Assessing non‐right‐handedness and atypical cerebral lateralisation as predictors of paediatric mental health difficulties

Research utilising handedness as a proxy for atypical language lateralisation has invoked the latter to explain increased mental health difficulties in left‐/mixed‐handed children. The current study investigated unique associations between handedness and language lateralisation, handedness and mental health, and language lateralisation and mental health, in children, to elucidate the role of cerebral lateralisation in paediatric mental health. Participants were N = 64 (34 females [52%]; MAge = 8.56 years; SDAge = 1.33; aged 6–12 years) typically developing children. Hand preference was assessed via a reaching task, language lateralisation was assessed using functional transcranial Doppler ultrasonography (fTCD) during an expressive language task, and mental health was assessed with the Strengths and Difficulties Questionnaire. As hypothesised, leftward hand preference predicted increased general mental health issues in children, with a strong relationship noted between leftward hand preference and the emotional symptoms subscale. Contrary to expectation, no relationship was found between direction of language lateralisation and general mental health issues, although exploratory analyses of subscales showed rightward lateralisation to predict conduct problems. Hand preference and direction of language lateralisation were also not significantly associated. The relatively weak relationship between manual and language laterality coupled with discrepancy regarding the predictive scope of each phenotype (i.e., hand preference predicts overall mental health, whereas language laterality predicts only conduct problems) suggests independent developmental pathways for these phenotypes. The role of manual laterality in paediatric mental health warrants further investigation utilising a neuroimaging method with higher spatial resolution.

and direction of language lateralisation were also not significantly associated.The relatively weak relationship between manual and language laterality coupled with discrepancy regarding the predictive scope of each phenotype (i.e., hand preference predicts overall mental health, whereas language laterality predicts only conduct problems) suggests independent developmental pathways for these phenotypes.The role of manual laterality in paediatric mental health warrants further investigation utilising a neuroimaging method with higher spatial resolution.
K E Y W O R D S fTCD, hand preference, language, paediatric, transcranial Doppler

