Of Trees, nodes, and minimal paths (was Re: Urheimat in Lithuania?)

[Robert Whiting]

On Wed, 29 Mar, Richard M. Alderson III <alderson at netcom.com>
wrote:

On Tue, 28 Mar 2000, Robert Whiting (whiting at cc.helsinki.fi) wrote:

<snip>

>> Our tools aren't that good yet that we can afford to ignore
>> potential improvements.

>Agreed.

>The issue I have is with the terminology "innovating" vs.
>"non-innovating": *This* is what I think is misleading to a
>number of non-linguists, who do not see "non-innovating" as
>equivalent to "not innovating in the same way".  *I'm* not
>altogether comfortable with it in that meaning myself, and I
>understand what is meant by it.

Agreed again.  But innovating/non-innovating is a by-product of
the tree model, and I expect that most of us are not altogether
comfortable with the tree model either.  The tree model has
severe limitations and it is important to be aware of what these
limitations are.  Perhaps all trees should be required to have
a warning label something like "WARNING:  This tree does not
reflect reality except in certain narrow areas.  Do not try to
apply this tree to real life situations."  or "WARNING:  This
tree is an abstraction based on limited data.  Prolonged use
without constant reference to the data may be hazardous to your
mental health."

Even so, the tree model is still useful for certain things so it
can't really be dispensed with.  And anything that provides a
better model of the features that the tree doesn't, will probably
distort the features that are made clear by the tree, as well as
running the risk of being too complex to be comprehensible (e.g.,
isogloss maps or dialect geography).

Bob Whiting
whiting at cc.helsinki.fi