Lexical phrases
l Every word has its own grammar. …‘knowing a word’ involves knowing its grammar --- the patterns in which it is regularly used. (Lewis)
l Language is not words and grammar; it is essentially lexical. (qtd. in Coady 235)
l Language consists of gramaticalised lexis, not lexicalized grammar. (qtd. in Coady 17)
Individual words often appear together on a regular basis. For example, native speakers of English when confronted with an economics article and see the word underdeveloped might predict that the next word will be nation or country. In other words, the choice of the next word is quite narrow. Other collocations, such as broad daylight, green with envy, and deep sigh are common in language and are often processed as single units. … Learners have to learn these multiword units as wholes.
Gass, S. and L. Selinker. 2008. Second Language Acquisition: An Introductory Course (3rd edition). Routledge. p459.
In some ways, the most uncompromising position was taken by Bolinger (1975) who proposed that language itself is much more memory-based than has been generally considered. He suggested that much of language consists of lexical elements, and that, on occasion, these may not even be easily descried by rues. He gave the example of expressions with ‘else’, like ‘somewhere else’, ‘somebody else’, but not ‘*sometime else’. Pawley and Syder (1983) similarly give the examples of ‘horror’, ‘horrid’, ‘horrify’; ‘terror’, * ‘terrid’, ‘terrify’; ‘candour’, ‘candid’, * ‘cadify’. All the examples cry out for completion of a rule-based system, and consistency, but this expectation is unfulfilled. Both Bolinger and Pawley and Sydwer use such examples to argue that a rule-based approach to language is an imposition of the linguist, and may not always be justified.
Bolinger, particularly, argued that the rule-governed basis of language itself may have been over emphasized, and that, in reality, instances of language use are much more based on lexical elements, of varying sizes, than used to be thought. … Bolinger proposed instead that much of language use is, in fact, repetitive, and not particularly creative. … most of the speech we produce is likely to have been produced before, probably by the speaker.
…
Native-like fluency
Pawley and Syder (1983) talk about the use of lexicalized sentence stems to achieve native-like fluency (one of the two ‘puzzles’ they propose for linguistic theory). As they say:
… what makes an expression a lexical item, what makes it part of the speech community’s common dictionary, is, firstly that the meaning of the expression is not (totally) predictable from its form, second, that it behaves as a minimal unit for certain syntactic purposes, and third, that it is a social institution. (ibid.: 209)
…
The average native speaker, Pawley and Syder (1983) suggest, knows hundreds of thousands of such lexicalized sentence stems, and these are then available as a repertoire of elements which may be used during ongoing conversation to achieve the degree of real-time fluency which we take for granted, and which would not be attainable otherwise. In addition, Pawley and Syder propose that the planning unit for speech is not very long. They advance the ‘one clause at a time’ hypotheses, to suggest that we only plan ahead this length of time to avoid having to engage in extensive structural planning while speaking. …
Native-like selection
Pawley and Syder … also discuss the puzzle of native-like selection, the capacity to sound idiomatic, and to say the sort of things in a second language that a native speaker of that language would say. They propose that many learners achieve native-like fluency without achieving native-like selection, in that they can produce the target language at a rate not particularly different from that produced by native speakers, but they are still not taken as native speakers (accent notwithstanding) because their choice of language makes it clear that they are operating a different system. They may, in other words, produce grammatical and fluent utterances, but still sound foreign. … to be accepted as a native speaker one has to acquire an enormous repertoire of lexicalize sentence stems; these stems have to be accessible, and an appropriate selection has to be made.
Skehan, P. 1998. A Cognitive Approach to Language Learning. Oxford University Press.（《语言学习认知法》上外社1999）pp31-9.I