Linguistic and Non-Linguistic Knowledge
Granted that a human being can have knowledge at all, it seems obvious that this knowledge can be classified in various ways. One classification would involve separating linguistic from non-linguistic knowledge. Following Chomsky, we want to say that such a classification is not only possible but correct: that it is not just imposed by the analyst, but has a basis in human mental organization. In other words, language, though only one among many cognitive systems, has its own principles and rules, which are different in kind from governing other cognitive systems, and for this reason must be studied separately.
Most linguistic theorists attempt to explain the acquisition of language in terms of whatever general learning theory they espouse. In approach, language is thus seen as acquired in the course of general intellectual development, and no language-specific endowment, apart from general intelligence and the ability to learn, will be needed for its acquisition. It is clear, though, that there is an alternative to this position. Language may be different in kind from other cognitive, different learning strategies, and different genetic programming. The two claims reinforce each other: if linguistic knowledge is different in kind from non-linguistic knowledge, then it is more likely that we need special programming to learn it; and if we have such special programming, then it is more likely that the result of language-learning will be different in kind from other systems not so programmed.
There are a number of rather obvious points that support the special programming view of language acquisition, and disconfirm the general-intellectual-ability approach. If we measure general intellectual development in terms of logical, mathematical, and abstract reasoning powers, these powers are still increasing at puberty when the ability to acquire native fluency in a language is decreasing rapidly. Particularly striking evidence for this mismatch between linguistic and general cognitive abilities comes from the case of the American girl Genie. Genie was discovered in Los Angeles in 1970, at the age of thirteen; she had been kept locked up in conditions of severe sensory deprivation from infancy. In particular, she had heard virtually no speech throughout the period in which children normally learn their first language. Despite this horrifying background, Genie’s intelligence turned out to be within normal limits in essential respects, and thus her progress with language-learning provides a useful basis for comparison with the language acquisition of more ordinary children. Her early language acquisition was typical of all children in that it passed through stages of one-word, two-word, three-word, and four-word utterances; however, Genie’s three- and four-word utterances typically displayed a cognitive complexity not found in the early speech of normal children, and her vocabulary was much richer than that of children at the same stage of syntactic development. In general, her ability to store lists of words is very good, but her ability to learn and manipulate rules has been minimal. This is reflected in the fact that whereas the ‘two-word’ stage lasts for about two to six weeks with normal children, with Genie it lasted over five months.
It seems, then, that language-learning abilities are not only different in kind from other intellectual abilities, but that they also become considerably impaired at a time when other intellectual abilities are still increasing.
It should be noted that not even all knowledge about language is to count as linguistic knowledge. The principle behind this decision is as follows: knowledge about language which is merely a special case of some wider generalization about human beings does not count as linguistic knowledge. Knowledge about language which does not emerge as a special case of some wider generalization about human beings is the only knowledge that we are prepared to call linguistic.
As an example of knowledge about language which does not count as linguistic knowledge in our sense, consider the following. Most linguists have a stock of odd items of knowledge about various languages: that most West African languages are tone languages, that Japanese has the verb in sentence-final position, that Turkish exhibits vowel harmony, and so on. This knowledge seems rather clearly to be encyclopedic, of the same type as knowledge that the capital of Ghana is Accra, and that elephants are found in East Africa. No special abilities would be required to acquire them.
Similarly, certain types of knowledge that native speakers have about their own language are not linguistic. For example, most native speakers of English can recognize the social or regional origins of others on the basis of linguistic cues such as accent, intonation, etc. They can also recognize such things as colloquial, deferential, and authoritarian styles of speech; they can tell whether a particular remark is socially or actually appropriate, and so on. Although there are certain rules and principles which make such judgments possible, we do not want to say that these judgments are evidence of linguistic knowledge, in our sense.
Strictly linguistic knowledge, then, will reduce to knowledge of those principles of sentence-construction and interpretation which do not fall together with wider generalizations about human non-linguistic behavior.
Intuitions
Native speakers of English have at their disposal a vast amount of fairly uncontroversially linguistic knowledge. For example, they know when two words rhyme; they know when two sentences are paraphrases; when a single sentence has two different meanings; when a change in word order results in a change of meaning, and when it merely results in ungrammaticality. The aim of writing a grammar is to give an account of all these facets of linguistic knowledge. How do we go about doing this?
Because, as we have already seen in the previous section, intuition lies well below the level of consciousness, direct questioning of speakers of a language is likely to yield little reliable information about their linguistic knowledge. If we approach a native speaker of English and ask him whether (1) and (2) below have the same syntactic structure, there is not the slightest chance of predicting what he will say:
1. I am leaving, for he makes me nervous
2. I am leaving, because he makes me nervous
He may have his own consciously worked out grammatical theory, or he may have no conscious idea of syntactic structure at all. In either case, there is no particular reason for believing the answer he actually gives us, and his knowledge of language will have to be investigated by rather more indirect means.
Whatever their conscious views on grammatical theory, most native speakers will be able to provide us with evidence of the following kind:
they will be able to tell us that each of the following sentence pairs has one grammatical member and one ungrammatical one:
3. (a) It was because he was nervous that he left (b) *It was for he was nervous that he left
4. (a) Because he makes me nervous, I’m leaving (b) *For he makes me nervous, I’m leaving
5. (a) Did you leave because he made you nervous? (b) *Did you leave for he made you nervous?
