I didn’t understand this. I had never found the words for left hand and right hand. The discovery of the Pirahãs’ use of the river in giving directions did explain, however, why when the Pirahãs visited towns with me, one of their first questions was “Where is the river?” They needed to know how to orient themselves in the world!
Only years later did I read the fascinating research coming from the Max Planck Institute for Psycholinguistics in Nijmegen, the Netherlands, under the direction of Dr. Stephen C. Levinson. In studies from different cultures and languages, Levinson’s team discovered two broad divisions in the ways cultures and languages give local directions. Many cultures are like American and European cultures and orient themselves in relative terms, dependent on body orientation, such as left and right. This is called by some endocentric orientation. Others, like the Pirahãs, orient themselves to objects external to their body, what some refer to as exocentric orientation.
Clearly the Pirahã way of giving directions is very different from the average American’s. But even in English we can use an “absolute” directional system similar to the Pirahãs’. For example, we might naturally say, “The United States is north of Mexico.” Or we might say, “When you get to the stop sign, turn west.”Compass-based directions are similar to the river-based directions of the Pirahãs in that they are anchored in the world external to the speaker. But in English and many other languages, unlike in Pirahã, there is also a system of directions oriented to our bodies. So we say things like “turn left,” “go straight ahead,” “turn right,” and so on, with terms that are based on body orientation.
This system can be useful, but it requires that the hearer know where the speaker is and how her body is oriented before the hearer can follow the speaker’s directions. This is harder than it sounds in many cases. Imagine a speaker facing you. Then her left is your right, her straight ahead is your straight behind, and so on. Or imagine a speaker on the phone or otherwise out of sight, where the orientation of her body is unknown. This “relative,” body-oriented system of directionals can work in some situations, but it is inherently imprecise and sometimes confusing.
So English has both an externally anchored, efficient system of directionals and a body-oriented, occasionally confusing system. It is largely facts of history and English-speaking cultures that are responsible for the persistence of the dual systems. The Pirahãs lack a body-oriented system and only have the nonambiguous, externally anchored system (true, the Pirahãs have the advantage of always being close to the river, with respect to which they orient themselves). So the Pirahãs need to think more explicitly and consistently about their location in the world than we do. This in turn means that the Pirahãs’ language forces them to think differently about the world.
The implication of this finding is that language and culture are not cognitively isolated from each other. At the same time, we must guard against drawing unwarranted conclusions from this. We wouldn’t want to conclude, for example, that Brazilians and Mexicans think that Coca-Cola is female, just because they assign it to the feminine gender in their grammar. Nor would we want to say that the Pirahãs are unable to perform counting-related tasks, such as tallying on their fingers or toes, because they lack number words. This could be a misapplication of the idea that language shapes thought.
This idea has always been controversial. It is known by various names—linguistic determinism, linguistic relativity, the Whorf hypothesis, and the Sapir-Whorf hypothesis, among others, although the hypothesis is mainly associated with Benjamin Lee Whorf these days, because he was one of the first linguists to write extensively about specific examples of language shaping thought.
But Sapir also continues to be associated with the idea that language can affect culture profoundly. Sapir was a founder of American linguistics. He was also a student, along with Ruth Benedict, Margaret Mead, and other American anthropologists, of Franz Boas, a physicist-turned-anthropologist at Columbia University, considered by some to be the father of American anthropology. Sapir’s conclusions and proposals on the language-culture-cognition interface were based on his vast field experience, studies of languages of North America, their structures, their cultures, their histories, and the relationship between their cultures and their languages. A famous paper of Sapir’s claims:
Human beings do not live in the objective world alone, nor alone in the world of social activity as ordinarily understood, but are very much at the mercy of the particular language which has become the medium of expression for their society. It is quite an illusion to imagine that one adjusts to reality essentially without the use of language and that language is merely an incidental means of solving specific problems of communication or reflection. . . . No two languages are ever sufficiently similar to be considered as representing the same social reality. The worlds in which different societies live are distinct worlds, not merely the same worlds with different labels attached.(From Sapir’s The Status of Linguistics as a Science [1929], p. 209)
According to Sapir, our language affects how we perceive things. In his view, what we see and hear in our day-to-day existence results from the way that we talk about the world. This would certainly help us understand how it is that when walking with the Pirahãs in the jungle, I might say I saw a branch move and they might say that they saw a spirit move the branch. Sapir even goes so far as to claim that our view of the world is constructed by our languages, and that there is no “real world” that we can actually perceive without the filter of language telling us what we are seeing and what it means.
If Sapir and Whorf are correct, the implications for philosophy, linguistics, anthropology, and psychology, among other fields of study, are vast. Whorf went so far as to claim that Western science is largely the result of the grammatical limitations of Western languages.
Could Kant’s a priori categories of morality be an artifact of the distribution of nouns and verbs in the grammar of German? Could Einstein’s theory of relativity? Unlikely though such hypotheses seem, they are raised by Whorf ’s suggestions.
