Íslenskt mál og almenn málfræði - 01.01.1982, Side 315

Ritdómar
313
— one can prove anything, i. e. nothing. I am sympathetic with Árnason in his attempt
to do something with the ill-behaved clusters, but rather than force them into the
straight jacket of some preconceived notion, one should perhaps admit that almost
every regularity is accompanied by some exasperating idiosyncracies (as Árnason him-
self is aware of, cf. p. 53). We should, by all means, try to see reason behind them
but also be prepared to concede defeat. Árnason has refused to do so, but I doubt
that the particular shape of his refusal will appeal to many readers.
2.2
The chapter devoted to explanation constitutes good reding although it would be
difficult to endorse all of Árnason’s claims either with respect to a general theory of
explanation or with reference to the specific case of the Icelandic quantity shift. Again,
however, Árnason displays his position with admirable clarity and what further compels
respect is the very fact of launching the question of how a specific historical phenonte-
non can be explained. As is well known, linguists tend to be wary of asking that ques-
tion, justifiably, perhaps, in view of the limited results reached so far. Árnason pointedly
notes that ,,a question that is never asked is rather unlikely to be answered" (p. 163).
All the more praise should go to Árnason for broaching the issue again.
The discussion of the historical change in Icelandic is prefaced by an examination
of the problem of explanation both in synchronic and in diachronic linguistics. Árnason
views explanation as a higher-order description, a formulation which seems very for-
tunate. He then considers possibilities of verifying linguistic descriptions depending on
whether these descriptions claim psychological and/or social reality or whether they
make no claims to reality at all. Theories should be testablc in principle although in
actual fact this is at prescnt normally impossible since our abilites to see the ,,real“
state of affairs are largely limited. Normally, then, theories end up as being ,,more
or less plausible guesses” (p. 171). In historical linguistics, explanations cannot be re-
stricted to claims about correspondences between two synchronic stages but should
involve considerations of the context (be it structural, social or phonetic) where the
changes take place. Árnason speaks about necessary and sufficient conditions of change,
where the former specify conditions under which a change may take place, while the
latter specify the conditions where it must take place. Árnason believes that in historical
linguistics we can realistically speak only about necessary conditions since sufficient
conditions are of too high an order. The same point was eloquently made by Kiparsky
(1973:169-170) several years ago:
,,the failure of a specific change to occur in a specific language at a specific period
means nothing, since no one has been able to show conditions under which a
change, however natural, must take place. Negative historical evidence must take
the form of a universal statement to be of any value. A statement of the form:
‘changes of type X never occur’ needs a linguistic explanation, but a statement
of the form: ‘change X did not occur in language L at time T’, in our present
state of knowledge, does not“.
Turning to the Icelandic quantity shift, Árnason tries to explain the process by brea-
king down the correspondences into a few steps; thus he assumes that the stressed