26

The free texts provided examples of previously
unattested forms or variations which were then sought
elsewhere. Although no attempt was made to obtain the
same folktales in different places, a few occurred more
frequently than others, providing readily comparable
data complementary to that obtained through the elicita-
tion list. All free texts were tape-recorded; in most
cases translations of the elicitation list were not.

A total of approximately five hours of tape-recordings,
plus another five hours of data transcribed directly
without recording, constitutes the basic corpus of this
study. (Included in the total are approximately 30
minutes kindly recorded on my behalf in Torata, Moquegua
by Francisco Gangotena and Carlos Saavedra.) Another
five hours of tape-recorded texts were used as background
only. (Included were some recordings made in Oruro and
northern Potosi by Javier Alb6 and generously made
available to me.) Several hundred hours were spent

in reviewing original transcriptions with one or more
native speakers, whenever possible the source of

the text or another resident of the same community.

(A text recorded in a trilingual area of Potosi was
checked with a Quechua speaker from Llallagua to see
whether a certain unusual form might be a Quechua loan,
but no formal attempt was made to compare Aymara texts

with Quechua.) Some texts were later checked with