For linguistics and anthropology, the Sapir-Whorf hypothesis suggests research questions for investigating how our languages cause us to think differently about the world.
The Sapir-Whorf view implies a symbiosis between language and thought. In the extreme version of this view (linguistic determinism), which virtually no one accepts, thought cannot escape the bounds of language. Speaking a particular language can give our thinking an immutable advantage or disadvantage, depending on the task and language involved.
A more widely accepted version is that while we can think “out of the language box,” we normally do not because we don’t even perceive how language affects the way we think. This version is even observed in practice among people who might explicitly reject the Sapir-Whorf hypothesis.
As an example of how intelligent people can be “conflicted” about the idea that the way we talk can affect the way we think, consider the views of members of the Linguistic Society of America. The LSA has strict guidelines against sexist language. This means that at least some members of the LSA think that the way we speak affects the way we think in ways identical or at least related to a mild form of the Sapir-Whorf hypothesis.
However, other LSA members reject almost every version of this hypothesis. What fascinates me is that both of these groups agree that the LSA should promote the use of gender-neutral language. One member, for example, might simultaneously give a paper against the hypothesis of linguistic relativity while being very careful to use only they or forms like s/he in his or her paper, as opposed to lapsing into use of he for both genders, as in “If anyone wants this job, they can have it”versus “If anyone wants this job, he can have it.”
This is not happening simply because gender-neutral language is more polite than gender-specific language. The pressure to change English in this way arose because people believe that the way we talk, whether offense is consciously intended or whether or not politeness
is at stake, affects the way we think about others.
cI have seen a sufficient number of psycholinguistic studies and have heard enough anecdotal evidence of the effects of language on thought to conclude that this weaker version of the Sapir-Whorf hypothesis is not an unreasonable idea.
At the same time, I don’t think the hypothesis does the work that some people have wanted it to do. In explaining the lack of counting in Pirahã, for example, it seems unhelpful. If the Sapir-Whorf hypothesis were taken as explaining the Pirahãs’ lack of counting (they don’t count because they lack number words), several facts would be left unexplained.
For example, many other groups around the world have had very impoverished numeral systems, but they have had counting and have borrowed numerals from surrounding languages as socioeconomic pressure builds to be able to use numbers in trade. The Warlpiri of Australia are an example. And the Pirahãs have been engaged in trade with Brazilians for more than two centuries. Yet they have not borrowed any numerals to facilitate their trade. Under a Whorfian account of Pirahã counting, there is little reason to borrow words to express concepts that become useful, because concepts couldn’t become useful without the words to begin with. That account would predict, wrongly, that without the words there cannot be the concept. In fact, this strong Whorfian account is incompatible with science, because science is largely about discovering concepts for which we previously had no words!
The Sapir-Whorf hypothesis fails to offer any unified account of the range of unusual facts about Pirahã culture and language, such as the absence of color words, quantifiers, or numerals, the simple kinship system, and so on.
Our quest for an account of the interaction of Pirahã language and culture needs to be placed in the context of the intellectual territory to be traversed. We need to map out some of the different relationships between grammar, cognition, and culture that have been proposed over the years. I summarize the leading ideas in the table opposite:
Cognition, Grammar, Culture Connections
We all know that any attempt to understand how culture, cognition, and grammar interact and affect one another must avoid simplistic solutions to the understanding of what shapes the “human experience.”At the same time, it is useful and necessary to begin with some idealization or deliberate simplification by which we can focus our attention on salient points of connection among these three domains while ignoring others momentarily. This is a useful way to come to grips with such complex material.
The first row in the table above expresses the case in which cognition, by which I loosely mean either the cerebral or mental structures necessary for thought or thought itself, exercises control over grammar. Noam Chomsky has focused exclusively on the effects of cognition, in this sense, on grammar for several decades, proposing the idea of a universal grammar as his idea of how cognition limits human grammar.
Universal grammar (UG) claims that there is in effect just one grammar for all the world’s languages, with variation allowed by a relatively small number of “principles and parameters.” The experience of growing up in one environment and hearing a particular language spoken will flip switches that call forth this or that grammatical property in the child’s emerging grammar. So suppose that you are born in Brazil and grow up hearing Portuguese. According to the UG line of reasoning, as a child you adopt a parameter called the “null-subject” parameter, the fact that sentences do not have to have overtly expressed subjects. So in Portuguese the equivalent of Saw me yesterday is grammatical, whereas in English it is ungrammatical. And Portuguese will have more information in its verbs about the nature of the subject (at least person and number) than in English. And so on. This has been easily the most influential of all the research traditions looking into the relationship between grammar and cognition.
Row two of the table symbolizes the Sapir-Whorf research tradition, which looks at the grammar-cognition interface from the perspective of how grammar, that is, the way our languages are structured, might affect the way we think.
