in-English jargon that no-one else knows (except ornithologists, among whom appreciation of the sublime chromatography of Chinese poetry rises dramatically).

The thing is, even as someone who is (by my estimates) about 1000% more interested in Latin than the average reader of classical Chinese poetry in translation, I did not know the Latin word for "kingfisher." Alcedine is completely impenetrable to me, preferable to a straight borrowing of ts'ui only because it's less obviously alien. The fact that it is no longer 1955 no doubt has something to do with this shocking disconnect from the roots of Western Civ, but I took heart from this passage in Eugene Chan Eoyang's Borrowed plumage: Polemical essays on translation, commenting on another of Boodberg's proposals ("WANG 王 ... may best be metonymized as BASILEARCH"):

I have quoted this dense exegesis at some length, not only because of what it says but also because of how Boodberg says it. It represents a variation of the "vehicular matching" that we encountered in Sternberg's scheme. We may note in passing that aside from the Greek in the exegesis, there are a number of infrequently encountered English words whose meaning can be fairly well adduced from the context but are disconcerting nonetheless: "coadunation," "affines," "protograph," "anthelion," "paragram." [...] One might not unreasonably ask why an exegesis of words requires its own exegesis.

As for the reliance on Greek, it's true that Boodberg came from a generation in which the educated were more likely to know Greek than not, but even so, why should it be reasonable to require of a reader who doesn't know Chinese that he know another foreign language — especially a language as hoary as ancient Greek — in order to understand the true meaning of Chinese?

[...] And what about the principle of familiarity: should the reader of a translation not be as familiar with a term in translation as the reader of the original with its counterpart in the original?

Boodberg takes the notion of "vehicular matching" even further than Sternberg develops it. He is addressing an audience at least as polylingual as he is; but he is more polylingual than the normal speaker of the English language — if the locutions which he uses in his own prose are any indication. Indeed, in his era, Boodberg expected his students to be post-Babelian; only the most gifted attempted the study of the difficult non-Western languages — Arabic, Chinese, and Japanese — and only after mastering the most formidable Western languages — Latin and Greek. The study of Chinese was therefore the enterprise of la crème de la crème, which by definition would be a very small elite. The consequences of this approach, natural as it was for Boodberg and for sinologists of his generation, makes no sense today, when the study of Chinese is no longer restricted to the classical philologist.

I love the way Prof. Eoyang's criticism of ostentatious polyglottery is seasoned with just a little French. Listen, we may not know Ancient Greek, but we're not barbarians here. (Nobody say it.)

Incidentally:

cedule (plural cedules): 1. (obsolete) A scroll; a writing; a schedule.