Tag Archives: proofreading

Translate meanings, not words

I’m not a fan of Translation Theory. Translation is a very practical activity: it involves making judgements about the meanings and connotations of a text and deciding on the best way to convey these into another language. A good solution for one problem in one situation may well be a terrible solution when applied to an apparently similar problem in a different situation. At best, Translation Theory helps translators identify starting positions from which to solve problems; at worst, it provides them with highfalutin justifications for bad solutions.

Personally, I’d swap all the Translation Theory in the world for a few good maxims. And if I was only allowed one maxim (my Desert Island maxim, so to speak) it would be this:

“translate meanings, not words.”

It would be great if we could train translators just by shouting this at new members of the profession, repeatedly and at ever higher volumes, like a British tourist on the Costa del Sol. Unfortunately, life is never that easy.

So the first thing we need to do is to add some clarification. The key word in this maxim is ‘meanings’ and by meanings here, I refer not just to the referential meanings but also to all the other connotations of language: style, register, literary effect and the like.

Even so, this maxim is hard to apply. Most novice translators default into literal translation. When you point out that they have ‘translated the words, not the meanings’ they often stare at you blankly. And when you suggest that some passages of their translation are awkward, unclear or even nonsensical, they are wont to reply along the lines of ‘but that’s what it says in the original.’

So I’d add some practical advice to this maxim:

“Read your translation as you produce it. Reread it once it’s complete. Read it again once you think it’s perfect. If the style is awkward, change it. If the meaning is unclear, clarify it. If you find yourself writing nonsense, either you’ve badly misunderstood the source text or expressed yourself poorly. And if the reader needs to understand the source language to decode your translation, you haven’t done your job.”

Any decent professional translator will recognise all of this as something that they do instinctively, so much so that it feels like common sense. How else could one translate? Well, if you want to see what happens when somebody translates without applying this approach, try reading Ann Goldstein’s translation of Elena Ferrante latest column for the Guardian.

It’s quite short (just 400 words – one or two hour’s work for a translator) but manages to pack in a series of errors, including at least one which is catastrophic.

The subject of the article is laughter and how this provides temporary relief from the constraints of our lives. Ferrante introduces her topic with a childhood anecdote.

“I remember a design that was very amusing to me as a girl. You have to imagine the sign that prohibits honking: a trumpet in a circle, crossed out by a diagonal strip. Next to it is a convertible, and a slow-moving pedestrian who keeps the car from proceeding. The driver is leaning out over the windshield and playing the violin in the pedestrian’s ear. I laughed, and my girlfriends said: ‘Why do you find it so funny?’ ”

I read this a few times and really couldn’t make out what was going on. What was this ‘design’ that so amused the young Ferrante? It sounded like a road sign. Did 1950s Italy specialise in humorous road signs? It seemed unlikely. And then I remembered that the word disegno in Italian doesn’t usually mean ‘design’ at all, but is more typically a drawing and, by extension, a cartoon. The next two paragraphs are devoted to further discussion of the ‘design’ but unless the reader has realised that Ferrante is actually referring to a cartoon, they don’t make much sense.

Now, it’s true that any translator can unthinkingly reproduce misleading cognates in their target language. Not for nothing are these words called false friends. A decent translator, though, quickly develops the habit of questioning such cognates and will often instinctively avoid them. Fortunately, when really treacherous ones slip through, they’re easy enough to spot. One review of Goldstein’s version above should have alerted even the most lackadaisical reader to the glaring error.

If you manage to struggle through the first four paragraphs of the article (and I’ll come back to a couple of other issues in a moment), you will then meet the following sentence:

“Laughter for me can do only this: stretch what is tense to the point where it is unendurable. Otherwise it seems to me overrated.”

At first sight, this sounds reasonable enough. Until you stop and think. Is Ferrante really saying that this (stretching what is tense…) is the only thing laughter can do? That’s what Goldstein’s version says, but laughter can, undoubtedly, do lots of other things as well. Either Ferrante doesn’t know about laughter’s other qualities, or what Goldstein was trying to say was:

“Only laughter can do this…”

(Laughter can do other things as well, but there are no other phenomena capable of stretching what is tense.)

That makes much more sense, and is confirmed by the next sentence:

“Otherwise it seems to me overrated.”

If, as Goldstein has it, there is only one thing laughter can do, then the ‘otherwise’ is somewhere between superfluous and just wrong. If, as I assume Ferrante had it, laughter has at least one unique quality, then the ‘otherwise’ makes perfect sense: despite this quality, laughter is not all it is cracked up to be.