For the third row, the names that come to mind are Brent Berlin and Paul Kay, both emeritus professors of the University of California at Berkeley. Their work purports to show that all cultures’ classification of colors follows restrictions imposed by the human brain’s physical constraints for recognizing tints, hues, and relative brilliance of colors. This cerebral-cognitive limitation imposes constraints on the classification of colors in all cultures.
Row four represents the perspective of linguistic anthropologists like Greg Urban of the University of Pennsylvania. Urban’s work makes the case that language can affect culture in interesting and subtle ways. One of the examples he discusses concerns the effect of passive (such as John was seen by Bill) versus active (Bill saw John) grammatical constructions on the concept of the hero in different societies.
Urban claims that in some languages the proportion of passive clauses may be much higher in natural oral or written discourse than active clauses, while in other languages active sentences may occur much more frequently. He further makes a case that when passives are the more natural and most frequent type of construction, heroes discussed in discourses will be perceived more naturally as having things done to them rather than initiating actions. These heroes will be perceived as having more passive personalities than the heroes of languages where active sentences predominate.
In the case of a language without passive constructions, we would encounter sentences like The man killed the jaguar and The jaguar killed the man, but not The man was killed by the jaguar. When an action is performed, the one doing the action is central to the telling of the story.
On the other hand, in a language that favors passive constructions the one doing the action is less central to the story. For example, if we compare closely the active versus passive contrasts of examples like The man killed the jaguar and The jaguar was killed by the man (or, even more likely in a passive construction, The jaguar was killed) occurring time and time again in stories, we would soon realize that in the passive, the role of “the man” is reduced in centrality. What becomes central is the object of the action, here “the jaguar,” and not the subject or doer of the action. Such contrasts can work hand in hand with the culture to produce either heroes that are central to the telling of stories or stories in which doers of action are not so crucial, not so central, and, hence, not so heroic.
Since Pirahã lacks passive constructions, its principal characters in stories, like the panther story, are active initiators of actions and much more heroic than their counterparts in languages that favor passives. (I will not give an example of the latter, since I am only offering a simple summary of Urban’s theory. In fact, I think examples of languages like this might be more complicated than this theory suspects.) In any case, this underscores how crucial it is to study language and culture together, rather than in isolation. Like my own work—though from the opposite direction—this goes against the traditions of both modern linguistics and much of modern anthropology.
Row five represents the research that investigates how culture can affect cognition. The Pirahã case is a good example. Pirahã’s lack of counting is a result of cultural constraints, as we discussed earlier. But this cultural by-product has cognitive effects—Pirahã adults find it nearly impossible to learn to count after a lifetime spent in a numberless environment.
Finally, the last row in the table represents the research that others, including me, are doing on local and global effects of cultural values on sentence formation, word structure, and sound structure. This is controversial work too and also goes against much received knowledge in linguistics. It is what the immediacy of experience principle, for example, is trying to get at.
15 Recursion: Language as a Matrioshka Doll
Theories affect our perceptions. They are part of the cultural information that constrains the way we see the world around us. There are many examples of culture-perception connections that don’t involve science but which illustra
te my point, such as the time I mistook an anaconda for a floating log. My culture told me to look for logs when traveling by boat (universally good advice!). And it gave me information on what floating logs look like in a river. But it didn’t have anything to tell me about what very large anacondas look like swimming toward you.
We were traveling out of the village in our own motorboat, all the way to Humaitá to catch the bus to Pôrto Velho. Keren had made tuna sandwiches with homemade bread and we had Kool-Aid to drink. As I piloted the boat down the Maici and then the Marmelos, everyone was relaxed. Shannon was reading the Brazilian comic book Monica while the others dozed or watched the scenery go by.
We came to my favorite part of the entire trip, the encontro das aguas, where the dark green water of the Marmelos meets the chocolate-milk-colored water of the Madeira. I shouted for everyone to look and we all watched as the two colors of water ran for a bit side by side, then we saw swirls of muddy water in the green water, then finally the green water was absorbed, about five hundred yards past the mouth.
I then turned my attention upriver, as we skirted the island that sits at the mouth of the Marmelos, motoring toward the Auxiliadora, where we would spend the night. The Madeira River is so named for the trees that are washed from its muddy banks and float down it toward the Amazon. There are huge trunks and branches in the river, especially dangerous when they float invisibly just beneath the surface. About two hundred yards upriver I saw a log floating in the fast current. It was twisted. When I first began to travel the Amazon system, I expected to see new things in this new world, so I mistook every log in the river for a snake, because the water makes wood seem to undulate. This log seemed to undulate as well, though by now I knew enough not to mistake it for a snake. And I also knew that snakes were not as big as logs. This one, as I watched it more closely, was perhaps forty feet long and three feet thick.
Don't Sleep, There are Snakes Page 26