Ed, your definition of T-unit is about as close as you can come.  It's an
independent clause with all its subordinate clauses and all its
nonrestrictive modifiers, regardless of punctuation. So a modifier
punctuated with a capital letter and period will be included in the T-unit
it attaches to.  There will be, here and there, a "garble," a constituent
you can't attach.  And from time to time, especially in student papers
written in class, you'll add a finite verb-if two counters agree that the
student left it out accidentally.  Usually you can tell whether a student
is writing ill-formed constituents or whether she just omitted a verb
because of the rush of time.  You see that sort of omission on emails all
the time nowadays.

Remember that Kelly was concerned with a measurable, definable unit.  He
had gone through all the research to the mid sixties and found none that
successfully defined sentence or could capture what happened syntactically
between 4th and 12th grade.  Sentences clearly get longer, but how?  And
4th graders can write surprisingly long sentences by just compounding short
clauses.  "We went to the store and Mommy bought some peanut butter and we
went home and Mommy made me a sandwich and . . . ."  Kelly needed a way to
differentiate a 12th grader's sentence length from a 4th grader's.  Earlier
researchers had identified subordination ratios: they knew that 12th
graders used more subordinate clauses.  But they were stuck at that point
because of their definition of sentence. Kelly's genius was to see that the
sentence couldn't be delimited, that he needed a grammatically definable
unit to capture the syntactic growth that took place in the 8 years between
4th and 12th grade.  Hence the T-unit (a minimal terminable unit).   The
beauty of the T-unit is that it captures the growth in clause length as
well as the subordination ratio.  In fact, a T-unit's length is clause
length x subordination ratio.  That was exactly his explanation of what
made T-units grow: as students grew older, they wrote longer clauses and
used more of them.  The major difference between 12th graders and "superior
adult" writers was that the clause length of professional writers
increased, not their subordination ratio.

Kelly came to regret that he had ever used the terms "maturity" and
"superior."  He didn't mean the terms in the way they came to be taken.
Too value laden.

I think you'll find, Ed, that T-unit length will vary greatly from text to
text,  within the same magazine, even within the writing of one individual.
It's just a way to define a certain kind of syntactic ability that develops
in the writing of students during the school years.  And it's only valuable
when you're measuring large groups.

Kelly and Francis Christensen had a running argument about how to measure
such growth (Kelly had a paragraph or two on that in, I think, his 1964
NCTE research monograph).  In essence, though, they were talking about
similar phenomena but using different terminology.  Christensen talked
about professional writers using cumulative sentences, sentences with lots
of free modifiers.  Hunt talked about professional writers using longer
clauses.  When you realize that most free modifiers (participial phrase,
absolutes, etc.), since they are phrases (that is, they have no finite
verb), will be counted in the length of the independent clause they are
attached to and will pump up clause length considerably, you realize that
Kelly's clause-length measurements would capture that use of free modifiers.

Christensen had more of a  prejudice toward narrative writing, which gives
you lots of free modifiers.  Kelly's measurements were more neutral.
Business and academic writing, with lots of nominalized phrases, will also
pump up clause length.  Of course, Christensen abhorred long noun phrases.
I don't think Kelly used them very much in his own writing (his style was
always clear and readable).  But he was aiming to do different things in
his research from what Christensen was aiming to do.

Kelly wanted to describe certain kinds of syntactic growth, statistically.
He did just that.  And he gave us useful insights into what happens to the
written syntax of students as they move through school. He was more of a
grammarian than a rhetorician. He laid the foundation for sentence
combining, because he saw the growth as being based on the principles of
early generative grammar (which he was very much influenced by).  He wrote
an article called "How Little Sentences Grow into Big Sentences." And he
gave dozens of presentations on the same topic throughout the country.

I don't know if I've answered all your questions, Ed.  Actually, I'm glad
you gave me a reason to remember a part of my life and career very dear to
me.  Kelly Hunt was a gentleman of the old school, always kind and generous
to his colleagues and students, even a graduate student who fell into
grammar accidentally from literature and took a long time learning to love
grammar and statistics.  Well, actually that graduate student never learned
to love statistics, though for a brief period of his life he could
differentiate t-tests from chi-squares.

I hope this has been useful.  Max