Table 9 Case study analysis: identifying language deficits in an examiner-patient dialogue

From: Exploiting large language models for diagnosing autism associated language disorders and identifying distinct features

Examiner (E) - Patient(P) Dialogue

 

E: So, I’m going to ask you a few questions about work and school

 

P: Yes.

 

E: Um, first of all, do you have a job?

 

P: No, I used to be laid off.

 

...

 

E: And that’s okay? Yeah Um, while you were working or now at school, or at high school, maybe before that, did you have a group of, any problems getting along with people You weren’t in high school?

 

P: Any school. Well, like, like, stupid schools for you when I was developing angry or high \({\text{school.}}_{{{\rm{F}}}_{2}}\)

 

...

 

E: What kind of things you used to bother you that other people did?

 

P: Like, uh, when I was in the school bus, I had students grabbing my backpack, whatever, and I didn’t mad it or \({\text{suck.}}_{{{\rm{F}}}_{10}}\)

 

...

 

E: And have you ever done anything so that other people wouldn’t teach soon?

 

P: Yes, but sometimes they just, it’s like they’ve been doing it for a while, so it’s just kind of like Hey, you know or what, \({\text{whatever,}}_{{{\rm{F}}}_{6}}\) we’ll just tease him about something else.

 

...

 

Extracted features based on ChatGPT response

 

Unconventional content (F2): There are instances where the patient uses unconventionally chosen phrases like “stupid schools for you when I was developing angry or high school”.

 

Superfluous phrase attachment (F6): The patient attaches redundant phrases or filler expressions to their speech without contributing any substantive meaning or context, such as “whatever” and or “whatever”.

 

Clichéd verbal substitutions (F10): The patient resorts to clichéd expressions when describing how he felt during certain situations: “I didn’t mad it or suck”.

 
  1. Phrases highlighted in blue indicate observed linguistic anomalies, while red underscores the specific feature category of language deficits.