| INTRODUCTION
The human brain exhibits a degree of asymmetry in its typical structure and activation patterns (Davidson & Hugdahl, 1995).Lateralisation of cerebral activity for certain neuropsychological functions, such as hand preference and language processing, is thought to be the evolutionary result of hemispheric specialisation (Toga & Thompson, 2003).A recent meta-analysis estimated that 89.4% of the world's population are right-handed (Papadatou-Pastou et al., 2020), which is shown to reflect laterality towards the contralateral (i.e., left hemisphere) motor cortex due to the decussation of the pyramidal tract (Dassonville et al., 1997).In addition, speech production and language comprehension are typically localised towards the left hemisphere (Rentería, 2012).
A relationship between non-right-handedness (i.e., left-or mixed-handedness) and atypical lateralisation of language function has been frequently reported (Knecht et al., 2000;Rasmussen & Milner, 1977;Szaflarski et al., 2012).This is characterised by approximately 30-39% of non-right-handed individuals exhibiting bilateral or right-lateralised language function, compared with only 10% of right-handers (Clarke et al., 2016).In the past, theoretical explanations for this association have alluded to a common neurodevelopmental aetiology for atypical hand-motor and language lateralisation, involving an underlying mechanism of genetic, biological, and/or sociocultural origin responsible for influencing the development of several lateralised functions in the brain (Annett, 1972;Geschwind & Galaburda, 1985).For instance, Annett (1972) proposed the existence of a 'right-shift' gene responsible for guiding the typical development of certain lateralised functions (e.g., language and handedness), and whose absence may therefore result in deviation from typical cerebral asymmetry.Other frameworks regarding the epi/genetic pathways underpinning lateralised functions included the Geschwind-Galaburda Hypothesis (GGH), which posits that, during foetal development, a genetic predisposition to testosterone sensitivity may impede maturation of the left hemisphere while potentially enhancing contralateral development (Geschwind & Galaburda, 1985).
However, findings from genome-wide association studies have identified a multitude of genetic variants associated with handedness (Cuellar-Partida et al., 2021), precluding the notion that a single gene governs the shifting of commonly lateralised functions (i.e., evidence against Annett's right-shift theory).Similarly, while the GGH was an admirable attempt to integrate endocrinological and neurobiological processes within a theoretical framework to explain certain phenotypes, from handedness and cerebral lateralisation, to autism and dyslexia, as well as to immune/autoimmune dysfunction, the model itself has received limited empirical support (Bryden et al., 1994).
Modern theories regarding the development of cerebral lateralisation move away from the notion of a critical laterality mechanism responsible for shifting commonly lateralised functions, such as handedness and language, and instead propose separable ontogenetic pathways (Bishop, 2013;Schmitz et al., 2017).However, there remain empirical findings that are unexplained by such notions, such as the capacity for handedness to predict a wide range of mental health dysfunction.Several studies have observed associations between nonright-handedness and a range of mental health outcomes in typically developing samples (Rodriguez et al., 2010;Rodriguez & Waldenström, 2008;van der Hoorn et al., 2010).One study found mixed-handed 8-year-olds to exhibit increased language difficulties, attention-deficit/hyperactivity disorder (ADHD) symptoms, and likelihood of psychiatric disturbance compared to their right-handed counterparts; both language difficulties and ADHD symptoms were reassessed at age 16 and were found to persist (Rodriguez et al., 2010).Another study utilising a sample of early adolescents noted positive associations between non-right-handedness and severity of depressive symptoms, social problems and thought problems, when compared with right-handers (van der Hoorn et al., 2010).These findings are further supported by Rodriguez and Waldenström (2008), who found links between mixed-handedness and a composite measure of conduct problems, peer problems, emotional symptoms and inattention/hyperactivity in a sample of 6-year-olds.
It seems somewhat puzzling how a phenotype such as handedness could reliably predict psychosocial and affective problems that are, presumably, rooted in neural circuitry that has limited overlap with basic hand-motor function.Notably, in adult samples, Cuellar-Partida et al. (2021) noted genetic correlation between left-handedness and neuropsychiatric traits (i.e., schizophrenia and bipolar disorder), leaving open the possibility of ontogenetic overlap between them.Other explanations for the association between handedness and mental health disorders in children emphasise nongenetic influences, including the notion that handedness reflects some unitary construct of atypical brain laterality that induces a wide range of behavioural dysfunction, and where atypical laterality is itself the result of prenatal neurological insults or other obstetric complications (Rodriguez et al., 2010;Rodriguez & Waldenström, 2008).In this case, it would be expected that not only should nonright-handedness and other expressions of cerebral laterality, such as atypical language lateralisation, be strongly correlated, but that they also predict a similar scope of mental health difficulties.
While no studies have examined associations between atypical language lateralisation and general mental health in non-clinical samples, to our knowledge, direction of language lateralisationoperationalised through various methods including electroencephalography (EEG), structural and functional magnetic resonance imaging (MRI and fMRI), and dichotic listening-has been linked to several psychiatric/developmental disorders.For those diagnosed with ADHD, research utilising EEG has shown abnormal alpha wave (Hale et al., 2009) and beta wave (Hale et al., 2010) asymmetries over core language regions, as compared with healthy controls.In addition, a longitudinal study using MRI to assess developmental trajectories of various cortical regions found overall slower rates of structural lateralisation towards the left inferior frontal cortex in children with ADHD, relative to controls (Shaw et al., 2009).
In clinical depression, the presence of atypical language lateralisation has been investigated using dichotic listening paradigms, with depressed adult patients exhibiting increased left-ear (right-hemisphere) advantage for different words presented to both ears, as compared with healthy controls (Bruder et al., 1999;Johnson & Crockett, 1982).In addition, another study noted an association between emotional withdrawal/ psychomotor retardation symptomatology and leftward ear advantage within a depressed group (Wale & Carr, 1990).However, findings appear to be mixed, with newer research showing no difference in ear advantage for depressed participants versus controls (Bruder et al., 2016;Hugdahl et al., 2003), and one study finding the reversed effect (i.e., increased left-hemisphere advantage) for depressed women and a null effect for men, as compared with healthy female and male control groups, respectively (Bruder et al., 2004).
Perhaps more consistently than other forms of psychopathology, atypical language lateralisation has also been observed in schizophrenia: an early meta-analysis of 58 studies investigating handedness, dichotic listening, and/or structural imaging data in schizophrenia concluded that there was strong evidence of increased atypical cerebral lateralisation for this patient population (Sommer et al., 2001).In particular, nonright-handedness was significantly more prevalent for individuals with schizophrenia compared with healthy controls, and congruently, dichotic listening performance was characterised by significantly greater righthemisphere bias for consonant-vowel or fused-word tasks.Sommer et al. (2001) also noted structural asymmetries in core language areas, such as the planum temporale and Sylvian fissure.
Further, studies on laterality and schizophrenia have shown consistent genetic overlap between these two phenotypes, although such research typically treats schizophrenia as a homogeneous trait rather than utilising a symptom-specific approach, the latter of which may improve understanding of the aetiology of schizophrenia, and subsequently, the precise role of laterality (Ocklenburg et al., 2015).That is, researchers are yet to elucidate whether genetic overlap between schizophrenia and language lateralisation is indeed disorder-specific, or instead, whether language lateralisation is associated more strongly with language-mediated symptom clusters (e.g., verbal hallucinations and disorganised speech).This idea seems to be partially supported by meta-analytic evidence from dichotic listening studies, where atypical language lateralisation was identified as a weak marker for schizophrenia itself but exhibited a much stronger effect among those who endorsed auditory hallucinations (Ocklenburg et al., 2013).
In any case, it appears that non-right-handedness, and potentially atypical language lateralisation, may predict a wide range of symptomatology and disorders.As such, the proposal that both atypical language lateralisation and atypical handedness may be caused by a common underlying mechanism seems reasonable, with again, a number of researchers having alluded to such a mechanism and implicated its atypical development as a driver of the relationships observed between atypical handedness and mental health observed in paediatric samples (Rodriguez et al., 2010;Rodriguez & Waldenström, 2008;van der Hoorn et al., 2010).
However, others have questioned the correlative strength between handedness and language lateralisation (Groen et al., 2013;Somers et al., 2015;Wilson & Bishop, 2018), and as such, the meaningfulness of such an association in the context of a transdiagnostic aetiology is contentious.One study compared associations between several widely used handedness measures and lateralisation for language in children (Groen et al., 2013).While Groen et al. (2013) did find significant associations between continuous measures of handedness and language lateralisation, the amount of variance in language lateralisation that was explained by each handedness measure varied from 8% to 16%, suggesting these relationships to be relatively weak.Further, Groen et al. posited their findings as evidence that lateralised functions, such as those relating to hand-motor control and language function, follow independent neurodevelopmental pathways.Other studies have reported similarly weak effects between handedness and language lateralisation (Badzakova-Trajkov et al., 2010;Gonzalez & Goodale, 2009;Hodgson et al., 2021), with some even reporting the relationship as nonsignificant (Mazoyer et al., 2014;Wilson & Bishop, 2018).Again, this perspective aligns with others who have proposed largely separable ontogenetic pathways for handedness and language lateralisation (Bishop, 2013;Ocklenburg et al., 2014;Schmitz et al., 2017).As such, the notion of a robust association between non-right-handedness and atypical language lateralisation derived from common neurodevelopmental factors is under dispute.
The current study assessed hand preference using a reaching task, language lateralisation using functional transcranial Doppler ultrasonography (fTCD) during an expressive language task and general mental health using a parental questionnaire, in a typically developing child sample, to facilitate three objectives.The first objective was to examine previously reported links between handedness and general mental health in typically developing children.The second was to investigate the relationship between language lateralisation and general mental health in children.The third objective was to characterise the strength of association between handedness and cerebral lateralisation of language function to contextualise research utilising non-right-handedness as a marker for atypical cerebral lateralisation (Rodriguez et al., 2010;Rodriguez & Waldenström, 2008;van der Hoorn et al., 2010).
It was hypothesised that: (H 1 ) atypical hand preference would uniquely predict increased mental health difficulties, as observed in past literature (Rodriguez et al., 2010;Rodriguez & Waldenström, 2008;van der Hoorn et al., 2010); (H 2 ) atypical lateralisation of language would predict increased mental health problems; and (H 3 ) hand preference would only weakly predict lateralisation of language function.

