Turkish sign language: elicited vs. natural data collection

[Asli Ozyurek]

Dear members of the sign language list,
 Thanks for your warm welcome to the list, and sorry for bothering you
with my  earlier   "testing" message.   I  am a psychologist/linguist
who  has been  working on Turkish Sign Language for two years .   With
my students at the Koc University, Istanbul   we have been collecting
elicited data from Turkish adult signers in several deaf clubs in
Istanbul.  Our aim is to first give a basic  description of how space is
used for arguement structure In Turkish Sign Language. We  have
probably the first  recordings  of TSL collected for scientific purpose
and there has not been any study conducted on the structure of this
language before.  There is no education of the TSL for  the deaf
population and thus no standardization of the language .  Thus we  have
a lot of variation ( probably as is the case for many sign  languages
).  So my question is this :  When you  collect data  from a  sign
language  that has not been analyzed before is it better to start with
elicited data ( ex :   with picture books ,  simple action movies ) or
transcribe natural data with a consultant?  Do you have any knowledge
about whether the results obtained from the elicited data might not
reflect the characteristics of their "real" language ? Do you know any
study that has compared the two?  Of  course  I think the best thing is
to  use both methods  but I wanted  to  have some opinion on this.   If
you also have  a reference that describes methodology  for describing a
sign language  for the first time ,  I would  also appreicate that.
Best wishes
Asli Ozyurek  , Assist. Prof.
Department of  Psychology
Koc University
Rumeli Feneri Yolu
80910, Sariyer Istanbul, Turkey
http://home.ku.edu.tr/~aozyurek