Mark Israel <misrael@scripps.edu> writes:
Quoting me (from a post that I've lost, alas) as ">>":

>> By the way, technically, commas aren't a part of "grammar" except
>> in the very broad sense of "some kind of rules about language"
>> that one learns in the third grade.  "Grammar" generally is restricted
>> to refer to morphology and syntax, [...]
>
>   If you wrote a generative grammar for written English, would it not
>produce commas?

In the unlikely event of anyone's actually doing such a thing
(it's about as likely as enumerating the primes, which is, of
course, theoretically possible; and we even know the algorithm),
it would be even more unlikely that the putative generative
(maybe that ought to be hyphenated?) grammarian would try to
produce commas.  S/he'd be doing very well indeed to produce
periods and semicolons, which actually deal with grammatical
phenomena.  I doubt that question marks could be generated.

I don't really want to get into an argument about generative grammar,
which is largely irrelevant to English usage.  Straw man, sorry.

>   I hold that the statement "Commas are part of the syntax of written 
>English" is true, for the same reasons that the statement "Commas are 
>part of the syntax of FORTRAN" is true.  You will immediately see the 
>truth of the latter; but Saussure's (false) dogma that the only raison 
>d'etre of written language is to represent spoken language may obscure 
>the truth of the former for you.

Saussure was certainly capable of making mistakes, though I doubt that
that particular phrasing represents what he *really* said; Saussure's
major linguistic work was compiled after his death from some students'
class notes (any professor would shudder at what their students might
do in similar circumstances). In any event, of course, the priority of
spoken over written language is obvious - there are no living languages
with only written versions and no spoken, but there are plenty of vice
versa.  At the individual (i.e, "usage") level, the majority of people
in the world are illiterate, either because of lack of education or
because there is simply no tradition of literacy in their language.  And
that's the way it's always been.

Which is not to say that writing has no other function beyond
representing spoken language.  Writing, though it's modern technology
compared with speech, is still pretty old, and we've accumulated a lot
of useful tricks and traditions (I don't try to distinguish) by now.  We
clearly couldn't do without writing, and much of what we do with it we
couldn't do in speech.

So never mind Saussure or my undoubted blindness.
Let's just look at the data.

We do agree, of course, that commas are a part of the syntax of FORTRAN.
That's because FORTRAN isn't a language.  It's a scheme for instructing
computers and the fact that it's called a "language" doesn't make it
one, any more than "reading" a file is actually reading, in the same
sense we use for human reading.  Analogies are great fun, and are
sometimes persuasive, but they're weak reeds to lean on.

And FORTRAN only has syntax, but no semantics (except in the most formal
and sterile sense possible), no phonology, no morphology, etc.
English (the ostensible topic of a.u.e) has all of the above, and
plenty more.

As for commas, you may consider them part of the "grammar" of
written English if you like -- it's your language and your grammar.
I was merely pointing out the traditional distinction between
phonetics/phonology, on the one hand, and grammar proper on the other.

This distinction allows me to agree with your phrasing in the case of
written German, where commas *are* in fact determined by syntactic
considerations, but to disagree with it in the case of written English,
where they are determined by phonological considerations, and therefore
fall outside the purview of "grammar" proper.

--- Followup:

(poster is not Mark from here on)
 
>If, say, a parenthesis is marked off by commas, is that phonological
>or syntactical?

If, for instance, one looks at your sentence (or, for that matter, this
one), one "hears" a comma intonation wherever a comma has been inserted.

 
The intonation curve is (roughly) up-down-back.up, graphically something
like the following:

      /\     /\
     /If\  /say\   a parenthesis...
         \/      \/

I have found that the best writers only use commas when they want
the reader to "hear" that intonation curve.  It seems to be part
of the phenomenon of "voice" in English writing.

No doubt there are plenty of special cases, including perhaps overt
parenthesization, where commas might not be used though the intonation
curve is apparent.  On the other hand, parentheses have their own
intonation curves.   What I'm quite sure about, though, is that good
(i.e, "effective", not necessarily "correct") use of commas
overwhelmingly evokes that intonation.

Which, by the way, is not a "pause", though one may optionally
pause at a comma intonation, if one wishes to.  But it's the intonation
that matters.

--- More followup (quoting me as ">>"):

>>This distinction allows me to agree with your phrasing in the case of
>>written German, where commas *are* in fact determined by syntactic
>>considerations, but to disagree with it in the case of written
>>English, where they are determined by phonological considerations, and
>>therefore fall outside the purview of "grammar" proper.
>
>Interesting. Could you give an example of how phonology determines
>usage of commas in English? I mean a "minimal pair" of constructions,
>differing only phonologically, which differ in how commas are used.
>

Sure.  Take, first of all, appositives.  One can say them with or
without comma intonation, especially if they're short:

    My son the doctor is coming to visit.
    My son, the doctor, is coming to visit.

