Culture Shapes the Neural Functional Alteration and Its Development in Children With Dyslexia: A Cross‐Linguistic Meta‐Analysis

ABSTRACT

How culture shapes brain functional alteration and its development remains poorly understood in dyslexia (DD). Hereby, we conducted a meta‐analysis comparing brain activation in children across Chinese (logographic), English (deep alphabetic), and German (shallow alphabetic) writing systems. A total of 48 studies involving 1554 participants were included, consisting of 695 typically developing controls and 859 individuals with DD. Results revealed universal neurobiological signatures of DD characterized by convergent hypoactivation in the left fusiform gyrus and the inferior parietal lobule across languages. Language‐specific reductions were identified: the left inferior frontal gyrus in Chinese, the left inferior temporal gyrus in English, and the left inferior parietal lobule in German. Developmental meta‐regressions demonstrated that in deep orthographic languages (English, Chinese), older children showed more convergent inferior temporal gyrus hypoactivation, suggesting atypical ventral stream maturation. Conversely, younger DD children exhibited more convergent hypoactivation in the superior temporal gyrus in alphabetic languages, indicating delayed dorsal stream development. These findings provided novel cross‐linguistic evidence of how neural atypicalities in DD differentially develop across writing systems, highlighting dynamic interactions between linguistic characteristics and neurodevelopmental plasticity in DD.

Summary

• Convergent hypoactivation in left occipitotemporal (OT) and temporoparietal (TP) regions across Chinese, English, and German reveals language‐universal neural signatures of developmental dyslexia (DD).

• Language‐specific hypoactivation was identified in the three languages: the left inferior frontal gyrus (IFG) for Chinese, the left inferior temporal gyrus (ITG) for English, and the left inferior parietal lobule (IPL) for German, indicating cultural constraints on reading and DD.

• Functional deficits in DD follow a distinctive developmental tendency shifting from dorsal decoding pathways to ventral orthographic pathways, manifesting variably across languages due to the interplay of maturational and linguistic factors.

1. Introduction

Developmental dyslexia (DD), a neurodevelopmental disorder characterized by specific reading impairments, affects approximately 5%–10% of children globally (Stevenson et al. 1982). Cumulative evidence has consistently associated DD with dysfunction in the left‐hemispheric reading network (Pugh et al. 2000; Richlan 2012). The core deficit hypothesis has long dominated the theoretical frameworks, positing that DD originates from specific cognitive impairments, most notably phonological processing deficits, which manifest as corresponding functional abnormalities in neural circuitry (Ahissar 2007; Zuk et al. 2018). This perspective is exemplified by well‐documented phonological deficits and associated temporoparietal cortex dysfunction (Marchesotti et al. 2020; Turker et al. 2025). However, emerging evidence challenges this unidimensional view, suggesting that neural signatures of DD exhibit significant variability across cultural contexts (Yan et al. 2021) and age differences (Richlan et al. 2011), thereby necessitating a more nuanced neurodevelopmental framework.

