Halliday's weather/climate analogy is drawing on the notion of time
depth. From a short time frame we can only observe the weather; but from
a longer time frame we can generalise weather into climate - and use
climate then as a way of predicting weather - probabilistically of
course. With language, he's saying that with a short time frame, what we
see of language are texts, but that given a longer time frame we can
generalise these into a system which predicts what might come next. In
these terms, langue is a kind of phylogenetic record - the store of
meanings that are imminent and relevant to the yet to be spoken. Parole - A
text
is the new meaning we make in relation to what has been - the weather we
experience as a result of our semiotic climate. And of course, every
experience of weather changes the climate - maybe only in small ways -
but the whole process of instatiating langue in parole is dynamic,
dialectical... . What I was querying was your association of the
cognitive with language and the social with parole, which I don't think
is wise. Or at least, we've had more than enough of it.
Jim Martin