Thanks for this great contribution from the world of linguistics, Jim. For me, your article was both as a quick 'refresher course' and a useful insight into how we translators can use the classifications of linguistics to refine our approach to our work. Your exposition of the different "appropriate levels of transfer" (the haiku example illustrates this nicely) was really useful.
In my experience, it is the pragmatic level that often matters most in translation (especially in translation for industry). As an independent translator, what I want is to keep my customers happy. What they want is to keep their customers happy, and so I help them achieve this by creating a text that functions in the required way so that the target readers are completely satisfied with the text. Translation discussions that centre on what the 'equivalent' word is in the target language rather than on the appropriate process to use in translating a specific text are putting the cart before the horse, IMO. (That’s not to say that we don’t have to translate at the word level too - just not before and to the exclusion of other levels.)
Your point about genre also playing a role in determining the linguistic level of a translation, and specifically, about the discourse level being the dominant level in technical translation (“translating the instruction manual for some piece of equipment”), was especially interesting in this regard. This is commonly perceived to be an area of translation where the syntactic and semantic levels matter most, based on the belief that a technical text is highly specific and therefore the source text should be closely adhered to in order to avoid mistakes or misrepresentations. Based on my experience as a technical translator, however, technical texts are often quite prone to vagueness and imprecision (this is possibly also a factor of German technical texts). My job as the translator, as I see it, is to figure out what the source author really meant (sometimes contrary to what he says), filter out the ‘padding’ so common in German, and re-express it in good English – in other words, to work largely at the discourse level, plus a bit of the pragmatic. The customer is then delighted when I hand him a high-quality, well-expressed text, with not one word too many, that tells the reader everything he needs to know.
Unfortunately, as translators, we sometimes do not to trust ourselves enough to make these kinds of decisions (and I include myself in this) - we sometimes cling to the source text rather than trusting our own judgment to work at the appropriate level and to create the target text as we see fit. Your article reminded me of the complexity of the translation process and the careful consideration that it deserves.

I'm a professional translator, working "only" in the tech/medical/chemistry areas. I agree with most of Orla's comments and the OP.

For Chinese, though, it should be mentioned that that language presupposes intelligent listeners.

haizi mai le shu
child buy PERF book
the child bought a book

That sentence might also mean "The child bought books". It's all about context.

And for the Swedish example, to

A: Tror du att han ska gifta sig?,

I'd in a conversation probably reply "I helvete heller!", which "literally" would be something like "In hell, either".

When I do a manual, I always try to imagine that poor fellow standing in front of the gadget/machine/whatever, trying to make it work using my translation.

My favourite customers are those who understand this kind of reasoning, and don't try to enforce some kind of EU legal principles, thinking that any future modifications might be "change the third word in the second sentence on page 6 to 'X'".

I've translated nmreous manuals for blood glucose meters. I don't bother one bit about what the original manual says, but understand how they are supposed to be handled, and write instructions for that procedure. The customer (and, provenly, the end user!) is as happy as me myself.

I don't agree with Orla's views on German padding. When discussing word counts, I often use this example:

English original: Now please press button A.
Swedish: Tryck på A.
German: A drücken.

Orla - I'm glad you enjoyed the post and that it was apparently comprehensible.
Anders - I agree of course about the role of context in Chinese (and indeed in all languages though in each case in slightly different ways). As to the Swedish example, it's slightly shameful that I managed to get the verb form 'tror' wrong, and misspell 'att'. I didn't know the expression 'I helvete heller' and it's good to see that the point about making the transfer at the level of pragmatics would still hold.

I was wondering when I wrote about translating something technical and appealing primarily to the level of discourse structure whether professional translators might argue that this would be more like localisation than translation. Do other people agree with the comments so far on that question?

Jim's post and Orla's response are both refreshing to see in a blog. I especially appreciate Orla's reference to the importance of pragmatics in translation. I hope that the day of assuming that translating can simply be done by changing one code (language) into another is gone. We have to dig below (and around) the surface to try to get at what the original author really meant by what he was saying. Fortunately, good insights in this area have been coming from Relevance Theory, with relevance (!) for translation, as Ernst-August Gutt has demonstrated in his writings.

Jim, I think the level(s) I use for my translations depend(s) entirely on the nature of the text. I'm not sure putting an emphasis on the pragmatic level would be the most appropriate thing to do for, say, literature works, where the form and style are just as important as the message that is conveyed. On the other end of the scale, the phonological and syntaxic aspects of "Do not push the red button or you'll die" are of little importance, what matters here is the semantic and pragmatic levels.

So what is the difference between localisation and translation? I think translation considers a document in isolation, whereas localisation places a document within a specific context. A document intended for a practical use (like a manual) will have to be localised, whereas a stand-alone, self-sufficient document, which operates within a world of its own (like a novel), will have to be translated.

My original background was in linguistics. I am not a practising translator at the moment, but I am interested in issues of translation. (Incidentally, the difference between being a practising translator and not being a practising translator is huge. Only if you are under pressure to produce a professional piece of work for a client can you appreciate the true pragmatics of the translation process. Translation is not an academic exercise; it is a serious matter of producing an acceptable text of a high standard.)

