Historical Linguistics Without Syn-chrony is Doomed to Di....

[Richard Janda]

----------------------------Original message----------------------------
 
[...With sincere apologies to G. E. Lessing...]
  "(Almost) no one will deny", implies Herr DeLancey, that 'there is in
principle no way to understand a lot of facts about synchronic structure
except in terms of grammaticalization'."
 
  Well, then:  I am that (almost) no one; I deny it thoroughly.  What ap-
pear to be diachronic grammaticalizational accounts of phenomena are in
fact essentially always just synchronically-based accounts involving two
or more different time periods.
 
  Further:  What is really "diachronic" (as opposed to synchronic) about
a child acquiring the VOT of voiced stops in English, say?  There have
to be voiced stops present in order for children to acquire them, but that
basically misses the point.  Similarly, English _dinner_ used to mean
'breakfast' (it's ultimately from Latin dis-jejun-are 'de-fast, break fast'),
but so what?  Paul's dictum--that you can't be scientific unless you're
diachronic--is just bad science, especially nowadays.  It seems as if
many grammaticalization theorists have leapt from the well-supported
conclusion that grammaticalization phenomena are characteristic of cer-
tain aspects of language to a more global but extremely premature and
ill-advised conclusion that the study of grammaticalization can essential-
ly replace grammar.
  Grammar doesn't always just emerge.  After it emerges, it often stays
around.  And Delancey's posting admits this:  e.g., it accepts that non-
discrete categories may continue to exist after emerging.
 
  As for myself, I seriously doubt whether any real mechanisms of lan-
guage change make reference to information or a vantage point which no
individual speaker could possess--like a perspective covering hundreds
or even thousands of years of language change.  (What linguists study
in order to collect their data is a different matter.)  If something is going
to keep a grammaticalization trend going over millennia, then it has to be
a chain of synchronic phenomena which are passed along in the manner
of a relay race.  An even better model for language transmission is a
children's "flip book" (the forerunner of motion pictures on film), in
which something that appears to move across one page is in fact really a
stack of distinct (albeit very similar) pages.
 
  It sounds so simplistic as to be platitudinous, but to deny it has conse-
quences that are fatal:
       Language change always takes place in the present, via synchronic
     mechanisms.  There are no diachronic "mechanisms"; there are only
     diachronic correspondences.  (The latter distinction has been most
     strongly emphasized by Henning Andersen.)
  But this has been said been said before, and well:  see especially Cose-
riu 1988 ("'Linguistic Change Does Not Exist'", reprinted in J. Albrecht
(ed.), _Energeia und Ergon..., Band I... [der] Schriften von Eugenio
Coseriu (1965-1987)_, pp. 147-157).
 
  There are, of course, idealizations involved here--e.g., in labeling data
collected over a period of 10 years as "synchronic"--but this clearly con-
trasts with, e.g., Greenberg's study of Aramaic suffixal -a "over a peri-
od of approximately 3,000 years" [in Traugott & Heine (eds.) 1991:
301-314].  I.e., the grammaticalization literature is astoundingly weak
on *synchronically* documented (non-reconstructional) studies where
*all* major stages are attested, even though the mechanisms involved
are admittedly going on around us all the time.  Some exceptions which
do focus on current synchrony are Romaine & Lange 1991 (in _Ameri-
can Speech_, on English _like_) and Joyce Tang Boyland's recent UC
Berkeley dissertation (which includes a focus on English would have).
But it's going to be a long time before centuries-long grammaticalization
as a sequenced complex of phenomena will have been truly established
empirically (though surely this will eventually happen).
  Hence one thing in particular bears emphasizing:  many studies which
claim to motivate grammaticalization theory in fact actually presuppose
it, because they rely crucially on unattested stages filled in using recon-
structions arrived at thanks to--you guessed it--grammaticalization theo-
ry.  "Don't disturb my circles!", indeed....
 
  The only time when anything ever happens is the present.  Think about
it:  the past is gone--or, rather, it exists only in present memory (or is
presently in the ground)--and the future isn't here yet.  Even if speakers
realize language change by pursuing functional teleologies, such goals
are synchronically present.  Synchrony is alive and well.  In the study
of change in the past, it is the only thing that has a future....
  The trouble with Paul's assumption that science must be diachronic is
that, as they say, "It's turtles all the way down".  Where do we stop?
Eventually, we reach phenomena we haven't observed, and then we
have to guess (and, at that point, our history can no longer be very
sure).  Then we realize that we're relying on *synchronic* typology as a
touchstone, anyway.
 
  To dwell on non-discrete categories is a red herring.  The reason why
non-discrete categories exist is that synchronic language-systems allow
them.  To conclude otherwise (by retreating into diachrony as the main
source for non-discreteness) is to be unnecessarily concessive to the
other side (the one that demands discreteness).  E.g., New Mexico
Spanish speakers at some point reanalyzed certain instances of the 1.pl.
verb-affix -mos as the bound-root subject clitic =nos (a degrammaticali-
zation showing an upgrading:  affix > clitic), whereas there were previ-
ously no personal subject-clitics in the dialect (though _se_ can be con-
sidered an impersonal subject-clitic).  The result was a ragged system
(with only =nos as a personal subject-clitic), but the people who inno-
vated it must obviously have thought that it was a possible organization
of human language.  And they certainly weren't just "grammaticalizing"
(cf. Janda 1995, in _CLS 31_, pp. 119-139, plus references there).
 
 This all ties in with one of Miguel Carrasquer Vidal's points in his post-
ing about Dixon's new book:  that many linguists lament the death of
undescribed languages without being willing to do anything to stop it
(through language description or language maintenance efforts).  Yes, I
accept the children's adage that, when you point one finger at someone
else, you're pointing three fingers at yourself.  But, still, what if even
just half of us refocused on the study of language change, including
grammaticalization correspondences, via a closer and longer-term look
at linguistic phenomena that are going on now?  Isn't that the real way to
understand grammaticalization?  Historical linguistics doesn't have to be
historic linguistics.
  Even Paul had something to say about variation and then-current inno-
vations in Modern German, and he himself accepted discrete categories
like inflection vs. derivation (and compounding).  In fact, his "Ueber
die Aufgaben der Wortbildungslehre" proposed a version of so-called
"blocking" (such that lexical particularities preempt general rules) in
1896 (published in 1897 as part of the _Sitzungsberichte_ ... [of the
Royal Bavarian Academy of Sciences for] 1896, pp. 692-713), though
Panini was much earlier on this one.
 
  We always have a/the present; we always have time.  And now is the
time to study grammaticalization--or perhaps also now and then.  But
not just/mainly then....  A historical linguistics that doesn't recongnize
syn-chrony is doomed to di....
 
[Some other references:
  Joseph, Brian  & Janda 1988 ("The How and Why of Diachronic Mor-
     phologization and Demorphologization", in Hammond & Noonan
     (eds.), _Theoretical Morphology_, pp. 193-210).
  Janda 1997-MS ("Beyond 'Pathways' and 'Unidirectionality':  On the
     Discontinuity of Language Transmission and the Reversibility of
     Grammaticalization").
- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -
 - - - - - - - - -
          Richard Janda