I was recently watching an episode of Raised By Wolves in which a character tells that good ol’ joke:
When is a door not a door? When it is ajar.
The subtitles read “When it is a jar [ajar]” to make the pun clear in written form. The joke only works in English, so it got me wondering: what is the general practice for translating puns/wordplay like this into other languages?
A punk song I listened to in the past was called “Gouvernemensonge”. It’s a bit of wordplay, “Gouvernement” (government) and “mensonge” (a lie). There’s… just basically no configuration that works in English. I mean I guess I’d do “Polieticians” or something if I wanted to translate that as reasonably closely as I could, but it doesn’t come close.
A classic, practical example of your question is in the Astérix et Obélix comics, where the names of most the characters are jokes, etc., etc. Many of these bits of wordplay are inherently untranslatable. You need something else, and that’s what they do.
For instance, the druid is named Panoramix in French. In English, he’s Getafix. (The English version is funnier and more relevant than the French original if you ask me.) The bard is named “Assurancetourix” (it puns off “total insurance”) and in English, he’s Cacophonix (and for a terrible bard, that name is a brilliant pun).
So the general thing is, I guess, come up with a pun that works in the language you’re translating to. If the pun is really important, you’ve got your work cut out for you. That sort of thing doesn’t really make it across language barriers, at least insofar as English and French are concerned.
Generally, you can’t; it simply cannot translate. That’s one of the main frustrations of interpretation.
However, in some languages, some words can become a different-but-still-similar joke, if the interpreter has quick wit. Example seen on Reddit:
English - “Where do cats go after they die? Purrrrrrrgatory!” (Spanish - “Where do cats go after they die? Pur-GATO-rio!”)
Thanks for the replies.
I get that ‘they’re often just not translatable, so you cannot’, but specifically what I’m wondering is: what do subtitlers and other translators do when they get to a pun like this – would they leave it blank? Or translate it literally (and thus lose the humour)? Or something else?
Short version: There’s no one hard and fast policy. It’s context sensitive. In general, you do the best you can given time/space constraints.
In a generic subtitling, you’d never leave it blank. I mean one where there was ample time to get the job done. You’ve got the mood of the scene, the intent of the writers, and the importance of the wordplay itself, as well as the time constraints. So you pick the most important elements and make it fit as best as you can. (I’ve seen a lot of English shows subtitled into Quebecois French in my youth, so I have an elaborate frame of reference, though a dated one.)
A more recent case in point: In the movie Amélie the mentally-challenged fellow is going off about the awful greengrocer fellow Colignon and getting some good rhyming burns in. “Colignon, tête de fion,” got translated to, “Colignon, big moron.”
The character was firing off sick burns in a rapidfire, rhyming kind of way. The English translation keeps that general feel but is a swing and a miss for meaning. But it generally fit the scene, all other things considered.
If you’re talking about realtime subtitling or live interpretation, I guess that’ll depend on the wit of the person doing it. In another thread (though it’s recently active it’s just faster to recall from memory), an interpreter might just say, “He has made a joke, please laugh,” rather than even try to get that in in realtime.
That’s about the best precision I can answer. I mean, other than, “They do the best they can,” which isn’t a satisfying response.
As an aside, my version of Velocity’s cat joke would have been, “purchatoire”.
The Asterix books are heavily laden with puns (including visual jokes), far beyond just the characters names. It is interesting to get copies of the same episode in various languages and compare the dialog, panel for panel, to see what they did with the puns.
They often just invent new puns to replace the originals, and modify the surrounding dialog to make them fit in. I noted one case where they appear simply to leave a piece of dialog blank. I found one case where they re-wrote an entire page of dialog, apparently because of one pun that they couldn’t otherwise work with.
In the episode Asterix in Switzerland, the entire first two pages consists of a long drawn-out set-up for a line with four puns in one sentence – a Shaggy Dog story, effectively. (I’ve only read that in the English. I suspect that it can be made to work in French and maybe other languages too, maybe.)
So to back up, in your Raised By Wolves example, I would have gone for a different, dad-grade groaner in French instead. The door/ajar joke wasn’t itself important per se (so I’m guessing, not having seen it), but the general dorkiness of the joke was. So in this case, swap in a dorky French joke. It fits the character and the scene.
This happens to be my main field of research, so I hope you’ll permit me to provide a very detailed (but interesting, I hope) explanation.
