Íslenskt mál og almenn málfræði - 01.01.1982, Blaðsíða 310
vowel is possible as in vitlaus 'crazy'". The second word, or actually its first element,
is given both the long and the short vowel by Einarsson (1967:489) and by Böðvarsson
(1962:866);' Árnason (p. 44) has a short vowel herc but a long diphthong in thc second
part whcre Böövarsson has a short díphthong and Einarsson a half-long one!2 Likewise
thc prcfix mis- (e. g.: mislíka 'dislike', mislitur 'vari-coloured') has a long vowel in
Einarsson's textbook but a short one in Böðvarsson's dictionary. Even worse, in Böð-
varsson's dcscription (1962:951), diphthongs ending in [i] are claimed to be somewhat
shortcncd before [j], and bogi 'bow' is transcribcd as [bjiji] while thc same word is
transcribed by Einarsson (1967:9) as [boi:jij with the commcnt that [j] has a lengthcning
Coming back to Arnason's interpretation, Iet us note that his length rule crucially
depends upon stress. The rule assigning stress is formulated as follows (p. 45):
Primary stress falls on thc first syllable of every word and a secondary stress falls
on evcry second syllablc, counting from there, except when the word is a com-
pound. If the word is a compound, then a secondary strcss falls on the first syllable
of every new constituent of the compound.
It is quitc obvious from his examples (pp. 49-50) that by stress as relevant to the length
rule, Árnason means not only primary but also secondary. Unfortunately, we are not
shown how this is supposed to work, neither for simple words nor for compounds.
The word for 'calendar' and its gen. pl. def. form arc assigncd stress as almanak and
almanakanna rcspectivcly (p. 44), hcnce one would expect the vowels bearing sccondary
strcss to be lengthened but they are not long: *alman[a:]k, *alman[a:]kann[a:]. Some-
thing more is clcarly necded here — even if one admits that ,,we are not talking in
absolute tcrms" (p. 49), but this Árnason does not discuss. Similary, if compounding
does not include derivational suffixation, then e. g. duglegur 'dúigent' should be assigned
stress and length as *diigleg[v.]r (unless some details are added to the description).
True compounds are equally problcmatic. Let us take hluttaka 'participation' and
follow Árnason's proccdures (pp. 49-50). The underlying /#hlut##taka#/ by strcss
assignmcnt, boundary wcakening and stress adjustment becomes /#hlút#táka#/. Leng-
thening produces/#hlú:t#ta:ka#/and syllabification yields/#hlu:tt$a:ka#/. Since this
form violates the output condition, the length rulc applies again deriving/#hlutta:ka#/
with the first vowel short (and preaspiration?) and the second long (at least it will
bc lengthened.) Although the pronunciation of the word may vary somewhat, I do
not think that this is what we get. Either the first vowel is long and the second short
or we get a short vowel in the first syllable followed by preaspiration and then a short
(unstressed) second vowel.
The compounds búmaður 'experienced farmer' and búkona 'good housewife' should
1 In what follows, I will, unless otherwisc stated, be relying on Árni Böðvarsson's
carcful phonetic transcription as found in thc leelandic-Russian Dictionary (Bérkov and
Böðvarsson 1962). Böðvarsson's outline of Icelandic grammar mcntioned below is ap-
pended to that dictionary.
2 The half-length of segments is undoubtedly a late-phonetic phenomenon which I
will disregard in what follows.