synchronic and diachronic "explanation".

[Koontz John E]

On Tue, 1 Oct 2002, Catherine Rudin/HU/AC/WSC wrote:
> At the risk of making myself unpopular on the list, I'd like to stick up
> for the idea that synchrony is NOT always just frozen diachrony.

I don't know that this is an unpopular stand.  I think everybody agrees
that syntax changes, by changes in interpretation, by addition of new
constructions, and by changes in order.  I think Bob's point was only that
one doesn't have to know and recapitulate the history of a form to produce
it.  Synchrony grows out of diachrony, but history is change as well as
inheritence.  Or putting it another way, history is as much a general
principle as any synchronic generalization.  I've always suspected that
universal principles are simply the aggregate effect of initial states,
phonological changes of several sorts, and reanalysis.

> some of John's examples of ordering exceptions in Omaha might do, and
> there are interesting cases in Romance and Slavic clitic orders etc.
> if only I could remember how they go...

I don't know how the Romance and Slavic clitic arguments go, either,
though I've heard the Romance ones cited as specific examples of
diachronic forms having been reordered by reanalysis.

I believe also that the "infixed" location of pronominals in both
Athabascan and Caddoan reflects to some extent phonological rules, to the
extent that the placement of pronominals may actually be in mid-morpheme
(and certainly is in mid-stem), in a historical sense.  So, while the
tendency to infix the pronominals may arise from diachronic considerations
such as original order, the location of infixation can be adjusted in
various ways and may be determined by subsequent principles, which, in the
case of Navajo, can be reduced to contemporary phonological principles.
I think my authority on this is a paper by Peggy Speas.  I'm not sure the
Caddoan cases are reducible to synchronic rules.  I heard a report of this
from David Rood, but he spared me the details.

In Siouan morphology or morphosyntax infixed pronominals always seem to go
nicely between morphemes, but the shape of morphemes is generally close to
the shape of canonical syllables (or sequences of them), and, of course,
there are a number of infixing stems whose constituency isn't actually
understood diachronically, so this apparent regularity may be a matter of
phonological chance and our ignorance.

However, one good class of examples of reanalysis in Siouan morphology
would be the common occurrence of pleonastic regular pronominals placed
over irregular ones, e.g., modern Omaha attaN'be, dhas^taN'be, daNba'(=i)
from earlier ttaN'be, s^taN'be, daNba=i 'I/you/(s)he saw'.  I believe
LaFlesche gives a comparable pattern of inflection for Osage paN 'to
call', showing to have been, originally, a p-stem, though Omaha baN has
been regular from the date of detailed knowledge of the language.
This tendency has reformulated whole paradigms in Winnebago and
Ioway-Otoe, and probably explains the limpid regularity of all but two
verbs in Mandan.

Another good class of examples is the migration of pronominals from
infixed position to initial position, e.g., while OP kku=...he is
infixing, I know that some languages treat the stem *hko=phe as prefixing
(I'm not sure which at the moment).  Related to this is the tendency to
reduce multiply inflected stems to singly inflected stems.  I believe the
various attested paradigms of Dakota hiyu reflect this.  Another example I
think I recall is that IO guNra is singly inflected, while OP gaN=dha 'to
want' has both roots inflected.

These tendencies aren't absolute.  I can think of probable reversals.
For example, I seem to recall that Dakota has converted the inner
instrumental naN 'by foot' into an outer instrumental, probably on the
model of the *Ra 'by heat' instrumental, which is outer.  And I'm pretty
sure that the pattern of double inflection with the Omaha-Ponca suus of
stop stems, agippaghe, dhagis^kaghe, gikkaghe ('to make one's own')
results from over generalization of the simple paradigm ppaghe, s^kaghe,
gaghe on some simpler original like *agikkaghe, *dhagikkaghe, *gikkaghe,
perhaps under the influence of the dative, which is eppaghe, dhes^kaghe,
giaghe.

I apologize for the fuzziness of some of the examples here.  I'm working
from memory.

JEK