I'm glad to see that others have joined in and added to the thread!

Has taken the "pressure" off of me a bit! (Ooh, was that a pro-drop just then?).
Coming back to it then (after plenty of mince pies in between flicking through a number of books), I reckon firstly that it would be best to not talk of pro-drop at all, or at least call English more a "pro-
add" or (better yet!) "
pro-retaining/maintaining/using" language than, negatively and potentially misleadingly-confusingly, '
not - José! - a pro-drop one'. (I know that all is a bit Englo-centric or whatever the word is, but EFL students are, well, more studying English than surveying the world's languages!).
Secondly, the few teaching-related UG-based bits and pieces that I have (Vivian Cook's paper on 'Universal Grammar' in Odlin's
Perspectives on Pedagogical Grammar, and Terry Shortall's on 'What learners know and what they need to learn' in Willis & Willis'
Challenge and Change in Language Teaching) really don't appear (
appear?) to offer very much (not in the eventual way of concrete teaching suggestions, at any rate). A more informed, sympathetic and ultimately tenacious reader might get more out of such stuff than I do, however. I guess I would prefer to read Carl James'
Contrastive Analysis, which Shortall quotes from, in that it apparently anticipates (presages? LOL

) UG, P&P etc. Or one can read general typological-functional stuff with profit still, for which Mairal & Gil's
Linguistic Universals seems a good guide to thought in both UG and functional camps - see for example the footnote at the bottom of its pages 16-17.* Then, there is the likes of the interesting section 3.7 (in fact, the whole of this third chapter on syntax is worth reading!), 'Unrealized Words and Ellipsis', of Hudson's
Language Networks, though this book is unfortunately no longer previewable even in part on Google Books.

(Aside to José: I've only dipped into Jordan's book, mainly to read what Piaget and Bates in addition to our old pal Sampson might've had to say about Chomsky. I guess I'm happy sticking with Sampson's
Empirical Linguistics and
The Language Instinct Debate as my main guides to "the science in linguistics", and would now have preferred to get e.g. Radden & Dirven's
Cognitive English Grammar, if I had to keep a book from John Benjamins, than Jordan's book.

As for Chomsky’s supposed indifference to AL concerns, see page 138 of Newmeyer’s
Grammatical Theory: Its Limits and Its Possibilities, alluded to/quoted here:
http://forums.eslcafe.com/teacher/viewt ... 7139#37139 ).
Talking of ellipsis, the actual examples that Carter, Hughes & McCarthy provide of it in their CANCODE-inspired
Exploring Grammar in Context: Grammar Reference and Practice (Upper intermediate and Advanced), like those they provide for 'Tails' (which I've moaned about on several occasions), unfortunately fail to make the phenomenon they are meant to be illustrating appear that essential, that "vital".
The most immediately practical stuff that I could find then (other than the likes of Murphy, who you are obviously familiar with from having mentioned him already, Srx!) is the following:
Ur's
Grammar Practice Activities simply if somewhat hoarily (hoaryally?) suggests that groups or teams of students give hints (
It's [made of] metal; It's sharp; It's for cutting up food; or
She's quite old and/or in bad shape; She certainly can't dance; She was likened to "a dalek in drag"; etc) to help others guess what is being referred to. Obviously this sort of "repetition-based" activity is for more "real/concrete" referents/reference, and can't be used much for dummy subjects (unless we permit conjunction:
It's raining and windy and generally horrible, but that's England for you! That, or nursery rhyme-like stuff:
It's raining, it's pouring, the old man is snoring...), not that the former couldn't provide a means of leading into and better apprehending the latter.
Thornbury's
Natural English (subtitled 'The keywords of English and how they work', so it's a somewhat lexicogrammary approach that T's taking) meanwhile has a unit on 'it', with four exercises (to complement the pattern, and collocational-phrasal, info preceding 'em), the first of which would seem the most useful: 1) add n instances of 'it' back into a story in which they have all been omitted, and in which there are no gaps or gap-fills (obviously, for here the question isn't
which form to use but simply
where to use/place it! It starts thus:
Was getting dark and the road was wet. Seemed to go on forever. 'Is another fifty miles,' said Tom. 'Will we make before nightfall?' Debbie asked. 'I hate when you keep asking that,' Tom replied. ...). (The other three exercises briefly are: 2) Rewriting sentences like
To miss the start would be a shame as
It would be a shame to miss the start; 3) Filling in gaps in sentences such as
It's no _____ you are hungry: you haven't eaten all day by choosing items from a range of n offered [
funny, aim, fault, wonder, obvious, etc]; 4) Transforming two sentences into one:
I didn't break the window. Rob did. >
It was Rob who broke the window, not me. (
It wasn't me who broke the window, but Rob?)).
Sorry for the long post...must've been bored from all the Xmas lounging around!
*Which I'll quote here due to it being currently unavailable in the limited preview on Google Books: "In the P&P model, speakers have a markedness theory which guides them in their choice of options represented by parameters. One of the two possible values of each parameter represents the unmarked or neutral option, which is the initial one that is set by default. This setting only changes if language acquirers deduce from contextual input that the language that they are learning has selected the marked option. They will thus adjust their grammar to the corresponding parameter value. In other words, what the generativist model has always maintained is the idea that markedness should be understood in binary rather than scalar terms. Accordingly, categories are defined by the presence or absence of binary features. In fact, it is argued on the basis of very clear phonological and syntactic evidence, as well as the implicit or explicit belief that cognitive processing in language is binary, that all ramifications must be so as well. Consequently, the binary tradition of marked and unmarked pairs for each differentiating feature is one of the most salient features of markedness theory within Generative Grammar. This vision contrasts with that proposed by functional models in which the concept of markedness in recent years has been revised by Givon (1984, 1990, 1995) and Croft (1990), among others. More recently, the functional approach to markedness has been linked to cognitive linguistics. For this reason, Croft (1990, 59-60) no longer believes in discrete binary categories, but rather proposes continuous categories that radially cluster around prototypical elements. Consequently, he does not interpret markedness in the traditional way - nor as the generativists do - by postulating that an element has or does not have a certain feature (thus making possible the opposition or neutralization of the opposition in certain contexts), but proposes a gradual differentiation of the marked element from the unmarked one. This differentiation is established through a progressive increase in the structural complexity (of marked elements), a lower textual frequency, and a greater cognitive complexity, as pointed out by Croft himself (1990, 59-60). Once more, this differentiation highlights the tensions between the use of internal or external criteria in the formulation of interlinguistic generalizations."