Language and cognition¹

The problem

That the capacity for naming has a biological dimension may be seen from the difficulties that animals experience in this respect. For instance, it is possible to train a hunting dog to ‘ point ’, and it may be quite possible to teach him to point to a specific set of objects in a specific environment upon appropriate command in a natural language. But it does not appear to be possible to teach a dog to do the ‘name-specific stimulus generalization’ that every child does automatically. The hound who has learned to ‘ point to the tree, the gate, the house ’ in the trainer’s yard will perform quite erratically when given the same command with respect to similar but physically different objects, in an unfamiliar environment. The correctness of the animal’s responses may even vary with such extra-linguistic cues as the geographical position, posture, and bodily movements of his master, the time of day, or the clothes that people are wearing while he is being exercised. There is no convincing evidence that any animal below man has ever learned to relate any given word to the same range of stimuli that is covered by that word in common language-usage. So- called proof to the contrary always lacks proper controls on interpretation. For instance, there is a report on a parrot who could say good-by (in German) and who supposedly knew what this word meant or when it was properly used. Once the bird was also heard to say good-by upon the arrival of some friends of the family; the proud owner judged this to be a sign that his pet did not merely know the meaning of the word but was even using it to produce a desired effect: to send the just- arrived friends away, presumably because he had taken a dislike to them.

It may be well to stress once more that our concern is with the capacity for (natural, human) language which, ordinarily, leads to the understanding of a definably structured type of utterance; or, in other words, with knowing a language. The infant who has a repertoire of three tricks (wave by-by, show me your tongue, show me how tall you are) which he can perform upon the appropriate commands but who can understand no other sentence of the same grammatical, structural type has not yet begun to acquire language. The essence of language is its productivity; in the realm of perception and understanding of sentences, it is the capacity to recognize structural similarities between familiar and entirely novel word patterns. Thus our criterion for knowing language is not dependent upon demonstrations that an individual can talk or that he goes through some stereotyped performance upon hearing certain words, but upon evidence that he can analyze novel utterances through the application of structural principles. It is the purpose to show that the understanding of the word-object relationship, the learning and acquisition of reference, is also dependent upon certain cognitive, analytic skills, much the way understanding sentences is. The problem of reference cannot be discussed without simultaneous considerations of the relationship between language and cognition.

Evidence for understanding language may be supplied by different kinds of response. It is not necessary that the subject has the anatomical and physiological prerequisites for actual speech production. In the case of man, we may cite children who have learned to understand language but who cannot speak; compare this to children who have the anatomical equipment for speech production but whose cognitive apparatus is so poorly developed that only the primordia for language are detectable but not fullfledged comprehension. In the case of animals, we have birds who can talk but who give no evidence of language understanding and we have the famous case of Clever Hans, the horse, who had a nonacoustic response repertoire (stamping of hoofs) that, unfortunately, gave the erroneous impression of a coding system for the German language. Had the horse actually had the cognitive capacity for acquiring a natural language, his motor response limitations would have been no obstacle to his giving evidence for language comprehension. A similar argument could be made for the physical nature of the input data. Language acquisition is not dependent in man upon processing of acoustic patterns. There are many instances today of deaf-and-blind people who have built up language capacities on tactually perceived stimulus configurations.

¹ This paper is reprinted from E. H. Lenneberg, Biological Foundations of Language, Wiley, New York, 1967.