| METHOD
This study received ethical approval from the University of Western Australia's Human Research Ethics Office (Reference: RA/4/1/5926).Research was completed in accordance with the Helsinki Declaration.

| Participants
Archival data from N = 86 participants of the Children's Activity Program (2014Program ( -2016) ) at the University of Western Australia were accessed.Participants for this program were convenience sampled from primary schools (Year 1-Year 7) around the metropolitan area of Perth, Western Australia.Participants' data were excluded from the current study if their parents reported past or current diagnoses of neurodevelopmental conditions such as ADHD, ASD (or Asperger's syndrome), or any other form of developmental, auditory processing, learning or speech disorder; nine cases (7 males, 2 females) were excluded from the study based on these criteria.Of the remaining 77 participants, n = 13 were removed because of unreliable fTCD data (i.e., <9 epochs available).The remaining sample comprised N = 64 typically developing children (34 females [53%]; M Age = 8.60 years; SD Age = 1.33; aged 6-12 years).This sample size exceeded the benchmark of N = 58 set by a power calculation (G*Power; Erdfelder et al., 1996) for a linear regression based on moderate effect sizes that have been previously reported between atypical language lateralisation and mental health issues (Dollfus et al., 2005;Hale et al., 2005;Whitehouse & Bishop, 2008).
Prior to commencement of the program, participants and their primary caregivers provided assent and consent, respectively, and were given the option to withdraw at any time during testing.

| Apparatus
Blood flow velocities in the left and right middle cerebral arteries were recorded using the Doppler-Box™ (DWL, Compumedics, Germany).The Doppler-Box™ received input from two 2-MHz transducer probes affixed to an adjustable headset and situated bilaterally over participants' temporal windows.During the language task, stimuli were presented using a PC running Presentation software (Neurobehavioral Systems).The raw signal from each probe was downsampled at a rate of 100 Hz.

