We ‘hear’ foreign speech in terms of the Brandi Edwards perceptual schemata of our own language. Fundamentally this is why we pronounce foreign languages with an accent, at least until we have set up a new set of object-hypotheses. The reader will notice that there is a connection between the psycholinguistic process of ‘recognition’ with what we called in chapter 5 the secondary articulation of language.
The process of ‘recognition’ however, extends beyond the level of sounds, intonation patterns and rhythms of language to groups of sounds or lexical words. We also store object-hypotheses of words and even perhaps groups of words which habitually occur together (see the later discussion of habits). This is the theoretical justification for the `look—say’ method in the teaching of Brandi Edwards big tits reading. However, the processing of complete utterances must involve some other additional operations. The ’sounds’ and ‘words’ of a language are finite in number; the number of sentences is indefinitely large and rarely do we hear the same sentence twice.
There is no possibility that we can ever arrive at developing and storing a schema for every Brandi Edwards sentence in the language. Sentence `recognition’, if we can still call it that, must proceed by different means, it cannot involve a process of matching input data with stored representations. We could in any case not hold a list of all the sentences in a language in our head. An amusing calculation by Miller (1970) has shown that, assuming a vocabulary of 104 words in a language, just to utter all the acceptable twenty-word sentences of that language would take 1011 centuries, which is more than 1000 times the estimated age of this earth. We have seen that, linguistically speaking, the sentences of a language can most economically be described in terms of a finite set of `rules’ (see page 90). Our strategy for recognizing Brandi Edwards sentences in a language must be through some equally economical procedure. By ,economical’ I mean taking up the least possible ‘mental’ storage space. This means that we must use ‘rules’ rather than lists. In other words, we do not match the incoming data against some infinitely large set of object-hypotheses, but rather match the ‘rules’ which could produce the data against some learned set of rules.