There are whole minefields here about what actually separates “word” from “collocation”, and where grammar takes precendence over lexis.
@SavoirFaire – The collocations you mention are indeed very strong. If you’re interested, take a look at the British National Corpus or the Michigan Corpus of Academic English. The BNC is a far, far larger corpus (and there are corpora in existence an order of magnitude larger), although the BNC web interface is, really, really annoyingly, limited to 50 random hits – what are they trying to do, annoy us into ignoring it? MICASE is much smaller, but it allows much more granular search. MICASE is very useful for university-level ESL students – they can see for themselves how English is spoken in an academic context.
There are many more corpora online. The sad fact is the major ones that are used to compile dictionaries such as the OED, COBUILD or Merriam-Webster are locked away behind academic-level paywalls and a misguided sense of propriety of the data and therefore of the language itself.
Anyway, when I have to explain collocation to my students, I tell them it’s a similar phenomenon to people having friends – there are some people who are just inseparable, while others only meet on the odd occasion to do very specific things.
@gasman – an isolating language tends to have shorter words that are not composed through affixing bits to the beginning, middle and end of pre-existing words. English tends towards this paradigm. Mandarin is an archetypical example of the idea – words are often composed of only one sound/character and nothing more – context supplies the necessary information to process the code.
Japanese and Basque, on the other hand, are rather strongly synthetic (or agglutinative) languages – the “words” of the languages tend to pick up affixes that perform certain grammatical funtions (although the definition of ‘word’ is problematic).
For example, the situation where a situation or event occurs through the inaction of an agent might be expressed through “it was allowed to happen” in English. Each individual word within the sentence adds to the meaning “it happened” to provide a granular statement about the occurence – we know that it wasn’t a natural fact like the sun rising; we know that there was a strong element of human (in/anti)agency and we know that the person uttering the statement probably believes that it’s true.
Japanese, on the other hand – a synthetic/agglutinative language – would handle the same sense in an entirely different way – “okiraresareta”:
> Okiru (verb) – to happen
> Okiraseru (verb; causative) to let/make happen
> Okiraseta (verb, causative, past) – let/made happen
> Okiraresareta (verb, causative, past, passive) – was allowed/made to happen.
(I might have gotten the intricacies of the Japanese verb wrong – I’m still learning, and my wife focusses on communication over accuracy when we talk, so I could be slightly off)