In Harmonium

Posted By Marc on May 25, 2009

Joel Garreau at the Washinton Post has a very interesting article posted yesterday called TONGUE IN CHECK: With Translation Technology On Their Side, Humans Can Finally Lick the Language Barrier (hat tip to Adam Russell).  The story is well worth reading for a number of reasons.

While the story starts by establishing the need for such a translator (I’m sure that Max will comment on this), where I find it most interesting is in reference to two things, both of them tied not to DARPAs universal translator, but to Google’s Translate project.  The first point is the reliability of the translations – they are still somewhat poor, but they are “satisficing”.  In other words, they can get the gist of a document, but none of the subtleties.  The second point relates to the first, but it has to do with how the tool is marketed by Google.

Get translation with a single click and make your webpage instantly available in other languages.

Sometimes, I feel like I am the reincarnation of Socrates (especially in his warnings about writing), and this is one of those times.

In my teaching and writing, I expend a lot of effort on hammering home the point that the map is not the territory (and the inverse – the classic Batesonian paradox of epistemology).  Google translate is, in my opinion, an excellent examplar which needs the same point hammered home.  A “satisficing” translation is certainly fine for ordering a beer, finding a washroom or trying to get food (you might be surprised, but at least you will probably get to eat).  Indeed, a “satisficing” translation, or at least the availability of one, is also probably a good way to leverage learning a new language in the field. What satisficing translations are not good for is when they become a crutch rather than an aide.

Years ago, Edward Sapir and his colleague Benjamen Lee Whorf (not a Klingon!), proposed the idea that language shapes perception; this is the somewhat infamous Sapir-Whorf Hypothesis.  Personally, I accept, and use, a “weak” version of that hypothesis (“shape” rather than “control”), and a fair bit of what I write and research assumes it.  I also use a lot of Canadian Communications Theory (McLuhan, Grant, Innes and their descendants) to look at how different technological extensions and genres of communications twist and warp messages.  Both the DARPA Universal Translator (coming soon in iPod size!) and the Google Translate project and other, similar efforts (e.g. Rosetta Stone) are certainly grist for the mill!

So, let’s think this through.  We now have technology that enables satisficing translations.  We also have a general social situation where people are structurally encouraged to adopt satisficing behaviours, partly as a result of time pressure, but also for other reasons.  With Google translate, I can arrange for this blog to be “translated” into over 40 languages.  Simply wonderful!… Not.

Satisficing behaviour is something that we, as a species, are encouraged to use simply because it is the evolutionarily easiest way to adapt to our environments.  Think about many of the memes such as “if it ain’t broke, don’t fix it”, “what was good enough for grandad is good enough for me”, etc., etc., ad nauseum.  This is the type of thinking that made the Roman Empire what it is today – dead.

Satisficing behaviour, while “natural” (quoth he with gritted teeeth), is also dangerous in many settings.  It evolved to meet the needs of stable environments and, probably, as an exaptation of physiological “learning” – think about learning to tie your shoes.  It did not evolve as an efficient way of meeting new challenges, and yet it is being marketed as such.

Twenty-four hundred years ago, Socrates refused to write down his teaching on the grounds that writing was a fallible extension of the human memory.  It was not that writing per se was fallible but, rather, that a) you wouldn’t have the book (or scroll) when you needed it and b) the action of reading was a non-interactive act of interpretation.  If you relied on books and writing to “store” your knowledge and skills, then they weren’t yours, they were the books.

Garreau’s article in the Washington Post starts witha heartwrenching story of an American soldier in Iraq unable to comfort a young child because he couldn’t speak the language.  This is an emotional claim for why a Universal Translator is neccessary.   However, we note in the same story, that he physically had all of his weaponry and armour and, presumably, knew how to use it.  As Garreau puts it

There he stood, helplessly, in full battle rattle, with his ballistic glasses and helmet, his weapon bristling, his body armor making him waddle like a bipedal rhino.

He spoke no Arabic. He couldn’t comfort her, he couldn’t tell her he wanted to get her medical help.

The “solution”, of course, is to add another piece of equipment.  This is “Golden BB” thinking.  If having the local language is “mission critical”, then why is it exteriorixed?  I can just imagine some poor Private telling his Lieutenant “Sorry Sir, I couldn’t recharge my UTiPod so things just sort of blew up!”.  Given that the current operations in Iraq and Afghanistan are what is called “population-centric COIN”, then language is mission critical, as critical as making certain that your weapons work.  Take that last comment and shift it to “Sorry Sir, I couldn’t find the support people to clean my weapon so it just sort of blew up!”, and you can see what I mean about “exteriorized”.

Shifting from the military realm to that of academia, how many of us would accept a student handing in as original work, a google translation?  Or, perchance, using a google translation of a “mission critical” disciplinary work? “Hey, yeah, Marc, I like couldn’t read that Weber dude so I like ran his stuff through Google translate”.  Hmmmm.

While I personally like the idea of Universal Translators and Google Translate, I view them as tools that could help me to learn languages (and when is Google translate going to get some of the languages I need, like Sumerian????).  For most “missions”, be they military or academic, satisficing behaviour just isn’t acceptable, and the reasons why are simple: it is not good enough to get at the actual information we need, and it is not part of “us”.

Category: Communications Theory, Epistemology
Tags: Sapir-Whorf Hypothesis, Satisficing Behaviour, Universal Translator