A Dog With Two Names?

It ain’t easy learning words—especially for an infant fuzzy about just how words work. Even after he’s miraculously concluded that some words refer to things in the environment, he’s left with some rather daunting questions: which words and which things?

So over the years, cognitive scientists have posited a variety of means by which a baby might whittle down the number of red herrings. Few theories have received as much attention as Stanford psychologist Ellen Markman’s theory of mutual exclusivity.

Show a baby an unfamiliar percussion instrument in one hand and a banana in the other. “Where’s the cabasa?” you ask. A four-month-old, if you can get her to look at the objects at all, will look at whichever strikes her fancy. A one-year-old, while fully aware that the banana is called “banana,” will be nonetheless flummoxed: What is this “cabasa” of which you speak? But an 18-month-old understands that since the banana already has a name—“banana”—then “cabasa” must refer to the unfamiliar object. That’s the theory of mutual exclusivity: the assumption that everything in the world has one and only one label. (Note that mutual exclusivity does not strictly preclude children from learning additional labels, like synonyms; it merely offers resistance.)

Mutual exclusivity has plenty of support—for monolingual babies, that is. But what about babies who must learn more than one language? For these infants, it seems, learning that “banana” has one and only one label wouldn’t be all that helpful.

Interestingly, the data bear this out: babies learning two or more languages are far slower to use mutual exclusivity to attach novel labels to novel objects. In a 2009 study, psychologists Krista Byers-Heinlein of Concordia University and Janet Werker of the University of British Columbia found that 18-month-old bilinguals showed a weaker preference for mutually exclusive object labels than did monolingual babies—while trilingual infants (the show-offs!) seemed to have no preference at all.

A new study by the same researchers paints an even clearer picture. Twenty bilingual 18-month-olds repeatedly saw pairs of objects, some consisting of a familiar object—a shoe—paired with a novel one. Would the infants understand that the novel label goes with the novel object when asked, “Where is the nil?” The researchers, who’ve seen bilingual infants struggle with this task in the past, were fairly confident the answer would be no (and they were right).

The researchers were really interested in whether some of the bilingual infants would struggle more than others. Specifically, would bilingual infants who had learned a large number of “translation equivalents”—that is, word pairs like the English word “dog” and the Mandarin word for dog, gǒu—be less likely to use mutual exclusivity than bilingual infants who knew fewer translation equivalents? This would confirm that babies only assume mutual exclusivity to the extent that “one-object, one-label” approximates the true state of affairs as they understand it. Sure enough, the researchers found just that.

Eventually, though, even bilinguals and multilinguals acquire mutual exclusivity, albeit of the more nuanced “one-object, one-label-per-language” variety. So where does mutual exclusivity come from then? Is it innate, a biological predisposition that manifests over time? Or is it a learned behavior? Oddly enough, this study doesn’t bring us much closer to an answer. Learning is clearly involved; it’s just not obvious what must be learned: To assume mutual exclusivity, or not to, at least not yet?