Whorf, B. (2012). Language, Thought and Reality - Selected Writings of Benjamin Lee Whorf. Second edition. Science and Linguistics, p. 271. Massachusetts Institute of Technology.
Watzlawick, P., Beavin Bavelas, J., Jackson, D. D. (1967). Pragmatics of Human Communication: A Study of Interactional Patterns, Pathologies and Paradoxes. Ch. 1: The Frame of Reference. W. W. Norton & Company, Inc. (2011).
Much more than in many other disciplines, translation is a field where practitioners' profiles vary greatly. Here is why.
Firstly, many people "speak" several languages without having graduated as translators or interpreters, or in any discipline mainly related to languages and/or writing: some learned languages as part of their work, some because they travel a lot, some because their parents are natives from another country, some because they simply love languages, etc.
This does not mean, of course, that all these people are perfectly bilingual or even trilingual.
In fact, true bilingualism is extremely rare. In most cases, one language tends to prevail in terms of mastery and fluency.
Speaking several languages does not mean either that a person is a good writer or will be a good translator.
Everyone usually agrees that speaking does not imply writing. But as soon as two languages are involved, quite paradoxically, many people assume that speaking implies translating, when this is a much more complex cognitive process.
Secondly, working from home has its appeals and translation is precisely a line of work that the Internet made particularly convenient in order to do so, with little to no investment: "all" you need is a computer, Internet access and your brains (i.e. mastery of at least two languages, excellent writing skills, an insatiable thirst for knowledge and learning, and a few other skills that will require zero investment if you already have them) — the latter "brain element" may go slightly overlooked by some, especially if fooled by the claims of machine translation software.
Web portals with worldwide reach are available, where anyone can create a profile in a matter of minutes — whether qualified or not. The translation market has a very low entry barrier: entering it is easy. Lasting, however, is another matter altogether. For example, as of March 2017, the Proz.com URL numbering system had passed 1,000,000 created profiles. However, if you run a search, for example, for English to French translators who are also Certified Pros, you get... 205 translators only. Same search for Russian to French returns only 4 translators (as of March 21, 2017). These figures definitely show the kind of appeal translation may have, and how misleading the low barrier to entry can be.
Thirdly, and most importantly, the sheer variety of translators' backgrounds means that they do not all:
- have similar training and knowledge (unlike what you may expect from a GP, an accountant or an architect); nor
- approach translation in the same way, in particular when faced with poorly written material, i.e. material "authored" by unprofessional "authors". This issue is especially salient for material "authored" in English, the language with the highest amount of non native speakers in the world — and that includes me ;o).
« Avant donc que d'écrire, apprenez à penser ». ~ Nicolas Boileau-Despréaux, L'art poétique.
In this era of compulsive "post-truth" communication largely induced by unscrupulous short-term attention/power seekers — or demagogues — who can now comfortably rely on a huge social-media-lobotomised mob, "writing" often gets neglected: this report is usually needed for yesterday and Bob will do (you know, Bob, the new guy from sales who just hates reading?). Hey, nobody will read it in full, anyway... But make sure it has nice images, impressive charts and "infographics"!
Even scholars are now caught in this compulsive post-truth communication spiral: they have to publish no matter what. This questions the very value and quality of such publications, and led Science correspondent John Bohannon to test peer review by submitting fake science papers to 304 open-access journals: they got accepted by 157 of them!
For many, to communicate has become more important than to communicate clearly, meaningfully and truthfully.
Also, compared to, say, a hundred years ago, written communication has become highly unstructured.
Before the Internet, smartphones and social media, any spoken or written material to be "published" or "broadcast" was created — and structured — more or less professionally. I could read and write before entering first grade and as a little girl, I remember I loved to read short stories (mostly real ones) published in Reader's Digest. That was around 40 years ago and by then, the quality of their contents was up to par with what you only seem to find, nowadays, in literary journals like Glimmer Train or in (some) published authors.
Today, anyone can "publish" and "broadcast". Granted: this is, beyond doubt, an invaluable progress for freedom of speech.
