🚀 go-pugleaf

: John Cowan writes:
::
::Something that's always puzzled me:  how do historical linguists decide
::what form should be the standard reconstructed form?

Several reasonable answers have been given. Rich Alderson says:

:In general, appeals are made to plausibility of development, based on such
:things as known histories of languages, proposed linguistic universals, etc.

My favourite answer is the one given by Louis Hjelmslev:
When one compares the various IE languages, one finds certain
correspondences, e.g.,
  Gothic f : Irish 0 : Latin p : Gr. p : Arm. h : OI p : Tokh. p
This correspondence gets a name, for instance *p.
This name is entirely conventional, and does not imply anything
about the phonetics of a parent language.
[Not even that it had as many distinct phonemes as we find correspondences.]

Similarly, de Saussure reconstructed a "quasi-sonante" *A without
trying to define its sound.

This point of view enables one to separate the more-or-less exact
science called IE comparative linguistics from the more speculative
theories that try to assign phonetic reality to formulas like *p
(or semantic reality to reconstructed roots).