Íslenskt mál og almenn málfræði - 01.01.2017, Blaðsíða 154

summary
‘Three variants of /ðk/’
Keywords: Icelandic, phonological variation, devoicing before /p,t,k/, dialect boundaries,
mixed pronunciation
As described in this paper, devoicing before the stops /p,t,k/ is a common feature of
Modern Icelandic. This devoicing most typically affects the sonorants /l,m,n,r/ but to
somewhat different degrees. Thus all speakers devoice /r/ in this context and virtually all
speakers devoice /l/ before /t/ belonging to the same morpheme. This devoicing has been
more limited than elsewhere in the North-East region of Iceland. It is also well known that
similar devoicing can affect the cluster /ðk/ turning it into [þk] and this devoicing is also
regional (/ð/ does not occur before /p,t/ in Modern Icelandic so the devoicing only
applies to the cluster /ðk/). This gives rise to regional variation in the pronunciation of
words like maðkur ‘worm’, namely [maþkYr] vs. [maðkhYr]. – As with most regional
phonological (or phonetic) variants in Icelandic, there is a lot of intra-speaker variation in
this kind of devoicing, partly because many speakers are more likely to devoice some
phonemes than others before /p,t,k/ but also because some speakers may vacillate
between voiced and voiceless pronunciation of the same sonorant in a given cluster. Hence
many speakers have traditionally been classified as having „mixed voiced pronunciation“
with respect to the relevant clusters (as opposed to „purely voiced“ or „purely voiceless“
pronunciation). – In previous reports on the boundaries of the „voiced pronunciation“ of
these clusters, it has been maintained that there are relatively large areas of „mixed voiced
pronunciation“ on either side of the core area of the „voiced pronunciation“ in the North-
East. Closer scrutiny of the data shows, however, that these mixed areas are of rather dif-
ferent nature. In the data on regional variation in Icelandic phonology collected by Björn
Guðfinnsson in the 1940s, it turns out that on the Eastern side (i.e. in Eastern Iceland)
about 40% of those who were classified as having „mixed voiced pronunciation“ fell into
that category only because they pronounced words like maðkur with a voiced [ð] whereas
they did not have any examples of voiced sonorants before /p,t,k/. The remaining 60% of
those classified as having mixed voiced pronunciation were so classified because of occa-
sional occurrences of voiced sonorants before /p,t,k/. On the Western side, on the other
hand, virtually everybody classified as having mixed voiced pronunciation fell into that cat-
egory because of occurrences of voiced [ð] in words like maðkur and they had no occur-
rences of voiced sonorants before /p,t,k/. It turns out that an important reason for this
difference between the Eastern and Western areas of mixed voiced pronunciation is the
fact that there exists a third pronunciation of the word maðkur, namely [maðkYr], with a
voiced [ð] and an unaspirated [k] (which is also what we should get for a word like maðgur
in Icelandic if it existed). This is mainly found in Western and North-Western Iceland.
This pronunciation has typically been counted as „voiced“ in studies of phonological vari-
ation in Icelandic, not only by Björn Guðfinnsson and his associates in the 1940s (and
later) but also in more recent studies. The result is a relatively high percentage of „mixed
voiced pronunciation“ in Western and North-Western Iceland, despite the fact that voiced
pronunciation of sonorants before /p,t,k/ is basically nonexistent there. This is very dif-
Margrét Guðmundsdóttir154