Íslenskt mál og almenn málfræði - 01.01.1984, Blaðsíða 150
148 Kristján Árnason
mation relatively easily. I will not commit myself as to the ontological
status of that information, what sort of ‘reality’ is being described
etc. Whether for example the constructs of this model can easily be
related to psychological models of speech-performance or some mo-
dels for social structure or dynamics. The model is relatively abstract,
and it is not certain what relation it will have to observable pheno-
mena in speech perception or production by Icelandic speakers.
As to explanatory force, there are certain predictions made (or ex-
pectations triggered) by this model that would not be made by more
segmentally oriented ones. The parsing of the rhyme into a nucleus
and coda (in the specific way described above) predicts that the
sounds in hest and helst, parsed s[o[h]o r[n[£S]n c[§]c]r]s ar*d
s[o[h]o r[n[e1]n c[sSc]r]s respectively, may behave differently as a
consequence of their different phonological status. These predictions
may or may not be borne out by later development or by the results
of some phonetic measurement. (The non-fulfillment of these predic-
tions would, however (and regretfully), not disprove them, although
it would make them suspect.)
Similarly, the status of [g] in vegna [vegna] should be different from
that of the [g] in kirkna [chlfgna], since in the first instance it would
be in the nucleus and in the second it would be in the coda. There
is a tendency among many speakers of Icelandic, particularly younger
ones, to pronounce forms of stop + nasal as in vegna and Bjarni
with a glottal stop instead of an oral one, to pronounce [ve^na] and
[þja^nl] instead of [vegna] and [þja^nl]. I have not heard e.g. the
pronunciation [chlj?na] for [chljgna] (kirkna). Similarly, it can be seen
as support for a theory of syllabic phonology, distinguishing (at a
higher level in the hierarchy) between onsets and codas, that initial
/gn/ as in gnýr does not show this tendency: [?ni:r] is definitely not
heard. The process is evidently sensitive to syllabic position.
In the present model, using manner and place as two parameters,
the process of glottalization can be expressed as a ‘neutralization’ of
the parameter of place in the category ‘closed’ in this position in the
syllable. The fact that the stop is realized as glottal is evidently related
to the fact that (unaspirated) stops in this position seem quite fre-
quently to be glottalized in the speech of speakers who still retain
a distinction between three places. (Cf. footnote 2.) It could then
be said that a component of the category ‘closed’ in a nucleus would