Boris Smus

interaction engineering

Is That a Fish in Your Ear? by David Bellos

David Bellos drops many interesting tidbits in this book about translation. Unfortunately, I feel unlikely to retain most of them. Each chapter stands on its own, which is both good for piecemeal reading, but also feels disjointed. The book struggles to tie together into an overall narrative. The author's erudition is apparent, dropping a million mostly relevant and interesting references. Sometimes this digresses into an obsession with word games, and the references can come off as gratuitous and self-serving. Overall I liked it, but there were also parts I found suspect.

What follows is a correspondingly disjointed account of what I found interesting from the book:

Finding a good match: Ultimately the goal of a translator is to find a good match. Nobody ever does word for word translation, but sense for sense to match the audience and medium. In translation, the things to preserve are information and force. Not number of words, sentence length, placement of punctuation. Speech interpreters can’t preserve tone, pitch, facial gestures, stamping of feet, etc. Hofstadter has a good example.

Fake translations: Some authors claim to produce a translation but there is no original. Don Quixote claims to be a translation from Arabic. Others claim to produce an original but just translate an existing work (Castle of Otranto). Sometimes to pass censorship, other times to try on an identity, to come off as more authentic. Dzhambul Dzhabayev was a folk singer who was “translated” into Russian but actually written from scratch. Soviet Union sought to translate great poets from Kazakh, Ingush, Dagestani etc, but had to invent them instead.

Decorative foreignism: When translating a character from one language to another, how do you preserve their cultural essence? One approach is to preserve interjections (eg. Gregor Samsa would have exclaimed "Ach Gott!"), but this only works when readers are assumed to know some phrases of the other language, so only for languages that share history (for English, German & French, but not Chuvash).

L1 and L2: Translating into your own native language (L1) vs. translating from your native language into one you are less familiar with (L2). Most literature is L1, but things like Engrish is definitely L2.

Up vs down translation: Up refers to higher prestige or more readers. Like translation into French from Russian during the 19th century. Or anything into English these days. Down is the opposite: translating the Bible from Hebrew to anything. Translating up tends to remove all traces of the original culture but down tends to preserve it because its perceived as important.

Languages and conquest: Sumerian was the up translation target for 3000 years. This despite the fact that Sumer was conquered by Akkad, who decided to keep the language and culture. Similarly, Greek language and culture became dominant despite conquest by Rome. So the economic military bad political expansion doesn’t directly relate to cultural dominance.

A culturally dominant language is one that maintains significant volumes of translation activity between itself and a significant number of languages

Languages are interesting: Kuuk Thaayorre from Australia lay out ordered objects from east to west rather than left to right. In Hopi there’s a concept of evidentials. Words change depending on whether they are in the field of view. So “the farmer killed the duck” doesn’t have enough info to translate to Hopi.

Mistrust and translation: When British PMs talks to French president they each bring their own interpreter out of fear of imperfect or deceptive performance. Turkish dragomans who would make the Ottoman decrees more palatable to other monarchs. Adage: traddutore/traditore. (But this may just be a coincidence).

Asymmetry of translation: There are few vehicular languages (regional lingua franca) between which vast majority of translation occurs. There are far more non-vehicular native tongues. Even among the vehicular there are great disbalances, eg Swedish gets as much attention as Chinese despite 100x fewer speakers. English is the dominant language today, 65% of all translations are from English and 7% are into English. To really do well globally you need to read and write English.

Hub languages: It’s common for English to be used as the inter-language too, for example for going from Chinese to English to {Finnish, Estonian, Latvian, etc}. Simultaneous translation also uses a hub and spoke model. Rather than going N to N languages at the EU, and requiring N^2 interpreters, use a hub language like German so everyone translates to and from German instead, only requiring 2N. Bizarre that most translators into English are amateurs, despite being important in the global translation flow (as a hub language). Translators into Japanese are revered and have almost author like status.

Translation vs. language learning: In India, there was historically very little translation between regional languages. (Translation is always a crutch for our inability to quickly learn other languages.)

Gestures as a language universal: Moving hands while speaking (not reading aloud) is done in all languages. Interesting commonality between eating and speaking, both requiring hands and mouth. Many other examples of universal gestures: slapping someone on the back, punch to the gut, shrug of shoulders.

Random tidbits:

  • Some languages make distinctions between spoken and written words, eg. paroles vs mots in French.
  • Adair translated Poe's Raven into The Black Bird without using the letter "e".
  • Eskimo has 40 words for snow, but English has 40 words for coffee.

Questionable claims:

  • in German, "spoken word" and "written word" are different words. This doesn't seem to be true.
  • Humans are unique in their aptitude to vocal learning. Also not true: bird songs do it, as do cetaceans and some bats.
  • Author feels that translation between modalities (eg. book => movie) is about as relevant as translating flour into noodles. Cool story.

I end with food for thought:

Babel tells the wrong story. The most likely original use of human speech was to be different, not the same. [...] Languages merge when peoples do. [...] Translating is a first step towards civilization.