Dear Aditya,
This would first require some type of word-level or utterance-level coding that meets your definition of circumlocution. Many of the transcripts have the utterance-level post-code [+ cir] to flag instances of circumlocution but those have not been subjected to any kind of strict reliability evaluations, and not all of the transcripts have those utterance level post-codes.
The command you would use to find utterances with that post-code for the Participant is:
freq +s"<+ cir>" +t*par *.cha
Best,
Davida