From a cross‐linguistic perspective, two competing theoretical frameworks remain central to dyslexia research: the culture‐specific deficit hypothesis (Siok et al. 2004) and the universal cultural deficit hypothesis (Paulesu et al. 2001). Proponents of the culture‐specific deficit hypothesis argue that language is inherently embedded within its cultural context. Each language possesses a unique writing system and distinct linguistic rules, which collectively shape and constrain neural responses to printed scripts across different languages (Bolger et al. 2005; Tan et al. 2005). Notably, even for individuals acquiring multiple languages with highly similar writing systems, cultural contexts still exert an independent influence on the neural and cognitive processing (Liu and Cao 2016). Specifically, orthographic transparency critically shapes reading pathway engagement. In shallow orthographic languages, such as Italian, German, and Spanish, the dorsal reading pathway (i.e. the temporoparietal areas) is more involved due to the regular orthographic‐phonology‐correspondence. Conversely, the ventral reading pathway (i.e. the occipitotemporal areas) and the left inferior frontal gyrus (IFG), are more involved in deep orthography, such as English, due to the existence of irregular words and subsequently the whole word reading strategy (Paulesu et al. 2000). Chinese, a nonalphabetic language, is even more contrastive. It has no consistent grapheme‐phoneme‐correspondence and each character maps to a whole syllable. In the brain, the left IFG is more significantly involved in Chinese reading than alphabetic reading and the temporoparietal areas are more involved in alphabetic languages (Bolger et al. 2005; Tan et al. 2005). These language‐specific neural adaptations necessarily predict divergent neurobiological signatures of DD across writing systems. Within alphabetic writing systems, orthographic depth significantly modulates neurobiological manifestations of DD. A meta‐analysis revealed that individuals with DD in deep orthographies (e.g., English) exhibited more consistent hypoactivation in the left IFG pars triangularis, the left precuneus, and the right superior temporal gyrus compared to those in shallow orthographies (e.g., Italian, German, Spanish). Conversely, shallow orthographies demonstrated greater convergence of hypoactivation in the left fusiform gyrus, the left temporoparietal cortex, the left IFG pars orbitalis, and the left frontal operculum (Martin et al. 2016). In contrast, Chinese dyslexia exhibits distinct neural characteristics. Siok et al. (2004) identified unique impairments in Chinese children with DD, particularly within the left middle/IFG, a pattern diverging from the typical alphabetic profile characterized by diminished activation in the left temporoparietal and occipitotemporal regions (Martin et al. 2016; Yan et al. 2021). This Chinese‐specific finding has been corroborated by a recent multimodal meta‐analysis, which confirmed greater convergence of functional and structural abnormalities in the left middle/IFG among Chinese readers with DD relative to alphabetic language readers (Yan et al. 2021). Notably, DD in alphabetic languages showed more pronounced hypoactivation in the left fusiform gyrus (FG) and the middle temporal gyrus (MTG) compared to DD in Chinese. Beyond the language‐specific brain differences, substantial evidence also supports the language‐universal neural deficits in dyslexia (Feng et al. 2020; Hu et al. 2010; Paulesu et al. 2001), This evidence highlights common functional deficits in brain regions such as the left temporoparietal (TP) cortex (Paulesu et al. 2001; Yan et al. 2021) and the occipitotemporal (OT) regions (Martin et al. 2016; Paulesu et al. 2001) across different languages. Collectively, these results indicate that key neural substrates subserve reading functions across languages universally, independent of disparities in linguistic architecture, orthographic complexity, and broader cultural contexts.

Learning to read is also a dynamic process, involving different core cognitive skills. According to the reading developmental model (Frith 1986), the process unfolds in three stages: A logographic stage, an alphabetic stage, and an orthographic stage, and different reading strategies are adopted in each reading phase. The initial logographic stage is marked by word recognition based on prominent visual features, while the alphabetic stage involves phonological decoding (letter‐sound mapping). The mature orthographic stage enables whole‐word recognition through automatized orthographic processing. Neurologically, reading development is associated with increased engagement of the left OT cortex and the left dorsal precentral gyrus, alongside decreased reliance on the left superior temporal gyrus (STG) and the bilateral supplementary motor regions (Martin et al. 2015), indicating a shift from phonological reading to orthographic reading. A longitudinal investigation further revealed that reading proficiency gains correlated with: (a) initial heightened connectivity within the dorsal stream (fusiform‐inferior parietal cortex) critical for phonological decoding, followed by connectivity reduction; and (b) sustained connectivity in the ventral stream (inferior occipital‐fusiform cortex) (Wise Younger et al. 2017), indicating a mature orthographic recognition. This dual pattern suggests a developmental trajectory where dorsal stream engagement diminishes as ventral stream automation emerges. These results provided robust verification of this developmental transition. However, cross‐linguistic comparisons reveal that these developmental trajectories vary systematically across writing systems. In English—a deep orthography—developmental studies demonstrate increased left frontal activation and decreased left extrastriate activation from childhood to adulthood (Schlaggar et al. 2002). Conversely, Polish—a shallow orthography—exhibits increased activation in both the left IFG) and the left OT regions during development (Chyl et al. 2018). Chinese, as a logographic writing system, shows a unique pattern characterized by increased the left IFG and the right middle occipital gyrus (MOG) activation alongside decreased the left STG involvement (Cao et al. 2010; Cao et al. 2011). Collectively, these findings underscore the profound influence of orthographic and linguistic properties on the functional neurodevelopment of reading, highlighting the necessity of cross‐linguistic approaches in reading research.

