syntactic complexity using MEGRASP?
Kimberly Mueller
I am continuing to search for an efficient way to analyze syntactic complexity, using cookie theft samples in a late-middle aged cohort at risk for Alzheimer's Disease. Some of these participants are declining in memory and language, but I expect deficits in syntax to be extremely subtle, if they exist at all.
I have quite a large number of transcripts, and we unfortunately did not hand code clauses beforehand, so I am hoping for an automated way to look at complexity. Does anyone have any suggestions?
Sincere thanks!
Kimberly Mueller, UW Madison
Brian MacWhinney
Dear Kimberley,
One measure people have used for AD is propositional density. There is an automatic syntactic complexity analyzer built into CLAN. However, you need to first have a %mor line created to use it. In general, most measures of complexity will require some kind of part-of-speech tagging or syntactic analysis.
--Brian MacWhinney
--
You received this message because you are subscribed to the Google Groups "chibolts" group.
To unsubscribe from this group and stop receiving emails from it, send an email to chibolts+u...@googlegroups.com.
To post to this group, send email to chib...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/chibolts/a8210670-a457-4c66-93d3-274f06b893a0%40googlegroups.com.
For more options, visit https://groups.google.com/d/optout.
Kimberly Mueller
Thank you! Yes, I have been using the propositional density measure which is derived from CPIDER within CLAN and that has been useful so far. I was hoping to obtain more specific syntax complexity measures, such as subordination index. And I do have the POS tagging from MOR, and did also run the MEGRASP command; I am just unsure of how to extract the right variables for syntactic complexity, such as a subordination index. If anyone has experience with that, I'd love to get advice!
Brian MacWhinney
Dear Kimberley,
Once you have a dependency structure from MEGRASP, you could tabulate the GRs (grammatical relations) the involve subordination (CMOD, XMOD, CPRED, XPRED) using FREQ and then divide by total #words. However, if you want to exactly replicate the methods of the subordination index as defined in SALT, you would have to really do this all by hand. My guess is that an automatic computation based on the codes on the %gra line would be just as accurate as the 16 step/criteria SI computation defined in SALT. It would be interesting to know for sure.
--Brian MacW
To view this discussion on the web visit https://groups.google.com/d/msgid/chibolts/b7721f15-3f07-499b-a105-2576da884927%40googlegroups.com.
Kimberly Mueller
Kimberly