Halliday & Matthiessen (1999: 446):
The view that syntax distorts semantics implies that the relation of grammar to meaning is indirect and arbitrary. This view became tenable in modern linguistics, where meaning was either excluded from its scope altogether, as among structural linguists in the U.S., or, with Chomsky, kept at a distance by the metaphor of deep and surface structure in the syntax, only the former being semantically responsible. This paved the way for a number of analyses on the model of ‘surface x is really deep y’. We find suggestions such as the following: adjectives are really verbs (e.g. Chafe, 1970), nouns are really verbs (cf. bach 1968), pronouns are really articles (Postal, 1966), negation is really a [higher] verb, tense is really a [higher] verb (cf. Huddleston, 1969), auxiliaries are really full verbs, verbal group complexes are really reductions of embedded clauses, moods are really separate clauses of saying, and so on.