Íslenskt mál og almenn málfræði - 01.01.1982, Side 316

314
Ritdómar
syllables developed a tendency to be of the same length. This was manifested in shorten-
ing long syllables and lengthening short ones; subsequent changes of long vowels (diph-
thongisations) stabilised the pattern. Thus the higer-order description is basically ratio-
nalisation from the phonetic point of view and as such it is eminently plausible. As
Arnason himself admits, however, there is nothing that could corroborate it and ,,in
the absence of relevant data, we can only make guesses which seem more or less likely
to be true“.
There are two riders which I should like to make with reference to this sort of expla-
nation. Despite the substantial support that phonetic rationalisations have gained in
recent linguistic literature, their validity is open to serious doubts (see Hellberg 1978,
1980; Dinnsen 1980; Anderson 1981). I challenge the claim that shortening a vowel
before, say, two consonants makes sense phonetically. Any historical grammar will
produce numerous examples where consonantal clusters evoke lengthening (e. g. the
Old English cluster lengthenings, c. f. Campbell 1969:120-121; Jordan 1974:43—47).
There is nothing inherent in consonantal clusters that would require a short vowel
before them; note that in Old Icelandic, vowels could easily be long there. For a state-
ment to be valid phonetically, it would have to make language-independent (i. e. uni-
versal) claims; vowel shortening before two consonants hardly counts as one.
It might still be possible to defend this view by stressing the tendency toward quan-
titative uniformity of syllables which was realised by what might seem a „phonetic
necessity", i. e. lengthening or shortening of vocalic nuclei (the phonetic necessity might
still have taken a different shape by affecting consonantal codas for example, but we
will leave this point). This brings me to my second rider.
Typical generative formulations ‘rule X was added to the grammar of language L’
have frequently been accused of failing to address the question of the origin of the
rule, i. e. a generative formulation is said to be unable to answer the question „where
did the speakers get the rule from?“ (Andersen 1973:766). I contend that Árnason’s
explanation, dcspite its phonetic garb, can be accused of exactly the same failure: where
did the speakers get the tendency from? Surely it is not a phonetic necessity, as Old
Icelandic itself, prior to the development of the tendency, clearly documents. All of
this is not meant to decry Árnason's higher-order description of the Icelandic quantity
shift. I only wonder whether the description could indeed be termed higher-order rather
than merely different-order.
2.3
The final chapter is a discussion of phonological length and quantity with special
reference to their history in Icelandic. Árnason presents past interpretations of segmen-
tal length which has been treated as a feature of individual segments, a separate pho-
neme, as derived from higher order units (syllables or even words) or as a cluster
of identical segments (a geminate). Further, Árnason takes up Lass’s (1976:45) sug-
gestion that length dichotomises vowel systems into two subsystems which can be pair-
based or set-based. In the former, each member in the short system has a corresponding
member in the long one, as in Modern Icelandic (i. e. the system would traditionally
be termed symmetrical) while in the latter the subsystems are opposed as wholes with
some vowels remaining unpaired, as in Old Icelandic. The question arises which of