Table 10 Comparison of human speaker diarization and Google speaker diarization with extracted language features

E: Okay. So you’ve mentioned a number of friends. Can you tell me about them a little bit? P: Well, my roommate’s name is Emily, and she’s from Minnesota, and she’s getting married this summer.

E: Okay. So, you mentioned a number of friends. Can you tell me? P: \({Um,she{\prime} sgettingmarriedthissummer.}_{{{\rm{F}}}_{2}}\)

P: She’s been in a long-term relationship of 6 years since, like, high school. So it’s, like, pretty young, but I think she’s ready for it. E: Mhmm.

P: She’s been in a long term relationship, a six years. Since, like, high school, so, uh, it’s, like, pretty young, but I think she was ready for it. E: Um-hum.

E: That’s cool. Okay. P: So MRE and others. Oh, and then there’s Carrie. E: How do you spell it? P: K a r I. E: Okay.

P: Cool. E: Yeah. It’s going to be good. P: \({Okay.So,.Andothers?}_{{{\rm{F}}}_{2}}\)E: Uh, and the miscarriage. P: \({KARI.}_{{{\rm{F}}}_{2}}\)E: Okay.

P: We met her freshman year, in math class and then we’ve just been friends ever since.

E: When I’m at her freshman year, um, in math class and then we’ve just been friends ever since.

...

...

P: I also like having people over to my place for dinner because I like cooking and that’s fun for me and for them too. E: What kind of stuff do you like to cook? P: I don’t know. We make like enchiladas is good and also, like, different pastas but, like, with homemade sauces.

P: \({Wealso,like,Ithinkso,like}_{{{\rm{F}}}_{6}}\), having people over to my place for dinner. Because I like cooking. E: What kind of stuff do you have to cook? P: \({Um,kindof,meal,like,like}_{{{\rm{F}}}_{6}}\), enchiladas good, and So, like, different pastas, but, like, like homemade sauces.

Extracted features based on GPT response

Unconventional content (F₂) superfluous phrase attachment (F₆)