>>I'm not sure how well these speech synthesizers do with commas and
>>inflection, so I don't know if the audience could necessarily 
>>discriminate between whichs that have commas and whichs that don't, 
>>but I know for dang sure they can tell between which and that. 

>I think you're exactly right about inflection and commas.

>    "Azaleas which are acid-loving plants should be planted in the
>     shade."  (restrictive/bound)
>    "Azaleas, which are acid-loving plants, should be planted in the
>     shade."  (nonrestrictive/free)

>With the nonrestrictive, not only is the pause at the first comma obvious 
>but the voice goes up on _z_ and than comes down to the original pitch to 
>end the word.  I think this (pitch and comma pause) may actually be a 
>more reliable indicator of the nonrestrictive/restrictive distinction 
>than *that* in speech.  The problem is in writing, where the trick is to
>make punctuation reflect speech.

>Note that in both sentences there is a pause after "plants" but in only
>the second should there be a comma.   John Lawler, I believe, wrote 
>earlier that one could safely insert commas where pauses in oral language 
>occur.  In general, I disagree with that statement, assuming I've
>correctly reported his advice.  But it does seem to work for 
>nonrestrictive relative clauses.
It surely does -- the distinction between restrictive and non-restrictive relatives is fairly important, as you demonstrated clearly. And that predates English punctuation rules by quite a lot, so there must be an oral analog to the comma. Which is not a pause, incidentally -- there usually aren't any discernible pauses in connected speech (and I think I'd disagree that the sentence you cite above without commas has a pause of any sort after the second /z/ in /@ze'ly@z/) -- but rather is a peculiar intonation curve: Mid-High-Low-Mid, about what we get (an exaggerated version of) in, for example, enumeration: One, Two, Three, ....

(By the way, I tend to use the term intonation for this stuff rather than inflection, since there's already a technical linguistic term "inflection" that refers to the variety of morphology that occurs in paradigms, like the /-z/ plural inflection in /@ze'ly@z/ or the Latin -a neuter nominative/accusative plural inflection. It's hard enough trying to deal with pronunciation in ASCII without running the additional risk of confusing phonetics with grammar. :-)

In my experience as a linguist studying English, and in my experience as a teacher of English composition and of English as a Second Language, this intonation curve and the use of the comma dovetail very nicely. So nicely that I've found it productive to teach students to use it as a rule.

That's a good point, though, about speech-synthesis software. We all know how complicated English spelling and punctuation is, and how terribly they represent English pronunciation, and it's not really much of a surprise to find that commas don't always come out with the appropriate contours. That speaks to the infancy of the field, however, and to the need for improving the software, rather than to the necessity for all writers of English to avoid the use of which in restrictive relatives.

I grant you, it's the safest course. But it's not always a wise idea to restrict oneself to the safest course in every case -- only when you need to.

  - John Lawler       Linguistics Department and Residential College     University of Michigan

    "Language is the most  massive  and  inclusive  art  we know,  a           - Edward Sapir
      mountainous and anonymous work of unconscious generations."       Language (1921)

More English Grammar   More About Language   The Eclectic Company   The Chomskybot