Page 82 - Encyclopedia of Philosophy of Language

Language and Mind
according to some maturational schedule also regu- lated by inherited material.
Another route towards a resolution involves a num- ber of linked ideas:
(a) that to characterize children's early language adequately a much reduced and simpler set of concepts is required;
(b) that children are more adept at constructing concepts than hitherto supposed;
(c) that to employ a concept explicitly, as a linguist does, is a very different thing from employing it tacitly, as a speaker does;
(d) that the contribution of genetic material to the process of acquisition is much more general, thus not encapsulated, or perhaps confined to some specific aspect of language, for instance to production and reception of speech.
The general theory of acquisition attempting this sort of resolution is known as 'semantic bootstrapping.' It has been explored by Pinker (1984).
So knowledge of children's categorization abilities, of the sorts of concept they are able to construct and of the innate resources that these abilities imply, is needed in order to make an adequate assessment of the plausibility of the program just outlined. Also, whatever theory of first language acquisition is proposed, such knowledge is needed in order to set upper limits to the possible content of children's utter- ances. The thought expressed may only partially reflect the thought that prompted expression, but it is surely absurd to propose that the former exceeds the latter in complexity of content. These two relation- ships are very programmatic and it cannot be claimed that much progress has been made with either of them. Thus far the most profitable relationship between chil- dren's categorization and first language has been the converse one; namely, that study of early language can reveal facts about early categorization (Vygotsky 1962), although here—by the same reasoning—con- clusions about the content of early language can only setlowerlimitstocategorization abilities.
The reader should perhaps be warned that there is no clear consensus amongst scholars about either cognitive or linguistic development during this period of childhood: in both fields a range of well-supported views is encountered spanning the two positions sketched above (Ingram 1989 provides a balanced review; Piattelli-Palmirini 1980 records a famous and instructive dispute). As noted above, there are even those who deny that the two fields of develop- ment are in any way connected.
2. Categorization
From a psychological perspective, categorization is involved whenever an individual treats distinct phenomena as if they were the same recurrent phenomenon. This arises in at least three different ways: (a) because the individual is biologically dis-
posed to treat the phenomena in this way—these may be called 'constitutional categories'; (b) because the phenomena form a natural cluster, isolated from other such clusters—'environmental categories'; or (c) because, arising from some purpose of the individual, it makes sense to treat the phenomena in this way— 'constructed categories.' Examples of constitutional categories might be certain regions of the color solid or certain classes of auditory event; a case has been made that some natural kinds such as lions and zebras are environmental categories (Mervis and Rosch 1981), although not all natural kinds are. It may be that artefactual kinds like/orfc or spoon provide purer cases of isolated clusters than do natural kinds. The best examples of categories that are clearly con- structed are perhaps those categories of number, quantity, relation, etc. whose development was inves- tigated by Piaget (1952). The term 'kind' is used as shorthand for what are sometimes called 'sortal cat- egories'—categories whose members are readily indi- viduated and, say, counted. Thus, dog denotes a sortal category. How many dogs are here? deserves an answer, and gets one. But How many red things are here? does not. For example, if there is a red handker- chief, should we count each thread, or molecule, etc.? The categories defined by qualities such as colors, shapes, etc. are thus clearly not kinds. This distinction is ancient, sortals corresponding to Aristotle's sub- sumtia secunda.
Mervis and Rosch's influential review (1981) pre- sented evidence that a particular level within natural kind hierarchies was psychologically privileged. Cat- egories at this level, called 'basic-level categories,' are environmental categories inasmuch as within- category similarity is maximal relative to between- category similarity at this level. They argued that this level is the point of entry to the hierarchy for children, pursuing an older insight of Brown (1958), and the level of category most easily manipulated by adults in a range of experimental tasks. For biological hier- archies, this level falls roughly at the level of the genus (e.g., tiger, as opposed to Felid, or Siberian tiger).
Whereas constitutional categorization is pre- sumably automatic, and environmental categor- ization may come about as the outcome of simple perceptual processes that detect some invariant prop- erty that distinguishes the isolated clusters with reasonable reliability or by other simple methods, con- structivecategorization ispresumedtoinvolvemental effort, at least in early stages of the categorization process. The operation of these processes by no means always progresses from the particular to the general, nor does it follow the same pathways in different com- munities. To illustrate, initial stop consonants may be allocated to numerically distinguished constructed categories of voice onset time (with the aid of suitable instruments). These categories are of course finer grained than the categories detected by unaided
60