Íslenskt mál og almenn málfræði - 01.01.1982, Blaðsíða 317
the possible interpretations of length fits best the facts of Icelandic and its history.
In brief, Árnason suggests that up to around 1200 Old Icelandic can be interpreted
as having geminates underlying its (long) vowels and diphthongs; then, in then 13th
century, a change took place whereby length came to be an inherent feature.
While feasible, this account is seriously weakened, if not vitiated, by its speculative
character and lack of convincing arguments. Added to this is the general absence of
a well-articulated theoretical framework, which makes Árnason's position difficult to
evaluate. Consider the geminate analysis of length coupled with the pair-based dichotomy
of the system. Árnason argues that the two phenomena go hand-in-hand: ,,a geminate-
analysis is basically such that it fits only a pair-based dichotomy, an inherent feature
analysis is more appropriate in a set-based dichotomy" (p. 194). The Modern Icelandic
pair-based dichotomy, however, cannot be interpreted as deriving from a geminate anal-
ysis since long diphthongs are not geminated short ones (p. 214) — presumably Árnason
would like to maintain that geminates imply a pair-based dichotomy but not the other
way around. On the other hand, the Old Icelandic vowel system, with its lack of sym-
metry, would not normally qualify as a candidate for a pair-based relationship and
hence it would not be an ideal candidate for a geminate analysis. But Árnason claims
that historically, in earliest or pre-Old Icelandic, such a relationship could possibly
be established. The generalisation for Old Icelandic is then called 'passive' or 'past-ori-
ented'. Also Árnason considers as a logical implication (p. 190) that languages with
a geminating system should display a great number of diphthongs (e. g. Finnish) since
the phonotactic principles would allow not only identical vowels to stand together but
in fact any two vowels; Old Icelandic, on the other hand, had only three diphthongs
and this then hardly supports the geminating rule. This point, however, is taken to
be another aspect of the 'passivity' of the gemination (pp. 209-210). Obviously, by
proceeding in this way one can go about arguing for or against any analysis of absolutely
anything and Árnason does not seem to want to adopt any one solution.
Furthermore, the historical discussion as a whole relies crucially on structural (i. e.
autonomous) phoneme inventories; this, as noted above, creates a theroretical confu-
sion which allows its author to make pronouncements frequently smacking of the lin-
guistic catch-as-catch can. Should we take seriously Árnason's belief „that phonological
entities are largely language-specific" (p. 198 — how largely, incidentally?), then setting
up general criteria for deciding, say, when length is an inherent feature and when it
is a cluster looks Iike an exercise in futility. I must also note that Árnason's cautioning
against formal simplicity arguments in linguistics (pp. 211-212) looks strikingly bleak
in view of the use he himself makes of the notion in the synchronic part of his study.
Quantity in historical phonology is a serious attempt to apply the non-realistic (or
what Lass (1976:213 ff) calls metaphysical) position in linguistics to a large body of
synchronic and diachronic data. I find myself disagreeing with much of the framework
and its specific descriptive results. However, Árnason deserves our thanks for having
explored the possibilities of such an approach and his monograph is a valuable addition
to the mainstream of current linguistic discussion.