That being said, the flip side is that finding professionally written (i.e. well thought-out) material is currently like looking for a needle in a haystack of total gibberish. Compulsive and unstructured communication is ruled by images, mostly unrelated short messages and instantness. This phenomenon often reminds me of a scene from 1993 movie Demolition Man, where the most popular radio station appears to be one broadcasting "Nonstop wall-to-wall minitunes", i.e. completely idiotic commercials. Hard not to draw a parallel with social media and the blogging rage, right? Nonstop wall-to-wall mostly vacuous and copy-pasted (er... "shared") "information".
Not exactly the kind of environment and stimuli that promotes structured and well thought-out communication.
Even among professions that mainly revolve around writing, skills seem to deteriorate.
For all these reasons, today more than ever, many translators have to deal with mediocre documents, that rather look like hasty drafts than properly reviewed final versions.
The question is: how do you professionally translate this ever increasing mass of poor writing?
« Traduire, c'est faire comprendre ». ~ Jean Delisle.
Whether a text is perfectly written and structured, or is a mere patchwork of (copy-pasted?) gibberish, reasoned translation rests on the following skills, which I have amply honed over the past 15 years:
- Ability to fully understand originals and their sociocultural context, which are essentially foreign to the target audience of the translation;
- Ability to express this foreign message so it is not perceived as foreign, or at least can be understood, in the target sociocultural context.
This second ability rests on:
- Ability to dissociate the two linguistic systems involved (so as to avoid interferences, i.e. unnecessary/incorrect introduction of non-native elements in the target language);
- Mastery and appropriate use of translation techniques (modulation, idiomatic expressions, transposition, etc.);
- Consideration of any relevant extra-linguistic information;
- Excellent writing skills in the target language.
This is the only way to create translations that are perfectly clear for the target audience and do not make the original author look stupid. This is the only way to serve both target audience and author.
At the opposite of this approach lies literal/dictionary/machine translation, whether statistic MT or neural MT. I emphasize lies because lies they tell indeed.
Contrary to what most claim to be, machine translation software is nothing more than the contemporary version of online dictionaries, except you can look up entire sentences or texts in one go, instead of looking up all words you do not know one by one. Absolutely unnecessary for professional translators but very tempting for non specialists.
Still, this "dictionary lookup in one go" happens mostly word by word, even if this is now transparent to the user. This means that just like any dictionary, these "solutions" do not have "writing skills". Just consider how often the Word spelling and grammar checker makes inappropriate suggestions (at least in French). If it does not even get spelling and grammar right, how could it master the "art of writing"?
As an aside, there were 43,380 professional authors and writers in the US as of May 2015 (this excludes freelancers), for a total population of over 300,000,000 (hence merely 0.014% of the population, or 1.4 writer per 10,000 inhabitants). Writing is far from being a "common" skill, and it has nothing to do with "computing", nor with what a search engine does.
Machine translation is, moreover, unreliable, because contents used are not systematically edited by professional linguists/writers. They may give you a correct translation for most isolated words, or for very short word combinations, but not every time. Here are two equally wrong examples, according to Google Translate and Systran. So how, as a non linguist, can you tell when the provided translation is right, and when it is wrong? How can you tell the difference between reliable and unreliable resources, unless you are a professional linguist and do not really need said resources to begin with?
As for longer sentences (and sometimes also for very short texts — see image below), all you will get is essentially word for word and context-blind translation, i.e. only what a dictionary is "capable" of.
Here is an example of poor English writing, very common in market research questionnaires (keep in mind that this was aimed at middle and high executives):
"In thinking about your company, in addition to the industry you participate in, which answer is closest to how you would describe your firm?
If you want to be taken seriously and look professional, you might obviously prefer avoiding this kind of redundant, beat-around-the-bush context-blind "French". My reasoned translation in this particular case was, quite simply: "Quelle est l’envergure de votre entreprise ?".
Here is another example:
"If the intention is that of practicing a profession, in terms of the aforementioned Act one has to submit an application to the relevant Designated Authority requesting authorization to practise the profession."
