Words are not labels attached to objects by conditioning; strictly associative bonds between a visual stimulus and an auditory stimulus are difficult to demonstrate in language. It is usually through the relative inhibition in the formation of such bonds that the labeling of open, abstract categories can take place, which is the hall-mark of the human semantic system. The meaning-bearing elements of language do not, generally, stand for specific objects (proper names are a special case), and strictly speaking, not even for invariant classes of objects. Apparently they stand for a cognitive process, that is, the act of categorization or the formation of concepts. Operationally, such a process may be characterized as the ability to make a similar response to different stimulus situations within given limits,¹ which rests on the individual’s capacity to recognize common denominators or similarities among ranges of physical phenomena. Even though there may be a large overlap among species in their capacity for seeing certain similarities, one also encounters species-specificities. No other creature but man seems to have just that constellation of capacities that makes naming possible such as is found in any natural language.

Natural languages differ in the particular conceptualization processes that are reflected in their vocabulary. However, since speakers use words freely to label their own conceptualization processes, the static dictionary meaning of words does not appear to restrict speakers in their cognitive activities; thus it is not appropriate to use the vocabulary meanings as the basis for an estimation of cognitive capacities.

The process of concept formation must be regulated to some extent by biological determinants; therefore, naming in all languages should have fairly similar formal properties. The basis for metaphorizing in all languages is usually transparent to everyone; it never seems totally arbitrary or unnatural. Also, we have excellent intuitions about what might be namable in a foreign language once one knows something about that community’s culture, technology, or religion. Even the demarcation lines of semantic classes are more frequently ‘ obvious ’ to a foreign speaker than a matter of complete surprise. When we learn new words in a foreign language we do not have to determine laboriously the extent of the naming class for every word we learn; there will be only a few instances where demarcations are not obvious. The incongruences between languages become marked only in certain types of grammatical classifications such as animate-inanimate, male-female, plural-singular, to’ness- from’ness, etc. But, curiously enough, on this level of abstraction, where the semantics of the form-classes often does take on an entirely arbitrary character, cognitive processes seem least affected. There is no evidence that gender in German or the declinational systems of Russian, or the noun-classification systems in the Bantu languages affect thought processes differentially.

Because of methodological obstacles, the relationship between language and cognition cannot be investigated experimentally except for a very restricted realm of the lexicon, namely the words for sensory experience. Three approaches were described that reveal (1) the dictionary meaning of sensory words: (2) the use speakers make of the dictionary meanings in a given situation; (3) the efficiency of communication in a given situation, granted complete freedom of expression. The first approach is the most faithful representation of the reference relationships that are peculiar to a given natural language, and the other two approaches are progressively less so; the third approach is the most faithful representation of the communicative freedom enjoyed by speakers of any language, whereas the other two approaches fail to reflect this aspect.

Semantic aspects of words may aid memory functions under certain special conditions, but the bias thus introduced by properties of a natural language are slight and they are minimized or eliminated by other, more potent factors (that have nothing to do with the semantic structure of a language) in most situations.

¹ I recognize that this formulation leaves the problems of synonymity and homonymity unsolved; but I am not aware of a satisfactory solution through any other formulation.