West Germanic vs. the Stammbaum

[Randy Presuhn]

Hi -

> From: "Anthony Aristar" <aristar at linguistlist.org>
> To: "John Cowan" <cowan at ccil.org>
> Cc: <ietf-languages at alvestrand.no>
> Sent: Monday, February 23, 2009 10:50 AM
> Subject: Re: West Germanic vs. the Stammbaum
>
> I don't think I'd go that far.  Language trees are meant to  show 
> genetic descent, and that they do very well.

As an old Germanist with a background in linguistics and computer science....

Trees are useful data structures for many purposes, not just for
representing genetic relationships.  Don't assume that all tree
data structures with elements labeled with language names are
necessarily "language trees", particularly in the sense used by
historical linguists.

>  They don't attempt to show 
> borrowing or areal influences,

This is true of "language tree" as used by historical linguists.
This is not guaranteed for of all tree structures with elements
corresponding to languages or language groupings.

> and nor should they.

Here's the disagreement.  Any programmer will tell you that trees
can be useful for organizing information in other dimensions as well.
For purposes of language tagging, we also have to recognize that
the perceptions of language communities (which may be at odds
with historical reality) will affect how varieties are tagged.  We've
already encountered this problem (how to handle Erzgebirgisch,
for example, not to mention the Chinese mess) and will enounter
it again in the future.  Likewise, since many of the applications for
language tagging come from the bibliographic community, it's
not surprising that extent of borrowing and geographic distribution
might have gotten more weight in how things were organized than
would suit the historical linguist.

Ltru (and by extension, ietf-languages at iana.org) was chartered to
use the ISO codes.  Complaints about those codes and their organization
should go to ISO, not here.  We just work with what we're given.
Since representation of genetic relationships per se really isn't
needed to  tag language varieties, whether the ISO codes do or
don't reflect the genetics isn't a problem for us here.

Randy