Let's understand what the debate about theories of language and teaching grammar is about.

Craig and I agree that the teaching grammar in the curriculum is being neglected.  He is trying to find a way to make it more important.

He believes, incorrectly I think, that claims that students already know grammar before they get to school is responsible for this sad set of affairs.  (People who cite claims that language is part of our biological endowment so grammar is already in place seriously misunderstand those claims and why formal knowledge of grammar is important in writing and reading.)

In his belief that claims that grammar is innate have lead to the sad state of grammar in the curriculum, Craig is looking for a theory of language which validates his view that grammar must be formally taught.  He finds such views in "cognitive grammar" and ties that into the systemic functional linguistics which is not interested in what people know about language but what are the range of grammatical choices in specific contexts.

In wanting to deny that students have to be "mentored" or "instructed" on grammar, Craig feels the need to reject any observations that EVERY native speaker knows a lot about the language and this knowledge is not the product of mentoring or instructing or whatever.

As a consequence, he is not even away that sometimes his evidence, such as it is, doesn't support what he wants it to support.

I cited the count/noncount distinction because it has never been, to my knowledge, part of the K-12 curriculum.  Despite Craig's claim that children are exposed to "radically different" language, every native speaker seems to get this distinction.

If that is the case, then there is  a problem with his view of language that what a child knows about language comes ONLY from interaction (and we know there is great variation on what that interaction) grammar needs to be more prominent in the curriculum).  Here is Craig's belief how "cognitive linguistics" can explain the count/noncount distinction.

   I think Langacker does a wonderful job handling count and non-count
(mass nouns) as cognitive categories. There are some very interesting
ways in which nouns shift from one category to another. "Yellow", for
example, is not normally a count noun, but we can say "I like the
yellow" or "I tried several yellows." "I don't like many wines."  "He
had too many beers." "I love diamond." (non-count). "She wore several
diamonds." (count.)  I suspect that these are learned as we pick up
vocabulary. There's no reason to believe the count/non-count
distinction is a purely formal system, separate from our interaction
with the world and our ways of talking about it and conceptualizing it.
Is "wood" count or mass?  "They broke through the plaster to wood." "He
tried several woods before he found one that looked right." The
question came up in my class just the other day about "trouble."  "He
got into trouble." "Nobody knows the troubles I've seen." These are
fairly dynamic categories.

All of the examples are known to all NATIVE speakers.  (I don't remember Langacker saying anything about the kinds of interaction children need to get all of these structures and Langacker identifies no speech community that does not have the full range of structures because certain kinds of interactions never take place.) Has ANYONE noticed a native speaker who is completely befuddled by "I tried several yellows"?  

The point is that all native speakers know almost all of English grammar, and they especially know about the count/noncount distinction in English nouns. NO one has to mentor them.  And, the best evidence I have for this is that dictionaries for native speakers say nothing about this distinction, nor do dictionaries consider the examples Craig has. I point out that this lack of description for native speakers means everyone seems to have the same knowledge.

If Craig thought about cognitive grammar a little harder, he would realize that this knowledge might be the result of the fact that humans all have the same cognitive capacities.  As a consequence, proposing a special language facility is not necessary.  However, he doesn't want to acknowledge that possibility because he believes that claim keeps grammar an insignificant part of the curriculum.  To my knowledge, cognitive grammar is interested in accounting for knowledge we all have about language without the need to cite some kind of special language facility.  

His commitment to systemic functional linguistics, a view of language that is only interested in characterizing the possible range of choices a language user has in a given context, means in citing Langacker and all of the variation in the count/noncount distinction, Craig is able to say: "Because language is the result of interaction and not everyone has the same kind of interaction, some speakers won't get all the variations.  It is just those speakers that need to be mentored to get those variations."

Craig's commitment to systemic functional linguistics means he is only interested in what is possible in a language.  Remember he likes to cite corpus data to account for what we know.  Because I really believe we need a competence (what is possible in a language) and performance (what we actually do) distinction, grammaticality judgments are one way to discover what are underlying competence is.  

In this discussion, it is unfair for me to use grammaticality judgments.  This is so because Craig has no theory of language that accounts for them.  Remember he claims our knowledge of language comes in interaction.  So, he writes this response which completely ignores the examples I used.

There can be a number of different reasons why certain structures may
strike us as "not possible" in the language. "Bob likes ice cream
and..." may very well come up in the right kind of context. "I like ice
cream and fudge. Sally likes ice cream and nuts. Bob likes ice cream
and...?" With the right inflection, it would be treated like a sensible
question. An utterance needs motivation, and new utterances need to pay
off before they will be accepted.  

I noted (1) and (2) are possible, but only (3) is a possible wh-question and not (4).  Craig did not explain why EVERY native speaker knows the normal wh-question is not possible for (4).  The problem that a strictly interaction account has when it comes to grammaticality judgments is that we have to notice structures we have never heard. Can anyone report what structures they have never heard and therefore they KNOW such structures are not possible in English?

1) Bob likes ice cream with what?
2) Bob likes ice cream and what?

3) What does Bob like ice cream with?
4) *What does Bob like ice cream and?

(Our judgments about 1-4 can be explained by the fact that in 1 and 3 "what" is the object of preposition and in 2 and 4 "what" is part of a coordinating construction.  However, that observation does not help a theory of language that claims there is no special knowledge that we have about language.  From that point of view, there is no such thing as a preposition or a conjunction. I have no idea how "interaction" makes it possible for a person to figure out the constraints on movement that exists for one and not the other.) 

The following statement is an empirical question.

   Cognitive and functional approaches are not naive. They give very
robust explanations for all the phenomena you bring up as "proof"
against them. 

However, to the degree that such approaches are robust explanations, those explanations are robust for EVERY native speaker.  

Bob Yates, University of Central Missouri

To join or leave this LISTSERV list, please visit the list's web interface at:
     http://listserv.muohio.edu/archives/ateg.html
and select "Join or leave the list"

Visit ATEG's web site at http://ateg.org/