Acoustic Deviances in Relation to Corpus Callosum
Volume in Schizophrenia Spectrum Disorder
N.G. Stoel
Date: 19-06-2020 Student number: 11295708
Supervisors: Bob Oranje & Alban Voppel
University of Amsterdam - Bachelor Psychobiology Bachelor project at University Medical Centre of Utrecht
2
ABSTRACT
Schizophrenia spectrum disorder is characterised by multiple symptoms and brain alterations, including acoustic aberrations and a decrease in corpus callosum (CC) volume. The CC seems to be a major brain area involved in language production, as some linguistic aberrations are shown in people with a decreased CC volume. Acoustic deviances seem to be an effective tool for assessment, diagnosis, and classification of schizophrenia. The present study examined whether acoustic alterations in schizophrenia are linked to lower CC (sub)volumes. Structural MRIs of 26 patients with schizophrenia spectrum disorder were compared to 30 matched healthy control subjects to investigate the CC volumes. Interviews with both groups were recorded to computationally extract acoustic parameters. Speech rate and mid-anterior CC volume were found to be decreased in patients and had a significant correlation between them, which was not shown in the control group. Additionally, negative symptomatology of the patients was significantly related to both the acoustic and CC volume alterations. Here, a neurological correlate of an acoustic deviance is confirmed in schizophrenia spectrum disorder. The present study adds to fundamental research as well as the promising future use of acoustic deviances as a marker of schizophrenia spectrum disorder.
3
1. Introduction
Schizophrenia spectrum disorder, often referred to as schizophrenia, is a complex psychiatric disorder characterised by a diverse collection of symptoms. These are commonly classified as positive, negative or cognitive. The first set of symptoms becomes apparent during a psychosis and include delusions and hallucinations, whereas the last two tend to show their effect on a long-term period (Owen et al., 2016). These symptoms may also manifest in language, as people with schizophrenia often show certain linguistic anomalies affecting their ability to communicate comprehensibly (Kuperberg, 2010). Linguistic aberrations are a core phenomenon of schizophrenia spectrum disorder. Along with other approaches, linguistic measures are used for investigating and diagnosing the disorder.
A broad range of studies employing neurological approaches contribute to a collection of various brain abnormalities in schizophrenia (Fatemi & Folsom, 2009). Several symptoms have been correlated with such brain abnormalities which give insight into the biological underpinnings of schizophrenia (Voineskos et al., 2019). The corpus callosum (CC) is shown to be anomalous in schizophrenia and is an underlying neurobiological substrate of language (Stipdonk et al., 2018; Agosta et al., 2013). Thus, this brain structure makes a suitable candidate when investigating the specific linguistic deviances in schizophrenia.
The yet incomplete model of schizophrenia can be expanded by a fundamental understanding of its linguistic symptoms. A coherent model of the disorder has the potential to improve diagnosis, monitoring and treatment. Investigation of linguistic aberrations as measures of schizophrenia can be better understood by taking into account the correlation with underlying pathology and symptomatology. Therefore, this study will examine the relationship between linguistic features in schizophrenia and the CC. Additionally, it is investigated how these features relate to the symptoms.
1.1 LINGUISTIC FEATURES IN SCHIZOPHRENIA
Language produced by people with schizophrenia is deviant from regular language production. Commonly reported deficits include poverty of speech, less appropriate content and decreased connection between sentences (DeLisi, 2001). Language production anomalies appear to be capable of indicating the presence and the severity of symptoms (Millan et al., 2014). These linguistic deviances can be subtle and should be carefully examined. Use of objective means would result in robust analyses of these deviances. Deviances established in this manner can be applied in clinical settings as a valuable marker of schizophrenia.
Acoustic properties of speech can objectively be analysed using computational tools. This can be effective in diagnosing psychiatric disorders, in particular psychosis (de Boer et al., 2018). A meta-analysis of Parola et al. (2019) clarified the acoustic features that significantly differ between patients with schizophrenia and healthy controls. These are pitch variability, speech rate, percentage of spoken time, and the duration of pauses. It is not yet examined whether these are suitable diagnostic tools. In addition, the relation between these acoustic deviances and other impairments in neurobiological substrates and clinical symptomatology is still unknown. The knowledge of these acoustic features and their usage can be improved by finding underlying neurological correlates.
1.2 THE INVOLVEMENT OF THE CORPUS CALLOSUM IN LANGUAGE
An important brain structure that is involved in language production is the CC. It is the major bundle of nerve fibres connecting the hemispheres in the human brain. It is crucial for the coordination and transferal of several sorts of information. As the CC is a part of the neural substrate of language, anomalies in this structure are correlated with linguistic deficits (Stipdonk et al., 2018; Agosta et al., 2013).
4 This correlation becomes apparent in preterm children. Multiple structural magnetic resonance imaging (MRI) studies showed an underdeveloped CC in these children and investigated their language outcome. A strong correlation was found between a decreased volume of the CC and difficulties in oral language skills and verbal fluency (Stipdonk et al., 2018). In addition, research on people who developed language impairments later in life shows that their CC is atypical, as it has microstructural white matter damage (Agosta et al., 2013).
This establishes the crucial involvement of the CC in language. A relationship between the CC and the specific acoustic features of speech that are described in schizophrenia has not yet been explored. Additional support to investigate the CC as a neural correlate of these acoustic properties is derived from a validated framework that shows the area to be anomalous in schizophrenia.