Anyway, I feel I must give expression to my very strongly-held feeling that _linguistics is not really very useful to translation_. I came to this conclusion after I did a webpage on the translation of relative clauses. What I found was that linguistic descriptions of the relative clause in English can go into minute detail on the syntax involved, but completely miss the way in which relative clauses are _used_ by writers.

The sentence I looked at was: 'This snake, which may reach gigantic size, and live many hundreds of years, is born from a chicken's egg, hatched beneath a toad'. It is from Harry Potter, if you are interested.

The only thing that linguistics can tell you is that 'which may reach gigantic size, and live many hundreds of years' is a non-restrictive relative clause. Granted, this is actually relevant and useful information. But what linguists don't seem to concern themselves with, and seems to me to be of vital importance, is the way in which this structure is used by living people.

In fact, the whole structure could be reversed and it would hardly make an iota of difference: 'This snake, which is born from a chicken's egg, hatched beneath a toad, may reach gigantic size, and live many hundreds of years.' In other words, the only purpose of using a non-restrictive relative clause here is to provide some syntactic variety! If you take this sentence at its grammatical face value when you translate it, it is this _stylistic_ intent, more than any linguistic analyis, which must be taken into account. In translating into languages that do not use relative clauses in quite the same way as English, it is imperative that the translator should grasp this stylistic fact and _not_ mechanically transfer the hypotactic syntactic structure into the foreign language when a paratactic (?) structure might be more suitable.

I am really looking forward to the day when linguists stop introspecting deeper and deeper into the minutiae of syntax and start describing how language is actually used.

Please feel free to disagree with me, but I am rapidly coming to the conclusion that 'traditional grammar' is about all the linguistic apparatus that a translator needs to have in order to pursue his/her trade.

It's difficult to know how to respond since you give such a good example of how understanding some facts about the languages you are involved in translating - in this case the use of relative clauses - can help you in the translation task. But I think maybe linguistics can tell you a bit more than whether or not some clause is restrictive or non-restrictive. It can tell you for example to what register such and such an expression belongs; whether or not its use correlates significantly with other factors; in principle if it is a fact about language or its use and can be described linguistics can describe it for you. There are not a few good books on stylistics - both in literary and non-literary texts - by linguists who work in that field; books by corpus linguists who build their descriptions from large banks of authentic data; sociolinguistic, discourse analytic, pragmatic and other approaches to the question of how and why people use particular linguistic structures.

Linguists are like all other scientists: some scientists look deeper and deeper into theoretical issues and some of them get their hands dirty on real-world data or even building things. But no scientist is under a particular obligation to do one or other of these.

It's difficult to know how to respond since you give such a good example of how understanding some facts about the languages you are involved in translating - in this case the use of relative clauses - can help you in the translation task. But I think maybe linguistics can tell you a bit more than whether or not some clause is restrictive or non-restrictive. It can tell you for example to what register such and such an expression belongs; whether or not its use correlates significantly with other factors; in principle if it is a fact about language or its use and can be described linguistics can describe it for you. There are not a few good books on stylistics - both in literary and non-literary texts - by linguists who work in that field; books by corpus linguists who build their descriptions from large banks of authentic data; sociolinguistic, discourse analytic, pragmatic and other approaches to the question of how and why people use particular linguistic structures.

Linguists are like all other scientists: some scientists look deeper and deeper into theoretical issues and some of them get their hands dirty on real-world data or even building things. But no scientist is under a particular obligation to do one or other of these.

In the ideal world, linguistics would be vitally important to translators. But in the real world I don't think it attains that kind of status.

I have not yet found any good stylistics books by linguists, although I am hampered by living in China.

Perhaps it is a book waiting to be written.

I've had a bit more time to think about your response.

You mention that I give "such a good example of how understanding some facts about the languages you are involved in translating ... can help you in the translation task".

I do not doubt that an understanding of the facts of the languages you are translating is vital. If you define 'linguistics' as 'any structured knowledge of the structure and workings of language' (as linguists are wont to do), I have no doubt that linguistics is essential to the translator. If, however, you define linguistics as 'a structured knowledge of the structure and workings of language according to people who call themselves linguists', that is when I become rather sceptical.

'If it is a fact about language or its use and can be described linguistics can describe it for you' -- yes, of course it can. My quibble is whether linguists actually do this in a way that is relevant to translators.

I would be genuinely interested in seeing the books on stylistics, books by corpus linguists, as well as sociolinguistic, discourse analytic, pragmatic and other approaches to the question of how and why people use particular linguistic structures that you refer to. I have not come across many that elucidate the problem of use of relative clauses that I mentioned in my posting, but that is not to say that they don't exist. I would be very grateful if you could point me in the right direction, I am very keen to see what is on offer as this is a problem that I would like to look into further.

Just as an aside, I'm not very impressed by corpus linguistics. In my admittedly limited experience, it seems able to come up with numbers without being able to tell you why the numbers are there. It is like doing an analysis of a painter's works and pointing out that 73.52% of the colours used in his works are based on the colour blue. This figure might be interesting, but it would be more useful to know that the painter mainly painted seascapes!