A comprehensive developmental framework elucidating the evolution of neural manifestations in DD across diverse writing systems remains conspicuously absent. However, emerging findings from the limited developmental neuroimaging studies conducted to date reveal a convergent yet complex neurodevelopmental trajectory. This trajectory is characterized by a dynamic balance of deficits within the brain's dorsal and ventral reading pathways. This developmental shift mirrors the known transition from phonological decoding to orthographic reading in typical development. Empirical evidence from diverse languages supports this model. In a shallow orthography, Chyl et al. (2018) demonstrated in a longitudinal Polish study that hypoactivation in the left IFG and the FG emerged only after formal literacy instruction began, suggesting reading acquisition triggers ventral stream atypicalities in at‐risk children. In deep orthographies like English, the divergence is more pronounced. While typical readers showed age‐related increases in the left ventral OT activation, those with DD exhibited persistent dorsal system reliance (evidenced by elevated IFG activation) (Shaywitz et al. 2007), indicating a failure in the typical ventral pathway automatization process. This pattern is further corroborated in nonalphabetic scripts. Yan et al. (2024), in a study of Chinese readers—a logographic system with deep orthography—demonstrated that adults with DD exhibited greater hypoactivation in the left inferior temporal gyrus (ITG) than children with DD, whereas the reverse pattern held for the left IFG. This finding powerfully illustrates the progression of deficits: an initial impairment in dorsal phonological processing in childhood that gives way to a more pronounced ventral orthographic deficit in adulthood. Complementing these findings in different language, Richlan et al. (2011) conducted a coordinate‐based meta‐analysis across alphabetic languages, revealing adults with DD exhibited greater left FG differences than children, while children showed more pronounced bilateral inferior parietal lobe deficits. These results further verify the developmental trajectory associated with DD. Collectively, these findings converge on a sophisticated neurodevelopmental model of DD. The core synthesis is that DD is not a static neural deficit, but a dynamic disorder characterized by a developmental shift from dorsal pathway atypicalities (phonological decoding) in childhood to progressive ventral pathway hypoactivation (orthographic processing) in adulthood. A critical challenge for the field lies in integrating these findings into a unified, cross‐linguistic developmental theory of dyslexia.

In the present study, we conducted a cross‐linguistic meta‐analysis to examine convergent and divergent patterns of brain functional alterations in developmental dyslexia (DD) across three orthographically distinct languages—Chinese (logographic), English (deep orthography), and German (shallow orthography). These three languages represent well‐established typological prototypes in dyslexia research and provide robust, comparable datasets for cross‐linguistic meta‐analytic synthesis. Although additional shallow‐orthography languages such as Polish and Spanish could theoretically enrich the comparison, available empirical studies on these languages are limited in sample size and methodological consistency. To avoid confounding mixed effects arising from small, heterogeneous datasets, we adopted a principled approach of selecting only one representative language per orthographic typology in the current analysis. Focusing on readers aged 16 years and younger, we selected this age range based on neurodevelopmental evidence that expert reading capabilities typically emerge beyond this age threshold (Wolf 2007). We hypothesized that: (1) neural signatures in phonological and orthographic reading areas would exhibit language‐specific variations across the three writing systems; and (2) regardless of linguistic background, older children with DD would demonstrate progressively greater functional differences in orthographic reading regions, while younger children would show more pronounced differences in phonological reading areas, reflecting distinct developmental trajectories. We adopted three complementary analytical strategies to address the research questions. Linear model‐based group comparisons were used to examine between‐group neural differences across languages. Meta‐regression was performed to explore age‐related variations in brain activation. Conjunction analysis was applied to identify overlapping neural patterns across groups. These methods were chosen for their suitability to disentangle cross‐linguistic, developmental and shared neural characteristics.

2. Methods

2.1. Literature Retrieval

Neuroimaging studies were searched through “PubMed” (http://www.pubmed.org) and “Web of science”. The key words were the combination of “dyslexia”, “reading disorder”, “reading impairment” or “reading disability” and “fMRI”, “PET”.

The inclusion criteria of the studies were: (1) studies were about dyslexia; (2) functional MRI was conducted; (3) the participants were native Chinese speakers, English speakers, or German speaker; (4) the age of the participants was under 17 years old; (5) the 3D stereotaxic coordinates were reported in Talairach space or MNI (Montreal Neurological Institute) space. The exclusion criteria of the studies were: (1) resting studies; (2) no whole brain comparation was reported; (3) only ROI or functional connectivity results were reported; (4) not language‐related tasks for the fMRI scanning.