1.3 CORPUS CALLOSAL ANOMALIES IN SCHIZOPHRENIA
Patients with schizophrenia show deviances in their CC. The interest in examining the CC in schizophrenia originates from several post-mortem studies that observed differences of the CC compared to healthy controls. This was expressed as either a decreased or increased volume (Rosenthal et al., 1972; Bigelow et al., 1983). With the advent of neuroimaging methods, MRI studies were able to clarify these contradicting findings and showed that the CC in patients is reduced in volume (Arnone et al., 2008; Downhill et al., 2000; Narr et al., 2000; Woodruff et al., 1995). As the CC connects interhemispheric areas (de LaCoste et al., 1985), this volume reduction may reflect decreased communication between areas (Symms et al., 2004).
Inconsistencies exist regarding which CC subareas are reduced, possibly due to differences in the characteristics of the patients. Although not exclusively demonstrated, results are pointing to a reduction in the anterior part of the CC in patients with an early age of onset (Woodruff et al.,1995; Arnone et al., 2008), whereas a later age of onset is related to abnormalities in more posterior CC areas (Downhill et al., 2000; Narr et al., 2000). Regardless, all studies confirm a reduction in CC areas in schizophrenia.
1.4 CURRENT STUDY
Previous studies have demonstrated an abnormal CC in schizophrenia and a link between disruptions in the CC and linguistic problems. However, the link between the CC in schizophrenia and the altered acoustic properties of produced speech has not yet been investigated. The current study examines this relationship.
Structural MRIs of people with schizophrenia spectrum disorder were used to determine CC volume. These patients were interviewed to record and analyse their speech. Subsequently, acoustic features were computationally analysed. The patients were compared to a matched healthy control group. Additionally, deviances in CC volume and acoustic features in patients were related to their score in the Positive and Negative Syndrome Scale (PANSS; Kay et al., 1987) to investigate how the deviances correlate with the severity of symptoms.
It is hypothesized that 1) the volume of subareas of the CC is decreased in patients with schizophrenia spectrum disorder compared to healthy controls, 2) there are differences in the acoustic features between these groups, 3) a correlation exists between anomalies of the CC volume and deviances in the acoustic features in schizophrenia, and 4) the anomalies in CC volume and acoustic features in the patient group are related to the symptomatology.
5
2. Methods
2.1 PARTICIPANTS
The subject group consisted of 26 patients with schizophrenia spectrum disorder and 30 matched healthy controls. Matching was based on sex and age. Handedness was not considered during matching as the CC is not subject to differences in lateralization. Patients were recruited from both inpatient and outpatient units at the University Medical Centre of Utrecht (UMCU). Their treating psychiatrist diagnosed them with either schizophrenia, schizophreniform disorder, schizoaffective disorder, or psychosis Not Otherwise Specified (NOS). Trained researchers confirmed these diagnoses using the Mini-International Neuropsychiatric Interview (Sheehan et al., 1998). Symptom severity was rated using the PANSS (Kay et al., 1987). Dosage of antipsychotics was converted to an equivalent dose of chlorpromazine (Leucht et al., 2014).
Brain imaging was performed as part of the SIMVA-study (Begemann et al., 2015), whereas the interviews were executed on behalf of the PRAAT-study (De Boer et al., 2020). Both studies were performed at the UMCU and were approved by the local medical-ethical review board. Patients and controls were recruited for the SIMVA-study according to their inclusion criteria. These patients were required to be between the age of 18 and 50 years old, and the onset of their first psychosis had to be less than three years ago. For the full list of inclusion and exclusion criteria, see the study trial registration Begemann et al. (2015). After inclusion, subjects enrolled in the SIMVA-study were approached for participation in the PRAAT-study if they were eligible. Inclusion criteria for the PRAAT-study were as follows; native speaker of Dutch, over the age of 18 years, and no uncorrected speech or hearing impairment.
All subjects gave written informed consent for participation in both studies. For the PRAAT-study, participants were only informed about the exact aim of this study after the procedure as to not affect natural speech. A debriefing occurred after the interview and subjects received a small monetary reward.
2.2 INTERVIEW PROCEDURE
The interviews consisted of a list of open-ended questions to retrieve 15 minutes of spontaneous speech. The theme of these questions was events in daily life, e.g. “Can you tell about a dream you recently had?” or “How are your experiences with the dentist?”. Participants could skip any question if they did not feel comfortable answering. The interviews were conducted in a silent room. For the full list of questions, see Appendix A.
Both the interviewer and the participant spoke into a head-mounted microphone at approximately 3-4 centimetres removed from the mouth. A TASCAM DR-40 was used as a digital recording device, with separate recording channels for interviewer and participant.
2.3 ACOUSTIC PARAMETERS
Acoustic features of the participants were extracted and analysed with the software Praat version 6.0.37 (Boersma & Weenink, 2013) using the Praat Script Syllable Nuclei v2 (Quené et al., 2011), and openSMILE version 2.3.0 (Eyben et al., 2010). Pitch was calculated with openSMILE using GeMAPS (Eyben et al., 2016). Table 1 summarises the parameters of each software and the according calculations for the acoustic features.
6
2.4 STRUCTURAL MRI
T1-weighted images of the participants were obtained using a 3T Philips Achieva scanner (Philips Medical Systems, Best, the Netherlands) with a fast field echo and a SENSE-Head-8 coil. Table 2 represents the parameters used during acquisition.
Table 2. Scanning parameters
Repetition time (TR) 10 ms
Echo time (TE) 4.6 ms
Min. inversion time (TI) delay 964.4 ms
Flip angle 8 degrees
Acquisition matrix 304 x 299
Field of view (FOV) 240 x 240 x 160 mm
Voxel size 0.79 x 0.80 x 0.80 mm
Slice thickness 0.8 mm
Number of slices 200
Plane orientation Sagittal
Quality control of the raw T1-images was done by two researchers. The next analysis step was performed utilising FreeSurfer software v6.0.0 according to the atlas from Destrieux, et al. (2010). The CC was segmented with the mri_cc command into the following five areas: posterior, mid-posterior, central, mid-anterior, and anterior. Images were obtained of the intracranial volume and the segmented CC. The quality of these processed images was manually examined by two independent researchers.
