The Status of Information Processing Models of Language

J. Morton
1981 Philosophical Transactions of the Royal Society of London. Biological Sciences  
An introductio n is given to the n atu re of inform ation processing models in psychology. It is then claim ed th at, so far as cognitive an d linguistic processes are concerned, there are no practical constraints from biology into psychological theory. Exceptions to this principle m ay be found w here there is a precise one-one m apping o f psychological and biological units. H ow ever, the study of the effects o f b rain dam age makes it im plausible th a t such relations exist for higher m
more » ... tal functions. Finally the possibility of relations betw een psychological and linguistic descriptions is explored by exam ining the results of experim ents on the influence of word m orphology in a n um ber o f tasks. I t is concluded th a t psychology and linguistics do not constrain each other, b u t m ight exchange useful pointers to possible advances. In this p ap er an attem p t will be m ade to place psychological theory in relation to the biological sciences on the one h an d and to linguistics on the other. In the course of doing this I shall also give a p articu la r view ab o u t the n atu re of inform ation processing models. L et me start by indicating the m agnitude o f the enterprise. Figure 1 illustrates the model w ith which I currently work. T h e boxes, in general, represent either a process th a t converts inform ation from one code to an o th er or a store of inform ation of some kind. T h e directed lines indicate the m ore im p o rtan t or m ore usual passage of inform ation. T h e model appears to be a m inim um structure for talking ab o u t word recognition in two modalities, object recognition, speaking and w riting single words, together w ith some concern for larger units of language and for phenom ena of m em ory for language m aterials over short time intervals. V irtually all the com ponents of the model are supported by a variety of observations from different laboratories. T he labels on the processes have been om itted since their m ode of functioning is not self-evident. T he model becomes m ore complex w hen the details of the com ponent processes are spelled out. In addition there appears to be evidence th a t the interconnections are actually richer than those shown. Also there is evidence th a t the instructions given to a subject in an experim ent, or the strategy he adopts spontaneously, can affect the mode of function of the system. Thus, one of the routes can dom inate u nder certain conditions and another u nder other conditions. O ne exam ple of this will suffice. It concerns perform ance in the Lexical Decision task. In this task, subjects are presented w ith a sequence of strings of letters and have to decide w hether or not each one corresponds to an English word. This technique has proved useful in exploring the storage and retrieval o f our vocabulary. D avelaar et al. (1978) performed an experim ent in which the crucial measure was the time th a t it took to respond affirmatively to a word like k e r n e l which had a horrfophone (c o l o n e l ) of higher frequency. This was done under two conditions. In one of these, the non-words used, such as g r o n e , would sound like real words if they were pronounced. In the other condition the non-words, like s l in t , did not have this property. T he d a ta showed th a t in the latter condition the response times were longer for the hom ophones than for control words. This indicated th at some phonological coding was being used in the task. W hen the non-words sounded like real words, the subjects performed the task
doi:10.1098/rstb.1981.0147 fatcat:e4i46usqznfvhoj23ddrn7hsqm