| Quantification of Hand Preference
Hand preference was evaluated using a reaching task developed specifically for assessment in children (Bishop et al., 1996).In a validation study by Musalek et al. (2016), the Quantification of Hand Preference (QHP) was appraised using a sample of N = 376 children aged 8-10 years.Musalek et al. (2016) reported high test-retest reliability (.80), convergent validity (.89) and classification accuracies (97% for right-handers, 90% for left-handers) when compared alongside popular measures of hand preference such as the Edinburgh Handedness Inventory (Oldfield, 1971).
This task involved seven stacks of playing cards placed in a semi-circular arrangement at 30-degree intervals.Each stack was equidistant to the child at reaching distance.Participants stood at the mid-point of the base of the semi-circle and were instructed to pick up each card and place it into a marked rectangular region positioned at the midpoint.A hand preference quotient (HPQ) was computed by dividing total right-hand reaches by total reaches and subtracting 0.5.HPQ scores ranged from 0.5 to À0.5 indicating exclusive right-hand preference to exclusive left-hand preference, respectively.
The SDQ is a 25-item questionnaire containing four 'difficulty' subscales pertaining to emotional symptoms, conduct problems, inattention/hyperactivity and peer problems.The fifth subscale assessing prosocial behaviours was not used for the present study.All subscales consisted of five items, scored on a 3-point Likert scale ranging from 0 (not true) to 2 (certainly true).A composite total difficulty score was computed as the sum of the four difficulty subscale totals.Higher scores on each of the difficulty subscales indicated more problematic behaviour and/or symptoms.

| Animation description paradigm
Cerebral lateralisation for language was assessed using an animation description paradigm taken from Bishop et al. (2009); this is a valid and reliable method of assessing language lateralisation based on correlation (.68) with a 'gold standard' silent word generation paradigm (Bishop et al., 2009), as well as high split-half reliability (.84-.88) noted across several child samples (Bishop et al., 2009(Bishop et al., , 2010;;Wilson & Bishop, 2018).The task is robust to varying literacy levels, can be completed in 15 min, and is generally found to be child-friendly (Groen et al., 2012).
The task comprised 30, 30-s trials, with each trial representing one epoch.Trials consisted of three phases: a 12-s video clip (presenting a short story segment with nonverbal audio), a 10-s expressive language period and an 8-s resting period (see Figure 1).After silently viewing the video clip, participants were prompted with an onscreen cue denoting the start of the expressive language period, at which point they verbally described the clip.Following this, participants were asked to rest quietly while cerebral blood flow velocity returned to baseline levels.The prespeaking phase of the trial was selected as a baseline given that no evidence of hemispheric laterality has been reported during this period in previous studies (e.g., Groen et al., 2013).Participants were required to minimise gross motor movement throughout the task to avoid distorting cerebral blood flow velocity measurement.

| Data processing
Raw fTCD data were processed offline using the software package dopOSCCI (Badcock, Holt, et al., 2012;Badcock et al., 2018) MATLAB toolbox (R2022b, Mathworks, Natick, MA, USA).The dopOSCCI v3.6.6 software (available from: https://github.com/nicalbee/dopStep)was used to normalise signals from left and right channels, correct for signal dropout (signals beyond 3-4 standard deviations from the mean affecting less than 5% of the data were set to the mean) and remove the effects of cardiac cycles (using a linear correction).Relative to each event marker, data were epoched from À12 to 14 s and baseline-adjusted from À12 to 0 s to correct for breathing and other movements.Epochs were excluded based on extreme values beyond 50%-150% activation, or a leftright channel difference of 20 times the inter-quartile range of the difference where more than 1% of the data were affected.
The fTCD recording equipment received a trigger from the Presentation software at commencement of the expressive language phase of each trial, marking the start of an epoch.A period of interest was selected based on past research (Bishop et al., 2009) and defined as 4-14 s following each trigger.Activation differences within the period of interest were averaged and used to compute a language laterality index (LI) for each epoch.Mean LIs were then averaged across all epochs for each participant to produce a final LI, with positive and negative LIs reflecting left-and right-lateralisation, respectively.