2.5 STATISTICAL ANALYSIS
The statistical analyses were conducted in R, version 1.1.463, including the packages ‘car’ and ‘ggplot2’. A significance threshold of p<0.05 was used. Comparisons of demographic information between groups were done with parametric t-tests and χ2-tests or a non-parametric Mann-Whitney U test when appropriate.
The differences in the CC volume between patients with schizophrenia spectrum disorder and healthy controls were assessed with a multivariate analysis of covariance (MANCOVA). The five CC volumes were dependent variables. Intracranial volume was considered as a covariate, as were age and sex.
Table 1. Software used to extract parameters for (calculation of) each acoustic feature
Software Parameters and
calculations Acoustic features PRAAT Number of syllables in total
spoken time Speech rate
PRAAT Spoken time as percentage
of full interview Percentage of spoken time
openSMILE Mean length of unvoiced
segments Mean duration of pauses openSMILE Standard deviation of pitch Pitch variability
MRI scanning parameters obtained by using a 3T Philips Achieva scanner
Acoustic features extracted by PRAAT and openSMILE software. All these features are extracted or calculated over the time the participant speaks.
7 Another MANCOVA was performed to analyse differences in the four acoustic parameters between the two groups, with sex and age as covariates.
Since most acoustic parameters are not normally distributed, a Spearman correlation test was used. This test is robust against normality violations. The correlation between CC volume and acoustic parameters between each group was examined using Spearman correlation tests for the two groups separately. Additionally, the CC volume and acoustic deviances in patients are tested in their relationship with PANSS-scores with a Spearman correlation test.
3. Results
3.1 DEMOGRAPHICS
Demographics of participants are displayed in table 3. Age and sex ratio did not differ between healthy control subjects and patients with schizophrenia spectrum disorder. The patient group completed a significantly lower amount of educational years compared to controls (W=527, p=0.0241). See Appendix B for the specific diagnoses and antipsychotic use of the patient group.
Table 3. Demographics of participants
SZ HC Statistics
n=26 n=30 t, χ2 or W p-value
Age (years) 25.5 ± 5.20 25.1 ± 5.41 t = -0.332 0.742
Sex (male/female) 20/6 26/4 χ2 = 0.901 0.343
Education (years) 12.7 ± 3.12 14.4 ± 2.13 W = 527 0.0241* Duration of illness (years) 2.01 ± 1.54
Age of onset (years) 23.5 ± 5.42 PANSS positive 10.9 ± 3.92 PANSS negative 14.2 ± 5.01 PANSS general 27.2 ± 6.32 PANSS total 52.3 ± 11.5 Chlorpromazine equivalent (mg/day,) n=21a 232 ± 145
Values are mean ± standard deviation. A student t-test (t), chi-squared test (χ2), and Mann-Whitney U test (W) were used.
a.22 patients of the SZ-group used differing doses of one of the following antipsychotics: quetiapine, aripiprazole,
olanzapine, clozapine, haloperidol, paliperidone. Doses of these antipsychotics are converted to a chlorpromazine equivalent. One participant used 800mg/d amisulpride, which cannot be converted (Leucht et al., 2014) and thus has not been taken into the calculation. SZ = participants with schizophrenia spectrum disorder; CTL = healthy control subjects; PANSS = Positive and Negative Syndrome Scale; n = sample size. The *-notation indicates significance (p<0.05).
8
3.2 ACOUSTIC MEASURES
Table 4 shows the average acoustic values of patient and control group and statistics. A MANCOVA was performed between these features with age and sex as covariates. This revealed a significant difference between groups (F(4,49)=6.52, p<0.001). Both age and sex had a significant effect (F(4,49)=
3.19, p=0.0211 and F(4,49)=10.0, p<0.001, respectively).
Univariate testing revealed significant group differences in the percentage of spoken time (F=20.0, p<0.001), when controlling for age. Sex had no significant effect on spoken time. Speech rate was also found to differ significantly between groups (F=4.36, p=0.0417). Age and sex had no significant influence on speech rate.
Table 4. Statistics of acoustic features of participants
SZ HC Univariate statistics n=26 n=30 F p-value Pitch variability 0.201 ± 0.0402 0.190 ± 0.0412 1.06 0.308 Percentage of spoken time 72.5 ± 10.2 82.8 ± 7.83 19.9 <0.001* Duration of pauses (s) 0.174 ± 0.0450 0.191 ± 0.0550 1.66 0.203 Speech rate (number of syllables / total spoken time)
2.73 ± 0.490 2.98 ± 0.390 4.36 0.0417*
Table 5. Comparison between brain structures
SZ HC Univariate statistics n=26 n=30 F p-value CC Posterior 950 ± 133 1.013 ± 165 2.01 0.162 CC Mid-Posterior 540 ± 92.1 560 ± 118 0.531 0.469 CC Central 511 ± 88.3 577 ± 125 5.15 0.0275* CC Mid-Anterior 526 ± 91.9 598 ± 139 5.07 0.0287* CC Anterior 958 ± 165 972 ± 165 0.100 0.753 Intracranial Volume 1.526 ± 1.605 1.606 ± 1.665
All values are displayed as mean ± standard deviation. A MANCOVA with age and sex as covariates was performed to determine differences between the groups. Here, the univariate outcomes are showed. Age had a significant effect on percentage of spoken time. Sex and age significantly influenced pitch variability. Group differences were controlled for these effects. SZ = participants with schizophrenia; HC = healthy control participants; n = sample size. The *-notation indicates significance (p<0.05).
