ARTICLE Brain structural associations of syntactic complexity and diversity across schizophrenia spectrum and major depressive disorders, and healthy controls

Abstract

Deviations in syntax production have been well documented in schizophrenia spectrum disorders (SSD). Recently, we have shown evidence for transdiagnostic subtypes of syntactic complexity and diversity. However, there is a lack of studies exploring brain structural correlates of syntax across diagnoses. We assessed syntactic complexity and diversity of oral language production using four Thematic Apperception Test pictures in a sample of N = 87 subjects (n = 24 major depressive disorder (MDD), n = 30 SSD patients both diagnosed according to DSM-IV-TR, and n = 33 healthy controls (HC)). General linear models were used to investigate the association of syntax with gray matter volume (GMV), fractional anisotropy (FA), axial (AD), radial (RD), and mean diffusivity (MD). Age, sex, total intracranial volume, group, interaction of group and syntax were covariates of no interest. Syntactic diversity was positively correlated with the GMV of the right medial pre-and postcentral gyri and with the FA of the left superior-longitudinal fasciculus (temporal part). Conversely, the AD of the left cingulum bundle and the forceps minor were negatively correlated with syntactic diversity. The AD of the right inferior-longitudinal fasciculus was positively correlated with syntactic complexity. Negative associations were observed between syntactic complexity and the FA of the left cingulum bundle, the right superior-longitudinal fasciculus, and the AD of the forceps minor and the left uncinate fasciculus. Our study showed brain structural correlates of syntactic complexity and diversity across diagnoses and HC. This contributes to a comprehensive understanding of the interplay between linguistic and neural substrates in syntax production in psychiatric disorders and HC. Schizophrenia (2024) 10:101 ; https://doi.org/10.1038/s41537-024-00517-6 INTRODUCTION Deviant thinking, language, and communication, collectively known as formal thought disorder (FTD), have been extensively documented within schizophrenia spectrum disorders (SSD), encompassing the diagnoses schizophrenia (SZ) and schizoaf-fective disorder (SZA) 1,2. Multiple studies highlighted the clinical importance of FTD in SSD as it predicts onset, recurring episodes, and hospitalization 2,3. Furthermore, FTD has been associated with poor (social)functioning 4 and perceived quality of life 5. More recently, this multifaceted construct has been investigated across SSD as well as major depressive disorder (MDD) and bipolar disorder showing transdiagnostic dimensions of FTD (i.e., disorganization, emptiness, and incoherence) and underlying neuroanatomical structures 6,7. FTD has a prevalence rate of up to 80% in SZ, 60% in SZA, 53% in MDD, and 6% in healthy controls (HC) 3. While investigations of speech and language disturbances (i.e., FTD) assessed with operational rating scales such as the rating scale for the assessment of objective and subjective formal thought and language disorder (TALD) 8 are of high clinical value, they do not facilitate the evaluation of nuanced phenomena or deviations concerning language structure including grammar and sound organization. Therefore, studies have used computational linguistic tool such as natural language processing (NLP) and derived latent features for investigating subtle alterations in language production and processing including