historical explanation of language structure

[Isidore Dyen]

----------------------------Original message----------------------------
 
I am not sure it is advisable to horn in on this discussion, but I will
anyway. It is my understanding that in science 'explanation' is tied to
the question 'How?'. With that as a premise the how of a natural language
is reasonably sought in its antecedents. Examination its state, no matter
how stated (including pseudo-histories) are only kinds of analyses. Some
scholars believe that pseudo-histories are illuminating because they
sometimes resemble historical inferences, but historical inferences are
qualitatively different from analyses, though analyses provide the basic
data for historical inferences.
 
I believe that you saw this point and I hope that my comments have helped.
 
 
On Fri, 10 Jul 1998, Robert R. Ratcliffe wrote:
 
> ----------------------------Original message----------------------------
> This relates loosely to a discussion initiated, I believe, by Scott
> Delancey, a few months ago about the extent to which diachronic
> explanations are necessary to explain synchronic structures.
>
> It strikes me that there is a fundamental contradiction between the
> de-facto use of diachronic explanations in generative theory and the
> 'orthodox' generative assumptions about language change. It is
> well-known and uncontroversial that the underlying forms of classical
> Generative Phonology are often identical with reconstructed forms. Many
> critics of GG in the 60s and 70s saw this in itself as problem.  But I'm
> willing to allow that ontology may recapitulate phylogeny as long as the
> theory of change and the theory of acquisition can be reconciled.  But
> in GG they are not.  Acquisition is assumed to involve simplification--
> the child contructs the simplest grammar possible-- while the adoption
> of reconstructed forms as underlying forms implies a theory of change as
> grammar complication-- addition of rules.
>
> I'm particularly disturbed by the way in which non-linear or 'process'
> morphology (ablaut, reduplication, etc.) is handled. Many generativists
> seem to be powerfully attracted to a theory of language as consisting
> only of words or morphemes (which must be conventional, thus
> artifactual, acquired) and syntactic relations which are presumably
> universal. But in order to sustain this view process moprhology has to
> be explained away. (Hence we have books like the recent ones by Lieber
> or Stonham showing how all process morphology can be reduced to linear
> morphology). The conventional way of dealing with an ablaut (like
> man/men) in this kind of theory is by postulating a zero affix which
> triggers a phonological change. This is a de facto historical
> explanation. It is often (not always) the case that ablaut or apophony
> develops from affixational morphology, because an affix (which may later
> be lost) triggers a phonological change in the stem. But this type of
> de-facto historical exaplanation in a synchronic account is really a
> non-explanation or an anti-explanation. If the assumption is correct
> that the simplest, most natural grammar (the one closest to UG) is one
> which involves morphemes and linear order relations only, and if the
> assumption that language change is simplification is correct, then
> ablaut type change should be impossible.  A child confronted with a pair
> like man/maener (or whatever it was in OE or Proto-Germanic, I am not a
> germanist) should simply factor out the 'noise' of the vowel change,
> identify the -er as a plural suffix and produce man/maner according to
> universal rules for combining morphemes. But in actual cases of language
> change it seems as often as not that it is the suffix which is factored
> out and the vowel contrast identified as the marker of plurality (or
> whatever semantic category), hence Eng. man/men. Thus either the theory
> that language change is simplification is incorrect, or the assumption
> that the simplest grammar is one which involves morphemes plus syntactic
> relations only is incorrect. (Of course both may be wrong.)  If the
> latter assumption is incorrect then there is no motivation to search for
> deep-structure de-facto-historical explanations of process morphology.
>
>     Thus while I am not yet ready to side with Paul, and say that there
> are no explanations of language phenomena except diachronic
> explanations, it seems to me there really hasn't been much progress in
> the direction of finding synchronic 'explanations', as opposed to
> synchronic descriptions, of language phenomena. Generative theory has
> failed to provide an adequate explanation of  language change, yet
> generative theorists contiune to rely on explanations based in language
> change to explain synchronic language structure.
>
>
>
>
> +++++++++++++++++++++++++++++++++++++++++++++++++
> Robert R. Ratcliffe
> Senior Lecturer, Arabic and Linguistics,
> Dept. of Linguistics and Information Science
> Tokyo University of Foreign Studies
> Nishigahara 4-51-21, Kita-ku
> Tokyo 114 Japan
>