The words that constitute the dictionary of a natural language are a sample of labels of categories natural to our species; they are not tags of specific objects. When names have unique referents, such as Michelangelo, Matterhorn, Waterloo, they may be incorporated into discourse but are not considered part of the lexicon.¹ Thus most words may be said to label realms of concepts rather than physical things. This must be true, for otherwise we should have great difficulty in explaining why words refer to open classes. We cannot define the category labeled house by enumerating all objects that are given that name. Any new object that satisfies certain criteria (and there is an infinity of such objects) may be assigned that label. It is easier to say what such criteria are not than to say what they are. They are not a finite set of objectively measurable variables such as physical dimensions, texture, color, acidity, etc. (except for a few words, which constitute a special case; these will be discussed under the heading ‘The Language of Experience’). We cannot predict accurately which object might be named house and which not by looking only at the physical measurements of those objects. Therefore, categorization and the possibility of word- assignment must usually be founded on something more abstract.

The abstractness underlying meanings in general, which has been the focus of so much philosophizing since antiquity, may best be understood by considering concept- formation the primary cognitive process, and naming (as well as acquiring a name) the secondary cognitive process. Concepts are superimpositions upon the physically given; they are modes of ordering or dealing with sensory data. They are not so much the product of man’s cognition; conceptualization is the cognitive process itself. Although this process is not peculiar to man (because it essentially results from the mode of operation of a mechanism that can only respond in limited ways to a wide variety of inputs), man has developed the behavioral peculiarity of attaching words to certain types of concept formation. The words (which persist through time because they may be repeated) make the underlying conceptualization process look much more static than it actually is, as we shall demonstrate presently. Cognition must be the psychological manifestation of a physiological process. It does not appear to be a mosaic of static concepts, or a storehouse of thoughts, or an archive of memorized sense-impressions. The task of cognitive organization never comes to an end and is never completed ‘ in order to be used later ’. Words are not the labels of concepts completed earlier and stored away; they are the labels of a categorization process or family of such processes. Because of the dynamic nature of the underlying process, the referents of words can so easily change, meanings can be extended, and categories are always open. Words tag the processes by which the species deals cognitively with its environment.

This theoretical position also elucidates the problem of translation or the equation of meanings across natural languages. If words label modes of cognizing, we would expect that all semantic systems would have certain formal commonalities. For instance, if we hear a given word used in connection with a given object or phenomenon, we are able to intuit the general usage of that word - it does not have to be paired with two hundred similar objects or phenomena before we can make predictions whether the name applies to a new object. Man’s cognition functions within biologically given limits. On the other hand, there is also freedom within these limits. Thus every individual may have highly idiosyncratic thoughts or conceptualize in a peculiar way or, in fact, may choose somewhat different modes of cognitive organization at different times faced with identical sensory stimuli. His vocabularly, which is much more limited and unchangeable than his capacity for conceptualizing, can be made to cover the novel conceptual processes, and other men, by virtue of having essentially the same cognitive capacities, can understand the semantics of his utterances, even though the words cover new or slightly different conceptualizations. Given this degree of freedom, it becomes reasonable to assume that natural languages always have universally understandable types of semantics, but may easily have different extensions of meanings, and that, therefore, specific semantic categories are not coterminous across languages.

It does not follow from this that differences in semantics are signs of obligatory differences in thought processes, as assumed by Whorf (1956) and many others. The modes of conceptualization that happen to be tagged by a given language need not, and apparently do not, exert restrictions upon an individual’s freedom of conceptualizing.

¹ For instance the proper name Benjamin Pfohl will ordinarily not be considered to be part of the dictionary of the English language.