Assumptions in Computing phylogenies

[Hans Holm]

"UPenn tree"

LA>I am always skeptical about assumptions getting hidden when computers
LA>are used.

.. < they must not nessessarily get hidden: in a standard scholarly work you
do have to make things clear. The UPenn tree is not clear.
- the one side is the so called "perfect phylogeny" of prof. Warnow. The
 truth is that biological taxonomists come out with new, increasing
 complicated  "perfect phylogenies" every 14 days. A professional overview is
 "Swafford/Olsen/Waddell/Hillis: 'Phylogenetic inference'. Molecular
 Systematics, 2nd ed. 1996." - have fun;-))
- the other side is Prof. Ringe's list; but: first, this is not published,
 and - still worse - the decisions fed into the tree are not published either.
 Or did I miss one of the many articles ?

So -  /what/ can we really argue about? Every question LLoyd stated furtheron
shows that the UPenn tree is not fully documented.
----------

LA>I would be happier if we had a technique that could give results as a
LA>combination of dialect net and tree, ...

.. < There is. See e.g. Forster/Toth/Bandelt 'Evolutionary network
analysis of word lists: Visualising the relationships between Alpine
Romance languages. in: Journal of Quantitative Linguistics, Vol 5-3/1998:174-
187. And this one in fact /is fully documented/, including the word list.
---------

LA>a third one I raised recently as a question, and do not think the one
LA>response I received got me further in my understanding:

LA>(c) are results sensitive to whether a dialect in a dialect net is near
LA>the center, surrounded by closely related languages, with many nearby
LA>characteristics to compare,

..< attention! We /must/ not mix up dialect geography with genealogy!  These
are quite different issues. Two languages may be direct offsprings of a
mother language, in spite that one of them has lost e.g. ~90% of the original
features (e.g. Albanian, mainly due to Latin influence) and the other, e.g.
Greek, lost only ~ 40%.
--------------

LA>that Ringe expressed "surprise" that the results of using the technique
LA>were highly consistent with traditional scholarship.  I found that
LA>expression of surprise itself surprising,

.. > indeed. It's commonplace in informatics: "garbage in = garbage out".
    And, vice versa, of course.
---------------
Regards
Hans J. Holm
D-30629 Hannover