Again, reading the translation for meaning would have picked up this problem immediately, and even an inattentive translator should have been alerted by the glaring non-sequitur that was a side effect of the initial error.

There are, in my opinion, at least two more clear errors in this text. In the penultimate paragraph, Ferrante/Goldstein writes:

“Ridicule, yes, annoys the powerful, but it doesn’t bury them. Yet for the moment we’re laughing, we feel their grip on our life relax a little.”

In context, we understand that ‘for the moment’ means ‘right now’. But it’s odd. ‘For the moment’ in English usually means, ‘at this time (and until something changes)’. For example, “I’m quite happy in my job for the moment (implication: but I might look for a new one next year).” Here, there’s a minor slip in meaning but context forces the reader to correct it almost without realising.

And in the final paragraph, Goldstein has:

“That must be why the laughter that interests me most, in the context of a story, is incongruous laughter, the laughter that explodes in situations where laughing is inconceivable, in fact seems an enormity.”

Here, the meaning is clear enough, but the grammar is all wrong. The simplest fix is to replace the comma after the second instance of ‘laughter’ with a semi-colon. Again, most readers will do this on the fly, reading the erroneous comma as if it were a semi-colon or a full stop, probably without even realising they are doing so. That’s fine – but there’s another way of looking at this, which is that a shoddy translation is forcing the reader to do work that the translator really should have done for them.

The problems with Goldstein’s translation, though, don’t stop at these errors of meaning and grammar. There is also something forced about the structure of many of the sentences. Let’s go back to that first error:

“I remember a design that was very amusing to me as a girl.”

Even if we correct ‘design’ to ‘cartoon’, we are left with a rather tortured structure: “…that was very amusing to me as a girl”. We know what this means but, as it stands, it’s not quite English. More natural ways of saying this might be that the young Ferrante found it funny, it made her laugh or even just that she loved it.

And once you start to notice this kind of thing, you will see that the text is peppered with these odd ‘English but not quite English’ constructions:

“…a pedestrian who keeps the car from proceeding”

“…I get on well with anyone who can come up with this type of idea”

“laughter for me can do only this”

“ridicule, yes, annoys the powerful”

Let’s call these constructions ‘Italianate’. What, you might ask, is wrong with that? After all, Ferrante is an Italian writer. The problem is that these structures are quite normal in Italian (the linguistic term is ‘unmarked’) but they are strange in English (‘marked’). There’s nothing wrong, per se, with strange or marked constructions. Indeed, without them original writing would be impossible. But there is a problem when the translator takes ‘unmarked’ constructions from the source language and routinely translates them with ‘marked’ ones in the target language. The effect is to make the original text seem stranger than it really is, and to render it unnecessarily difficult to read and to understand. It’s what we call ‘translationese’.

If you think I’m being too harsh or indulging in cherry-picking, I’d reiterate that all of this occurs in a 400-word translation that would normally take a translator between one and two hours to produce. Or, to set it in a professional context, if I was doing quality control for a client and they asked for my opinion on this text, my verdict would be that the translation, as provided, is not fit for purpose, and that they should remove the translator from their database of suppliers.

I would also suggest they pass the following feedback on to the translator:

“Read your translation as you produce it. Reread it once it’s complete. Read it again once you think it’s perfect. If the style is awkward, change it. If the meaning is unclear, clarify it. If you find yourself writing nonsense, either you’ve badly misunderstood the source text or expressed yourself poorly. If the reader needs to understand the source language to decode your translation, you haven’t done your job.”

Translator queries: more than just questions

I’ve just been helping a fellow translator to submit queries about her translation to a direct client, and it made me think about why we send queries to our clients.

Maybe the question seems stupid. Surely, we send queries when there is something we haven’t understood in the source text or if we are unsure about target language terminology, particularly if it might involve in-house language?

Not quite. Before submitting queries, I would ask the following questions.

Can the client reasonably be expected to help?

Seems obvious, but there’s no point asking a client who operates exclusively in the target language to help you decode the source text. Or asking a client who operates exclusively in the source language to help you resolve target language terminology issues.

How many queries do you have?

If your source text generates a vast quantity of queries, then either you have taken on a text which is beyond your capacities (your problem, not the client’s!) or the source text is seriously flawed. The latter is quite common, but the best solution is simply to make a general comment to this effect rather than to rub the client’s nose in every spelling mistake, grammatical error and logical non-sequitur. And if the text is genuinely untranslatable, then say so.