Volume values are displayed as mean ± standard deviation. The unit is mm3. MANCOVA’s with age, sex, and
intracranial volume as covariates was performed to determine differences between the groups. Here, the univariate outcomes are showed. None of the covariates had a significant effect. SZ = participants with schizophrenia spectrum disorder; HC = healthy control participants; n = sample size. The *-notation indicates significance (p<0.05).
9
3.3 STRUCTURAL MRI
The MANCOVA was performed to determine differences in the five CC volume measures between patients with schizophrenia spectrum disorder and healthy control subjects. Sex, age, and intracranial volume were considered covariates. No significant effect between groups was found in the multivariate test (F(5,47)=1.66, p=0.161). However, the univariate test revealed that the central and mid-anterior part
of the CC were significantly decreased in the patient group (F=5.15, p=0.0275 and F=5.07, p=0.0287, respectively). Age, sex, and intracranial volume were no significant confounders for these measures. See table 5 for the values of the CC volumes and univariate statistic outcomes.
3.4 CORRELATION ANALYSIS
Spearman correlation tests were performed to investigate the relationship between the acoustic features and the segregated CC volumes. The group of patients with schizophrenia showed a correlation between speech rate and both the mid-anterior CC volume (rs=-0.422, p=0.0331) and the mid-posterior
part (rs=-0.411, p=0.0362). These correlations were not found significant for the control group
(rs=0.0265, p=0.890; rs=0.275, p=0.142, respectively). Figure 1 displays the scatterplots of the
significant correlations.
Speech rate in relation to mid-anterior and mid-posterior CC volumes a.
Figure 1. Correlations between speech rate and both the mid-anterior and mid-posterior CC volumes. The acoustic feature
is compared to CC volumes for patients with schizophrenia (resp., a. rs=-0.422, p=0.0331; c. rs=-0.411, p=0.0362; n=26) and
healthy controls (resp., b. r=0.0265, p=0.890; d. rs=0.275, p=0.142; n=30). The correlation is visualised by the blue line with
95% confidence intervals represented by the dark-grey areas. The *-notation indicates significance (p<0.05).
b.
c. * d.
10 On account of these results, a Spearman correlation test was performed in the patient group to see if speech rate and the CC volumes correlated with symptomatology. This revealed significant correlations between speech rate and the negative PANSS score (rs=0.553, p<0.01). Only the
mid-anterior part of the CC was significantly correlated with the PANSS negative score (rs=-0.401,
p=0.0434). One other acoustic parameter, the duration of pauses, is significantly correlated with negative PANSS scores (rs=-0.690, p=<0.0005). Figure 2 displays these significant correlations.
Negative PANSS-scores in relation to acoustic parameters and mid-anterior CC volume
Figure 2. Significant correlations between negative PANSS-scores and acoustic parameters and mid-anterior CC volume in patients with schizophrenia spectrum disorder. The correlation between the negative PANSS-scores and speech rate (a. rs=0.553,
p<0.01), duration of pauses (b. rs=-0.690, p<0.0005) and the mid-anterior CC volume (c. rs=-0.401, p=0.0434) is analysed for
patients with schizophrenia (n=26). The correlation is visualised by the blue line with 95% confidence intervals represented by the dark-grey areas. The *-notation indicates p<0.05, ** indicates p<0.01, *** indicates p<0.0005.
a. b.
c.
** ***
11
4. Discussion
The current study investigated the relationship between the CC volume and acoustic parameters in patients with schizophrenia spectrum disorder. First, the central and mid-anterior CC volumes are significantly decreased in patients compared to the control group. Secondly, patients spoke for a significantly lower percentage of the interview and had a lower speech rate. Finally, correlations are found in the patient group between speech rate and both the mid-anterior and mid-posterior CC volumes. Patients with a lower volume of these CC areas talked faster. Both speech rate and mid-anterior CC volume reduction are linked to negative symptomatology. Specifically, speech rate was increased and mid-anterior CC volume was decreased in patients with higher negative PANSS-scores. Thus, these results provide insight into a neurological correlate of acoustic deviance in schizophrenia spectrum disorder.
Evaluation and implication
The relationship between mid-anterior CC volume and speech rate is exclusively found in patients with schizophrenia. Both independent measures significantly differ from control subjects. This suggests that a healthy person with a relatively low mid-anterior CC volume can rely on different brain areas to maintain a normal speech rate. However, these brain areas might also be anomalous in patients. Thus, the mid-anterior CC volume might not be the only neural correlate of an altered speech rate. Not having found a correlation of the CC volume with percentage of spoken time and duration of pauses provides additional support for another underlying factor.
The results contribute to the hypothesis that reductions in the CC manifest depending on the age of onset (Woodruff et al., 1995; Arnone et al., 2008; Downhill et al., 2000; Narr et al., 2000). This study confirms speculations that CC volume reductions tend to be more anterior in early-onset schizophrenia. The patients in this study have a relatively early average age of onset (23.5 years). Whereas their anterior CC is reduced, they do not show reduction in posterior areas. However, the sample size is small and excludes late-onset patients. To fully support the hypothesis, studies should be conducted which include both early- and late-onset patients.
