19

single-dialect descriptions of English, Labov and his
followers have developed the concept of variable rules
and a methodology to collect and analyze data reflecting
them, using statistical measurements. There is an
obvious need for such studies of other languages con-
ducted by native speakers trained in linguistic field
methods. In early stages of research, however, what is
needed are structural descriptions upon which to base
later studies in greater depth. Such descriptions,
while limited in accuracy and completeness and relatively
informal in presentation, may be considered scientific
if they meet certain criteria. These are (1) use of
sound field methods for gathering and recording not only
linguistic, but also relevant social and cultural data;
(2) collection of sufficient data to insure the identi-
fication of significant features; and (3) adherence to
analytical methods that respect the structure of the
target language or dialect and that base the description
on that structure.

The last criterion implies a willingness to
experiment with different models and to select models
that best fit the data, while avoiding the temptation
to choose for description only those aspects of a
language or dialect that lend themselves to description
in terms of currently popular models, or worse, to force

the data into a distorting mold. Ideally, this approach