Alex Speaks

Humans have been fascinated with bird songs probably since our ancestors first evolved the capability for fascination. Early on, listening to and appreciating birds paid off in food—either the bird itself, its eggs, or some other reward—in the case of the honey guides, the bird’s preferred food, honeycomb. Later, birdsong became to humans a natural art form that we still highly prize.

But to a species as self-absorbed as humans, nothing beats hearing your own words. And since the time of Aristotle, people have known that parrots can talk back. Several species can mimic human speech, from “Hello” to “Polly want a cracker.” But does imitation imply understanding? Irene Pepperberg, of the University of Arizona, set out in the spring of 1977 to use the ability of parrots to speak as a tool in understanding what they have to say. She chose a champion speaking species, the African Gray parrot, in particular an African Gray named Alex.

Alex, Pepperberg has found, can do much more than learn 100 words, including 90 names of objects. He can also answer questions. Pepperberg asks “What’s this?” (holding up a green key) “Green key,” Alex responds. Holding up a blue pentagonal wooden block, Pepperberg asks “What color?” and Alex answers “Blue wood.” But if she instead asks “What shape?” Alex knows to respond “Five-corner wood.” (Apparently “pentagonal” is a bit beyond Alex.) Alex can also pick a red key out of a collection of red non-keys and non-red keys. Asking Alex whether he can count required Pepperberg to invent a careful experimental design.

Birds, it turns out, keep track of numbers of things quite well. Crows keep track of both the number of caws they hear and their rhythm to identify each other, scientists think. Pepperberg likens this ability to singing “Deck the Halls” at Christmas without counting the number of fa-la-las. Timing is a critical component of singing for songbirds, involving the number, sequence, and duration of notes. In lab tests, pigeons beat humans at determining the ratio between durations of short events.

Pepperberg attempted to prevent Alex from relying on this kind of all-at-once perception of number with a strategy similar to one used with humans. She made the task confusing. She set out a bewildering (at least to humans) collection of objects from wooden tongue depressors and spools to metal, paper and plastic cups. Each object fell into a color category (rose, green, purple, blue, yellow, orange and gray). Pepperberg and her students then began to ask Alex “How many blue cups?” or “How many orange keys?” For each test, the experimenters had arranged the objects so they could test Alex on a particular number, starting from one object that met the criteria and advancing to six. Alex achieved a “B” average, answering more than 83 percent of the questions correctly. He made about the same number of mistakes no matter what the number of objects being asked about. Unless Alex’s ability to visually segregate objects that met the criteria and then perceive their total without counting greatly exceeds that of humans, Pepperberg writes, Alex was counting in the same sense that we do.

Pepperberg cautiously draws some parallels between Alex’s learning and the learning of preverbal children. For example, children use monologues. The monologues rarely appear as an attempt to communicate content, even when spoken in company. Furthermore, children also speak monologues when they are alone, often practicing new words in this non-judgmental environment.

Alex also practices while alone. Pepperberg recorded all Alex’s vocalizations each evening after the trainers had left. Alex included in the monologues words his trainers had been teaching during the day, but that he had never spoken before. Alex often practiced these to himself before using them in replies to trainers. And he always practiced privately before his use of a new word became reliable.