Greg Detre
Monday, 09 October, 2000
Pinker�s lengthy quotation from Orwell�s 1984 highlights perhaps its most alarming legacy, Newspeak, the barbarised form of English where every term, idiom and phrase had been robbed of connotations damaging to the Party�s cause. The premise is that without the means to formulate antithetical expressions and concepts in words, the populace would be robbed of the means to even think of such notions, bringing a chilling concreteness to the idea of linguistic determinism. Such a strong form of linguistic determinism can be contrasted with strong cognitive determinism, where it is our cognitive processes which wholly shape our words. Somewhere between the two, there must be a more plausible way of understanding the interaction between thought and language.
Pinker draws attention to the pivotal phrase, �At least so far as thought is dependent on words�. Orwell is perhaps overstating the case for Newspeak when he considers that the connotations of intellectual and political freedom can be eradicated and lost forever from the word, �free�, in the same way that �queen� and �rook� fail to conjure up checkerboards to someone who has never played chess.
Pinker gives various examples of creoles being born adult from fledgling pidgin dialects in the space of a single generation. This is strong evidence that human languages grow to full maturity and complexity, however stunted and restrictive their birth may be. His examples of the Atlantic slave trade and the South Pacific give strong cause for hope that the Oxford English Dictionary could not one day stage a totalitarian coup simply by repressing the means to vocalise rebellion. For after all, though the emergence of writing was a laborious, multi-step process requiring luck, genius and circumstance to gestate, spoken language seems to well up irrepressible and rich wherever humans congregate. Though the extent of universality and differences between human languages is highly debatable, some genetic predisposition to learn or form human language seems incontestable. It is also doubtful whether a language could be constructed (let alone enforced) which remained functional having had so many integral concepts stripped away.
Let us instead consider the weaker form of linguistic determinism, usually termed �linguistic relativity� after Benjamin Whorf�s phrase. We will first understand his use of the term, before discussing alternative forms which linguistic relativity might take.
Whorf was an amateur scholar in a variety of Amerindian languages, especially Hopi (Arizona), Aztec, Maya and Nahuatl (Aztec). His ideas originated from his respect for Fabre d�Olivet�s, work which characterised certain Hebrew letters as having high-level connotations (e.g. goodness, stability). This led him to posit that the Amerindian languages were fundamentally different from �Standard Average European� languages. He believed that the Amerindian languages are �oligosynthetic�, i.e. they are comprised of just 35 root fundamental element ideas which underly the whole language. With Sapir�s help, he extended these lexical differences to incorporate grammatical structure, growing into the Sapir-Whorf hypothesis of linguistic relativity.
His two major premises were that:
language shapes our higher-level thought; and
the structure of your first language influences the way you understand and behave in your environment.
His contemporaries, Lenneberg and Feuer, sharply criticised Whorf�s methodology and conclusions on a number of grounds. Lenneberg points to the translation techniques Whorf used to demonstrate differences in languages, e.g. just because two languages describe an event very differently does not mean that they have a different way of seeing that event; it might simply be because the language has developed metaphorically in one direction (e.g. we are unaware that �breakfast� comes from literally �breaking a fast�). Lenneberg also points out the basic requirements of the scientific method dictate neutral and separate descriptions of the linguistic and non-linguistic events before attempting to correlate them. Otherwise, the argument for linguistic relativity is either circular or tautological, if the evidence pointing to the differences in �world view� on the basis of language structures is the differences in language structures. Feuer also claims that our perceptions of reality cannot differ greatly, because a wholly incorrect perception of reality, i.e. our environment, would not give us much chance at survival. However, this seems a weak criticism, since many of the differences Whorf points to, especially some of the subtler praises of Hopi language structure, are not differences which would affect our survival. They relate to our perception of the �Cosmos� (to use Whorf�s term), rather than our internal representation of the immediate environment.
Many of Whorf�s ideas hinged around his interpretation of the Hopi language as having a more precise and less misleading grammar and terminology than English, and better fitting a modern physicists understanding of the Cosmos. He gives examples of the Hopi use of a word akin to �slosh� instead of �wave�, the lack of subject-predicate distinctions and the Hopi inclusion of both space and time in a sentence. Pinker appears to demolish Whorf�s examples, notably Whorf�s conjecture that they understand time in a fundamentally different way because they have no linguistic means of dividing time into units. This about a culture with advanced calendars, knotted ropes to represent days passing and other expected chronological paraphernalia peppering their sentences.
Strong initial support for some form of linguistic relativity is suggested by Brown and Lenneberg�s 1954 experiment which showed that differences in our ability to recognise and remember colours were associated with availability of specific colour names. If this experiment were to be accepted, it might demonstrate a weak form of linguistic relativity, since it directly relates a simple lack of vocabulary with elementary perception. In the same way, we might not expect a caveman to be able to remember or distinguish between a truck and a car. Perhaps though, familiarity and a simple understanding of their separate functions would help, as would being taught separate words for them. A large vocabulary forces fine distinctions upon us, but Whorf�s stronger form of linguistic relativity emphasised grammatical structure, a much finer and potentially more fundamental influence on cognition.
