Books : reviews

Daniel L. Everett.
Don't Sleep, There are Snakes: life and language in the Amazonian jungle.
Profile. 2008

rating : 3 : worth reading
review : 30 April 2011

Dan Everett was a missionary off to convert the Pirahã, an Amazonian tribe. His approach wasn’t preaching, but to translate the New Testament into their language, and let it speak for itself. So he had to learn their language. This involved him getting training in linguistics, living with the tribe, immersing himself in their culture, so that he could understand the language in context. In turn, this led to some startling discoveries about their culture, their language, and the relationship between them. This is his account of those discoveries, and the effects they had on him.

The first part of the book is about the learning process, of the people, their culture, and their language. The Pirahã are remarkable for being very happy: smiling, laughing, talking, all the time, even through the night. They are a proud, tough, hard working, peaceful, contented people, who see no need to import external inferior ideas into their lives. Oh, they will ask for matches, and alcohol, and canoes, and medicines, but they are not interested in acquiring lots of possessions. There is a scene where they ask Everett for a canoe, and he brings in a local craftsman to teach them how to build one. They work hard, build the canoe, are proud of what they have done, and off goes the teacher. Then they ask Everett for another canoe. They have no interest, it seems, in assimilating the external canoe-building skill into their own highly developed skill-set. I wonder, though, from other discussions, if it is as simple as that. There is another story of how, despite eight months of highly motivated lessons (they wanted to stop being cheated by the traders), they nevertheless failed to learn how to count. Their language and worldview simply don’t support the crisp concept of exact number. Why then should their language and worldview support recalling a long complex building process? Just because they had once managed to perform a complex highly-skilled task under instruction is no reason to expect them to be able to recall and repeat the task with no instruction. After all, they appear to have happily learnt a rudimentary form of farming to supplement their mainly hunter-gather lifestyle.

So, what is the Pirahã language and worldview, and how are they related? Having set the scene, the second part of the book then goes more deeply into this. The Pirahãn language is much "simpler" than any other known language (where the measure of simplicity is based on linguistic theories of language structure, not on any measure of expressiveness). There are very few phonemes (sounds): three vowels, seven or eight consonants, and two tones (although this simplicity allows several forms of the language: spoken, hummed, whistled, sung, yelled, used in different contexts). Nouns have no number:

pp195-6. Aside from this [suffix meaning "my own"] , Pirahã nouns are for the most part very simple. … they do not have plural or singular forms ... It’s as though every Pirahã word were like the English words fish and sheep in having no plural.

Although they have no concept of "number", they do have a concept of "amount". Two small fish, or one big fish, are described as the same "amount".

Verbs are somewhat more complicated.

p196. Each verb can have ... up to sixteen suffixes in a row. ... Since a suffix can be present or absent, this gives us two possibilities for each of the sixteen suffixes---216, or 65,536, possible forms for any Pirahã verb. The number is not this large in reality because some of the meanings of different suffixes are incompatible and could not both appear simultaneously. But the number is still many times larger than in any European language. ...
     Perhaps the most interesting suffixes ... are what linguists call evidentials, elements that represent the speaker’s evaluation of his or her knowledge of what he or she is saying. There are three of these in Pirahã: hearsay, observation, and deduction.

But it’s not the individual sounds or words that make this language simple. Everett discovered that the sentence structure is surprisingly --- astoundingly --- simple. All languages were thought to use recursion, with noun phrases and verb phrases that can be nested ("I saw a dog in the hall chewing a big meaty bone with enjoyment"). Not so Pirahãn, where the sentence structure is essentially "noun (subject) noun (object) verb" with a few embellishments, but no recursive phrases. This means there is a fixed upper limit to the length of a Pirahãn sentence (and quite a short limit at that), compared to the unbounded length of a sentence in a recursive language (although I have commented previously on the the difference between being technically unbounded, but pragmatically very bounded). But this lack of recursion doesn’t stop the Pirahã expressing complex ideas.

p228. the evidence I was collecting was beginning to build support for two ideas I later came to hold about Pirahã sentence structure. The first was that Pirahã sentences lacked recursion. The second idea was that recursion wasn’t all that important---apparently, whatever you could say with recursion in one language, you could say without it in another. Linguists have long believed, though not always using the same terminology, that recursion is very important in language.

The reason they don’t need recursion in single sentences but can still tell complex stories is that they use multiple linked sentences to build up the structure. But this doesn’t fit with how classical linguistic theory says single sentences should have such structure.

So, the Pirahã can express recursive thoughts, just not in single sentences. Why is their language structured this way? Everett believes that it is intimately connected with their culture. Not only does grammar affect culture, but culture affects grammar. The sentence, the unit of linguistic analysis, doesn’t happen in a vacuum, but in a physical environment and a culture and a shared context. Have you never said a phrase or sentence fragment, seemingly out of the blue, to you partner or a colleague, that in isolation makes no sense, but been perfectly well understood, because of your deep shared context? Everett uses the example of giving directions. We might say to turn left at the lights; we don’t say to stop and wait for them to go green first: that’s shared culture. Moreover, Pirahã don’t use relative directions like "left" and "right", but, as in some other languages, they use absolute directions based on their physical environment, here "upriver", "downriver", "into the jungle". In order to understand the language as used, you have to understand the environment and culture within which it is being used to communicate. You have to live in, be embedded in, the culture. (This is well-known in Artificial Intelligence: to get computers and robots to understand natural language, they either need an enormous amount of cultural and contextual background, or need to be restricted to highly constrained situations.)

