Íslenskt mál og almenn málfræði - 01.01.1982, Qupperneq 306

304
Ritdómar
(i. e. it is either a geminate or a consonantal cluster). Excluded are then combinations
of a short vowel followed by a short consonant or a long vowel and a long consonant
(or a consonantal cluster). This relatively simple situation, with a couple of glitches
to be mentioned below, presents a number of descriptive and theoretical problems.
Firstly, what should be considered distinctive: vocalic length or the length of the conso-
nantal coda? Or both? Arc we dealing with a process of lengthening or of shortening?
Or both? What is the exact context where the lengthening (and/or shortening) rule
operatcs? What is the relation between a consonantal cluster and a long consonant?
The major theoretical issue is the status of the syllable, both in Icelandic phonology
and in phonological theory. Directly connected with it is the question of syllable bound-
aries ($) and the assignment of intervocalic consonants to appropriate syllables. Árna-
son takes up all of these issues in the monograph under review.
Following Benediktsson (1963), Árnason argues that it is the length of the coda that
is underlying; in this he explicitly relies quite heavily on morphophonemic and morphol-
ogical information (cf. the discussion of thc genitive -x marker on pp. 20-21). The
reasoning here is unimpeachable; it is not, however, applied consistently throughout
the monograph. While recognising the relevance of morphological considerations to
phonology here and there, Árnason falls back on autonomous phonology in a way
which is often quite annoying. Thus in his discussion of the u- and /-umlaut in Old
Icelandic (98-101), Árnason appears to define the realm of phonological regularities
as governed by the if-and-only-if relation; within such an approach the two umlauts
would have to be regarded as morphophonemic processes because the segments al-
legedly causing the umlauts sometimes fail to do so, e. g. tali ‘speech’ (dat. sg.) and
sometimes the processes take place although the segments evoking thcm are absent,
e. g. btprn ‘children’, k0mr ‘he comes’. This is, to all intents and purposcs, a taxonómic
position or its modern garb (so-called natural generative phonology) and thus, if applied
consistently, should rule out the (morphophonemic and morphological) arguments for
consonantal codas as distinctive in Modern Icelandic since the arguments have nothing
to do with the if-and-only-if relation or principles derivable from it. But if Árnason
is not simply trying to present a consistent phonological analysis but considering the
value of different ways of arguing, he should make his position and conclusions clearer.
This is not an isolated case but recurs in a number of places (p. 10-11), where mini-
mal pairs are adduced only to admit that, on morphological grounds, they are no mini-
mal pairs at all. I believe that the explicitness of Árnason’s framework helps reveal
one of its basic weaknesses, namely the reluctancc to take a firm stand; thus stronger
statements are frequently watered down in an apparent attempt not to get committed
(as in the example just mentioned: a minimal pair is given but an immediatc morphologi-
cal withdrawal follows). The reader is left high and dry as to Árnason’s exact position:
if he considers minimal pairs to be crucial to his analysis, then why bother with mor-
phological evidence contradicting it? If, on the other hand, phonology is not just the
if-and-only-if relation, then why bother with minimal pairs in the first place?.
Beforc proceeding to the length rule, I want to make a couple of riders concerning
Árnason’s system of transcription which, at times, is quite confusing. Tfiis applies in
þarticular to stop consonants which the Icelandic linguistic tradition divides into ‘hard’
/p t k/ and ‘soft’ /b d g/; they are all phonetically voiceless and differ in that the hard