A total of 121 university students participated in the study. The sample had a mean age of 20.43 years (standard deviation (SD) = 2.40; range: 18–34). Participants with neurological or medical conditions known to compromise brain function and participants who used medication were excluded.

All participants were assessed using the SPQ, which is a self-report questionnaire with 74 forced choice items based on the Diagnostic and Statistical Manual of Mental Disorders, Third Edition, Revised (DSM-III-R) criteria for schizotypal personality disorder [33].

Based on SPQ total scores, the sample was divided into three groups: the low-SPQ group (scores below the median (21); n = 60), the intermediate-SPQ group (scores above the median ($>$21) and up to 40; n = 44), and the high-SPQ group (scores $\ge$41; n = 16). The cut-off of 41 was selected to define the high-SPQ group in accordance with the previous research [33], suggesting pronounced schizotypal traits or an increased probability of schizotypal personality disorder at this threshold. It is important to emphasize that, in our study, schizotypal traits were determined using a questionnaire, and not based on clinical data. The participant with an SPQ score exactly equal to the median (SPQ total score = 21) was excluded from the analysis (n = 1) to maintain clear group separation. Group comparability was assessed using one-way ANOVA (F) for normally distributed continuous variables, the Kruskal–Wallis H test (H) for non-normally distributed continuous variables and the chi-square test ($\chi$²) for categorical variables. The results of these analyses, along with the sociodemographic and clinical characteristics of the participants are summarized in Table 1.

Age and sex did not differ significantly between groups. As ANOVA revealed significant between-group differences in SPQ total scores, post-hoc comparisons were performed. These analyses showed that all three groups differed significantly from each other (p 0.001).

The experimental stimulus set comprised 180 Russian word pairs consisting of nouns and their corresponding attributes. Stimuli were distributed across three categories: 60 literal phrases, which were semantically congruent and had a direct meaning (e.g., Grey jacket); 60 idiomatic phrases, which conveyed figurative meaning (e.g., Indian summer); and 60 semantically incongruent word combinations without semantic meaning (e.g., Birch parrot). To ensure comparability across conditions, attributes, and nouns were matched for average length. The mean length of attributes was 6.85 letters (SD = 1.13) in the literal phrases, 7.27 letters (SD = 1.22) in the idiomatic phrases, and 7.18 letters (SD = 0.96) in the incongruent phrases. The mean length of nouns was 5.62 letters (SD = 1.45) in the literal condition, 5.73 letters (SD = 1.02) in the idiomatic condition, and 5.63 letters (SD = 0.78) in the incongruent condition.

Participants were seated comfortably in a slightly lit room at a viewing distance of approximately 1 m from a computer screen. Stimuli were presented using PsychoPy (v1.83.01, Open Science Tools Ltd., Nottingham, UK) as word pairs displayed sequentially at the center of the screen in white, lowercase letters against a gray background.

Each trial commenced with a fixation cross displayed for 1000 ms, followed by a blank screen for 200 ms. The first word of the pair was then presented for 1200 ms, followed by another 200 ms blank interval before the second word appeared, resulting in a stimulus onset asynchrony (SOA) of 1400 ms. The second word remained on the screen for 1200 ms, followed by a 200 ms gray screen, which was then replaced by a question mark prompting a response. The overall experimental procedure is illustrated in Fig. 1.

Fig. 1.

The experimental procedure. “+” is a fixation cross, “?” is the question mark. The horizontal arrows indicate examples of the first word (word 1) and the second word (word 2) in the word pairs.

Participants were instructed to withhold responses until the question mark appeared. The question was whether the phrase that appeared on the screen made sense (literal or figurative) or not. Responses were made via the keyboard, with the right button indicating that the phrase was meaningful (“yes”) and the left button indicating that the phrase was meaningless (“no”). Presses of both the right and left buttons on the remote control were performed with the right hand. The question mark remained on the screen until a response was registered.

The experimental stimuli were presented in a randomized order for each participant and for each condition. The task comprised a total of 180 trials, with a short break provided at the midpoint of the experiment.

Electroencephalographic (EEG) activity was recorded using a Neuron-Spectrum-4/EPM analog amplifier with 16-bit ADC (№0494КТ; Neurosoft, Ivanovo, Russia) with 21 Ag/AgCl electrodes mounted in an elastic cap (MCScap 10–20, 205 19; Medical Computer Systems, Moscow, Russia) according to the extended international 10–20 system. Signals were referenced to the left and right ipsilateral mastoids, and a cephalic electrode positioned between the midline frontopolar (FPz) and midline frontal electrodes (Fz) served as ground electrode (GND). Electrode impedances were maintained below 10 k$\mathrm{\Omega }$. Data were continuously sampled at 1000 Hz.