The volume reductions in the mid-anterior and central CC may indicate fewer fibres and thus abnormal interhemispheric communication between these areas (Symms et al., 2004). The mid-anterior and central areas relate to the posterior part of the genu and the body of the CC. These areas roughly represent the interconnection between the frontal lobes and the motor cortices, respectively (de Lacoste et al., 1985). The mid-anterior CC volume decrease may result in an altered speech rate through faults in motoric language production, as there could be a lower communication between motor cortices. Functional instead of structural scans should be performed to confirm this decreased communication. The relationship found here between the mid-anterior CC volume and speech rate prompts further interest in examining the use of speech rate of patients with schizophrenia as a tool in clinical settings to evaluate severity of the disorder. This study showed that speech rate is linked with two existing correlates of schizophrenia. Specifically, an increased speech rate indicates a higher severity of both the CC pathology and a higher negative PANSS-score. If further studies substantiate these results, measuring the speech rate could ultimately provide a low-cost and simple tool in estimating the severity of negative symptoms patients diagnosed with schizophrenia.
Next to measuring the severity of symptoms, speech rate is a candidate that could be deployed as a diagnostic tool. Patients had an overall lower speech rate than control subjects. On the other hand,
12 within the patient group the mid-anterior CC pathology and negative symptomatology seemed to be more severe when speech rate was higher. Thus, if a low speech rate would be a criterion of diagnosis, these severe patients would go undetected. Nevertheless, a low speech rate could indicate less severe forms of schizophrenia. As these forms may not be as visible as severer ones, a supplementary tool would be required. Therefore, speech rate could be used as a diagnostic tool for a subset of people with schizophrenia.
Furthermore, negative symptomatology seemed to decrease with longer duration of pauses. The duration of pauses could supplement the negative symptom-severity estimation tool, whereas the percentage of spoken time could be helpful in a diagnostic tool. A shorter duration of pauses seems to be indicative of a higher negative PANSS-score. However, this measure does not seem as useful for distinguishing a patient from a healthy subject as it does not significantly differ between both groups. Percentage of spoken time could provide this distinction but does not correlate with symptomatology. These tools require prudence as acoustic parameters are likely to be related to each other. Tools that are based on measurements of these parameters could therefore have a low specificity. This should be considered in the creation of such tools. Additionally, unlike speech rate, the CC volume is not correlated with duration of pauses or percentage of spoken time. The neural underpinnings of these parameters should be investigated to understand its relationship with the pathology.
Limitations and future research
The current results cannot provide information on the relationship between a positive PANSS-score, language disturbances and CC volume. The positive PANSS-scores in this sample were low as nearly all patients took antipsychotics and were not actively psychotic during the study. Therefore, it was not possible to demonstrate any relationship with these positive symptoms.
Individual differences between patients can have significant effects on both acoustic features and CC volumes. This study is limited by its sample size in overcoming these effects. The population of people with schizophrenia is characterised by several differences, for example in antipsychotic use, symptom severity and educational levels. Still, the influence of variances could be high and should be investigated and considered. To increase statistical power, future research with limited sample sizes should consist of repeated studies with more participants.
The other two acoustic features found by Parola et al. (2019), i.e. duration of pauses and pitch variability, are not confirmed to significantly differ from control subjects with this sample size. In addition to the characteristic variances, this might also be the result of inconsistencies between studies in how the measures are obtained. Finally, different interview structures may require a different cognitive load, which could affect the found measures. Uniformity of these measures is important for further research. Uniform methods and analysis can decrease fragmentation of the field which is important for comparing studies. This way, additional vocal markers can be identified which could be used in predictive models to help in diagnosing as well as treating the disorder. The markers would also provide insight into the heterogeneous symptomatology of schizophrenia spectrum disorder. Ultimately, finding vocal markers that distinguish different psychiatric disorders would be of high clinical relevance. This may be more valuable than making a predictive model that can identify a healthy person from a person with schizophrenia spectrum disorder. Although psychosis alone might be relatively easily identified, the course of the disorder only becomes apparent over months if not years. If distinguishing acoustic features are found at the onset of the disorder, a more suitable treatment can be offered earlier.
13 The current study connects an underlying neurological mechanism with linguistic deviances in schizophrenia. It contributes to the understanding of anomalous acoustic processes and shows how the acoustic deviance changes in relationship with the severity of corpus callosum pathology. Thereby, it proves promising use of acoustic deviances as a marker of schizophrenia spectrum disorder and contributes to schizophrenia research regarding both its symptoms and its mechanisms.
14
REFERENCES
Agosta, F., Galantucci, S., Canu, E., Cappa, S., Magnani, G., et al (2013). Disruption of structural connectivity along the dorsal and ventral language pathways in patients with nonfluent and semantic variant primary progressive aphasia: A DT MRI study and a literature review. Brain and language. 127. 10.1016/j.bandl.2013.06.003.
Arnone, D., Mcintosh, A., Tan, G. & Ebmeier, K. (2008). Meta-analysis of magnetic resonance imaging studies of the corpus callosum in schizophrenia. Schizophrenia research. 101. 124-32. 10.1016/j.schres.2008.01.005. Bigelow, L. B., Nasrallah, H. A., & Rauscher, F. P. (1983). Corpus callosum thickness in chronic schizophrenia. The British Journal of Psychiatry, 142, 284–287. https://doi.org/10.1192/bjp.142.3.284
Begemann, M.J., Schutte, M.J.L., Slot, M.I.E., Doorduin, J., Bakker, P.R., et al. (2015). Simvastatin augmentation for recent-onset psychotic disorder: a study protocol. BBA Clinical, 4, pp. 52-58. https://doi.org/10.1016/j.bbacli.2015.06.007
de Boer, J.N., van Hoogdalem, M., Mandl, R.C.W. et al. (2020). Language in schizophrenia: relation with diagnosis, symptomatology and white matter tracts. npj Schizophr 6, 10. https://doi.org/10.1038/s41537-020-0099-3 de Boer, J.N. & Voppel, A.E. & Begemann, Marieke & Schnack, Hugo & Wijnen, Frank & Sommer, I.E.C.. (2018). Clinical use of semantic space models in psychiatry and neurology: A systematic review and meta-analysis. Neuroscience & Biobehavioral Reviews. 93. 10.1016/j.neubiorev.2018.06.008.
