commas and restrictive clauses

[Laurence Horn]

At 11:40 AM -0700 8/4/07, Arnold M. Zwicky wrote:
>
>you thoroughly misunderstand what i was saying, or perhaps i was
>thoroughly unclear.  your scholarly qualifications (which, of course,
>i was familiar with) are entirely beside the point.  what *anyone*
>thinks
>
>>>... about who uses which variants,
>>>for how long, how often, on what occasions, and for what purposes
>
>is not at all reliable; these judgments are distorted by a collection
>of reasonably well-known unconscious biases.  this is true of
>*everyone*, including the most distinguished scholars of language.
>as i've admitted several times in public in the past few years, it's
>true of me.
>
>a scholar's impressions can serve as a source of research questions.
>but they are not evidence, and they should not be framed as factual
>observations -- even when they come from someone with long experience
>in dealing with texts and from someone who is absolutely certain that
>their impressions are accurate.  you have to do the counts,
>carefully, and when you do, you might well be surprised.  (i have
>been, on several occasions.)
>
This principle, as Arnold eloquently enunciates it, is remarkably
similar (mutatis mutandis) to that purveyed in the pages of the
(defunct? still active?) rec.sport.baseball newsgroup I used to read.
The "sabermetricians" on that list, and their counterparts in the
brick-and-mortar SABR and even in the real world, with the gradual
establishment of beachheads of such approaches into the front offices
of some (but by no means all) major league franchises, have been
maintaining for some time that the real value of the inevitably
subjective impressions (colored by "unconscious biases") of even the
best "baseball men"--scouts, reporters, contemporary and ex-players,
general managers--is vastly overrated, compared with the results
obtained from verifiably significant statistical analysis ("doing the
counts").  The impressions of "experts" still count for more in
player personnel decisions (as well as decisions in evaluating
managers and GMs, strategies and tactics for running games and
building teams, Hall of Fame and All-Star votes,...) than empirically
supportable conclusions.  This isn't to say that it's entirely
obvious which of various competing statistical models are the most
reliable or valid--in this respect the arguments among
sabermetricians have always struck me as rather too familiar for
comfort.

LH

------------------------------------------------------------
The American Dialect Society - http://www.americandialect.org