How easy/difficult is it for the client to answer your query?

Direct clients are often not language people. Vague questions (can you explain this to me?) tend to generate vague answers.

I always try to frame queries as one of the following:

  • a yes/no questions
  • a choice between a short list of alternatives
  • a request for a very specific piece of information.

And remember that clients may not have access to the individual responsible for generating the source text. For example, I quite often translate public tender specifications. My client is an LSP; their client is a (target-language) company considering submitting a bid; the author of the text is a (source-language) government defining the conditions for that bid. Nobody in my chain has access to the author, and the timeframe doesn’t usually allow for it, anyway. So it would be pointless for me to query ambiguous wordings in the source document in this instance.

Hidden benefits

It’s also worth thinking about what benefits you can get from the query process (besides the obvious one of obtaining information to improve your translation).

Start a conversation

One of the big challenges for any translator, whether working for direct clients or LSPs, is to build a relationship. Sparing use of queries will help you to start a conversation with your client. And feel free to use the exchange as an excuse to build a more personal relationship. (I often throw in little personal titbits – where I’m off to on my next holiday, a comment about my pets or whatever.)

Add quality

I’d be very wary of openly criticising a source text from a direct client, but do use queries as an opportunity to identify errors in the source text, in a non-judgemental way. For example, when translating financial reports you might be surprised at the frequency of material errors. I even alerted one client to a potentially catastrophic confusion between ‘millions’ and ‘billions’ in one text!

Blow your own trumpet

As translators, we are often invisible. People take our work for granted and only notice when things go wrong. I work for an LSP which often provides me with quite ‘rough’ source material (project documentation produced by field officers working for NGOs). I don’t generally query much in these texts, but I do usually include a comment to the effect that the source material was a little rough and ready but I think I’ve done rather a good job of turning it into something clear and readable.

Or you can use a query to draw attention to a particularly neat piece of work. I recently translated a tourism text which included the line ‘Salchichas alemanas, el arte del embutido gourmet‘ (Literally: German sausages: the art of gourmet sausage-making’). I translated this as ‘Germany – where wurst is best!‘ and checked with the client: ostensibly, to make sure they were happy with my creative approach, but really to flag up a piece of inspired brilliance in order to remind them why they needed me to translate their texts and not someone else.

Cover your back

I’m slightly wary of including this one. In fact, I think a lot of the over-use of queries by translators is probably sub-consciously prompted by a desire to avoid taking responsibility for the final product. (You can’t – it’s what you’re being paid for!) However, there are some exceptions to this rule. Where the source text is truly ropy (see above) then it’s worth mentioning it, in the politest possible terms – and perhaps rephrasing it as praise for one’s own work rather than criticism.

And where there is a specific issue which you really can’t resolve and can’t be expected to take responsibility for, then you may want that on record. I have one client who often asks me to translate software documentation while the software itself is still at the localisation stage. As a result, the target language screenshots are not necessarily available. I always flag this issue up to the client and specifically remind them that they will need to check my translations (with relevant sections highlighted) against the final target language version of the software, website etc.

George Orwell’s six rules of writing: adapted for translators

In his essay, Politics and the English Language, George Orwell railed against confusing and unclear writing.


He summarised his advice in six rules, which have been the mainstay of English style guides ever since:

  1. Never use a metaphor, simile, or other figure of speech which you are used to seeing in  print.
  2. Never use a long word where a short one will do.
  3. If it is possible to cut a word out, always cut it out.
  4. Never use the passive where you can use the active.
  5. Never use a foreign phrase, a scientific word, or a jargon word if you can think of an everyday English equivalent.
  6. Break any of these rules sooner than say anything outright barbarous.

Unlike writers, translators are constrained by their source text. (I also bridle at the absolute nature of Orwell’s rules.) So I have adapted them for translators:

  1. When translating a metaphor, simile, or other figure of speech, you don’t always need to use the direct equivalent in the target language. You may not wish to use a figure of speech at all. And if you do, try to avoid using any figure of speech that feels tired and worn out.
  2. When translating a source text that is packed with long words, remember that the best translation will often involve shorter words.
  3. If it is possible to cut a word out, consider doing so. It may make your translation clearer and more elegant.
  4. Avoid using the passive where the active would be a more natural choice.
  5. Only use a foreign phrase, a scientific word, or a jargon word if the field and register of the text mean this is the best choice or if there is no everyday English equivalent.
  6. Never write anything that is outright barbarous, even if the source text reads as if it was written by Attila the Hun.