Boersma, P. & Weenink, D. J. M. Praat: Doing Phonetics by Computer Version 6.0.37 (2013). Institute of Phonetic Sciences of the University of Amsterdam, Amsterdam. www.praat.org.
Delisi, L. (2001). Speech Disorder in Schizophrenia: Review of the Literature and Exploration of Its Relation to the Uniquely Human Capacity for Language. Schizophrenia bulletin. 27. 481-96. 10.1093/oxfordjournals.schbul.a006889.
Destrieux, C., Fischl, B., Dale, A. & Halgren, E. (2010). Automatic parcellation of human cortical gyri and sulci using standard anatomical nomenclature. Neuroimage. 2010;53:, pp. 1–15.
Downhill, J.E., Buchsbaum, M.S. & Wei, T. (2000). Shape and size of the corpus callosum in schizophrenia and schizotypal personality disorder. Schizophr Res 2000; 42:193–208
Eyben, F., Wöllmer, M. & Schuller, B. (2010). openSMILE - The Munich Versatile and Fast Open-Source Audio Feature Extractor“, In Proc. ACM Multimedia (MM), ACM, Florence, Italy, ACM, pp. 1459-1462, October 2010. www.audeering.com/opensmile/.
Eyben, F., Scherer, K. R., Schuller, B. W., Sundberg, J., André, E., Busso, C., et al. (2016). The geneva minimalistic acoustic parameter set (gemaps) for voice research and affective computing, IEEE Transactions on Affective Computing, vol. 7, no. 2, pp. 190-202, 2016.
Fatemi, S. H. and Folsom, T. D. (2009) The neurodevelopmental hypothesis of schizophrenia, revisited. Schizophrenia Bulletin, 35, 528-548. doi:10.1093/schbul/sbn187
Kay, S.R., Fiszbein, A., Opler, L.A., (1987). "The positive and negative syndrome scale (PANSS) for schizophrenia". Schizophr Bull. 13 (2): 261–76. doi:10.1093/schbul/13.2.261. PMID 3616518.
Kuperberg, G.R. (2010), Language in Schizophrenia Part 1: An Introduction. Language and Linguistics Compass, 4: 576-589. doi:10.1111/j.1749-818X.2010.00216.x
de Lacoste, M.C., Kirkpatrick, J.B. & Ross, E.D., (1985). Topography of the human corpus callosum. J.
15
Leucht, S., Samara, M., Heres, S., Patel, M. X., Woods, S. W., & Davis, J. M. (2014). Dose equivalents for second-generation antipsychotics: the minimum effective dose method. Schizophrenia bulletin, 40(2), 314–326. https://doi.org/10.1093/schbul/sbu001
Millan, M., Fone, K., Steckler, T. & Horan, W. (2014). Negative symptoms of schizophrenia: Clinical characteristics, pathophysiological substrates, experimental models and prospects for improved treatment. European neuropsychopharmacology : the journal of the European College of Neuropsychopharmacology. 24. 645-92. 10.1016/j.euroneuro.2014.03.008.
Narr K.L. , Thompson P.M. & Sharma T. (2000). Mapping morphology of the corpus callosum in schizophrenia. Cereb Cortex 2000; 10:40–49
Owen, M. J., Sawa, A., & Mortensen, P. B. (2016). Schizophrenia. Lancet (London, England), 388(10039), 86– 97. https://doi.org/10.1016/S0140-6736(15)01121-6
Parola, A., Simonsen, A., & Bliksted, V., & Fusaroli, R. (2019). Voice patterns in schizophrenia: A systematic review and Bayesian meta-analysis. 10.1101/583815.
Quené, H., Persoon, I. & de Jong, N. (2011). Praat Script Syllable Nuclei v2 [Praat Script].
Rosenthal, R. & Bigelow, L. (1972) Quantitative brain measurements in chronic schizophrenia. British Journal
of Psychiatry, 121, 259–64.
Sheehan D.V., Lecrubier Y. & Sheehan K.H. (1998). "The Mini-International Neuropsychiatric Interview (M.I.N.I.): The development and validation of a structured diagnostic psychiatric interview for DSM-IV and ICD-10". J Clin Psychiatry. 59 (Suppl 20): 22–33. PMID 9881538.
Stipdonk L.W., Franken M.-C.J.P &, Dudink J. (2018) Language outcome related to brain structures in school-aged preterm children: A systematic review. PLoS ONE 13(6): e0196607. https://doi.org/10.1371/journal.pone.0196607
Symms, M, Jäger ,HR, Schmierer, K. & Yousry, T.A. (2004). A review of structural magnetic resonance neuroimaging, Journal of Neurology, Neurosurgery & Psychiatry, 2004(75), 1235-1244.
Voineskos, A.N., Grace, J. & Ameis, S. (2019). Neuroimaging Heterogeneity in Psychosis: Neurobiological Underpinnings and Opportunities for Prognostic and Therapeutic Innovation. Biological Psychiatry. 10.1016/j.biopsych.2019.09.004.0
Woodruff, P. W., McManus, I. C., & David, A. S. (1995). Meta-analysis of corpus callosum size in schizophrenia. Journal of neurology, neurosurgery, and psychiatry, 58(4), 457–461. https://doi.org/10.1136/jnnp.58.4.457
16
Appendix A: Interview questions
Table A1. Questions from the PRAAT-study for the interviews.
