A tutorial on multiblock discriminant correspondence analysis (MUDICA): a new method for analyzing discourse data...

Area of Science:

• Linguistics
• Psychology
• Statistics

Background:

• Clinical groups in communication disorders are often defined by performance patterns.
• Standard inferential statistics struggle with datasets featuring few participants and many variables.
• This necessitates novel analytical approaches for complex communication data.

Purpose of the Study:

• Introduce multiblock discriminant correspondence analysis (MUDICA) as a method for analyzing complex datasets in communication disorders research.
• Demonstrate MUDICA's utility with conversational data from individuals with dementia of the Alzheimer's type (DAT).
• Address limitations of standard inferential tools for small-N, high-variable datasets.

Main Methods:

• Applied MUDICA to analyze spontaneous conversations from 17 participant/spouse dyads (controls, early DAT, moderate DAT).
• Examined co-occurrence of trouble-source repair and topic maintenance variables.
• Utilized categorical data analysis suitable for linguistic and discourse variables.

Main Results:

• MUDICA identified significant associations between trouble-source repair sequences and topic transitions.
• Performance patterns in early and moderate DAT groups significantly differed from the control group.
• The analysis successfully differentiated clinical groups based on conversational variables.

Conclusions:

• MUDICA is well-suited for analyzing language and discourse data in communication disorders.
• The method can identify and predict clinical group membership using performance patterns.
• MUDICA accommodates datasets with few participants and many variables, offering inferential statistical rigor.