Re: Introducing myself to the list

On Mon, 19 Jun 2000, Nik Taylor wrote:

> "Thomas R. Wier" wrote:
> > A lot of the arguing going on in the PIE community about PIE's
> > obstruent series seems to center around whether to admit typological
> > evidence like that.
>
> I don't understand why that's controversial - if you have two possible
> proto-langs for a group of languages, shouldn't you consider the one
> that's more typologically common to be the more probable of the two?  A
> linguistic application of Occam's Razor.
An example from Shoshoni might illustrate the pitfalls of
relying on typology in reconstruction.

Gosiute Shoshoni has an interdental affricate [tT] which
corresponds to an alveolar affricate [ts] in other dialects.
Thus:

        Gosiute         Western Shoshoni

        [tTuhni]        [tsuhni]        'bone'
        [huttTi]        [huttsi]        'grandmother (FaMo)'

        etc.

Now, typologically speaking, the normal state of affairs might
be to suppose that a non-strident would become a strident (think
Ashkenazic Hebrew [Sabbos] for [SabbaT]). There aren't many
phonologists or historical linguists who would blink at that
generalization. However,  applying this typologically sound
principle leads to the conclusion that Gosiute has preserved an
original [tT] while all other dialects of Shoshoni have
innovated a [ts] from that. This is not only demonstrably
wrong (the shift has happened within the past 100 years based on
documentary evidence), but flies in the face of common sense in
linguistic reconstruction.

Now, granted we don't have anything like that clear of a
situation in PIE, typological generalizations should still be
approached with caution in reconstruction; there might be
Gosiute-like pitfalls lurking. In the absence of any kind of
evidence, speculation based on typological tendencies may carry
a bit more weight than "untethered" speculation, but it remains
speculation nonetheless.

Dirk

--
Dirk Elzinga
dirk.elzinga@m.cc.utah.edu