Rit (Vísindafélag Íslendinga) - 01.06.1970, Blaðsíða 98

96
mutual relationships, paradigmatic and/or syntagmatic, ofthe
phonemic units directly involved, and also the relationships
between these and other units. Thus, it has been claimed, for
example, that the so-called progressive umlaut and the í-um-
laut are in every respect, except the material properties of
the vocalic elements concerned, too different from the um-
lauts proper to be classed together with them (Diderichsen
1950:75; see also, e.g., Penzl 1951:14-15, Sonderegger
1959;2-3)-
It is clear, of course, that these considerations are nothing
but a diachronic reflection of the distinction introduced by
de Saussure, in synchronic linguistic analysis, between form
and SUBSTANCE.
A particularly lucid example is furnished by certain in-
stances of vowel lengthening, compensatory or not, in Proto-
Nordic (see H. Benediktsson 1967/68:54-55). The changes in
question are traditionally formulated as follows: (1) Short
vowels in final position in monosyllables were lengthened, cf.,
e.g., ‘that’ (Goth. sa) >sd; (2) certain inter- or postvocalic
consonants were lost, and their loss resulted in the compen-
satory lengthening of a preceding short vowel, e.g. *sehan ‘to
see’ (Goth. saihwan, OHG sehan, etc.) > *séan > séa (>sjá).
However, the traditional formulation of these changes meets
with certain serious objections; in particular, there is in fact
no evidence whatsoever to sustain the common assumption
that the root vowels, for instance, of sá and séa were long im-
mediately after, and as a result of, these changes. The essence
of the change, for instance, of *sa to sá was not that the vowel
was lengthened, but rather that it ceased to be distinctively
short. The change therefore aífected not only, for example,
the masc. *sa, but also the fem. *sð: the difference in quan-
titative structure which no doubt existed between these two
in Proto-Nordic (cf. Goth. sa, so < *sa, *sö, Gk. ó, q, etc.)
disappeared; the vowel of *sö ceased to be distinctively long
(which, of course, is not synonymous with its being shortened).
Similarly, when *sehan became séa, the essence of the change