Lots of interesting questions/comments to respond to.

First, I disagree strenuously with Connie's comment: "But I think the most
important thing is for school systems to assess the "grammar" needs of their own
students, based upon writing samples, and plan teaching experiences based on
that assessment." How does one define "Grammar needs"? Is it the errors students
make? Is it a lack of sophistication in their writing? How is one going to
define this? From my perspective, what Connie is asking for leads to a
"band-aide" approach to teaching grammar. It also leaves the decisions about
what is an error and what is good style in the hands of the teachers. What I am
arguing for, on the other hand, is an approach that would enable students to
understand and discuss ANY sentence in English. Once they have this ability,
students will understand what is and what is not an error (and what kind of
error it may be). They could also understand all of the most important aspects
of style -- clause length, placement, variety, verbals, parallel constructions,
etc. But if we are going to give students this, the easiest way to do so is
systematically. Unlike Jeff, I propose starting in third grade, not first, but
as I suggest in the description of the KISS curriculum, if we begin in third, it
will not take much class time each year, and it can be integrated with what
students read and write.

     Next, I agree with Bob Yates -- I have serious questions about grammars
that are built on five, seven, ten, or twenty "basic patterns." First of all, I
find the sheer number of options confusing. (And I think that English teachers
do too.) How can there be such a variety in the number of "basic patterns"?
Second, I have found every such book that I have looked at very limiting. They
focus on teaching the patterns, and not on analyzing real texts. (Please note
that I am not saying that the people who propose them shouldn't do so. Rather,
I'm saying that every such text that I have seen focusses on teaching students
some basic grammar, rather than on how to use grammar to analyze ALL texts.)
This is, by the way, one of several questions to which I will be devoting a
specific "Guest Book" on my web site. I'm hoping to set up several such books
(now that I'm learning how to make them) such that teachers' comments about them
can be collected in one place. The books should be ready within a week. (I was
going to work on them this morning, but it's 5:28 a.m., and I need some sleep.)

Last, at least for now, I'll turn to Johanna's question about human subjects.
First of all, Johanna, you are lucky to have a school to work with. Since (yes,
"since," although I could use "because" here) I no longer teach teachers, I have
to ask specific schools. They haven't been interested primarily because of the
work involved and the problem of parents' permission. I'll be interested in
knowing how that works out for you. What you really want is a sample from
everyone in the class, such that you do not miss the weaker students, i.e.,
those who might most be helped by the results of the study. (The parents of the
better writers will probably be less likely to object.) Then, of course, there
is the problem of deciding on what the writing sample will be. Primary here is
selecting a topic/prompt that does not evoke explanations of personal family
problems. (You might want to post your plans for that on this list to get
feedback before you invest the time into collecting the samples.) We know that
mode (narrative, expository) affects syntax.
     To be honest, however, I haven't tried very hard to get sample over the
last few years. I've been working out what it is that I want to look for, and
how the analysis will be done. As I'll try to show when I post the Aluminum
study, there are a lot of things to consider. That reminds me, if you can, get
permission to post the samples on the web. One of the things I am going to argue
in the Aluminum study is that almost all the previous research is suspect
because the writing samples are not available for review. There are major
questions about how the researcher defines and identifies various constructions.
In the Aluminum project, I am putting out three copies of each students'
"revision" -- the original, a copy with "idea units" identified, and a copy with
all the syntactic constructions I am counting identified. That way, even though
some people may not agree with my decisions about what is what, at least they
will know what the originals are and what I counted. Did you know that Mellon,
for example, defined clauses differently than Hunt did? As a result, their
results are not really comparable, but without the original texts, it is
impossible to determine how the difference in definition affected the results of
their studies. Good luck with your project.

   I know I said that the preceding was my last subject for today, but I just
remembered -- I wasted a fair amount of time today reading some research on
language development in pre-schoolers. If I remember correctly, there are people
on this list who have said that such research offers guidance for us. With very
few exceptions, I don't see it. Would the people who advocate studying this
stuff please explain what it offers. Don't just give us bibliographical
references. Today, for example, I learned that four and five year olds
comprehend relative clauses with inanimate direct objects before they comprehend
relative clauses with animate direct objects. By the time they are six or seven,
they comprehend both. For the life of me, I can't understand how this is
relevant to the grammar that we should be teaching in the schools.
Ed V.