I think most native English speakers would pronounce these
differently (at least, the commas are intended to represent
two different pronunciations -- whether all English readers
interpret them that way is another question).  In the first,
the intonation is relatively steady on the first 4 words,
rising to the stress on DOCtor and gradually lowering to the
full-stop fall.  In the second, there's the characteristic
mid-high-low-mid contour on the words preceding the commas.

It's much the same as what one gets in enumerative clauses:

    We'll have to invite Bill, Carol, Ted, and Alice.

and is also the source of some of the confusion about "commas
before 'and'", since one hears (and wishes to have the reader hear)
that intonation curve, pace the traditional rules, so it's often
used regardless.

Oh, and to forestall potential objections about "parentheticals", note
that the two appositives above can each be used either restrictively
(i.e, "... not my son the dentist") or non-restrictively (merely to give
additional information, not to restrict from a set of possible choices).
Comma intonation is more likely on non-restrictives, but that's a
tendency and not a strict rule -- there are *very* few strict rules
regarding English intonation, since it's a continuous rather than a
discrete phenomenon.

--- later, more followup on a different thread:
 
>>> The facts of the matter are these:
>>> 
>>> 1.  "Which" with a comma introduces a non-restrictive clause.

A non-restrictive relative clause.  Very important qualification.
Gets you away from that 'that' that introduces subject and object
complement clauses, another can of worms entirely.

>>> 2.  "That" and "which" without commas introduce restrictive clauses.

Well, that's correct, but a somewhat easier way to organize it is to
focus on clause type instead of words:

   1. Non-restrictive relative clauses are introduced by a WH-class
      relative pronoun, and usually include parenthetic commas.

   2. Restrictive relative clauses are far commoner and may be introduced
      by a WH-class relative pronoun *or* the word 'that'. Commas may
      be used to set off the clause, but reflect intonation rather than sense.

The "WH-class" includes who(m), which, where, when, and sometimes why.

>>> 3. Many writers, publishers, and editors of technical or other
>>>    materials where the penalties for misinterpretation could be very
>>>    high always use "that" rather than "which" for restrictive clauses
       ...
>
>>Then there's the *other* difference...  "which" becomes "who" (or
>>"whom", when applicable, for whose who accept "whom") when used for
>>people, and occasionally for animals or things being thought of as
>>people.  "That" doesn't.
>
>I just don't think in terms of "'which' becoming 'who' (or 'whom')"
>because my brain always gets there first with the "who" or "whom"
>and "which" doesn't enter into the picture at all.
>

Right on.  There's no reason to assume one word is basic and the others
are derived from it.  As noted, it doesn't work that way in practice,
and making an assumption like that burdens you with all kinds of
pointless arbitrariness.

Not that there aren't cases of derivation -- 'whom' is clearly derived
from 'who', for instance, and so is 'whose'.  But there's paradigmatic
evidence for that.  'Which' --> 'who' is another story.  Better to keep
them in the same paradigm and use an abstract cover term for both.

--- Still more followup:

>I've read in two different stylebooks that it's all right to omit the
>comma even in a nonrestrictive context when the two words are "unusually
>close' to each other.  I have no idea what that means.  Can I say "My son
>John" and omit the comma, even if I have only one son?

The answer is in your question.

   "Can I say 'My son John' and omit the comma, ...?"

If you're talking about talking -- viz, "say" -- then commas don't come
into it, do they?   You say 'my son John' however you actually say it.
You might, if you wanted to, include a dip in your intonation (as I
indicate twice in the line above with commas) between "son" and "John",
but most native speakers wouldn't, I suspect, because the two words are
too short for a double intonation curve and because they're already both
stressed.

But if you're *not* talking about talking, but rather writing
about writing, then you use commas as one cue for your audience
that such an intonation curve is to be understood in reading
what you write.  Which means that if you *hear* such a dip
in the intonation, especially if it helps make clear what you
want to say, then *please* use a comma.  And if you *don't*
hear such a dip when you hear what you write, please *don't*
use a comma.

And if you don't hear any intonation in what you write,
neither will anybody else, and you better start thinking
either about getting better at it or taking up some other
line of work.  Or getting an editor.

  --------------------------------------------
 -John Lawler                     More grammar
  Linguistics Program   University of Michigan
 "..and, who knows? Maybe the horse will sing."