How to handle puns and wordplay is one of the most extensively studied problems in the field of translation; there have been entire books written on the subject. Translation theorists have categorized the possible strategies as follows:
Pun to pun. In this strategy, you find a pun in the target language that (ideally) matches the meaning of the pun in the source language, or if this is not possible, you find a new pun in the target language that still makes sense in the context of the document you’re translating. Needless to say, this strategy can be extremely difficult to apply, and in many cases it’s simply not possible to use at all. An example of a successful application of this strategy can be found in the film Top Secret: in the original English version, the main characters encounter a coughing horse and ask its driver if he’s OK. The diver replies, “Oh, he caught a cold the other day and he’s just a little hoarse.” One subtitler rendered the driver’s reply instead that the horse’s voice was just a little “rostig”, meaning “rusty” in English. But in German one of the words for “horse” is Ross, so pronouncing “rostig” with a slightly overstressed first syllable makes it easy for German speakers to spot the pun. Now, of course, “rusty” isn’t exactly the same meaning as “hoarse”, but it’s close enough that the joke makes sense in the context of the scene.
Pun to conventional language. In this strategy, you simply ignore the double meaning of the pun and translate only one of the meanings. The humour is thereby lost.
Pun to non-punning wordplay. In this strategy, the translator substitutes some other type of wordplay for the pun. This can be, for example, rhyming, alliteration, or a malapropism. So in this case, the double meaning may be lost, but the whimsical nature of the text is at least preserved.
Omit the language containing the pun. Here, the translator excises not only the pun but also the surrounding context. That is, the sentence or paragraph containing the pun is simply omitted from the target document. Needless to say, this strategy also loses the humour. It is generally used only as a last resort, such as when a work of literature includes a very long string of puns that are so language- and culture-specific that they couldn’t possibly be translated using a humour-preserving strategy, and rendering them without the double meanings would be nonsensical.
Leave the pun in the source language. This strategy can work if you’re reasonably sure that your readership/audience is familiar enough with the source language to get the joke.
Introduce a new pun elsewhere. In this strategy, you would normally adopt strategy #2 for the pun in question, but then make sure to add a new target-language pun somewhere else in the document. This way you preserve the quantity of humour in the document, even if it may appear in a different place in the target document.
Introduce new material containing a pun. This strategy is essentially the opposite of strategy #4; you add some entirely new material into the target document that contains a pun. Again, this strategy preserves the amount of humour, but not its location. But it’s definitely not feasible for certain types of documents, such as films: no director is going to shoot new scenes for the foreign release just because the translator handling the dubbing or subtitling came up with some great new jokes. But when translating something like a novel, this strategy might be available to the translator.
Editorialize. This strategy involves explaining the source-language pun, using something like a footnote or endnote where the translator directly addresses the reader. Again, this is something that doesn’t work in dubbing or subtitling but might be appropriate for printed literature.
The first strategy might be viewed as the “holy grail” of successful translation, though in practice almost never works, and translators fall back to some other strategy. The main factors determining whether you can translate a pun with a pun are how related the source and target languages are, how deep a knowledge of the source and target languages you have, what sorts of resources you have available to you (dictionaries, thesauri, etc.), and how good you are at coming up with puns.
My work involves developing digital translation tools to help translators faithfully translate puns (i.e., using strategy #1). Computers can already locate puns and identify their meanings with a high degree of accuracy, so the next step is to get them to propose translations that can be checked and modified by a human translator. We’ve actually built a prototype system for this and will be testing it with real-life translators in a user study next month. If you’re interested in learning more, you can check out the website for the research project, Computational Pun-derstanding.
That’s excellent!
@psychonaut: Funny you had to include a pun in the name of your project. Can you elaborate on how computers “locate puns and identify their meanings”?
In 2017, I ran a competition where I produced a collection of punning jokes and challenged my fellow AI researchers to build systems to locate and interpret the puns. I got systems implementing a huge variety of approaches, and in the years since then others have published even more. One relatively simple way of finding puns is to have a computer look up every word in the joke in a dictionary and compare the full text of the joke with the full text of each dictionary definition. For words used unambiguously, there will usually be only one definition that has high similarity with the rest of the joke. But a pun will usually have two very different definitions match the rest of the joke. The same procedure can be used to actually interpret the pun: the two dictionary definitions that are most similar to the rest of the joke represent the two meanings of the pun. Of course, my description of the method here is glossing over a lot of the important details (such as how you determine whether any two texts are “similar”, how you handle puns based on things like proper names or idioms that aren’t in the dictionary, etc.), and there methods that take an altogether different approach. But I hope this gives you at least a general idea.