1 Kun je vertellen over je zwemles van vroeger? Hoe vond je dat? Wat vond je het moeilijkst? En het leukst?
Can you tell about your swimming lesson from when you were young? How did you like it? What did you find the most difficult? And what did you like the best?
2 Kun je vertellen over je tandartservaringen (slechte en goede ervaringen)? Kun je bijvoorbeeld vertellen over de laatste keer dat je bent geweest? En hoe heb je dat als kind ervaren?
Can you tell about your dentist experiences (bad and good experiences)? For example, can you tell me about the last time you've been? And how did you experience going as a child?
3 Heb je een rijbewijs? Zo ja, kun je wat vertellen over hoe de lessen gingen en hoe het examen ging?
Do you have a driving license? If so, can you tell me how the lessons went and how the exam went?
4 Ben je wel eens in een pretpark geweest? Kun je daar wat over vertellen? Heb je een favoriet pretpark, vertel daar eens over? Waarom is het je favoriete park?
Have you ever been to an amusement park? Can you tell me about that? Do you have a favorite amusement park, tell me about it? Why is it your favourite park?
5 Naar welke Nederlandstalige Tv-programma’s kijk je vaak? En aan welke heb je een hekel? Waarom?
Which Dutch TV shows do you often watch? And which do you hate? Why?
6 Kijk je wel eens naar sport, zoals voetbalwedstrijden? Zo ja, naar welke wedstrijden en wat vind je daarvan? En ben je een fan van een bepaalde club of sporter?
Do you ever watch sports, like football matches? If so, what matches and what do you think about it? And are you a fan of a particular club or athlete?
7 Kun je je laatste droom beschrijven?
Can you describe your last dream?
8 Hoe was je laatste verjaardag? Hoe vier je normaal je verjaardag?
How was your last birthday? How do you usually celebrate your birthday?
9 Wat zou je doen als je een miljoen zou winnen?
What would you do if you were to win a million?
10 Als je voor altijd een bepaalde leeftijd kon hebben, welke leeftijd zou dat dan zijn? Waarom?
If you could have a certain age forever, what age would it be? Why?
11 Als je overal ter wereld heen mocht, waar zou je heengaan?
17
12 Welk klusje in huis vind je echt vreselijk om te doen? Waarom? Doe je het dan?
What kind of household chore do you really hate? Why? Do you do it then?
13 Ga je wel eens op vakantie? Wat is je favoriete vakantiebestemming? Waarom?
Do you go on vacation? What is your favourite vacation destination? Why?
14 Welke oorlog in de geschiedenis heeft indruk op je gemaakt? Waarom?
What war in history has impressed you? Why?
15 Wat voor werk doen/deden je ouders?
What kind of work do your parents / did your parents do?
16 Wat voor werk doe je/heb je recent gedaan? Wat sprak je daarin aan?
What kind of work do you / have you done recently? What did you talk about?
17 Kun je wat vertellen over je laatste sollicitatiegesprek? Was je zenuwachtig? Ben je aangenomen?
Can you tell me about your last job interview? Were you nervous? Were you hired?
18 Wat is je favoriete jeugdherinnering? Wie waren erbij? Waarom vind je dat zo’n fijne herinnering?
What is your favourite childhood memory? Who were there? Why do you think that's a nice memory?
19 Hoe werden bij jou thuis de feestdagen gevierd?
How did you celebrate the holidays back home?
20 Waar ben je geboren? Welke stad, ziekenhuis of thuis? Wie waren daarbij?
Where were you born? Which city, in hospital or at home? Who were there?
21 Wat voor soort kind was je toen je klein was?
What kind of child were you when you were small?
22 Had je vroeger een lievelingsknuffel? Of iets anders dat je altijd bij je wilde hebben?
Did you have a favourite toy when you were young? Or something else that you always wanted to have with you?
23 Wat was je favoriete boek toen je klein was? En waarom was je daar zo dol op?
What was your favourite book when you were small? And why were you so fond of it?
24 Lijk je op je ouders? Op wie lijk je het meest, waarom?
Do you look like your parents? Who do you resemble the most, why?
25 Heb je broers of zussen? Op wie lijk je het meest, waarom? Met wie had je het meest ruzie?
Do you have any brothers or sisters? Who do you resemble the most, why? Who did you argue with most?
26 Wat wilde je vroeger worden (wat voor baan)? Heb je dat lang gedacht? Zou je dat nu nog willen?
What did you want to become when you were young (what kind of job)? Did you want that for a long time? Would you still like that?
18
27 Heb je wel eens een huisdier gehad? Zou je een huisdier willen? Wat voor een? Hoe was je band daarmee?
Have you ever had a pet? Would you like a pet? What kind? How was your relationship with it?
28 Welke mensen waren belangrijk voor je in je jeugd? Waarom?
What people were important to you in your youth? Why?
29 Was je weleens ziek als kind? Heb je weleens in het ziekenhuis gelegen?
Were you sometimes sick as a child? Have you ever been to the hospital?
30 Wat was je favoriete eten toen je klein was?
What was your favorite food when you were small?
31 Wat deed je vroeger waardoor je in de problemen kwam? Werden je ouders dan boos?
What did you do in the past that caused you to get into trouble? Did your parents get angry?
32 Wat voor spellen speelde je vroeger? Waar speelde je meestal? Buiten/binnen? Met wie?
What games did you play when you were young? What did you usually play? Outside inside? With whom?