The worse here is that this artificial, machine French is grammatically correct. Most non specialists and non writers will see nothing wrong with it. My reasoned, native and unredundant translation was, however, quite different: "La loi susvisée impose de soumettre une demande d’agrément à l’autorité compétente pour l’exercice d’une profession." Impression made on an educated French audience will be quite different too.
Another, rather redundant, example:
"A misuse of privileges occurs when intentionally legally or illegally obtained permissions are used outside of the scope of intended use."
My reasoned, native and unredundant translation here was: "L'utilisation abusive des autorisations obtenues (légalement ou non) est évidemment possible". Believe it or not, that is what was meant.
One last genuine example I once had to translate:
"We ask that if you are going to cancel a scheduled session that you do so with 48 hours notice."
And its machine translation, as well as "pure" neural machine translation. Very efficient word for word "translation" indeed, but it unfortunately verges on kindergarten speech and is definitely NOT how we would put it in native French.
My reasoned, native and unredundant translation here was: "Toute séance doit être annulée au minimum 48 heures à l'avance."
Of course, you are "asked to". Of course, you can only cancel a "scheluded session". And of course, you should only cancel "if you are going to cancel"!!! This kind of beat-around-the-bush phrasing simply does not work in French (does it even "work" in English?).
Whether dealing with pragmatic or aesthetic material, it is obvious no "author" deliberately writes texts that are confusing, poorly structured, grammatically incorrect, incoherent, painfully redundant or plagued with any other writing flaw.
No "author" deliberately tries to look stupid or to prevent the target audience from getting the message, evidently.
We live, however, in an era of compulsive and unstructured communication ruled by images, mostly unrelated short messages and instantness, all stimuli not conducive to but rather antinomic with the reflection and coherence required for skilful writing and clear, meaningful communication.
In all likelihood, poor writing is here to stay, too often combined with poor (i.e. indiscriminate) research, if any...
This is why, today more than ever...
« Le traducteur doit rectifier les erreurs de l'auteur, éclaircir les passages qui offrent quelque obscurité et fournir des notions qui conduisent à la parfaite intelligence du récit ». ~ William McGuckin de Slane (Baron de Slane)
This is why the most cost-efficient and sustainable option is reasoned translation, i.e. human and native professional translators, who understand your message and protect your reputation with the utmost care.
Not tools that take months to set up and "fine-tune", only to realize they are not "fine-tuned" at all and reach a result by far inferior to what any professional translator attains instantly. Not tools that may tarnish your reputation because they are unreliable. Not tools that actually discourage language learning and reflection, and are bound to propagate artificial, machine language, that will ultimately come feed the systems that created it in the first place, in a vicious loop of context-blindness.
 "A true bilingual is someone who is taken to be one of themselves by the members of two different linguistic communities, at roughly the same social and cultural level." ~ Thiery, C. (1977). True Bilingualism and Second-language Learning. Université Paris III - Sorbonne Nouvelle. In Language Interpretation and Communication, p. 146. Back to text
 "It is a mistake to believe that anyone knowing two languages can necessarily translate. This is a very common misconception." ~ Delisle, J. (2013). La traduction raisonnée - Manuel d'initiation à la traduction professionnelle de l'anglais vers le français, p. 19. Les Presses de l'Université d'Ottawa. 3rd edition. Back to text
 "Before you even consider writing, learn how to think." (I would add that copy-pasting is definitely not thinking). Back to text
 "To translate is to help the target audience understand the message". Ibid 2, p. 633. Back to text
 Ibid 2, p. 19. Back to text
 For example, many references to Common Law may be perceived as totally foreign (and as lacking logic) in Civil Law countries. See the Dicodex project, p. 261 (footnote 17): "Un notary public américain peut n’être qu’un coiffeur, voire une simple machine à timbrer installée dans un drugstore". Hence conditions for setting up an authentic act are far from being met per French or Belgian law. Back to text
 « Deux grandes langues de civilisation comme l'anglais et le français sont, malgré l'apparence, très différentes de structure et d'esprit » ~ Robert Le Bidois. Back to text
 Ibid 2, p. 631. Back to text