There’s a realty chain here called 我爱我家. I think it’s a great name. The pun is in the 家, which can mean ‘home’ or ‘family’. So the full thing translates as ‘I love my home’ or ‘I love my family’. To continue the punning, their website is 5i5j.com. ‘5’ is pronounced similarly to the Chinese word for ‘five’ and the letter ‘i’ is pronounced similarly to the Chinese word for ‘love’.
Now, how would you translate those, and still pun?
Interesting. It sounds a bit mechanical to me, but if you say it works reasonably well, or with a high degree of accuracy, I rest marvelled. Of course words like Punderstanding would not be found in a dictionary, would they? And identifying a malapropism requires some understanding (computers are not there yet, are they?), or would a statistical tool be enough to identify a use that is not common as wrong on purpose?
Well, for the purposes of this thread I deliberately chose a method that lends itself to explanation. There are statistical techniques based on machine learning (neural nets, etc.) but these tend to be black-box approaches; you just throw a bunch of examples at a computer and ask it to figure things out for itself. Sometimes the computer comes up with a good model, but when it does, it can be hard to know why. This approach is also hard to use for puns for two reasons. First, it tends to be extremely data-hungry: some human needs to spend an awful lot of time manually locating tens of thousands of puns and marking up their meanings so that the computer can use them as training examples. (I know this because I had to do it myself for the competition!) Second, it’s debatable whether humour is something that can be genuinely learned from shallow statistical processes, since it’s all about violating expectations.
Correct, but there are ways you could extract the two word forms. I’m familiar with work in machine translation on automatically translating neologisms formed by processes such as blending, which is also used in puns like “punderstanding”. I don’t know if anyone’s actually applied this work to puns yet.
Actually, computational linguists have been studying malapropism detection/interpretation for a lot longer than they have pun detection/interpretation. I have a colleague who was doing it 20 years ago for his graduate research project, and even he might not have been the first. I don’t know offhand what the state of the art is with respect to accuracy for this task, but I would imagine that by now at least it’s quite respectable.
Well, nice to talk to one of the people who are trying to make me redundant 
I guess you have seen this other recent thread? It is an enormous market and it is changing fast, even more so with this Covid.
Ha! I get that a lot, but it’s not at all what I’m doing. I firmly believe that humour isn’t something that can be translated with a computer alone; the best a computer can do is to make a human translator’s job easier. My research is all about building support tools for literary translation, not end-to-end translation systems that eliminate the human element. If you’re a literary translator, then you should rest easy knowing that your job is secure for the foreseeable future.
No, I am a conference interpreter. And no, the job will not disappear soon, but there are things a computer or a smartphone will do that we do not do and it will be good enough for some people in some situations.
I remember watching a dubbed version of “Spaceballs” (La folle histoire de l’espace) on the French CBC channel. There’s a scene where the good guys launch a jar of raspberry jam at the enemy ship followed by puns about “jamming the radar” and “giving us the raspberry”. The dubbed version with “de la confiture” and “framboise” wasn’t quite as funny.
I also thought the translated title was a bit bland (“The crazy history of space”) until I realised that it was a play on the translated title of “History of the World, Part 1” (“The crazy history of the world”).
Governmendacious.
Of course, in some cases (generally very closely related languages), the translation is easy, since the same pun exists in both. But that can be difficult with a language like English, which has extensive roots in two different language families: A pun in English might well derive one of its meanings from a Romance language and the other from a Germanic language, so a simple translation to either a Romance or Germanic language will lose one of the meanings. And also of course, this won’t happen at all for unrelated languages, except by extreme luck.
When I was a teenager I read a bit of Freud’s Jokes and their Relation to the Unconscious. The jokes translated into English made little sense at all. My absolute favorite was, and still is, " wie du dem Inder hast verschreiben, in der hast verschreiben." In translation is becomes, “in your rush to write the prescription for the Indian patient, you have made a mistake.” I have often presented the English translation as a typical German joke, although perhaps a little highbrow - it is from Freud after all.