Rit (Vísindafélag Íslendinga) - 01.06.1970, Page 579
577
Four possibilities come to mind, all of which do not differ too
greatly from one another: (i) Divide the lexicon into two
groups, native and foreign, by means of a diacritic feature
(or more than two groups, e.g., native, Greco-Latin, English,
etc.). (2) Allow long vowels in the lexicon only for foreign
words in which the stressed syllable has no more than one
consonant after the long vowel, e.g. studére ‘to study’, býrá ‘bu-
reau’. The stress rule will then place stress first on long-vowel
foreign words, then on foreign words with consonant length in
a non-initial syllable, and finally on the first syllable of native
words. (3) Mark some words as foreign, viz., those with stressed
long vowels in non-initial syllables; mark others as native,
viz., those with consonant length in non-initial syllables
which, nevertheless, have stress on the first syllable, e.g. blýant
‘pencil’; and leave a third group unmarked, viz., those which
may proceed through the rules without difficulty. (4) Allow
stress in the lexicon only for foreign words in which the stress-
receiving syllable has no more than one consonant after the
vowel; the stress rule is then not necessary for this type of
word, but places stress on other words as in (2). (3) may be
rejected immediately. The use of [ + native] and [ + foreign]
is equivalent to a tripartite division of the lexicon by means of
two binary features. Such a solution is at least as complicated
as marking each word for stress in the lexicon, which requires
only one additional feature to be specified. (2) can be rejected
in that it interferes with the marking convention proposed
for vowels. According to this convention all the vowels of
studére will be marked as long since they all occur in open
syllables. This would erase the distinction in vowel length set
up for this word in the lexicon. Solution (1) differs little from
simply marking stress in the lexicon for each word; in either
case one feature per word is added. In order to justify this
solution other motivations must be found for a foreign—native
bifurcation of the lexicon. The behavior of e before r can be
mentioned. In most native words, e > œ before r, while in many
foreign words this is not the case (§3, rule (9)). The placement
Proceedings — 3 7