~~~~~~~~~~~~~~ Topic: " Competence and Performance

   ~~~~~~~~~~~~~~
Topic: " Competence and Performance
~~~~~~~~~~~~~~~
  *The limitations of current language processing systems are not surprising: they follow immediately from the fact that these systems are built on a competence-grammar in the Chomskyan sense. Chomsky made an emphatic distinction between the competence of a language user and the performance of this language user.
.
*The competence consists in the knowledge of language which the language user in principle has; the performance is the result of the psychological process that employs this knowledge (in producing or in interpreting language utterances). The formal grammars, that theoretical linguistics is concerned with, aim at characterising the competence of the language user. But the preferences that language users display in dealing with syntactically ambiguous sentences constitute a prototypical example of a phenomenon that in the Chomskyan view belongs to the realm of performance. 
There is ambiguity-problem from an intrinsic limitation of linguistic competence-grammars: such grammars define the sentences of a language and the corresponding structural analyses, but they do not specify a probability ordering or any other ranking between the different sentences or between the different analyses of one sentence. This limitation is even more serious when a grammar is used for processing input which frequently contains mistakes. Such a situation occurs in processing spoken language. The output of a speech recognition system is always very imperfect, because such a system often only makes guesses about the identity of its input-words. In this situation the parsing mechanism has an additional task, which it doesn’t have in dealing with correctly typed alpha-numeric input. The speech recognition module may discern several alternative word sequences in the input signal; only one of these is correct, and the parsing-module must employ its syntactic information to arrive at an optimal decision about the nature of the input. A simple yes/no judgment about the grammaticality of a word sequence is insufficient for this purpose: many word sequences are strictly speaking grammatical but very implausible; and the number of word sequences of this kind gets larger when a grammar accounts for a larger number of phenomena.
* To construct effective language processing systems, we must therefore implement performance-grammars rather than competence-grammars. These performance-grammars must not only contain information about the structural possibilities of the general language system, but also about ‘accidental’ details of the actual language use in a language community, which determine the language experiences of an individual, and thereby influence what kind of utterances this individual expects to encounter, and what structures and meanings these utterances are expected to have. 
.
*The linguistic perspective on performance involves the implicit assumption that language behaviour can be accounted for by a system that comprises a competence-grammar as an identifiable sub-component. But because of the ambiguity problem this assumption is computationally unattractive: if we would find criteria to prefer certain syntactic analyses above others, the efficiency of the whole process might benefit if these criteria were applied in an early stage, integrated with the strictly syntactic rules. This would amount to an integrated implementation of competence – and performance – notions. 
.
*But we can also go one step further, and fundamentally question the customary concept of a competence-grammar. We can try to account for language-performance without invoking an explicit competence-grammar. (This would mean that grammaticality-judgments are to be accounted for as performance phenomena which do not have a different cognitive status than other performance phenomena).