Human translators' imminent demise greatly exaggerated

Reports of the imminent demise of human translators/interpreters have been greatly exaggerated - a debate over machine vs. human translation.

Echo Tang 语言作坊MyLanguageLab 29.08.2016


Orginally written in 2011, now slightly revised for publication on this site, this article is based on a random debate which loosely focuses on man versus machine in translation, then shifts to languages and cultures in general. It's by no means a logically coherent and structured essay, nor is it exactly in keeping with current developments in the field as the past 5 years have seen rapid proliferation of smart phones and tablets along with tremendous technological advances in the world of auto-translation tools that have resulted in ever-improving machine translation output. Nevertheless, it's still interesting to see how things changed from 5 years ago.

A: I dare predict that machine translation tools will soon replace human translators/interpreters. So those who make a living out of translating/interpreting, you'd better watch out! Well, there's no reason to panic just as've got a couple of years left to retire or change jobs until that invention hits the will probably be an extension of the apps on the iPhones that are currently translating text only (like menus in restaurants) just add voice recognition.

B: The belief that machine translation will eliminate human translators is overly optimistic. For one thing, machines can translate words, but not meaning. A machine-translated text normally requires tedious post-editing by a human translator, who, in most cases would rather create a translation from scratch than fix a clumsy and sometimes incomprehensible jumble of words put together by a non-thinking machine. But then, if accuracy is of only secondary importance, automatic translation tools are generally able to allow the reader to get a quick grasp of foreign texts like Emails, Facebook comments, etc.

A: Your point is made well, but iPhone already has an app that translates text rather well and immediately by "waving" over it, very handy for traveling (i.e. restaurant menus, tour brochures). These technologies will improve, and add to this the notion of English as the lingua franca and half of the world's 7000 languages disappearing by the end of this century, we are embarking on the move towards a linguistic monoculture that will render translating jobs obsolete.

B: I’m glad that you mentioned menu cards and travel brochures. These are exactly the sort of texts that automated translation can handle, and probably the best it can hope for. If the translated texts are to be taken really seriously, even the most simply structured ones are subject to human quality control before they get published. Not to mention texts of higher level of complexity - the level that most human translators work on - whose dynamics and subtlety can never be captured without human intervention. As a practising translator myself, I’m constantly frustrated by the fact that, even with so many online translation aids and language resources out there, I still have to spend hours toiling over little nuances and searching for that perfect word to get the job done.

Language death and monoculture are irrelevant to my argument – machine vs. human translation. But since you brought that up, I'm of the opinion that although globalisation may be able to turn our world into one big village, it won’t stop people from doing everything they can to preserve their identity, which is partly represented by languages. This makes translation even more important. Language means a lot more than just communication – it’s power, politics, ideologies, culture, etc. There’s even translation from French French to Quebec French, Swiss German to High German. Cosmopolitan trends will not lead to an absolute monoculture.

A: Of the estimated 7,000 languages spoken in the world today, linguists say, nearly half are in danger of extinction and are likely to disappear in this century. In fact, they are now falling out of use at a rate of about one every two weeks (Source: The New York Times, September 18, 2007)

Your point about spoken/written language is valid although I take issue with the equation of Quebec French and Swiss German being separate languages from their respective High Standard version and would argue that these are merely regional dialects. Notions of culture, identity, political persuasion, personal beliefs have always found expression through visual means (individual's dress code, hair style...) and are not the exclusive domain of language...visual representations and visual literacy are becoming increasingly more pervasive in digital media, especially since text based expressions can never adequately reflect a speaker's modulation (tone of voice, mood, etc.)

B: I’m afraid there’s a world of difference between our understanding of language and translation. First, there’s no such thing as mono language. The use of English moves to International Englisches. Second, translation may take place among different sets of codes: dialects, same language of different eras (see Bible translations, e.g., Old King James to colloquial; Classic Chinese to Modern Chinese, etc.). Your definition of translation is very strict, which is not what we accept in Translation Studies.

Language equality is a sensitive issue. But as a linguist, I’d say they are all equal, although they can be used for different purposes. Just make sure you don’t offend the Quebec French or German speaking Swiss.

A: I have not given a definition of translation, not yet, neither did I assert the notion of a mono language, but noted the diminishing of language varieties and the dominance of a linguia franca...preferably Klingon :-) Of course there is a difference between language and translation, never disputed that either, check my posts, and yes, translations can take place between different codes, never argued otherwise. I merely pointed out that inaccurate use of terminology in your statements. As a trained applied linguist myself I was taught that regional dialects are not languages but varieties of the high standard and in academic terms this still holds although Quebec French and Swiss might see it otherwise... but then again, Bavarian ain't a different language either although the Bavarians would like to think so.

B: I didn’t use any terminology to describe that phenomenon where two variations of a language coexist, one of higher prestige, and the other lower. ‘Diglossia’ is a ‘correct’ terminology if that’s what you’re looking for. Nor did I claim that they’re two ‘separate’ languages, although they’re both equally legitimate and complex. The whole time I’ve been arguing that your statement ‘we’re moving towards a linguistic monoculture that will no longer require translations’ is flawed. There’ll be always translations, be they between languages, a ‘standard’ form and its vernacular varieties, and same language at different points in time.

Many languages are dying generally because their speakers are few and marginal (e.g., tribal languages in Africa, indigenous languages in Australia, etc). But think about languages whose speakers are abundant? Chinese, Arabic, Indonesian? Even less known languages like Javanese? We teach local languages to preserve them. Crazy to think all these languages would be wiped out. It’d be the end of the world.

A: Well, dinosaurs died out and so did native speakers of Latin, and yet, the world keeps on turning.

B: Absolutely! Hence the birth of today’s Romance Languages. Thanks for proving the absurdity of the notion of a ‘linguistic monoculture’.

Dialects may also form new languages through time. Mandarin used to be a northern dialect spoken by the Manchurians, but as a result of socio-political-cultural change over time, it has become a prestige variety and eventually assumed official status. Translations occur between variations of the same language because there’re different levels of mutual intelligibility between dialects. Sometimes it’s very low, and that’s where translation comes in.