2.2. Voxel‐Wise Meta‐Analysis

After all the data were collected, a voxel‐wise meta‐analysis was conducted using AES‐SDM 5.14 (http://www.sdmproject.com). Specifically, the peak coordinates and the corresponding t values were used to recreate the effect size map and their variance map of each study (Radua et al. 2014). During the processing, the anisotropy Gaussian kernel was set to α = 1, and the full width at half maximum (FWHM) was set to 20 mm (Radua et al. 2012b). After the map recreating, random‐effects modelling was conducted to gain the mean effect size map. Finally, 500 permutations were conducted. The threshold was set at SDM‐Z >1, p < 0.005 uncorrected at voxel level and extended 100 voxels, and the threshold was deemed to ensure sufficient sensitivity and stability (Radua et al. 2012a). To verify the robustness of this meta‐analysis, we performed two supplementary analyses. First, a jackknife sensitivity analysis was conducted by sequentially removing one study at a time to assess the stability of the pooled results. Second, publication bias was assessed for all significant clusters collectively and for each individual cluster separately.

2.3. Group Comparisons

To explore the group difference between different languages, we conducted group comparison using linear model for each pair of languages, separately. After 500 permutations, the resulting map was also thresholded at SDM‐Z >1, p < 0.005 uncorrected at voxel level and extended 10 voxels.

2.4. Meta‐Regression

To explore the age‐related changes in the brain functional deficit associated with dyslexia within each language, we conducted a meta‐regression analysis with age as a covariate. The threshold was set at SDM‐Z >1, p < 0.005 uncorrected and extended 10 voxels.

2.5. Conjunction Analysis

Conjunction analysis was performed to identify commonalities across different comparisons. Specifically, the intersection of the thresholded resulting maps was computed using the image calculation function in SPM 12 (https://www.fil.ion.ucl.ac.uk/spm/software/spm12/).

3. Results

3.1. Demographic Information of the Included Studies

After data screening, there were 52 experiments from 48 studies included in the meta‐analysis (Table 1), including 14 experiments (from 12 studies) in Chinese, 22 experiments (from 22 studies) in English,16 experiments (from 14 studies) in German (Figure 1). In total, 1554 participants were enrolled (695 typically developing controls and 859 children with DD), the mean age of the Chinese participants was 11.31 years, with a total of 198 age‐controls and 223 DD readers, the mean age of the English participants was 10.93 years, with a total of 215 age‐controls, 361 DD readers, and the mean age of the German participants was 11.71 years, with a total of 282 age‐controls and 275 DD readers.

TABLE 1.

Studies included in the meta‐analysis.

