Natural language processing for defining linguistic features in schizophrenia: A sample from Turkish speakers.

Area of Science:

• Computational linguistics
• Psychiatry
• Cognitive science

Background:

• Schizophrenia is associated with alterations in language and thought processes.
• Previous research has primarily focused on Indo-European languages.
• The role of language structure in schizophrenia requires further investigation.

Purpose of the Study:

• To investigate linguistic feature alterations in Turkish-speaking individuals with schizophrenia using NLP.
• To determine if these linguistic features are language-dependent or independent.
• To explore the potential of NLP for objective schizophrenia assessment.

Main Methods:

• Natural Language Processing (NLP) applied to speech data from schizophrenia (SZ) and healthy control (HC) participants.
• Analysis included mean sentence length, lexical diversity (type-token ratio), pronoun usage, and parts-of-speech tagging.
• Word2Vec and K-Means clustering were used to assess semantic similarity and group differentiation.

Main Results:

• Schizophrenia patients exhibited reduced mean sentence length and lexical diversity, with increased first-person singular pronoun usage.
• Parts-of-speech analysis indicated fewer coordinating conjunctions in SZ.
• Word2Vec and K-Means accurately differentiated between SZ and HC groups with 86.84% accuracy, highlighting language-independent patterns.

Conclusions:

• NLP effectively identifies significant linguistic alterations in schizophrenia.
• The observed language features in schizophrenia are largely language-independent.
• NLP offers a promising avenue for rapid, objective measurement of language patterns in schizophrenia.