| Statistical analyses
To ensure reliability of the LIs, split-half reliability analyses were conducted on odd and even epochs from the fTCD paradigm used to generate LIs (i.e., animation description paradigm).Reliability coefficients were computed using the Spearman-Brown method.
For descriptive purposes, participants were categorised as left-lateralised, right-lateralised or bilateral for language.This was achieved by calculating standard errors across epochs and conducting one-sample twotailed t-tests on each participant's mean LI to determine whether it was significantly different to zero, as done in previous studies (Badcock, Nye, & Bishop, 2012;Bishop et al., 2009;Groen et al., 2013).Significant positive tvalues indicated left-lateralised language function, significant negative t-values indicated right-lateralisation, and nonsignificant t-values indicated bilateral activation.In addition, given that direction of lateralisation has been found to vary with age (Szaflarski et al., 2006), a Pearson correlation was conducted between LI and age to check for this confound.
To test unique effects of both hand preference (H 1 ) and direction of lateralisation (H 2 ) on general mental health difficulties, HPQ and LI were entered as predictors of SDQ total difficulty score in a multiple linear regression model; age and gender were controlled for, given past literature demonstrating covariation with SDQ scores (Vella et al., 2019).Squared semi-partial correlations were computed for HPQ and LI in each model to indicate the proportion of unique variance explained by either variable.Exploratory analyses were similarly conducted for each SDQ difficulty subscale; these were not linked to hypotheses, and therefore, no correction for multiple comparison was performed.To test the effect of hand preference on direction of lateralisation (H 3 ), a linear regression analysis was conducted with HPQ as a predictor of LI, again controlling for age and gender.A coefficient of determination was computed to discern shared variance for each predictor in each model.
Regression models were checked for assumptions of normality, linearity, homoscedasticity and influential cases.Standardised residual distributions were assessed for normality by dividing skew and kurtosis statistics of each distribution by their corresponding standard errors and comparing the quotients to a critical threshold of j3.29j (Kim, 2013).Scatterplots of independent versus dependent variables corresponding to each regression model were deemed sufficiently linear based on an absence of curvilinearity noted upon visual inspection.Homoscedasticity was assessed by visually inspecting a scatterplot of standardised residuals and standardised predicted values.To identify influential cases, Cook's distances were computed and compared with a threshold value of 1 (Field, 2013).All analyses were conducted using IBM SPSS v26 software.
Unreliable p-values for heteroscedastic regression models were adjusted using a method detailed in Hayes and Cai (2007).This method generates betaweight standard errors independently of the homoscedasticity assumption of multiple regression, allowing F I G U R E 1 A single 30-s trial of the animation description paradigm with three phases depicted.Participants silently viewed a 12-s video clip, vocally described the clip from 12 to 22 s and were reminded to stay silent for the rest period from 22 to 30 s.
reliable p-values to be calculated.To achieve this, syntax was taken directly from Hayes and Cai (2007) and ran with default configuration using IBM SPSS software.

| Data preparation
Split-half reliability was found to be good (ρ = .80,p < .001)based on analysis of odd and even epochs from the fTCD paradigm used to compute mean LIs.In addition, regression models satisfied relevant assumptions, including absence of outliers, linearity and normality of residuals; however, two regression models were found to be heteroscedastic, namely, those with emotional symptoms and conduct problems as the dependent variables; these models were adjusted accordingly using the aforementioned Hayes and Cai (2007) method.

| Descriptive statistics
Summary statistics regarding HPQ, LI and mental health difficulty score are presented in Table 1.For descriptive purposes, t-tests were computed for each participant's mean LI to determine whether it was significantly different to zero: 63% (n = 40) were significantly leftlateralised, 23% (n = 15) right-lateralised and 14% (n = 9) showed bilateral activation.There was a significant difference between the number of left-lateralised versus bilateral and right-lateralised participants (χ 2 [1, 64] = 4.00, p < .05).
Baseline-corrected grand-averaged waveforms for the animation description paradigm can be seen in Figure 2. Waveforms were plotted for cerebral blood flow velocity changes averaged across all participants in both left and right channels, as well as the difference between channels.All three waveforms were plotted over the length of one animation description trial running from À12 to 18 s, where the zero-point marks the end of baseline and onset of the expressive language period.Peak amplitude of the difference wave appears to occur within the period of interest, agreeing with past research (Bishop et al., 2009) and confirming the period of interest parameters used for LI extraction.

| Hand preference, language lateralisation and SDQ total difficulty score
A multiple linear regression model showed HPQ to significantly and negatively predict SDQ total difficulty score while controlling for the effects of LI, age and gender (β = À0.38,SE = 0.12, p = .003,sr 2 = .13).This implied a unique association between leftward hand preference and increased parent-reported overall mental health difficulties on the SDQ (H 1 ).In contrast, LI did not significantly predict SDQ total difficulty score while controlling for the effects of HPQ, age and gender (β = 0.19, SE = 0.12, p = .112,sr 2 = .03),suggesting that direction of language lateralisation was not uniquely associated with parent-reported mental health difficulties on the SDQ, addressing H 2 .A follow-up analysis with HPQ removed as a predictor yielded similar results, confirming no significant relationship between language lateralisation on SDQ total difficulty score even when the effects of HPQ were uncontrolled (β = 0.12, SE = 0.13, p = .334,sr 2 = .01).
Regarding H 3 , a regression model predicting LI from HPQ, controlling for age and gender, showed a smallmedium effect that failed to reach statistical significance (β = 0.19, SE = 0.12, p = .144,R 2 = .04),suggesting no significant association between hand preference and lateralisation of language.A scatterplot between LI and HPQ can be seen in Figure 3. Note: N = 64.Abbreviation: CI, confidence interval.

