Íslenskt mál og almenn málfræði - 01.01.1992, Page 98

96
Höskuldur Práinsson og Kristján Árnason
we “tested” people that BG had tested in his study about 40 years
earlier, we would sometimes ask them afterwards if they had ever been
involved in anything like this before. Most of them did not remember
that they ever had, although a few remembered BG’s visit to their
school.
Although the selection of the subjects was basically random, we
tended to visit homes that had family members of different generations.
Since we were interested in the geographical distribution of dialectal
features, we did not interview people who had recently moved into the
area. We also made a conscious effort to find people that BG had tested
in his study. Finally, we sometimes tried to “correct” the samples we
had obtained in a given district by adding more subjects to balance
the age groups, number of male vs. female subjects, socio-economic
classes, etc.
In our studies of the distribution and “strength” of the dialectal
features, we have not simply divided the speakers into three groups
as BG did (pure, mixed, absent). Rather we assign a particular value
to each type of pronunciation in each case. Suppose there are seven
instances of /zv-words in the speech sample of a particular speaker and
(s)he uses the [xv] pronunciation two times and the [khv] pronunciation
five times (cf. (3c) below). We will then compute an [xv] “grade” or
index for this speaker by multiplying the [xv] instances by two, the
[khv] instances by one and divide by seven: (2x2 + 5x1 )/7 = 1.29 — or
129 as we will usually write it (cf. Ámason & Þráinsson 1983:86; see
also Chambers & Trudgill 1980:59-64). Grades or indices for whole
groups (age groups, social groups, geographically defined groups) can
then be computed as simple means of the indices for the relevant
subjects.
Ingólfur Pálmason (EP) also interviewed subjects from all genera-
tions. He used the reading method almost exclusively and recorded
the data on tape. Although his study was confined to a rather small
geographic area (cf. Pálmason 1983), it provides a very useful com-
parison for our purposes. IP used the BG scoring system, classifying
the pronunciation of his subjects into three categories with respect to