If you think English words are hard to translate, look at some foreign ones
From The Economist, June 12th-18th 2004 (pg. 55):
A leapfrog appeal is an interesting bit of the English legal system: in a case of sufficient public importance, the appeal may go straight from the High Court to the House of Lords, bypassing the Court of Appeal. For those familiar with the children's game of leapfrog, it's a useful metaphor. But how to translate it? The game of leapfrog is Bockspringen in German (literally deer-jumping), saute-mouton in French (sheep-jump).
There are circumlocutions and rough equivalents -- Sprungrevision (literally "jump-appeal") in German, for example. But in a survey of 1,000 professional legal translators by Today Translations, a specialist London firm, "leapfrog appeal" was voted the hardest English legal phrase to translate, closely followed by "toxic tort" (harmful exposure to a poisonous chemical). Other troubling legal arcana included "Michaelmas term" (court sittings between November 2nd and 25th).
It cuts both ways. Untranslateable words, especially from big widely used languages, usually migrate untranslated: panache and schadenfreude are now English words, le weekend and das Briefing are embedded in French and German. But from smaller languages, things can be tricky. Esperanto, for example, has a verb krokodili meaning "to speak your native language when everyone else is speaking Esperanto" -- more euphonious than useful, perhaps.
The survey highlighted ten particularly difficult such words. Top of the list came ilunga, a word in Tshiluba (a central African language) for a person who tolerates abuse only twice. That was closely followed by shlimazl, Yiddish for a chronically unlucky person. Like a legal translator, perhaps.
We have a subscription to the Economist, and I've been doing some catch-up reading of recent issues to sharpen my view of current events. This was one 1/4 page article that really caught my eye, probably due to my somewhat amatuerish interest in linguistics. Please note that the Economist doesn't run by-lines for any of its articles, which helps explain the reason I did not cite the author's name. (This practice leads me to prioritize my reading by subject matter rather than authorship, which is both good and bad. I'd prefer to follow specific columnists, and then fill in the holes on the subject matter, but I'm certain this way is exposing me to a more balanced view of certain current event situations.)