Study	TD (N)	DD (N)	Mean age (month)	Tasks
English
(Booth et al. 2007)	13	13	126	Visual word judgment
(Cao et al. 2008)	12	12	148	Word rhyme judgment
(Desroches et al. 2010)	12	12	137	Auditory rhyming task
(Gaab et al. 2007)	23	22	128	Sound discrimination task
(Hoeft et al. 2006)	10	10	134	Word rhyme task
(Hoeft et al. 2007)	19	19	173	Word rhyme task
(Landi et al. 2010)	13	13	158	Rhyming task and semantic categorization task
(Temple et al. 2001)	15	24	127	Letter rhyming task, letter matching task
(Rimrodt et al. 2009)	15	14	141	Sentence comprehension task
(Meyler et al. 2008)	12	23	130	Sentence comprehension
(Farris et al. 2016)	16	15	112	Object rhyming task
(Hancock et al. 2016)	11	16	125	Word rhyming task
(Cutting et al. 2013)	19	20	147	Lexical decision task
(Horowitz‐Kraus et al. 2016)	9	10	120	Narrative comprehension task
(Kovelman et al. 2012)	12	12	108	Auditory words rhyming task and auditory words matching task
(Langer et al. 2015)	15	15	119	Sentence reading task
(Olulade et al. 2015)	12	16	121	Implicit word reading
(Perrachione et al. 2016)	25	26	95	Spoken words listening, Written words, objects, and faces viewing
(Zuk et al. 2018)	13	11	114	Phonological processing
(Hu et al. 2010)	10	11	164	Sematic match task, word/ picture naming task
(Borghesani et al. 2021)	14	26	125	Word reading
(Al Dahhan et al. 2022)	15	21	133	Word rhyming judgements
German
(Georgiewa et al. 1999)	17	17	168	Word processing
(Grunling et al. 2004)	21	17	163	Slash patterns matching task, letters matching task, words matching task, pseudoword matching task and pseudoword rhyming task
(Heim et al. 2010)	20	16	114	First sound detection task, motion detection task, Posner attention task, auditory discrimination task
(Kronbichler et al. 2006)	15	13	188	Sentence reading
(Kronschnabel et al. 2013)	22	13	192	Word and pseudoword reading
(Maurer et al. 2011)	16	11	137	Word matching
(Schulz et al. 2009)	15	15	138	Sentence reading
(Schulz et al. 2008)	21	12	138	Sentence reading
(van der Mark et al. 2009)	24	18	136	Phonological decision task
(Bach et al. 2010)	18	14	100	Covert reading
(Grande et al. 2011)	25	20	115	Picture naming
(Heim et al. 2015)	10	33	119	Word reading
(Kronschnabel et al. 2014)	22	13	191	Target detection task
(van Ermingen‐Marbach et al. 2013)	10	32	117	phoneme deletion
(van Ermingen‐Marbach et al. 2013)	13	17	117	phoneme deletion
(van Ermingen‐Marbach et al. 2013)	13	14	116	phoneme deletion
Chinese
(Siok et al. 2008)	12	12	132	Character rhyming task
(Liu et al. 2013)	14	14	142	Lexical match task and character rhyming task
(Liu et al. 2012)	11	11	143	Word rhyming task and semantic judgment task
(Cao et al. 2018)	19	23	133	Visual spelling task
(Siok et al. 2009)	12	12	132	Font size judgment task
(Feng et al. 2017)	20	14	123	Character spelling task and character rhyming task
(Cao et al. 2017)	13	17	134	Auditory rhyming task
(Siok et al. 2004)	8	8	132	Homophone judgement task and lexical decision task
(Hu et al. 2010)	8	8	172	Sematic match task, word/ picture naming task
(Cao et al. 2021)	17	16	137	Visual rhyming task
(Yang and Tan 2020)	16	16	124	Homophone judgments task and component judgments task
(Yan et al. 2024)	16	39	132	Auditory spelling task
(Yan et al. 2024)	16	16	133	Visual spelling task
(Yan et al. 2024)	16	17	133	Visual rhyming task

FIGURE 1.

PRISMA flow diagram for study selection.

3.2. Convergent Deficit in Three Different Languages

Meta‐analysis found Chinese‐reading children with dyslexia showed convergent lower activation than typical readers in the left IFG, which extended to the TP regions and the left ITG. They also showed convergent higher activation than typical readers in the right precentral gyrus and the right temporal gyrus (Table 2 and Figure 2).

TABLE 2.

Convergent deficit associated with DD in each language (JK represents the results of jack‐knife sensitivity analysis).

