Íslenskt mál og almenn málfræði - 01.01.1982, Page 312

310
Ritdómar
change to [f] in colloquial speech, a phenomenon which normally seems to occur across
+ , the vowel is shortened.5 Similarly, the failure of the gen. pl. -a ending to prevent
destressing of the stem final vowel in almanaka may mean that the ending is separated
from the stem by #.
The prefix á- in e. g. áskrift ‘subscription; address’ (cf. skrifa (utan) á), ástand ‘posi-
tion, situation’ (cf. standa á) is long in contradistinction to words such as askur ‘ash
tree’ or ást ‘love’ whose first segments are short. The complex forms áskrift and ástand
can be regularly derived if the prefix is separated by #; consider áskrift:
(2) á#skrift
stress assign. 1 1
destressing 0
length-rule á:
The negative prefix ó- can be handled in the same way. Consider óbjörgulegur ‘futile,
sterile’; if stress were assigned to the word as a whole, we would end up with
óbjörgidegiir which the lcngth rule would convert into the grossly ungrammatical
*óbjörg[Y:]leg[\’:]r. If the prefix is followed by #, then stress assignment will produce
a different pattern, viz. ó#björgulegur. Stress deletion and the length rule derive the
correct [ou:]björguI[e:]gur, without additional modifications.
The idea that suffixes carry boundaries is not particularly novel (Aronoff 1976, Booij
1977). It needs to be integrated with a coherent picture of the lexicon and with ruies
of word-formation, a task that at present is as remote for Icelandic as for any other
language. It seems clear, however, that we cannot study phonological phenomena intel-
ligently without some idea of both the former and the latter. Consider the pair vitlegur
‘wise’ and vitlaus ‘crazy’, based on the stem vit ‘intelligence, reason’; the former de-
rivative has a long vowel while the latter either short or long. Judging by the phonologi-
cal behaviour, we might say that the suffix -legur is preceded by # while laus by +.
This would seem to tally with the greater semantic transparency of vitlegur, which may
not be Iexicalised at all, as against the more opaque semantics of vitlaus, which may
be on its way to lexicalisation, and hence the fluctuating pronunciations.
Morphological structure of compounds and their lexicalisation seem all-important to
an adequate interpretation of length phenomena in Icelandic. Consider first the pair
dagmálaskeið ‘the time round 9 a. m.’ and námskeið ‘course (at the university etc.)’.
The noun skeið ‘period of time’ appears, on semantic grounds, to be separated by
# in the former compound (cf. dagmál ‘9 a. m.’) but by + in the latter. On this assump-
tion, the diphthong in skeið is predictably short in námskeið and predictably long in
dagmálaskeið. Furthermore, however, the short diphthong in nám- and the short vowel
in dag- become an automatic consequence of the following boundary: the cluster m+sk
and x+m inhibit lengthening. Numerous other compounds with a short vowel in dag-
5 This may be a more complex phenomenon if Árnason is right in claiming that
a short vowcl can occur in these words even when the stops do not change to fricatives
(see a review by him in Islenskt mál 2:233).— Note also that in genitives like vors
‘spring’ the vowel is always short, but there, interestingly, the r is devoiced — a process
which otherwise seems to be blocked by #.