“Words don’t mean; people mean.” – commonly attributed to S.I. Hayakawa
In the Indian adage about a group of blind people feeling different parts of an elephant, each of them gives a very different account of what they learn: the one feeling the trunk says that an elephant is like a branch, the one feeling the tail says an elephant is like a rope, and so forth. It’s a parable about subjectivity and the limits of perception. What this story does not tell you is that all of the elephant-feelers immediately fall insensible to the floor, paralysed by a new appreciation of the sucking abyss that surrounds the elephant for light years in every direction.
Because one of the terrible problems with learning is that you become ever more acutely aware of how ignorant, prejudiced and arrogant you have been in the past, and the logical extension of this realisation is that you probably still are. It’s not much comfort to be reliably informed by people who know considerably more than me that this process never stops and even picks up speed, like a personal version of cosmic inflation, you at the centre of your own increasingly dark universe with an ever-expanding void between each tiny red-shifted island of certainty.
The pay-off, the saving grace, is that you are getting much better at what you do, even if it feels like the opposite. So it becomes increasingly frustrating when you see references to the few glittering specks you do know something about misrepresented again and again in mass media.
By way of example, you can’t study the language-creating brain without running quite quickly into case studies of Christopher, a man with autism, often described as a “linguistic savant” or “polyglot savant”. Brought to popular knowledge by Smith & Tsimpli (see also video above), Christopher has severe cognitive deficits to the extent that he is unable to live independently, but he also has an unusual talent for language: he can, the literature claims, “read, write, translate and communicate in fifteen to twenty different languages”. It’s just the specific assertion that Christopher can “translate”, a term that is sometimes used very loosely, that we’re going to scrutinise here. Our questions are the same as they’ve ever been: what do you need in order to be able to translate? What makes a good interpretation good?
Christopher’s case is often cited in discussions around modularity of mind, a school of thought which argues that brains must have dedicated, functionally discrete, specialised “modules”, which may have evolved in response to evolutionary pressures. One (or, in some versions of the modular-mind story, an interconnected group) of these modules might be “responsible” for language and would therefore be unique to humans, since we still have no solid evidence that any other species has made the leap from communicative signalling to fully-fledged language. There are many competing accounts for what a “module” might actually be – perhaps we are not talking about a distinct “compartment” in the brain that you can point at or easily label on a chart (although there are people who do believe this), but a sub-system within the neural network which is functionally or connectively compartmentalised but very diffuse physically. There are also further debates on whether the cognitive assets required to handle language are domain-specific (ie. useful for processing language but nothing else) or domain-general (useful for other cognitive functions as well) or a mixture of the two.
The modular-mind view of language seems to lend itself very quickly to the study of autistic “savants”. In Christopher we have a well-studied example of a human being whose mental architecture and perception are atypical in many ways but whose cross-linguistic skills are unusually well developed. It could therefore be argued that whatever it is that causes autism, in this case it must mostly sit “outside” of Christopher’s “language module(s)”: that’s quite the hypothesis, and to explore it, Christopher’s command of his 20-plus languages was examined in detail: he was also taught new natural languages, the assumption being that they would be based on principles of a theoretical Universal Grammar, alongside an artificially-constructed language which violates those same principles. (A further study involved him learning British Sign Language: while his performance generally was as good as the rest of his cohort, he had specific difficulties with spatial reasoning which impaired his ability to produce valid BSL syntax. What does this say about the field’s assumptions of what a “language module” is made up from?)
But returning to the claim that Christopher can “translate”, if you look at the 21 translations he produced in Smith & Tsimpli (1991), many of them fail to capture the intentions of the original text, and a few make no sense at all. Just one example: in the Russian excerpt, he translates something which was intended to mean “wet weather” or “rainy weather” as “unhappy”. A first-language Russian colleague tells me that nenastnyi can indeed have a sense of “moody” as well as that of “rainy” or “bad weather”, perhaps in a more metaphorical sense that English-speakers can easily appreciate (“sunny disposition”, “stormy expression”), which might explain Christopher’s choice. But he did not select the intended sense – actual meteorological conditions – perhaps because his pragmatic competence, the ability to infer from and imply with information which is not specified semantically or syntactically, is severely impaired, like most other people with a diagnosis of autism. Translation and interpreting (as real work done in the real world beyond the confines of an experiment) are a good deal more than just semantics: they are concerned with designing the target text so that it is accessible to a well-defined audience, with the reactions a specific choice of terms might evoke in the minds of that audience; this absolutely requires you to establish (directly or otherwise) the intentions and goals of the source language author.
None of that is possible for someone with a missing or developmentally delayed Theory of Mind, the ability to mentally visualise or “sense” what other people might know and think. Translators and interpreters know that the semantic field of any term – the scope of meanings that users of the language typically encapsulate with a sign or a word – does not necessarily transfer tidily across languages. This phenomenon manifests itself endlessly in clickbait media as a relentless battery of articles with titles like “Seventeen UNTRANSLATABLE words that hilarious foreigners use”, each of which is then always followed by a perfectly adequate translation into English. Different languages are not “codes” for each other, despite the commonly-held and entirely erroneous belief that translation is all about selecting one-to-one or “literal” correspondences and that somehow it “isn’t a translation” if you need more than one symbol to render a target language equivalent. Christopher excels at memorising vocabulary and creating mental “look-up tables”, and he has a good appreciation for syntax, but he does not deal with meaning in the way that most “neurotypical” people do. Translation is not just a matter of joining the dots.
This is not meant in any way to belittle Christopher’s achievements. The fact that he can demonstrate competence in 20+ different languages is stunning, and puts my own feeble achievements in just two or three into their proper context. But despite his natural talents and his hard work, he does not “translate” by any measure of fidelity that applies to the actual work of translators and interpreters, and to be fair, I’ve seen no evidence that he ever claimed to. The assertion that he can “translate” is made on his behalf: much of the literature written about his case that I’ve seen, especially on dust jackets and in magazines or their electronic equivalent (all written by non-interpreter and non-translator editors), says so again and again.
It makes little sense to me to treat pragmatic competence as a “module” which is “separate” from language. To be a successful language mediator even in just one language, you ideally need to be a people person, or at least be able to simulate it very convincingly.