Brain regions	MNI Coordinates	SDM‐Z	P	Voxels	JK	Bias
Chinese results
TD>DD
Left precentral gyrus	−48, 10, 32	4.736	0.0000	3147	14/14	1.62
Bilateral median cingulate gyrus	6, 24, 36	2.826	0.0000	2652	14/14	−1.13
Left insula	−34, 26, 2	2.756	0.0000	739	13/14	−0.76
Left inferior temporal gyrus	−46, −56, −14	2.021	0.0004	416	14/14	4.90 ***
Left supramarginal gyrus	−58, −42, 26	1.906	0.0008	319	13/14	−0.38
Left superior parietal gyrus	−26, −66, 48	1.745	0.0018	186	12/14	−1.11
DD>TD
Right precentral gyrus	38, −24, 46	−3.395	0.0000	4737	14/14	1.94
Left posterior cingulate gyrus	−4, −46, 32	−2.081	0.0001	1188	14/14	1.43
Right superior frontal gyrus	2, 40, ‐8	−1.413	0.0032	180	11/14	−0.96
English
TD>DD
Left inferior temporal gyrus	−48, −46, −22	3.003	0.0000	2790	22/22	0.74
Left inferior parietal gyri	−46, −42, 46	2.502	0.0000	1041	22/22	1.97
Left middle frontal gyrus	−42, 14, 38	2.22	0.0001	279	21/22	0.94
Left superior occipital gyrus	−14, −92, 14	1.931	0.0009	246	21/22	0.52
DD>TD
Right supramarginal gyrus	62, −50, 24	−1.197	0.0002	1172	20/22	−0.73
Right hippocampus	30, −14, −14	−1.141	0.0003	284	20/22	−0.82
Right supplementary motor area	4, 8, 50	−1.007	0.0008	217	20/22	−0.55
German
TD>DD
Left inferior parietal gyri	−52, −40, 42	2.581	0.0000	3273	16/16	6.14 *
Left fusiform gyrus	−40, −42, −20	2.110	0.0000	871	16/16	2.62
DD>TD
Right cerebellum	38, −56, −28	−2.012	0.0002	1938	16/16	−4.20
Overlaps
TD>DD
Chinese & English
Left middle frontal gyrus	−42, 12, 36			257
Left inferior parietal lobule	−56, −44, 38			119
Left fusiform gyrus	−46, −56, −16			403
Chinese & German
Left supramarginal Gyrus	−56, −44, 32			240
Left fusiform gyrus	−42, −54, −16			159
Left inferior temporal gyrus	−52, −60, −10			12
English & German
Left inferior parietal lobule	−48, −44, 46			529
Left fusiform gyrus	−42, −42, −22			640
Left inferior temporal gyrus	−54, −62, −6			406
Chinese & English & German
Left fusiform gyrus	−42, −54, −16			159
Left inferior temporal gyrus	−52, −60, −8			12
Left inferior parietal lobule	−56, −44, 38			119
DD>TD
Chinese & English
Right supramarginal gyrus	56, −44, 30			546
Chinese & German
—	—			—
English & German
—	—			—
Chinese & English & German
—	—			—

^* p < 0.05.

^*** p < 0.001.

FIGURE 2.

Convergent functional deficit in each language.

English‐reading children with dyslexia showed convergent lower activation than typical readers in the left ITG, the left inferior parietal gyri, the left middle frontal gyrus, and the left superior occipital gyrus. They also showed convergent higher activation than typical readers in the right supramarginal gyrus, the right hippocampus, and the right supplementary motor area (Table 2 and Figure 2).

German‐reading children with dyslexia showed convergent lower activation than typical readers in the left inferior parietal gyri, which extended to the left TP regions and the OT regions, as well as the left FG. They also showed convergent higher activation than typical readers in the right cerebellum (Table 2 and Figure 2).

Conjunction analysis showed the three different languages showed consistent deficits in the left inferior parietal lobule, centered at −56, −44, 38, with a cluster of 119 voxels, the left FG, centered at −42, −54, −16, with a cluster of 159 voxels, and the left ITG, centered at −52, −60, −8, with a cluster of 12 voxels (Figure 3).

FIGURE 3.

Convergent functional deficits associated with dyslexia between languages.

For the robustness of this meta‐analysis, the results of the jackknife analysis were reported in Table 2. For the publication bias, Egger's test was conducted, and there was no overall publication bias detected (Bias = 1.07, t(12) = 2.08, p = 0.059 for Chinese studies; Bias = 0.46, t(20) = 0.65, p = 0.523 for English studies; Bias = 2.43, t(14) = 1.11, p = 0.286 for German studies). The detailed publication bias for each significant cluster was reported in Table 2.

3.3. Group Difference of Functional Deficit Between Each Language

Group comparation using linear modelling revealed that Chinese‐reading children with dyslexia showed more convergent brain hypoactivation in the bilateral supplementary motor area, the left IFG, the left insula, the left superior parietal gyrus, and the left cerebellum, and more convergent brain hyperactivation in the right postcentral gyrus, and the bilateral posterior cingulate gyrus than English‐reading children (Table 3 and Figure 4). Chinese‐reading children with dyslexia showed more convergent brain hypoactivation in the bilateral median cingulate/paracingulate gyri, the left precentral gyrus, the left superior parietal gyrus, the left cerebellum, and the right insula, and more convergent hyperactivation in the left posterior cingulate gyrus, and the right postcentral gyrus than German‐reading children (Table 3 and Figure 4). Conjunction analysis showed Chinese‐reading children with dyslexia showed more convergent brain hypoactivation than English‐reading children and German‐reading children in the bilateral supplementary motor area (peak coordinate: 0, 14, 30, cluster size: 2386 voxels), the left IFG (peak coordinate: −42, 38, 10, cluster size: 2433 voxels), the left insula (peak coordinate: −28, 20, −4, cluster size: 537 voxels), the left superior parietal lobule (peak coordinate: −20, −70, 54, cluster size: 12 voxels), and the left cerebellum (peak coordinate: −14, −70, −14, cluster size: 168 voxels), and more convergent brain hyperactivation in the left post cingulate gyrus (peak coordinate: −8, −52, 28, cluster size: 389 voxels) and the right precentral gyrus (peak coordinate: 46, −18, 44, cluster size: 12 voxels) (Figure 5).

