The UPenn IE Tree (the stem)

[Richard M. Alderson III]

On 15 Sep 1999, Steve Long wrote:

>We've been told that no reconstructions were used in the raw data.  So are we
>to assume that the program reconstructed not only the pre-attestation shared
>innovations that represent the nodes, but also your unrepresented (B, B')
>innovations?  Well, it better have reconstructed both, because your (B, B')
>innovations are also "unshared innovations" as far as A' is concerned.

Yes, and you have been told so, though perhaps not in so many words:  *Both*
sides of each fork represent innovations, as has been pointed out to you
several times.

>We've been told that the only chronological information that was used was the
>dates of attestation.  So how did the program determine "innovations" from
>before the date of first attestation?  And how did it determine the "relative"
>chronology of those innovations?  Did it have to go through the painful path
>that Sean Crist took in his ci/ki example?  Does this very smart program know
>that ki>ci "rarely" occurs in the world's languages?

We were also told that the chronological information was not added to the data
from which the program drew its conclusions until after a relative chronology
had already been worked out.

And as I have pointed out in a previous message, the data used by the program
are what *you* want to call reconstructions, although that is not how the word
is used by linguists.  The program was not handed a list of Sanskrit, Greek,
Latin, and Germanic words and expected to derive Grimm's Law therefrom, but was
rather given the information that Grimm's Law encodes (where Grimm's Law could
be substituted with Verner's Law, or Siever's Laws, or whichever change(s) you
like).

>We've been told that it is stemless, only reflecting relationships.  So how
>did the program know what is an innovation and what is merely is merely a
>vestige of the previous state of the language in question?  And how did it
>know which branching innovation came first?

If it is written as I would have done it, it contains a large number of
heuristics derived from 200 years of linguistic theory, such as "palatal stops
develop into (alveolo-)palatal affricates much more often than such affricates
develop into palatal stops", and presented with the data gleaned over the last
two centuries of IE studies.  It would then process the data through the rules
(heuristics) it had been given, and provide the most likely output in the form
of a tree of innovations.

Perhaps Mr. Crist knows if it was done significantly differently?

>The latest answer to this point is I believe that both lines coming out of the
>node can be considered innovating.  That's convenient, but chronologically
>absurd.  Unless both happened on the same day, the diagramm should show a
>branch off a branch, illustrating a significant innovation/divergence in the
>"proto-language" - the stem.

What is absurd is your insistence on seeing this tree not for what it is, but
for what you think it ought to have been.

The innovations in question are not single-point phenomena, nor does any branch
on the tree represent a single change:  Germanic, for example, is defined not
only by Grimm's Law and Verner's Law, but by having PIE *m rather than PIE *bh
in the plural oblique cases, and PIE *dh as a past-tense formant unlike any
other in all of IE, and PIE *e > PG. *i, PIE *o > PG. *a, PIE *a: > PG. *o:,
PIE *e: > PG. *a:, etc. ad nauseam.  *ALL* of these, *each and every one*, is a
part of the set of innovations that make Germanic different from the ten or so
other branches.

								Rich Alderson