Offline preprocessing included band-pass filtering between 0.1–30 Hz. Ocular and movement artifacts were detected and removed using Independent Component Analysis (ICA), and bad channels were manually identified and interpolated. Epochs exceeding $\pm$150 µV were automatically rejected. Trials with flat signals (1 µV) were also excluded. Data processing and analysis were performed in MNE-Python (v1.8.0, https://mne.tools/) [34]. For each participant, EEG waveforms were averaged separately by condition (literal, idiomatic, incongruent). Only correct responses were included; the average rate was 89%. Based on visual inspection of the grand averages, the N400 was quantified as the mean amplitude within the 250–450 ms post-stimulus window, and the PNP was measured within the 450–800 ms window. Baseline correction was applied using the 200 ms pre-stimulus interval; that is, the interval from –200 to 0 ms relative to the onset of the critical stimulus (second word).

The Shapiro–Wilk test and Q–Q plots were used to check for normality. Based on the results, parametric tests (repeated measures ANOVA) were used for normally distributed data, whereas nonparametric tests (the Kruskal–Wallis test and Spearman’s correlation coefficients) were used for non-normally distributed data.

Mean N400 and PNP amplitudes for correct responses were analyzed using repeated measures ANOVA, with Group (low-SPQ, intermediate-SPQ, high-SPQ) as a between-subjects factor and Condition (idiomatic, literal, incongruent) as a within-subjects factor, conducted separately for the Fz, midline central electrode (Cz), and midline parietal electrode (Pz). Violations of sphericity were corrected using the Greenhouse–Geisser adjustment, original degrees of freedom reported. Significant main effects and interactions were followed by pairwise comparisons corrected using Holm’s method. Statistical significance was set at p 0.05. Effect sizes for significant effects were reported as partial eta-squared ($\eta$²p).

To assess the N400 effect among the three groups, we calculated the amplitude differences between the conditions for each group (N400 amplitudes under literal condition minus N400 amplitudes under incongruent condition, N400 amplitudes under literal condition minus N400 amplitudes under idiomatic condition, and N400 amplitudes under incongruent condition minus N400 amplitudes under idiomatic condition). After that, we assessed these differences in ERP amplitudes between groups using the Kruskal–Wallis H test due to the non-normal distribution of the data. For significant effects, post-hoc pairwise comparisons were conducted with Holm’s correction. The same analysis for the PNP effect was performed.

Finally, correlation analyses were performed to explore the relationship between schizotypal traits and ERP amplitudes. Spearman’s correlation coefficients ($\rho$) were calculated between total SPQ scores as well as relevant subscales and the amplitude differences between conditions (idiomatic vs. incongruent, literal vs. idiomatic, and literal vs. incongruent) with Holm correction because of multiple comparisons.

A Kruskal–Wallis H test revealed no significant differences in the accuracy of semantic judgments across groups (Table 2).

A Kruskal–Wallis H test revealed no significant differences in reaction time across groups (Table 3).

Consistent with prior literature, idiomatic expressions elicited less negative N400 amplitudes than both literal and incongruent word pairs, reflecting facilitated semantic access for familiar figurative expressions. At the frontal site, only the low-SPQ group exhibited clear differentiation between idiomatic expressions and both incongruent and literal conditions. In contrast, intermediate-SPQ and high-SPQ groups failed to show significant N400 distinctions between idioms and literal expressions, suggesting reduced sensitivity to figurative meaning during early stages of semantic processing in individuals with increased schizotypal traits. Moreover, participants from intermediate-SPQ and high-SPQ groups did not show differences between idioms and incongruent expressions in frontal area. It is known that frontal N400 is associated with familiarity, and the recruitment of executive resources during language comprehension [35, 36]. So, insensitivity of the frontal N400 to differences between stimuli in individuals with higher questionnaire scores may indicate impaired executive control during the processing of idioms and incongruent phrases, possibly due to a failure to inhibit irrelevant information, which is one of the characteristics of schizotypy [37, 38]. The absence of the N400 effect in the frontal region is consistent with a previous study, which also reported no differences in N400 amplitude between literal, figurative, and incongruent phrases specifically in the frontal area in schizophrenia patients that was accompanied by a general reduction in cerebral activation within the left and right inferior frontal cortex [39]. It is known that the inferior frontal gyrus (IFG) is connected to the semantic control networks, and also to regions important for lower order semantic processing and integration [40]. Abnormalities of IFG functioning in schizophrenia patients are widely reported in literature [41, 42, 43].

At central and parietal sites, participants with higher SPQ scores demonstrated preserved differentiation between idioms and incongruent phrases, indicating that the processing of highly associative idiomatic words relative to incongruent stimuli remains intact. However, unlike low-SPQ participants, these individuals did not show facilitation of idioms relative to literal phrases. These results are consistent with findings from studies on idiom processing in patients with schizophrenia and may suggest that elevated schizotypal traits are associated with diminished automatic processing of figurative language [14]. At the parietal site, the high-SPQ group failed to distinguish between literal and idiomatic expressions, as well as between idioms and incongruent phrases, which may reflect more profound context processing deficit associated with schizotypy [6, 44]. However, this result should be interpreted with caution due to the small number of participants in the high-SPQ group.