33 Kun je je de laatste keer herinneren dat je goed nieuws kreeg? Wat voor nieuws was dat? Wat voor effect had dat op je?
Can you remember the last time that you got good news? What kind of news was that? What effect did that have on you?
34 Hoe heb je je beste vriend/in ontmoet?
How did you meet your best friend?
35 Ben je wel eens verliefd geweest? Weet je nog hoe jullie in gesprek kwamen? Was je verlegen? Weet je nog hoe jullie eerste afspraakje was? Was je zenuwachtig?
Have you ever fallen in love? Do you still know how you got talking? Were you shy? Do you remember how was your first date? Were you nervous?
36 Wat zijn belangrijke momenten in je leven geweest?
Looking back, what were important moments in your life?
37 Welke mensen spelen nu een belangrijke rol in je leven? Waarom?
Which people now play an important role in your life? Why?
38 Wat was een van de beste feestjes waar je ooit bent geweest?
What was one of the best parties you've ever been to?
39 Als je een tijdmachine had, naar welke tijd in de toekomst/verleden zou je gaan? Waarom?
If you had a time machine, what time in the future / past would you go? Why?
40 Als je met één persoon op aarde nu mocht spreken, met wie zou dat zijn?
19
41 Als je een dier was, welk dier zou je dan zijn en waarom?
If you were an animal, what animal would you be and why?
42 Wat is je favoriete bezigheid in de zomer? Waarom?
What is your favourite activity in the summer? Why?
43 Stel je wordt gedropt op een onbewoond eiland en moet daar een jaar blijven, welke 3 dingen zou je dan meenemen?
Imagine being dropped on an uninhabited island and staying there for a year, what 3 things would you bring?
44 Als je één van je zintuigen op moest geven welke zou het dan zijn? (horen, zien, voelen, ruiken of proeven). Waarom?
If you had to give up one of your senses, what would it be? (hear, see, feel, smell or taste) Why?
45 Wat is het leukste cadeau dat je ooit hebt gekregen? Van wie kreeg je het?
What is the best gift you ever received? Who did you get from?
46 Wat is je favoriete stripfiguur? Waarom die?
What is your favourite cartoon character? Why that one?
47 Wat is het beste/leukste dat je deze week is overkomen?
What is the best / fun thing you have come across this week?
48 Wat is het vreemdste dat je ooit hebt gegeten? (bv. slak, oester)
What is the strangest thing you have ever eaten? (e.g. snail, oyster)
49 Als je vandaag één wereldprobleem mocht oplossen, welk probleem zou dat zijn? Waarom?
If you could solve one world problem today, what problem would that be? Why?
50 Als je nu een auto mocht kopen, wat voor auto zou dat dan zijn?
If you were to buy a car, what kind of car would it be?
51 Wat voor soort huis is je droomhuis? Waar zou dit huis staan? Met wie zou je er willen wonen?
What kind of house is your dream home? Where would this house be? Who would you like to live with?
52 Welke taal zou je nog willen leren spreken? Waarom deze taal?
What language would you still like to learn? Why this language?
53 Als je voor één dag God zou zijn, wat zou je dan doen?
If you could be God for one day, what would you do?
54 Hoe ziet jouw perfecte pizza eruit? Wat zit er allemaal op?
20
55 Wat is het raarste kledingstuk dat je ooit hebt gedragen?
What's the strangest piece of clothing you've ever worn?
56 Wat zou je doen als je een dag onzichtbaar zou kunnen zijn?
What would you do if you could be invisible one day?
57 Van welk beroep droomde je als kind? Wat trok je hier toen aan aan?
What profession did you dream of becoming as a child? What did you like about it?
58 Als je getuige zou kunnen zijn van elke gebeurtenis in het verleden, heden of toekomst, welke zou het dan zijn?
If you could witness any event in the past, present or future, what would it be?
59 Als je elke willekeurige fictieve persoon zou kunnen zijn, wie zou je dan kiezen?
If you could be any fictional person, who would you choose?
60 Als je uit iedereen in de wereld kon kiezen, met wie zou je dan uit eten willen?
If you could choose from anyone in the world, who would you like to have dinner with?
61 Heb je liever een privévliegtuig of een privé-eiland? En waarom?
Would you rather have a private plane or a private island? And why?
21
Appendix B: Diagnoses and antipsychotic use of
patients with schizophrenia spectrum disorder
Table B1. Specific diagnoses and antipsychotic use of patients with schizophrenia spectrum disorder SZ n=26 Medication n, mg/day Haloperidol 1, 5 Paliperidone 3, 5.59 ± 5.58 Quetiapine 2, 450 ± 212 Aripiprazole 8, 10.6 ± 3.84 Olanzapine 3, 12.5 ± 4.33 Clozapine 4, 388 ± 155 Amisulpride 1, 800 No antipsychotics 4, - Chlorpromazine equivalent a 21, 232.2 ± 145.4 Diagnosis, DSM-code n Schizophrenia, 295.90 5
Schizophrenia – Paranoid type, 295.90 3
Schizophreniform, 295.40 2
Schizoaffective. 295.70 3
Schizoaffective – Bipolar type, 295.7A 1
Psychosis NOS, 298.9 12
Values of medications are mean ± standard deviation. a.22 patients of the SZ-group used differing doses of one of the following
antipsychotics: quetiapine, aripiprazole, olanzapine, clozapine, haloperidol, paliperidone. Doses of these antipsychotics are converted to a chlorpromazine equivalent. One participant used 800mg/d amisulpride, which cannot be converted (Leucht et al., 2014) and thus has not been taken into the calculation. n = sample size; SZ = participants with schizophrenia spectrum disorder