| Analysis of SDQ subscales
The results of four exploratory linear models pertaining to the unique effects of HPQ and LI on each SDQ subscale, while also controlling for age and gender, as well as corresponding squared semi-partial correlations, are seen in Table 2.A significant negative association between HPQ and the emotional symptoms subscale was observed.This result indicates a unique relationship between leftward F I G U R E 3 Scatterplot of hand preference quotient (HPQ) versus language laterality index (LI).
F I G U R E 2 Grand-averaged cerebral blood flow velocity change elicited during the expressive language task for the left (blue dashed line) and right (solid red line) channels, and their differential (solid black line) over time, from 12 s prior to the onset of the speech cue (time 0) to 18-s post-speech cue.Baseline and period of interest are depicted in grey, and the shaded areas represent the 95% confidence intervals.
hand preference and emotional symptoms as reported by parents on the SDQ, which was characterised by a medium-large effect.HPQ was not found to be significantly associated with the hyperactivity, conduct problems, or peer problems subscales.
Analysis of SDQ subscales also revealed a significant positive relationship between LI and the conduct problems subscale, suggesting that rightward language lateralisation increased with the degree of conduct problems reported by parents.No effect was found between LI or any other subscales of the SDQ.

| DISCUSSION
The aims of the current study were to examine both hand preference and language lateralisation in terms of their association with general mental health issues in a typically developing sample.A positive association was observed between hand preference and parent-reported mental health issues in children (i.e., right-handedness was associated with better outcomes; H 1 was supported).In contrast, the relationship between direction of lateralisation for language and mental health failed to reach statistical significance (i.e., H 2 was not supported), although exploratory analysis of subscales revealed an association between rightward lateralisation and the conduct problems subscale of the SDQ.As expected, a weak and statistically nonsignificant association between hand preference and direction of lateralisation was observed (H 3 ).

| Leftward hand preference and mental health
Support for H 1 was garnered from a multiple linear regression analysis wherein a continuous measure of hand preference derived from a reaching task was found to positively predict mental health problems, independent of direction of language lateralisation, age, and gender.That is, as participants exhibited reduced right-hand preference (or greater left-hand preference), they were also reported to experience increased general mental health problems as rated by parents on the SDQ.Regarding SDQ subscales, additional analyses revealed a significant association (that survived Bonferroni correction) between leftward hand preference and increased anxious/ depressive symptoms, with a moderate effect size.That emotional symptoms would be a key domain linked to atypical handedness is consistent with a recent genomewide association study evidencing overlap between genetic loci associated with both left-handedness, as well as a plethora of psychiatric phenotypes, most of which revolved around neuroticism and mood disturbances (Wiberg et al., 2019).
These findings are interesting in suggesting an important link between hand preference and child mental health, and empirically, they are consistent with research noting links between non-right-handedness and total difficulty score on the SDQ (Rodriguez & Waldenström, 2008), as well as other composite measures of paediatric mental health such as the Rutter Behaviour Scale (Rodriguez et al., 2010).Rodriguez and Waldenström (2008) found mixed-handed children aged 5-6 to exhibit higher total difficulty scores on the SDQ compared with a right-handed group.Similarly, in a study by Rodriguez et al. (2010), mixed-handed 8-year-olds were significantly more likely to score within the probable psychiatric disturbance range of the Rutter Behaviour Scale compared with their right-handed counterparts.When comparing a group of right-and non-right-handed adolescents aged 12-15 years, van der Hoorn et al. ( 2010) did not find a significant association between nonright-handedness and total problem score of the Child Behaviour Checklist, although selected subscales did Note: N = 64.β = standardised beta coefficient; standard error = 0.13 for all terms.sr 2 = squared semi-partial correlation.All models were adjusted for age and gender.Heteroscedastic-consistent standard errors were used to generate p-values for models involving emotional symptoms and conduct problems as outcome variables.
exhibit significant associations, namely, thought problems, social problems and withdrawal/depression.Notably, significant findings from Rodriguez and Waldenström (2008) and Rodriguez et al. (2010) were driven by mixed-handed versus right-handed group comparison, whereas a relationship between nonright-handedness and mental health in van der Hoorn et al.'s (2010) study largely reflects the effects of lefthandedness given a small proportion of mixed-handers (1.1%) compared with left-handers (13.4%) in their sample.Thus, some doubt exists as to whether mixedhandedness, left-handedness or both are implicated in mental health issues.Indeed, Cuellar-Partida et al. ( 2021) reported weak genetic overlap of left-and mixedhandedness, possibly suggesting somewhat independent mechanisms between the two constructs and, perhaps, differential relevance to the development of psychopathology.In addition, previous findings regarding handedness and mental health are confounded by the arbitrarily selected thresholds for left-, right-and mixed-handed groups: this method often gives rise to unreliable classification of handedness subtypes (Papadatou-Pastou et al., 2020) as well as spurious findings in psychiatric research (Bishop, 1990).As such, future research would benefit from the employment of neuroimaging to directly assess lateralisation in the hand motor cortex.Classification of mixed-handedness could then be based on a motor laterality index not differing significantly from zero, which is a protocol that adheres to a widely adopted a priori method for categorising laterality indices (Badcock, Nye, & Bishop, 2012;Bishop et al., 2009;Groen et al., 2013;Hodgson et al., 2021;Lohmann et al., 2005).Ultimately, this would facilitate research aiming to delineate whether a certain mode of handedness (i.e., left versus mixed) might be more salient in terms of a link to paediatric mental health, while also enabling more rigorous investigation into the neural underpinnings of this relationship.