The PNP revealed pronounced group differences in frontal and parietal areas. The intermediate-SPQ participants exhibited larger frontal PNP differences between incongruent and idiomatic conditions, whereas the low-SPQ participants did not show any differences between conditions at the late stage of words processing. Visual inspection demonstrated that late positivity is not expressed in the frontal area in the low-SPQ group and statistical analysis showed that the frontal positivity amplitude in the incongruent condition was larger in the intermediate-SPQ participants than in the low-SPQ group. Correlation analysis further supports these findings. Total SPQ scores and the interpersonal subscale positively correlated with PNP differences between idiomatic and incongruent conditions at Fz, indicating that greater schizotypal traits, particularly interpersonal dysfunction, are associated with enhanced late-stage differentiation of figurative versus anomalous stimuli. It can be assumed that correlation with the interpersonal subscale can provide a reflection of the sensitivity of the late positivities to interpersonal communication difficulties [6] or even to emotional load [24], which accompany the interaction with people. The role of the frontal PNP remains a matter of debate: while some research groups associate it with reanalysis processes [2], others link it to the inhibition of preactivated, highly probable sentence completions, emphasizing that the frontal PNP is specifically related to plausible but unexpected endings, whereas the parietal PNP is associated with anomalous ones [22]. It can be assumed that the increased PNP at later stages may reflect compensatory processes for incongruent information processing in the intermediate-score group, since these processes appeared to be attenuated at the N400 stage in this group of participants. This may reflect processes of semantic reanalysis or the suppression of excessively activated representations, which are required for successful information processing in the group with increased questionnaire scores. Differences in PNP amplitude in the central and parietal leads should be discussed as trends or exploratory comparisons, since the interaction has not reached statistical significance. In view of this, it should be noted that in the high-SPQ group, there was a tendency in the parietal leads to differences between incongruent phrases and idioms. Based on this, we hypothesize that in the group with pronounced schizotypal traits, the compensatory process may be shifted to the parietal region. Comparisons of PNP amplitude differences between conditions further support these findings, as it was revealed that at Pz electrode, the group with the highest questionnaire scores exhibited more pronounced differences between incongruent and idiomatic conditions.

In our study, we did not observe an increase in frontal PNP amplitude during idiom processing, which is inconsistent with Canal’s findings [2]. This may be due to the fact that, in our study, the idioms did not have an additional literal meaning and were not embedded in the sentence context, as they were in Canal et al’s study [2]; therefore, active reanalysis processes at later stages of word processing were not required. So, differences in stimulus material may have contributed to these competing results.

In accordance with the aim of the study, we analyzed the key stages of semantic processing of figurative, literal, and incongruent phrases, as well as the influence of schizotypal personality traits on the neural mechanisms underlying these processes. In the neurotypical sample, we observed facilitated processing of idioms, which can be explained by their status as stable and highly associated word combinations. Typically, a pronounced N400 potential was elicited during the processing of incongruent words, illustrating the classical N400 effect, which indicates increased difficulty in the semantic integration of incongruent stimuli. At later stages, neural resources appeared to be allocated roughly equally across all stimulus types processing.

Among individuals with higher questionnaire scores, the processing of figurative, literal, and incongruent stimuli, as reflected in the frontal N400 component, did not differ significantly. Abnormal modulations of the frontal N400 in groups with elevated scores can be explained as abnormalities in semantic control network. Analyses of the central and parietal N400 components suggest that idiom processing was not facilitated relative to literal phrases, although a typical enhancement of the response to incongruent stimuli was still observed. Such abnormalities in semantic processing at the N400 stage in individuals with increased SPQ scores may lead to more resource-intensive processes, involving semantic reanalysis or suppression of excessively activated representations, which are reflected in PNP differences compared to the normotypical sample.

It is important to emphasize that the present study investigates psychometrically defined levels of schizotypal personality traits and it did not take into account the clinical aspects, therefore, the obtained findings primarily reflect neural-cognitive differences at the level of personality traits but not clinical differences.

Overall, it can be concluded that our hypothesis was only partially confirmed: while we obtained the expected results for the N400 component, we observed unexpected effects for the late positivity. This may be due to the fact that the PNP component has been scarcely investigated in studies on idiom processing and has not been reported in research on schizophrenia and schizophrenia-spectrum disorders. So, the PNP response observed during the processing of idioms and incongruent word pairs in individuals with schizotypal traits was a novel and unexpected finding. This evoked response requires further investigation and may represent a potential neurophysiological marker that could contribute to the general ERP pattern characterizing schizotypy.

Several limitations of the present study should be acknowledged. First, the small sample size of the high-SPQ group leads to limited statistical power, which may influence the stability of some effects, so findings in this group should be interpreted with caution. Additionally, some of the findings relied on exploratory post-hoc comparisons, which should be interpreted with caution, particularly in cases where overall interactions did not reach significance.

Second, the groups with schizotypal personality traits were defined on the basis of psychometric scores rather than clinical diagnostic criteria, therefore this limits the interpretation of the results in terms of clinical prediction.

Third, the stimulus paradigm encompassed only a restricted subset of figurative language phenomena (idioms). Also, the specific nature of the stimuli (Russian word pairs) and potential cultural or linguistic influences should be mentioned.

Finally, longitudinal studies with clinical evidence are needed to determine whether the observed patterns represent stable cognitive characteristics of schizotypy or potential markers of psychosis vulnerability.