A Psycholinguistic Model of Language Structure and Change
The overall objective of this work has been to establish a relationship between language processing and language structure. Thus far, all the attention has been focused upon identifying linguistic patterns which are predictable on psycholinguistic grounds. The weakest interpretation of this procedure would be to say that correspondences have been uncovered. For example, iconicity corresponds to parallel information processing. In a similar vein, the regularization of irregular verbs in history corresponds to the 'majority rule' in language processing. While interesting in themselves, it might be held that these correspondences do not suffice to demonstrate the relationship between process and product. What has to be specified is the exact mechanism whereby processing principles impinge upon the linguistic system. This is what Clark and Malt ( 1984) termed the linkage problem. According to them, an essential requirement of a theory is that it specify precisely how a given psychological constraint brings about a particular linguistic phenomenon. How important this requirement is can be seen from the fact that a processing principle may be a necessary though not a sufficient cause. As it alone cannot be held responsible for the existence of a given linguistic pattern, a certain indirectness between assumed cause and effect has to be acknowledged.
This indirectness is particularly apparent in attempts to relate non-psychological factors to language. Let us take two examples. In discussing the impact of sociological variables on linguistic behaviour, Dressler and Moosmüller ( 1991) make the claim that it is not enough to set up sociological parameters in a first step, linguistic variables in a next step, and finally to correlate the two. They caution that there is no direct relationship between social role or status and language use; rather, this relationship is an indirect one which is mediated by psychological factors. The authors claim that, whatever the social role or status of a speaker, what counts is not so much the role that is assigned to an individual by society but rather the speaker's internalization of this role. Clearly, only what is part of the speaker's mental representation can exercise an influence on language.
The other example comes from the analysis of colour terms. In section 2.1, Kay and McDaniel ( 1978) were cited as an attempt to explain linguistic patterns in neurological terms. As will be recalled, the evolution of colour terminology was