However, on closer inspection the experiment�s results imply somehow that speakers of certain languages are robbed of the biological capacity to perceive these different colours. It is not made clear how our knowledge of language could somehow affect the connection� between our retina and visual cortex, or wherever in our brains is affected. It seems inexplicable that such a subtle physiological change would result from the vocabulary of the language we speak. This confusion has been cleared by more modern experiments on the eight most common colour words, whose commonality appears strictly ordered throughout the world�s languages, suggesting that the likelihood that a language has a word for a particular colour mirrors the ease with which our eyes appear to be able to perceive and learn to identify that colour. Although experiments over the years appear to demonstrate that subjects are better at identifying the colour of different paint chips from memory if a word in their language approximates the colour, this makes sense if we imagine that our verbal and visual memory together are more reliable than one without the other. Hence, having a colour word would give us a convenient tag for a colour, without necessary precluding other colours from being perceived any less easily.
In a similar fashion, Alfred Bloom�s experiment intended to demonstrate that the the Chinese language�s lack of any basic grammatical construction mirroring the subjunctive means that the Chinese have great difficulty understanding counterfactual situations is a straw man. Researchers have since criticised the experiment on the basis of the standard of the Chinese in the story he used, and ambiguities in the science mentioned. Unsurprisingly, the results have been shown to be invalid by more recent experiments.
Linguistic relativity takes on many forms. Whorf�s hypothesis was based on a humanist hope that understanding the barriers language erects between cultures would allow us to better understand cultural differences, rather than excuse them. He even had hopes for building a universal language which transcended the gulfs between people�s perceptions of reality caused by language structure. It may be that gender biases in words like �mankind� perpetrate gender biases in attitude, but this is a moot point irrelevant to the discussion here.
The Sapir-Whorf hypothesis of linguistic relativity holds that our conceptual space is mapped, shaped and bounded by our language structure. Folk psychology points to evidence like the eskimos� fabled twenty words for snow. Again, Pinker is quick to attack such fallacies with the facts. However, the point remains that surely the language grew up from the constraints of that culture�s reality (hence this particular idiosyncracy of the Eskimo tongue), though each person�s grasp of it will shape their own thought. As Feuer noted, as we all share fundamentally the same reality, so should we largely expect our languages to have converged in similar directions. But, just as an individual person�s personality is shaped by the dimensions and condition of his/her physical body, while sharing the same DNA and race with those geographically close, so does our individual grasp of language shape us even as we share it with others. Language and genes differ across cultures, but need we expect our different languages to affect our conception of reality any more than our racially differing genes do? It may be that research to corroborate or dismiss Luigi Cavalli-Sforza�s claim that our historical language tree mirrors our gene tree would shed further light here.
An alternative theoretical escape from linguistic determinism lies in the idea of a language of thought, or �mentalese�. It could be that mentalese is our true mother tongue, but requires translation into words that can be articulated by our vocal cords. After all, how could we mean to say something slightly differently to the words we use if language and thought are inseparable and unmediated? How could we translate from one language to the other or remember the gist of a text or speech without remembering it verbatim? The answer lies in our internal representations, which need not be stored in our spoken language, or even in a mentalese. We know that at a low level they are stored by long-term potentiation as synaptic weights, but we don�t know how the brain�s structure aggregates up to a higher level, i.e. how the neural maps to the conceptual level. (So we could of course say that, in a sense, the existence of mentalese is a certainty, if we choose to see the digital firing patterns of our neurons as its vocalisation. Like an insanely complex series of bushman�s clicks, our brain forms as an aggregate a silent electrochemical voice box, communicating internally at colossal bandwidth. However, the term �mentalese� usually refers to a high-level representation of concepts and propositions which can be translated back and forth into words.) If some form of mentalese is the way we form words from thoughts, everyone�s inner speech is phrased in this same invisible, behind-the-scenes mental language, and attempting translation from it into restrictive Newspeak will result in a mute but fully-comprehending populace. But not for long.
Does this leave us with hope that Newspeak, were it to exist and be imposed upon us, could really quench fires in our minds by starving them of the lexical oxygen to burn? It does, on the two counts considered above. Strong linguistic determinism, where the vocabulary (and grammar) we use determines the paths and boundaries of our thought can be largely discounted. A weaker form of linguistic determinism, such as Whorf�s linguistic relativity, suggests that language might be sufficient to colour our picture of the Cosmos, but insufficient to remove the means to think heretical thoughts. Moreover, it seems likely that a language as deficient as Newspeak must surely succomb (like a pidgin to a creole) to untraceable, effervescent neologisms and idioms which grow to capture the expressive niche of the day. As for mentalese, the evidence for an accessible, universal, comprehensible high-level encoding of linguistic concepts is small � it may be that here methodological difficulties bar us from knowing definitively.