| Atypical language lateralisation and mental health
Regarding H 2 , support was not found given that a nonsignificant association was noted between LI and total difficulty score on the SDQ.These results imply that the direction of language lateralisation did not significantly predict general mental health difficulties in children as reported by their parents using the SDQ.To our knowledge, no other studies have examined an association between language lateralisation and a general measure of mental health such as the SDQ; thus, direct comparison of this result to past literature is not possible, although our results run counter to the notion that atypical cerebral lateralisation underpins the relationship between handedness and mental health issues, as purported previously (Rodriguez et al., 2010;Rodriguez & Waldenström, 2008;van der Hoorn et al., 2010).In addition, null effects of language laterality on SDQ subscales pertaining to emotional symptoms, hyperactivity/inattention, and peer problems may suggest that implications surrounding research observing atypical language lateralisation in ADHD (Hale et al., 2009(Hale et al., , 2010;;Shaw et al., 2009), depression (Bruder et al., 1999;Johnson & Crockett, 1982;Wale & Carr, 1990) and schizophrenia (Ocklenburg et al., 2013;Sommer et al., 2001) may be specific to those disorders.That is, aberrant language processing may result from, or exhibit an independent dynamic with, hyperactive/inattentive, depressive, or schizotypal psychopathology in each respective disorder (e.g., direct involvement of language with auditory hallucinations in schizophrenia; Ocklenburg et al., 2013), rather than implying a generalised neuropsychological dysfunction and transdiagnostic relevance of atypical language lateralisation.Further, our research agrees with the notion posited by Berretz et al. (2020), in that although the described forms of psychopathology correlate well together, genetically (.40;Anttila et al., 2018), this shared variance does not overlap with genes associated with hemispheric asymmetry.In addition, while results from the ENIGMA Laterality working group confirmed associations between structural asymmetry across several cortical and subcortical regions and psychiatric disorders, such as ADHD, ODD, and ASD, the effect sizes were small (Kong et al., 2022) and, again, implicate several other factors in their aetiologies above laterality itself.Congruently, in the current study, effect sizes between language lateralisation and facets of mental health were also quite small but may have been detectable with a large multi-database such as that of ENIGMA-this is speculation, however, as structural laterality has shown a poor ability to track functional and behavioural measures of laterality (Kong et al., 2022).
On another note, a positive relationship was found between language laterality and conduct problems, where atypical (rightward) lateralisation predicted higher levels of conduct problems as reported by parents on the SDQ, independently of hand preference, age, and gender.Notably, the relationship did not survive Bonferroni correction ( p = .06),although the effect was moderate in size.Indeed, the finding is in line with research demonstrating links between reduced lateralisation for adolescent psychopaths (Raine et al., 1990) as well as associations between antisocial behaviour and atypical development of neural specialisation for verbal processing (Raine & Yang, 2006;Raine et al., 2000).Atypical specialisation for verbal processing has been linked, in turn, to language difficulties (Badcock et al., 2018), and indeed, language disorders show high prevalence among those involved in the justice system, ranging from 40% to 70% compared with a base rate of 10% observed in the general population (Norbury et al., 2016).Together, these findings suggest that atypical language development may confer considerable risk for rule-breaking behaviour.Importantly, our finding was observed in a typically developing sample without any reported functional impairment by parents; as such, atypical language lateralisation not only joins literature in implicating core language regions in conduct problems among antisocial cohorts but may also serve as marker for detecting children at risk of future conduct issues during periods where intervention is most beneficial (i.e., where conduct problems are still subthreshold and confer less functional impairment).Future research would benefit from using a longitudinal approach, where atypical language lateralisation is assessed as a risk factor for later development of conduct and/or antisocial personality disorder, with measures of language function at each time point to elucidate the precise nature and role of linguistic deficits in the development of these disorders.

