What Have the Corpus Linguists Done for Us?

[metal56]

Intuitions are fine for confirming something is correct, less so for saying something is incorrect.

If a native speaker, in person or in a language learning publication, had stated that "used to" is more common that "would" to talk about past habits, would your native speaker intuition have led you to agree with that statement?

Those learned folks at Aston say:

Get into the habit of looking at samples of natural language for insights into how language really works. Learn to mistrust ‘native speaker intuition’, and to rely more on the evidence of what people actually say and write.

http://www-users.aston.ac.uk/~roepj/pjrhome/lex.htm

I think that's good advice and teachers should be telling it to their students.

More:

"Firstly, empirical research is needed to replace the intuitive approach of most Business English practitioners. Secondly, the empirical data collected must be large enough to offer a representative picture of the language it is studying. The small size of research data and inability of the results of these works to be generalised have been noted throughout this review. In order to create a data bank of sufficient size, the use of computer-based technology and corpus linguistic methodology is necessitated. It has been noted during the review that many writers have suggested the need for this approach, for example, Cowan (1974), Pickett (1986b) and Yang (1986). It is, however, not suggested that intuition be done away with entirely, nor would it necessarily lead to a better study if it were. However, the validity of informed choices made on the interpretation of empirical data is of much greater value than when choice of language is made entirely intuitively. One further positive aspect in regard to the use of corpora and computer technology is that they can facilitate work discussed in the final dichotomy, that of research knowledge and classroom practice. "

So, I'd say that students should have a healthy mistrust of native-speaker intuition on what is incorrect as well as on what I claim is correct.

[fluffyhamster]

there's a lot of questionable English out there in Googleland.

And so there is o will be in any other corpus.
Which is why the performance based CL is questionable.

But the more you strip away the things that supposedly get in the way of or obstruct "understanding", the less language-like (is) the "product".

[fluffyhamster]

Sorry but I have to ask, metal, was the thread title in any way inspired by that line in Monty Python's The Life of Brian?

Edit: Oops, I should really be asking Dave Willis!

[metal56]

And so there is o will be in any other corpus.

Really? Would you then trust Google more than say, the British National Corpus, The American National Corpus, the London-Lund Corpus, Brown, etc?

[metal56]

But the more you strip away the things that supposedly get in the way of or obstruct "understanding", the less language-like (is) the "product".

Can you expand on that? I'd say you could strip away some teachers and the language would become more language-like.

Sorry but I have to ask, metal, was the thread title in any way inspired by that line in Monty Python's The Life of Brian?

Are your posts inspired by the "which room for an argument" sketch?

[Anuradha Chepur]

Really? Would you then trust Google more than say, the British National Corpus, The American National Corpus, the London-Lund Corpus, Brown, etc?

Perhaps not.
But I would'nt trust corpus linguistics either.

[metal56]

Perhaps not.
But I would'nt trust corpus linguistics either.

As a learner or teacher, what would you trust?

[Anuradha Chepur]

My reservations about CL are from the point of view of a linguist.
As a learner or a teacher, I would trust CL to some extent, and perhaps
common sense, general awareness of current usage, guides on modern usage.

[metal56]

My reservations about CL are from the point of view of a linguist.

So what is it about CL that you so mistrust?

[Anuradha Chepur]

I would think CL is of the arm-chair variety, it is aimed at being machine readable/ searchable.
It is finite and enumerable, whereas language is infinite.
It is a performance-based approach, and not a competence-based one.

[fluffyhamster]

I would think CL is of the arm-chair variety, it is aimed at being machine readable/ searchable.
It is finite and enumerable, whereas language is infinite.
It is a performance-based approach, and not a competence-based one.

Finite? Haven't you heard of monitor corpora? In theory, they could be expanded indefinitely (assuming computer storage and processing speeds will keep pace), and there'll never be a shortage of new material. As for that whole performance-competence dichotomy, do a search for 'dribble' (although I originally typed 'drivel'). Think (anti-)Pavlovian.

[fluffyhamster]

But the more you strip away the things that supposedly get in the way of or obstruct "understanding", the less language-like (is) the "product".

Can you expand on that? I'd say you could strip away some teachers and the language would become more language-like.

Sorry but I have to ask, metal, was the thread title in any way inspired by that line in Monty Python's The Life of Brian?

Are your posts inspired by the "which room for an argument" sketch?

Well, we all know the dead stiff dry dissected animal as opposed to bouncy fluffy living bunny analogy, right?

I'd say that if you stripped, you'd become more stripper-like. Would your students appreciate it, though?

I don't remember (maybe have never seen) that sketch, but I'd like to see it! Can you describe it for us? Ah wait, I think I've found it on Google (searched for anything containing the words 'monty python which room for an argument sketch').

[metal56]

Not sure which room you're in at the moment, Fluff, but I can direct you back to the one most of us are in.

It's here:

Hugh Deller has asserted "that much of what corpus linguists tell us can actually be retrieved simply by examining our own intuitions about language, and he seemed to be saying that if something cannot be retrieved from intuition it is probably not worth teaching anyway.·

From; What Have the Corpus Linguists Done for Us? By Dave Willis.

Do you agree with Dellar?

[metal56]

It is finite and enumerable, whereas language is infinite.

How can we tell that language is infinite? And, if it is infinite, is the infinite nature of language useful in the classroom? And, as Fluff will keep reminding us, this is the applied linguistics forum after all.

For me, the traditionalists, the prescriptivists, had education authorities, language teachers and students over a barrel for a few hundred years because they made us believe that "quality" language organisation was indeed finite. Thanks to folks such as the CLers, we now are only just beginning to have trust that it was all a big lie, a massive conspiracy.

We at least are beginning to understand that the finiteness imposed upon us by traditionalists is far less finite than we originally believed. I haven't yet seen CLers impose a new finiteness on us all.

[fluffyhamster]

I've finally gotten to read Willis' article (couldn't find let alone access it before on my mobile). One of Willis' strengths is that he rarely deals with total 'minutiae' (perhaps Deller's criticisms are aimed more at Sinclair), and goes beyond the odd "interesting" fact or two (e.g. Carter's speech on 'like' in direct reported speech - hardly unknown, and what if learners don't want to act too "reactionary" about what they or others said*. This wouldn't be the first time that the CANCODE boys could be accused of implying that native norms are the only way to fly) to give teachers outlines of possible syllabuses based on frequency and meaning criteria.

Generally I'd say that Deller's 'To hell with the boring details, just let's speak dammit' argument rather underestimates the range of items in combination(s) that learners in general might conceivably need to perform more subtle functions, especially in a second potentially more "fiddly" or "cumbersome" , seemingly grammar-encumbered language than their first. It is one thing to have a general idea of "what you want to say", and another in actually being able to achieve it (then there is the interesting phenomenon of students not actually seeming to be aware that certain functions that they take for granted in their L1 are actually possible in the L2 too, simply because they have not yet learned how to do it in the L2! It's up to us teachers to expand the horizons of the syllabus and show 'em how to do some "crazy" stuff!).

*What I'm trying to say is that 'was like' functions to break down the line between simple words and facial expression, gesture etc. But perhaps it would be hoping too much for its substituting 'said' to have much of an effect on completely robot-voiced students.