This seems fairly obvious, but Everett goes a step further. He says, based on his understanding of the Pirahãn language, that not only does the environment, culture and context affect what is said, but it affects the grammar of the language used to say it. A very constrained culture, and a deeply shared context, leads to a very constrained grammar. (This would presumably imply that in cases where there is little shared context, such as wanting to express new ideas or new discoveries, you would need a more sophisticated grammar to express it.) This leads Everett on to decide that (recursive) grammar need not be hard-wired into out brains, but can naturally emerge from context, culture, and simpler structures.

p238. If you can’t talk about things and what happens to them (events) or what they are like (states), you can’t talk about anything. So all languages need verbs and nouns. … if a language has these, then the basic skeleton of the grammar largely follows. The meanings of verbs require a certain number of nouns, and those nouns plus the verb make simple sentences, ordered in logically restricted ways. Other permutations of this foundational grammar follow from culture, contextual prominence, and modification of nouns and verbs. There are other components to grammar, but not all that many. … there really doesn’t seem to be much need for grammar proper to be part of the human genome

p241. The crucial application of [Herbert] Simon’s proposal for studies of language is that hierarchical structures found in languages that have so long been the focus of Chomskyan research are "emergent" properties, rather than basic properties of language. That is, they appear in languages in response to the interaction of the brain’s ability to think recursively and problems or situations in the culture or society that are more efficiently communicated recursively.

This constrained grammar, resulting from a constrained culture, does not mean that the Pirahã are "primitive", or that they can’t say what they want to say. All cultures have restricted domains of discourse.

pp257-8. the Pirahãs love to talk. ….
     They just don’t talk about many things. But neither did my family in Southern California when I was growing up. … most university professors I know, have if anything as narrow a range of conversational topics as the Pirahãs. Linguists talk about linguistics and other linguists much of the time. Philosophers talk about philosophy and philosophers and wine.

The Pirahã can get away with a simple language because they have only the one environment, culture and context. We have more "complex" languages because all our different subcultures, each with their restricted domains of discourse (cattle rearing, linguistics, philosophy, whatever), all use the same language.

pp258. Of course, to do all of this talking within the confines of a single language, our language has to be adequate for all the academic disciplines, professions, trades, and so on.

One important aspect of the Pirahã culture that leads to this simple grammar is what Everett identifies as the "immediacy of experience principle". The Pirahã are only interested in things they have experienced directly, or in first hand eyewitness accounts. If you didn’t experience it yourself, if you have no direct evidence for it, they are not interested in hearing about it. So they have no fiction, essentially no history beyond living memory, no creation myths, no future planning, and no fear of death: they line "in the moment". (Although, again, it is not quite as simple as that. When they know Everett is going to town, they ask him to bring back supplies. There is some forward thinking here.)

This is why they don’t need complex phrases to identify particular things ("the dog that ate the bone belongs to Dan", where the phrase "that ate the bone" identifies which particular dog I’m talking about); immediacy means it’s that dog there in front of you (or the one being talked about by the eyewitness). If you want to link facts, use consecutive sentences. ("That dog ate the bone. That dog belongs to Dan.") There is a reinforcing loop between culture and grammar: the Pirahã are only interested in immediate experience; their grammar both supports and enforces this interest.

This explains why it has been impossible to convert the Pirahã to Christianity. No first-hand experience of this Jesus? Were you there? No? So we’re not interested in hearing about him then. This demand for evidence not only insulates the Pirahã from missionaries’ stories, it made this missionary question his own beliefs.

p270-1. The immediacy of experience principle means that if you haven’t experienced something directly, your stories about it are largely irrelevant. This renders them relatively impermeable to missionary efforts based on stories of the long-ago past that no one alive has witnessed. .... Creation myths are no match for this demand for evidence.
     … The Pirahãs’ refusal to believe something just because I said they should … caused me to question my own faith. This surprised me. I was no novice, …. I was well trained in apologetics and personal evangelism.
     But I had now also been trained as a scientist, where evidence was crucial, where I would demand for any claim evidence similar to what the Pirahãs were now requesting of me. I did not have the evidence that they wanted. I only had subjective support for what I was saying, my own feelings.
     All the doctrines and faith I had held dear were a glaring irrelevancy in this culture. They were superstition to the Pirahãs. And they began to seem more and more like superstition to me.
     I began to seriously question the nature of faith, the act of believing in something unseen. .... The Pirahãs’ values of immediacy of experience and demand for evidence made all of this seem deeply dubious.

Eventually, Everett realised he was an atheist. They had converted him.

This is an fascinating account of the effect of language on culture, and vice versa. It is frustrating in some places where I don’t think the issues are followed through sufficiently rigorously, but that is more because those particular points are peripheral anecdotes rather than part of the central linguistic investigation. Where things are central to the science, there is a fair amount of detail. There is no bibliography to follow up these details, however. (And why didn’t he try to teach the children to count?)

update, 2 October 2016

E. J. Spode’s frank review of Tom Wolfe’s book The Kingdom of Speech suggests that Everett’s findings may be somewhat overstated.