Don’t Sleep, There Are Snakes is an absolutely fascinating, compelling, humane and curious book.
It is a mix of personal memoir and linguistic field study and theory by Dan Everett, who spent a good chunk of the past thirty years living and learning with the Piraha people, who inhabit the land around a tributary to a tributary of the Amazon (“Piraha” should have a squiggle over the terminal “a” but I can’t coax my keyboard to produce it…)
He began his journey as a young Christian Chomskyan and ends it as an atheist naturalist. The lessons he learned along the way about language and culture have apparently stirred up a good deal of controversy in the linguistics community.
I’ve never liked the Chomskyan view of language, which has always seemed to me to put far too much emphasis on grammatical formality and far too little on the role of general intelligence, so I found Everett’s account of how the Piraha language lacks fundamental features that Chomsky and his followers insist must be part of all languages entirely collegial.
But then, I’m a poet, and I’m not aware of any school of modern linguistics that studies poetry. Why this should be so escapes me, because we learn far more from the extremal cases than common ones, as linguists tacitly acknowledge in their apparent fascination with nonsense sentences like, “Oysters that oysters eat, eat oysters”. For some reason they insist on presenting this an similar sentences without punctuation and without the “that”, as if they wanted to make it more confusing and difficult to understand: “Oysters oysters eat eat oysters.”
Poetry is speech in which grammatical structure is subordinate to rhythmical structure (this is the definition of poetry as I understand it.) In the most extreme cases poetry can evoke–and therefore communicate–without any grammar at all:
rain spring rain spring rain
moss roof gutter overflows
rag-ball soaked alone
That’s a very free translation of Buson’s famous haiku:
rooftop spring rain falls
in the gutter all alone
lies a child’s rag ball
But both evoke the same scene, and therefore both communicate. Anyone who claims grammar is necessary for communication doesn’t understand poetry. Like many human conventions and inventions, grammar is useful as hell–it allows non-poets and even linguists to communicate–but it is not necessary.
Everett finds that Piraha lack counting numbers, colour-terms, and many other things besides. At its deepest point, the Piraha language lacks recursion at the sentence level. It is not possible to within a grammatically correct Piraha sentence to insert a sub-clause. To say something like, “Yogurt, my big cat, is on my lap” a Piraha would say, “Yogurt is my big cat. He is on my lap.” Two sentences, rather than one with a subordinate clause. This is not a small thing, as this and other aspects of Piraha grammar imply that there are a finite number of valid sentences that the Piraha can speak, theoretically.
Chomsky (which I am continually wanting to write “Chompsky”, after Ryan North’s dog, Noam Chompsky) et al want recursion to be the fundamental locus of linguistic creativity, and seem to think that having a potentially unlimited number of sentences (of unlimited length) is somehow an interesting feature even though practical considerations do in fact limit sentences in recursive languages like English to a few thousand words. A sentence of more than a few hundred words becomes awkward in English, and as someone who has written a thousand-word sentence I can say with certainty that the ability or lack thereof for a language to contain longer sentences is completely uninteresting.
As such, it isn’t clear why anyone would think that having an infinite number of potential sentences that no one under any circumstances whatsoever would or could actually utter is a remotely interesting feature of a language.
But then, I’m a poet.
Everett reports on the musicality of Piraha speech, including the use of various forms: humming speech (their equivalent of whispering), shouting speech (modified for long-distance communication in storms), and so on. Despite the simplicity of their language–it has just eleven phonemes for men, ten for women–they sing and express recursive ideas by the larger structure of their stories.
I once wrote an essay on the nature of poetry in which I argued that the rhythmic structure of a poem existed at all levels, from the metrical foot to the divisions of stanzas and whole books in longer poems. It seemed to me obvious that no particular level of structure had exclusive dominion over any particular type of expression: one could have recursive expression between books as easily as between phrases within a sentence, and we have enormously structurally rich novels like Samuel R. Delany’s Dhalgren for example, whose highest level of organization is a fundamental part of the meaning of the book. This is not news… except apparently to linguists, who haven’t actually been studying language very much, but rather theoretical descriptions of language.
Like the sterile excursion into analytic philosophy that the English-speaking world took in the mid-20th century, the excursion linguistics took through formalist grammatical concerns is not without some value. But Everett makes the case for a more naturalistic (my word, not his) study of language, a greater focus on field studies to complement and control the theoretical excesses. This would make linguistics much more like the other sciences, in which controlled experiment and systematic observation, founded on deep natural histories, are used to guide and correct theoretical explorations.
During Everett’s work with Piraha, whose profound cultural and linguistic conservatism have made them resistant to conversion to Christianity for centuries, he found himself increasingly drawn to their world-view and focus on the immediacy of experience. And as a scientist, he was sympathetic to their skeptical response to his tales of Jesus. Piraha are loathe to take seriously things that they have not experienced themselves or heard about from people who have experienced them. Everett could claim no such experience of Jesus, and that led him to question and in time abandon his faith.
He also argues against the idea of “truth” as Descartes would have envisioned it, and any good Bayesian will back him up on that. The world is real and at least somewhat knowable, but “truth” is not the goal of science. Knowledge–contingent, empirical and less than certain–is.
Finally, he delightfully and correctly points out that the strong form of the Sapir-Whorf Hypothesis is incompatible with science, as science consists in precisely creating ideas for which we do not have any language. As someone who has been arguing this for decades it is truly wonderful to see someone who knows far more about these matters than I do weigh in on the side of the angels. While it is true that in very carefully controlled experiments designed to be incredibly sensitive to such effects it is possible to detect an influence of language on thought, the very effort that goes in to such measurements is proof that any but the very weakest form of the Sapir-Whorf claim is not worth serious consideration.
The relationship between language, grammar, biologically reasoning capacities and culture is not simple. I disagree with Everett that we are best to start off with trivially simplified models and correct them as we go. I favour messier, more realistic theories from the start. They are less likely to mislead people into simplistic complacency by continually reminding them that they are–or should be–reasoning about the world in all its glorious complexity, not some elegant but incorrect theoretical representation. This is an error we fall into all too easily, and embracing complexity has been the road to success in the sciences from the beginning. Hopefully Everett’s work, and the work of others like him, will continue to move linguistics in this direction.
