There
are some practical points in connection with palatography that should be
noted. Firstly, care should be taken in selecting appropriate
words. We are often interested in comparing the places of articulation of
different sounds. Accordingly words must be chosen that contain these
articulations, and do not contain any other similar articulations that might
overlap with them. Thus when investigating the difference between s and
sh in English one should use words such as "sop-shop" rather than
"sot-shot." Similarly one should use either a range of vowels
("seep-sheep, sip-ship, same-shame, Sam-sham, sop-shop, etc.") or, if this is
not possible, just open vowels which might be expected to have less effect on
the consonant articulation. As with all instrumental phonetic
investigations, time spent selecting suitable words is a good investment.
When
doing palatography, one should allow the speaker to practice the task
extensively. It is important to get the speakers to relax after the
tongue or upper surface of the mouth have been painted, so that when they say
the word being investigated they do so naturally. It also requires
practice to stick the tongue out of the mouth the same way every time. It
is obviously important to date and label the photographs as soon as they are
taken. In addition, again as with all instrumental data, it is preferable
to make records of several different speakers saying a few utterances rather
than one or two speakers repeating a large number of different
utterances. Ideally one would like to get a dozen speakers of the same
dialect each repeating a dozen times all the contrasts to be
investigated. But making palatographic records is fairly time consuming,
and in a world in which time and effort are limited one may have to be
satisfied with half a dozen speakers saying each word once. We hope,
however that gone are the days when phoneticians such as Ladefoged made general
statements about some West African languages based on the palatographic records
of a single speaker of each language. We need to find out the properties of the
language that a group of speakers have in common, rather than the details of an
individual's pronunciation.