| Atypical language lateralisation and hand preference
Congruent with evidence against the notion of a shared laterality mechanism are the results pertaining to H 3 ; namely, that hand preference and language laterality were weakly but nonsignificantly related.While several studies have demonstrated a significant association between manual and language laterality (Badzakova-Trajkov et al., 2010;Gonzalez & Goodale, 2009;Groen et al., 2013;Hodgson et al., 2021;Knecht et al., 2000;Somers et al., 2015), observations to the contrary are relatively common (Bishop et al., 2009;Elias & Bryden, 1998;Mazoyer et al., 2014;Propper et al., 2010;Wilson & Bishop, 2018).The current study lacked sufficient statistical power to detect a small effect, powered for a medium effect size.However, it should be noted that research with larger samples have also failed to replicate a significant association between language lateralisation and handedness (e.g., Wilson & Bishop, 2018).
Nevertheless, the magnitude of shared variance observed between language lateralisation and handedness in the current study-approximately 4%-is comparable with the low-moderate effect sizes typically reported in past research (Badzakova-Trajkov et al., 2010;Gonzalez & Goodale, 2009;Hodgson et al., 2021).Indeed, the current result coincides with research utilising a relatively more comprehensive methodology, with Groen et al. (2013) finding different measures of hand preference and skill to explain 8%-16% of the variance in language lateralisation.Despite the larger effect size noted in Groen et al. (2013) compared with the current study, they posit that even 16% shared variance between handedness and language lateralisation is insufficient to support a common neurodevelopmental aetiology for functional hand-motor and language laterality.Thus, together with Groen et al. (2013), the findings of the current study contribute to literature suggesting that deviation from typical hand-motor and language lateralisation patterns does not arise from a global laterality mechanism (Rosch et al., 2012;Whitehouse & Bishop, 2009).As advocated by others, manual and language laterality may instead exist as complex phenotypes with limited overlap regarding etiological factors (Johnstone et al., 2020;Ocklenburg et al., 2014;Schmitz et al., 2017).

| Limitations and future research
A potential limitation of the current study relates to our finding that only 63% of the sample exhibited leftlateralisation of language.This contrasts with a consensus among past literature delineating that approximately 90% of typically functioning individuals exhibit left hemisphere dominance for language (Clarke et al., 2016).Although a scatterplot of hand preference and language lateralisation in the current study appears to resemble the typical profile, a weakened linear association appears because of the presence of several individuals showing both right-lateralisation and right-hand preference.An explanation may relate to the employment of visuospatial working memory during the descriptive phases (periods of interest) of the animation task, as this function has been identified as right-lateralised in past research (Badzakova-Trajkov et al., 2010;Groen et al., 2012;Jansen et al., 2004).Given that the middle cerebral arteries perfuse regions in the temporoparietal cortex related to visuospatial working memory, engaging this function may have led to skewed blood flow velocity differentials based on corresponding interindividual variability, although this is difficult to confirm given the low spatial resolution of fTCD.Future research would benefit from utilising a functional imaging modality with higher spatial resolution, such as fMRI or functional near-infrared spectroscopy (fNIRS).
Alternatively, a disproportionate number of leftlateralised children in our sample may suggest that lateralisation had not yet reached full development; while age was not associated with direction of lateralisation in our sample, there is some evidence that lateralisation continues to develop throughout childhood (i.e., age 13), with significant activation noted in both hemispheres up to this age (Olulade et al., 2020).In any case, more research is needed to determine what constitutes a typical base rate of left-lateralisation in paediatric samples.
Another limitation of the current study involves the primary measure of mental health in the current study, namely, the SDQ.Previous research has shown that although reliability and validity of SDQ total score range from satisfactory to excellent, the same cannot be said for its subscales, which typically show more variable convergent validity (Mieloo et al., 2012) and generally poorer reliability (Stone et al., 2010).As such, this may hamper interpretability of exploratory analyses concerning SDQ subscales.Future research should aim to investigate links between handedness and mental health subdomains that are operationalised by more comprehensive psychological measures.Replication with these measures will serve to provide more insight into the unique nature of the relationship observed between handedness and mental health.

| CONCLUSION
Most notably, our findings replicating a link between hand preference and overall mental health suggest that hand-motor lateralisation itself warrants further investigation as a salient variable in paediatric mental health, rather than merely serving as a proxy for lateralised hemispheric development.In addition, although language laterality predicted conduct problems in our sample, which may suggest a role of atypical language processing within this mode of psychopathology, there was no association found between language lateralisation and general mental health difficulties.Together with a weak effect size noted between hand preference and language laterality, these findings suggest that theoretical explanations for the link between non-right-handedness and mental health issues in children should be modified to accommodate separate neurodevelopmental aetiologies for atypical hand-motor and language lateralisation.
Summary statistics for laterality index (LI), hand preference quotient (HPQ), total difficulty score of the SDQ and individual SDQ subscales.
T A B L E 1 T A B L E 2 Results of multiple regression analyses and semi-partial correlations with hand preference quotient (HPQ) and laterality index (LI) as predictors of each SDQ subscale, controlling for age and gender.