Posted: Jul 09, 2012 9:18 am
by nunnington
Good stuff, Zwaarddjik. I find the material about grammar and lexicon very interesting, as there has obviously been a big shift in recent years, away from the slot and filler approach. Construction grammar is just one of the contemporary accounts, which are bleeding grammar and lexicon into each other.

I suppose the Chomskyan approach to syntax had a series of empty slots, into which lexical items could be inserted. However, as you say, this began to fall apart, as researchers realized that lexical items carried with them their own syntactic operations, as it were. Thus, as a simple example, we can observe that a verb like 'postpone' will tend to predict a preposition such as 'until'. In fact, you can then describe phrasal verbs and so on.

One of my memories from being at UCL was quite a lot of research into syntactic 'blends', that is, where one syntactic construction bleeds into another (I seem to have blood on the brain today), although I think this was first put forward by Bolinger. But this also militated against the 'slot' approach. An obvious example is 'different than' which presumably blends the use of comparative plus 'than' with 'different from'.

But the really interesting thing about this lexicalist turn, is that the purely formal approach of Chomsky has been abandoned. From this paradigm of extreme formalism, Chomsky made all kinds of inferences, for example, that language was not really about communication! And also, that acquiring syntax was separate from semantics and the lexicon.

Presumably, this turn to collocation grammar, or whatever you call it, will have big implications for psycholinguistics, language acquisition, language pathology, and so on. I used to do research in a stroke clinic, where patients with aphasia and the like were trained to re-establish some of their speech and language skills, so I think this new approach could have a considerable impact here. For example, purely syntactic drills seem to be the wrong way to go.