Rit (Vísindafélag Íslendinga) - 01.06.1970, Page 128

126
(Olcel. gleggr, glaggre), or inf. sökkva ‘to sink’, 3rd sing. sekkur
(Olcel. sekkva, sekkr).
5.2. A lucid example of the difficulties inherent in the
interrelationship of sound change and analogy is the z-umlaut
of o > 0, e.g., in inf. koma ‘to come’, 3rd sing. komr. Since the
secondary variant o did not develop before an i (as in the
3rd sing.), 0 should not have developed by i-umlaut. Since it
nevertheless did, one had to assume that the phonetic variant
[0] first acquired phonemic status (e.g., in the inf. koma),
after which it was introduced by analogy into the position
before i, and only then underwent z-umlaut as a regular sound
change. The difficulty involved appears hard to surmount.
However, if we instead interpret the f-umlaut, as done above
(§4.1), as a change of distinctive feature structure, taking
place under certain, strictly phonological conditions, then the
problem appears in a new light. Once z-umlaut has taken
place in a form like 3rd sing. tekr ‘he takes’ < *takÍR, resulting
in an umlaut alternation
[ - front] -> [ + front]
under certain morphological conditions, such as in taka, tekr,
the same regular alternation tended to become dominant also
in other cases, such as koma, taking place not as a regular
phonological change, but as a morphophonemic process, re-
placing the less regular alternation o~y.
Finally, in connection with a case like this, the question is
bound to be raised whether a morphophonemic process of
this type, operating in terms of the distinctive features in-
volved, could result in an increase in the phoneme inventory.
In the above case, the unit 0 arose also through w-umlaut of e
(as well as through combined i- and w-umlaut of a), so that
in this case the morphophonemic process cannot be said to
have been decisive. But an unambiguous case of a similar
kind of morphophonemic genesis of a new phonemic unit has
M