Biological Function of Language
IV. NAMING AND COGNITIVE PROCESSES
(1) General Strategies
How certain can we be that naming is actually the consequence of categorization, as claimed in the introduction, instead of its cause? If there is freedom ( within limits) to categorize, could the semantic structure of a natural language restrict the biological freedom? Is our cognitive structure influenced by the reference relationships of certain words? What would cognition be like in the absence of language?
Questions of this sort may be partially answered by following either of two strategies. We may use various features of natural languages as the independent variable and study how these affect certain features of cognitive processes; or we may use the relative presence or absence of primary language as the independent variable and see to what extent the development of cognition is dependent on language acquisition. Congenitally deaf children are the most interesting subjects if the latter approach is used. The former approach harbors a few hidden difficulties that are worth mentioning here.
Superficially it may appear as if the most direct procedure in language and cognition studies were the comparison of the performance of say Navaho and English speakers on some cognitive task. If their language background is irrelevant to their test performance the two groups should have equal scores. Unfortunately, it does not follow that if their test scores are different, such differences must be due to their native language. When we compare groups of different native speakers, we usually compare at the same time individuals of different cultural background. In most instances, it is very difficult to work with two perfectly matched groups, differing in nothing but their native language. But even if these conditions were met, the logic underlying this experimental procedure would still leave much to be desired. It is a rather weak hypothesis that predicts nothing more specific than a general difference in cognitive processes due to a general difference in language background. The argument would not become convincing unless we could specify the semantic or structural peculiarity of a given language that we expect to affect one or another cognitive process.
In other words, we ought to make more specific predictions on the nature of the difference in performance between the two groups of speakers. But this requirement conjures up new difficulties. Suppose thestructural peculiarity were the presence of inflection in one language vs. its absence in the other. First, it is difficult to intuit what type of cognitive process might be affected by this structural feature. Second, we would not know whether, among the multiplicity of other structural and semantic features of either of the languages, there may not be other aspects that fully compensate, cognitively, for the structural difference noted. Thus one language may introduce redundancy by inflections while the other language has an equal degree of redundancy which is introduced through the obligatory use of function words and word order. Even if we could match speakers for their nonlinguistic variables, we cannot match language structures and peculiarities in such a way that the two languages differ in only one respect. Since language are highly integrated patterns, one can do little more than rank-order the global degree of difference between given languages( English has greater affinity with Dutch than with Arabic), but it is very tricky and usually quite meaningless, to single out specific features and compare these in disregard of the total structural complexities( cf. Lenneberg and Roberts, 1956).
These considerations should induce us to shy away from experimental designs that call for the comparison of speakers of different languages. Instead it seems more promising to follow what I have called ( 1953 ) the intralinguistic approach. Here the language variable is some peculiarity within one and the same natural language, for instance, some feature of its lexicon. Suppose a language has in its dictionary words that refer to one set of physical phenomena, say different types of snow, but not to another set of phenomena, say different types of clouds; can we make predictions about the speaker’s perception of or memory for those things? In this approach we need not match groups of speakers nor different languages. We could simply compare different responses ( to snow and clouds respectively ) of individual native speakers, so that every subject is in effect his own control. This is the basic strategy followed in most of the experiments described below.
There is, however, a further problem that must be dealt with. It is necessary that the relationship between individual words and natural phenomena can be studied empirically so that we have an objective measure of how well or how poorly the language actually deals with one or the other phenomenon. In the previous section we have discussed this matter in detail and have given reasons why the best types of words to be used in this kind of study are those that refer to sensation, in sort the language of experience.
Colors have been the favorite stimulus material because their phys-