> If laryngeals did not affect the vowel quality of adjacent o-vocalism,
> then internal reconstruction can offer no guarantees as to where they
> were and weren't. Only where the o-grade alternates with some other
> grade can we see where there were laryngeals, based on the evidence in
> the daughter languages.
We can recover the presence of a laryngeal also from:
(a) the effect on consonants (e.g. aspiration in Indo-Iranian)
(b) the effect on prosodic patterns (e.g. Brugmann's Law in I-I)
(c) the effect on accent (e.g. in Serbian, Croat, Lithuanian etc)
(d) the outcome of voiced resonants (e.g. Greek /a/ versus /na:/ ~ /ana/)
etc.
Change in vowel quality tells us which laryngeal it is, and since this
evidence is sometimes lacking, some of our reconstructed roots have merely H
instead of h1 or h2 or h3.
Peter