Fróðskaparrit - 01.01.1964, Page 98

106
Toward the phonetic description of Faroese vowels
represented on an y*axis pointing downwards (see Fig. 2).
For the same reason F2 must be represented on an x*axis
pointing to the left, because F2 is associated with point of
articulation and ]ip*rounding in such a way that unrounded
front vowels have a high F2, and rounded back vowels
have a low F2. — The data presented below are only given
in the form of such diagrams. These are, of course, less
exact than tables but are more immediately informative,
because the relative location, i. e. the geometrical arrange*
ment of the vowels in the chart seems to be more im*
portant phonetically than the absolute formant frequencies
of the vowels. The latter may vary rather much from one
person to another, whereas the mutual arrangement of the
vowels is supposed to be fairly stable for different persons
speaking with “the same” pronunciation (it is evident that
this creates difficulties in the statistical processing of data
from different persons).1)
A further problem in formant measurements is that the
acoustic quality of vowels, especially of short vowels, is
strongly influenced by surrounding sounds, also in cases
where the ear does not notice any appreciable difference.
It seems fruitful to postulate the existence of a target posi*
tion for each vowel type, the best approximation to which
is obtained when the vowel occurs in the most neutral
surroundings (ideally in isolation) and is spoken rather
slowly (as in reading aloud).2) Vowels spoken under these
ideal conditions are, therefore, the best kind of data at the
initial stage of an acoustic investigation of formant frequ*
encies. The sample material consists partly of “idealized”
material, partly of material chosen more at random, and it
is thus not altogether in accordance with the principle just
mentioned, but in return it illustrates quite well the diffe*
rence between the two kinds of data and the range of
variation in natural (though not very fast) speech.
*) Cp. Peter Ladefoged: The Nature of Vowel Quality (Coimbra 1962).
2) A theory working with “target positions” has been proposed by
Bjórn Lindblom: On Vowel Reduction (RIT Report 29, Stockholm 1963).