Paul 1880 vs. 1886 vs. 1920...; Bloomfield 1933/1965

[manaster at umich.edu]

----------------------------Original message----------------------------
 
 
On Sat, 9 May 1998, Scott DeLancey wrote:
 
[snip]
 
> And it's been convincingly suggested (by Givon, Aristar, and my
> humble self, and probably others) that many of the famous word-
> order correlations are in fact just the synchronic projection
> of diachrony.  Why do adpositions fall on the same side of their
> NP as verbs do of their object?  Why, because adpositions commonly
> originate diachronically in serialized transitive verbs.  There is
> in fact no need for any synchronic account of this correlation (or
> many of the others--and thus, inter alia, no need for much of X-
> theory) apart from the very simple diachronic one.
 
[snip]
 
Although I am always uncomfortable disagreeing with Scott, with
whom I think I see eye to eye on so many bigger issues, I am not
comfortable with this argument.  It does have some force, of course,
but it does not go nearly far enough.  Linguistic changes, if allowed
to go their merry way, would be capable of producing all kinds
of startling results.  For example, the attrition of various
word-final phonemes should have led, in French and in many other
languages, to the rise of synchronic systems in which many
grammatical categories (e.g., number or case) would be left
unexpressed in the vast majority of lexical items.  Instead,
usually either the categories themselves disappear (e.g.,
case in French nouns and indeed in most IE) or new ways
of marking them come out of the left field (like the plural
marking on articles and other prenominal words instead of
on the nouns themselves in French, the rise of adpositions
in place of vanished case marking in many languages, etc.).
Similarly, but without reference to sound change this time,
the loss of active verb forms in certain tenses or moods
in Indo-Iranian languages (and elsewhere) would seem to
lead to a system in which we would have no expresion of
active voice in those tenses or moods.  But instead what
we find is that the formerly passive forms which take over
are reinterpreted as active (and lo we have ergativity).
 
So, unless I am completely mistaken, there IS after all
some need for recognizing that certain logically possible
linguistic systems are not in fact possible or at least
are difficult to acquire or maintain, for some
psychological (for lack of a better term) reasons,
and that language changes which seem to be bound to
produce just such systems are either inevitably or
at least usually accompanied or followed in short
order by other changes which "fix" things up again.
Indeed, this seems to me to be the traditional view
of language change, going back to Saussure and other
Indo-Europeanists of that time.
 
As far as the order of adpositions and other function
words (conjunctions) is concerned, unless I am mistaken,
there are in fact well-known data indicating that in
situations where they end up in the "wrong" place
there is a strong tendency to "fix" the order.  If
memory serves, is it not the case that in Indo-Iranian
again (and in Ethiopian Semitic and elsewhere) we do
find languages which are OV but have inherited
prepositions or clause initial subordinating conjunctions
etc., and that there is a strong tendency (manifested,
predictably enough, much more in colloquial than in literary
languages) to "fix" this.  This is a sweeping generalization
of course and with many exceptions, but I think it is broadly
right.  And if so, then again we would seem to have to acknowledge
an invisible hand (alias UG or the like) which seems to
guide language change in the direction of replacing
systems that are impossible or unnatural by ones which
are possible or natural.
 
Or am I missing something?
 
AMR