Does language determine thought? Are there concepts in some languages that can’t be understood in others because that language doesn’t have the word for it?
It seems like a reasonable question. But selection has equipped us all with the same tools, regardless of where we grow up. The Dani people of Indonesia don’t have words for colors, but that doesn’t mean they can’t understand the concept of color. Humans have special cells in the retina that are tuned to light at the wavelengths corresponding to red, green and blue.
What about more abstract concepts? For many years, the Hopi in North America were thought by some academics to have no words in their language for time, past or future. This idea was put about by Benjamin Lee Whorf, a fire-prevention officer and linguist, who trained at Yale University in the 1930s. Once it had taken hold, the idea quickly followed that the Hopi had no concept of time, or at least not the concept that most of us have.
However, it turns out that the Hopi have plenty of words and metaphors for time, and complex and accurate methods of tracking and marking it. But Whorf — who also propagated the myth that Eskimos have many words for snow (they don’t) — had a huge influence on linguistic thought, going as far as to hypothesize that language determines thought itself.
The idea has survived despite being unsupported by evidence.
Yet last week in the journal Science, Peter Gordon of Columbia University in New York published a paper suggesting that — as far as numbers are concerned — it is necessary to have words to be able to comprehend them. If corroborated, Gordon’s work would confirm, for numbers, Whorf’s hypothesis on thought.
Gordon, a bio-behavioral scientist, was reporting on his work with the Piraha, a tiny hunter-gatherer tribe living in Brazil on the Maici River, a tributary of the Amazon. Their language is the simplest known, having just 10 different sounds (phonemes). In comparison, English has 40 to 45 phonemes.
For numbers, the Piraha language only has words for one and two, and even those words are vague, meaning “about one,” and “more than one,” respectively. To refer to numbers above two, the Piraha say their word for “many.”
Gordon gave seven Piraha men simple tasks to determine their ability to count. He placed objects (such as nuts and small batteries) in a row and asked each subject to duplicate the pattern. The Piraha subjects responded accurately with up to two or three items, but their performance declined when challenged with eight to 10 items, and dropped to zero with larger sets of objects.
Gordon says that his experiments show that having the right linguistic resources determines a person’s reality. “Whorf says that language divides the world into different categories,” he said. “Whether one language chooses to distinguish one thing vs. another affects how an individual perceives reality.”
Can this be true? Elephants, which seem to mourn their dead babies, can apparently think about death, but they don’t have words for it. Baboons and chimps, it has been shown, are capable of abstract thought. Can it really be that just because the Piraha don’t have words for numbers, they can’t count, when monkeys, chimps and birds have been shown to be capable of counting?
“This is not my interpretation of the results, though it is quite a plausible one,” said Daniel Everett, a linguist at the University of Manchester in northwest England, who studied the Piraha with Gordon in the Amazon.
“My own view of the matter is that Piraha counting is part of a larger series of culturally based constraints on grammar and semantics in Piraha, which reflect a value against quantifying beyond the immediate experience of living members of the community.”
The Piraha are able to recognize quantity, Everett stressed, but only in a relative sense. They are not precise. They used their fingers in addition to their verbal statement of quantity, but this practice, too, was found to be highly inaccurate even for small numbers below five.
It still seems strange that some animals can count and humans can have trouble.
“But the data showing that animals can count is only obtained after weeks of training and conditioning,” Gordon said in an e-mail interview. “This is not counting, in the sense of tagging a numeric value that increments by one to get the cardinality of a set.
“If you look at the Piraha data, they also perform above chance [just as trained animals do]; you just expect humans to be a whole lot better,” he said.
“So, if you get 70 percent performance in matching five items from a pigeon, it looks great; if you get it from an adult human, it looks pretty crummy.”
Does the study vindicate Whorf’s ideas?
“My claim,” Gordon explained, “is not that language determines thought in general — you can clearly think without words — but that this is proof that it is possible for linguistic differences to radically limit the concepts that can be entertained.”
If more research supports Gordon — and more research is needed — then the idea that language can influence thought will have gained some much-needed respectability.
But Gordon had a final caveat: “For all I know, this may be a very rare exception to the rule.”