The native speakers of English clearly have some linguistic knowledge which enables them to distinguish for-clauses from because-clauses, in spite of their similarity in meaning. An adequate grammar must provide some way of replicating this linguistic knowledge. For our present purposes, it is not the actual rules which explain the speaker’s linguistic judgments that are of interest. What is interesting is that such judgments give us good ground for imputing a particular type of linguistic knowledge to the speaker: in this case, knowledge of syntactic structure.
What we have just been suggesting is that one good way of investigating linguistic knowledge is to ask the native speaker for judgments about the sentences of his language: not directly but indirectly, by eliciting a range of judgments about, say, grammaticality, ungrammaticality, paraphrase, and ambiguity, and then constructing a set of rules which will account for these judgments. The relevant judgments are generally called intuitions.
It is often felt, by both philosophers and linguists, that reliance on native-speakers' intuitions is an extremely suspect part of Chomskyan theory: intuitions are ‘unscientific’, not amenable to direct observation, variable, and untrustworthy. It seems to us that this is not a theoretical objection: discovering linguistic rules seems to us exactly analogous to discovering the rules of an invented, uncodified children’s game by asking the children concerned whether certain moves are permissible or not, good moves or not, dangerous or not, and so on. How else would one go about discovering unwritten rules?
This is not to say that there are not considerable practical difficulties in deciding how much reliance should be placed on native-speaker intuitions on any given occasion. We can only argue for caution in dealing with native-speaker intuitions.
Competence and Performance
English sentences may usually be assessed on two quite different levels. First, do they conform to the principles of correct-sentence formation in Standard English: are they grammatical? Second, on an actual occasion of utterance, how appropriate, felicitous, or comprehensible would they be: are they acceptable? The first level of assessment is a linguistic one; the second involves knowledge and abilities that go well beyond the purely linguistic. Within Chomskyan theory, the first is called the level of competence, the speaker’s knowledge of language, and the second is called the level of performance, the speaker’s use of language. The study of competence, then, is the study of grammars that are psychologically real, and which contain all the linguistic knowledge, whether innate or acquired, possessed by a given speaker of the language. The study of performance, by contrast, is concerned with the principles which govern language use: here such dimensions as appropriateness to context, ease of comprehension, sincerity, truth, and stylistic euphony all play a part. Performance would have to include, as competence would not, some account of the principles by which sentences are actually produced and understood – and hence occasionally misproduced or misunderstood. Competence, therefore, is the innate linguistic knowledge that a native speaker has that enables him to produce and understand utterances in his language, while performance is what the native speaker does with this knowledge, that is how he uses the language in actual situations and contexts. Like the notion of intuition, the competence-performance distinction seems to us a theoretically valid one, although like the notion of intuition, it raises certain practical difficulties.
Types of Linguistic Knowledge
We have argued that a language is a set of sentences described by the rules of a competence grammar, and that knowing a language is, essentially, knowing these rules. In fact, a complete description of linguistic knowledge must invoke a number of different types of rule, corresponding to the different types of linguistic knowledge that speakers control.
Just as the speaker’s ability to detect and correct mistakes indicates the existence of linguistic rules, so his reactions to mistakes of different kinds indicate the existence of different types of linguistic rule. Consider the following examples:
6. (a) John’s car is old
(b) ? John’s car is not a car
(c) *John’s car does be blue one
(d) *John’s rca is dlo
(e) ? John’s car is made of straw
Whereas 6(a) is fully acceptable in every respect, all of 6(b)-(e) are odd in some way. 6(b) is contradictory and makes no consistent assertion, although it is perfectly well-formed in all other respects. 6(c) is meaningful somehow but is syntactically ill-formed since it contains a sequence of words, "does be", which does not occur in Standard English. 6(d) is defective in neither of these ways but includes a word, "rca", which begins with a sequence of consonants that is impossible in English. 6(e), while admittedly peculiar, is not in the same category as any of the three preceding examples. It appears to be grammatically well-formed, and unless we stipulate arbitrarily that a defining criterion for cars is that they not be made of straw, it displays no oddity of meaning either. In this case, the apparent deviance is not linguistic at all but is due solely to knowledge of the world. Thus we can imagine situations where 6(e) would cease to be odd at all, for instance, in a fairy story, or in a community with advanced technology for hardening straws for car manufacture.
The intuitive differences among these sentences are reflected in the typical reactions to them of native speakers. 6(b) is likely to be met by a blank and baffled ‘what?’ 6(c) is likely to be corrected to (7):
7. John’s car is a blue one
6(d) is likely to evoke a request for the repetition of the offending non-word, and 6(e) will probably be greeted with incredulity.
The various examples given here indicate that we probably need to make separate statements of well-formedness reflecting knowledge of meaning (or semantics), knowledge of pronunciation (or phonology), and knowledge of what is traditionally referred to as grammar (or syntax).
Despite this separation of linguistic knowledge into different types, there is one domain in which all of them converge: the lexicon or dictionary. While a language cannot be considered simply as a set of words, but must also contain principles of sentence-formation and interpretation, it is obvious that knowledge of a language includes knowledge of its vocabulary.
CONCLUSION
In this unit, we have tried to explain some aspects of the knowledge of language. These aspects are universal and everyone who knows a language knows them, albeit unconsciously.
Leave a comment behind