Page 70 - Encyclopedia of Philosophy of Language

Language and Mind
at the University of Oklahoma and then at Central Washington College. All three of these projects used manual signing based on the American Sign Language for the Deaf. Primarily through use of various modes of trial-and-error teaching methods entailing molding of the ape's hand, young chimpanzees learned at least when to make a given manual sign or gesture in relation to each of a variety of exemplars and events (Fouts 1972). Whether the apes knew what they were signing was, at that time, not recognized as a central question. The emphasis was upon production, i.e., the use of signs by the apes.
In retrospect, all of the foregoing projects shared a common error. All project leaders naively assumed that, if an ape appropriately produce a sign, that it would also comprehend or understand that sign when used by other social agents. Since 1980, however, it has been learned that the skills of production do not by themselves warrant the conclusion that under- standing, by the user, is in place. In 1992 data and arguments also strongly support the conclusion that it is comprehension or understanding that is the critical ingredient for language, not the ability to produce signs, or to make specific sounds.
As discussed below, language understanding can be instated in chimpanzees by rearing them from birth much as human children are reared—in an environ- ment where language is used to announce and to coor- dinate social activities (Savage-Rumbaugh, et al. 1993).
A review of research, enabled by the computer- monitored lexigram keyboard that has led to our present-day understanding of language and the pro- cesses whereby apes can acquire it, is in order.
1.2 TheLANA Project
Research with Lana was designed so that she could exercise substantial control over her environment and life through use of her keyboard. On that keyboard, each key was embossed with a distinctive geometric symbol, called a lexigram. Each lexigramwasintended to function as a word, just as words generally do in our vocabularies. The first symboliccommunications learned by Lana were 'stock' sentences, i.e., they were specific sentences which had to be used by her in order to activate a variety of devices controlled by the computer. The software program in the computer specified certain relationships between words that had to be honored if Lana's multilexigram productions were to have any effect. Examples of stock sentences were, 'Please machine give milk,' 'Please machine make window open,' 'Please machine give piece of apple (or bread, banana, etc.),' and so on. Lana's use of stock sentences availed to her a variety of foods and drinks, music, slides, movies, a view out of the window, and so on. In addition, two-way symbolic communication between Lana and her caretakers became possible at least to the extent that specific
kinds of tests could be conducted to determine her skills of naming objects and their colors (e.g., 'What name-of this that's blue [or red, orange, green, black, etc.]?,' 'What color of this box [or ball, cup, shoe, etc.?]').
Lana demonstrated that an ape could learn to use more than 200 lexigram symbols, either singly or in stock sentences, with relatively high levels of accuracy. Lana also made interesting modifications in several portions of her stock sentences and phrases in a man- ner that suggested insight on her part that such could be used to achieve special purposes. For example, a cucumber was asked for as 'the banana which is green,' an orange-colored commercial soft drink was asked for as the 'Coke which is orange,' and a whole orange (i.e., fruit) was called the 'apple which is orange (color)' and the 'ball which is orange.' She also modi- fied several stock sentences (Rumbaugh 1977) so as to achieve unique communications with caretakers. For example, Lana readily modified her stock sentence, 'Please machine give coffee,' to 'You give coffee to Lana,' and 'You give this which is black,' in contexts where the well-mastered stock sentence, for various reasons, failed to net her a cup of coffee (e.g., where people, and not the machine, had the coffee).
Lana also learned to differentiate valid versus inva- lid stems or beginnings of stock sentences that were up to five lexigrams long. Given a valid stem, con- structed by the experimenter, she would complete it appropriately. For example, if the experimenter gave Lana the sentence stem, 'Please machine...,' Lana could a d d , ' . . . give C o k e , ' ' . . . give piece of banana,' and so on. Given an invalid stem, such as 'Machine please...,' or 'Please give machine of piece...,' rather than 'Please machine,' or 'Please machine give piece o f . . . , ' she would erase it rather than waste effort with it. (She erased it through use of the 'period' key, which served to clear the incorrect stem from the keyboard. She learned early on, and by herself, to use the period key to erase her own errors of production.) In this and other ways, Lana exhibited competence with the grammar of her language system, a grammar which had to be complied with if her requests were to be 'honored' by the computer that monitored her pro- ductions and that controlled the activation of the incentive-vending devices of her room.
After five years of training, Lana could produce sentences with up to 11 lexigrams in length. However, in the final analysis, Lana failed to show skills of comprehension that one would expect of her, given her impressive skills of 'production.' Simply learning associations between lexigrams and exemplars and learning rudimentary rules of grammar were not sufficient to produce a competence commensurate with a young child's language development, which reflects a far greater degree of understanding language. Consequently, an additional program of research that focused upon pragmatics and semantics or language understanding was begun.
48