Internal Reconstruction [was Re: Proto-Klingon?]

Padraic Brown wrote:

> On Fri, 27 Jul 2001, Thomas R. Wier wrote:
>
> >Not really.  When you only have one dialect to look at,
> >you can do internal reconstruction based on alternations
> >in that form alone.  You can do this easily with English:
> >look at allomorphs like "divine" /dIvaIn/ ~ "divinity" /dIvInIti/
> >and then, based on the data, determine what the protoform
> >must have been for that dialect.
>
> Without knowing Latin, what proto form do you get?
You know, I really should not write responses to email when
I haven't slept in nearly 24 hours.  Okay, so internal reconstruction
works by looking at variant forms within one language and then,
operating on the assumption that there was original invariance,
formulate hypotheses about which form was original. When doing
this, you have to bring in knowledge about what kinds of soundchanges
are likely to occur.  The answer here happens to be /divin/, IIRC, but
the Great Vowel Shift is rather complicated, and for purposes of
showing the validity of internal reconstruction, I didn't give a very
good example to do that.  Let's say you have an alternation like the
one in Latin between voiced and voiceless obstruents between the
NomSg and the GenSg forms:

    NSg        GenSg
    urp-s        urb-is        'city'
    re:k-s        re:g-is       'king'

There are two possibilities here.  (1) The voicless forms are original,
and they become voiced intervocallically, or (2) the voiced forms are
original, and they become devoiced by assimilation in voicing to the
adjacent segment:

    (1) *urp-s    *urp-is            (2) *urb-s    *urb-is
          *re:k-s    *re:k-is                *re:g-s    *re:g-is

Now, based on this information alone, there's no way for you to tell
which one was original.  But that's not the end of it:  you can look for
other words which don't show that alternation, and see if they follow
one or the other pattern.  So, in Latin you also have forms like

    NSg        GenSg
    wo:k-s     wo:k-is        'voice'
    stirp-s      stirp-is         'root'

These forms show that you don't have intervocalic voicing going on,
and, assuming that soundchange operates in a regular fashion,  that
you can then confidently say that the original forms must have been
like those in (2), and voicing assimilation occurred.  That, in a nutshell,
is how you do internal reconstruction.  Because by definition internal
reconstruction can show you a lot about the language's own internal
history, historical linguists always use this method first, before any
attempt to use comparative method, since shifts that happened just
to one language had to have occurred after any posited break with
a related sister tongue.

===================================
Thomas Wier | AIM: trwier

"Aspidi men Saiôn tis agalletai, hên para thamnôi
  entos amômêton kallipon ouk ethelôn;
autos d' exephugon thanatou telos: aspis ekeinê
  erretô; exautês ktêsomai ou kakiô" - Arkhilokhos

Reply