TABLE 3.

Group comparation between languages.

Brain regions	MNI Coordinates	SDM‐Z	P	Voxels
Chinese>English
TD>DD
Left inferior frontal gyrus	−46, 8, 26	3.701	0.0001	2843
Bilateral supplementary motor area	−2, 10, 44	3.210	0.0001	2562
Left insula	−30, 24, −2	2.573	0.0001	716
Left cerebellum	−12, −68, −14	1.375	0.0015	170
Left superior parietal gyrus	−20, −70, 54	1.209	0.0034	12
DD>TD
Right postcentral gyrus	42, −26, 54	4.143	0.0001	3280
Bilateral posterior cingulate gyrus	−8, −52, 30	2.080	0.0001	759
English>Chinese
TD>DD
Left inferior temporal gyrus	−50, −42, −24	1.494	0.0001	963
Left middle occipital gyrus	−14, −92, 14	1.200	0.0007	452
Left middle occipital gyrus	−32, −74, −4	1.080	0.0017	15
DD>TD
—	—	—	—	—
Chinese>German
TD>DD
Left precentral gyrus	−44, 10, 32	3.926	0.0001	3342
Bilateral median cingulate / paracingulate gyri	−2, 8, 40	3.024	0.0001	2506
Left cerebellum	−16, −72, −16	1.706	0.0002	537
Left superior parietal gyrus	−22, −68, 50	1.515	0.0006	200
Right insula	36, 24, 2	1.244	0.0026	42
DD>TD
Right postcentral gyrus	26, −28, 62	3.051	0.0001	3719
Left posterior cingulate gyrus	−10, −52, 30	1.932	0.0001	424
German>Chinese
TD>DD
Right superior temporal gyrus	62, −30, 18	1.412	0.0003	1451
Left middle temporal gyrus	−52, −66, 14	1.451	0.0002	690
Left insula	−42, −2, −2	1.358	0.0001	524
Right middle occipital gyrus	38, −74, 12	1.633	0.0001	419
Left anterior cingulate / paracingulate gyri	−4, 48, 10	1.239	0.0009	452
Left fusiform gyrus	−36, −40, −22	1.023	0.0037	17
DD>TD
Right insula	38, −8, 10	1.512	0.0001	1242
Right cerebellum	20, −68, −28	1.333	0.0005	1191
Left inferior frontal gyrus	−50, 16, 18	1.426	0.0002	636
Left superior temporal gyrus	−56, −16, 8	1.162	0.0016	167
Right inferior frontal gyrus	50, 40, 16	1.204	0.0012	103
English>German
TD>DD
Left fusiform gyrus	−34, −72, −12	1.298	0.0007	443
Left inferior temporal gyrus	−54, −46, −18	1.300	0.0007	405
Left superior parietal gyrus	−30, −60, 46	1.251	0.0010	199
Corpus callosum	−14, −92, 10	1.130	0.0023	71
Left middle frontal gyrus	−38,10,36	1.111	0.0026	27
DD>TD
—	—	—	—	—
German>English
TD>DD
Left anterior cingulate / paracingulate gyri	−6, 42, 14	1.342	0.0010	455
Right middle occipital gyrus	38, −74, 12	1.673	0.0001	235
Left supramarginal gyrus	−58, −54, 30	1.319	0.0011	115
Left insula	−42, 2, 2	1.213	0.0021	103
Left middle temporal gyrus	−56, −64, 14	1.223	0.0020	52
Right superior temporal gyrus	62, −34, 18	1.145	0.0031	20
DD>TD
Right cerebellum	28, −64, −32	1.636	0.0002	1814
Left inferior frontal gyrus	−50, 18, 18	2.029	0.0000	884
Right insula	38, −8, 10	1.556	0.0004	659
Right inferior frontal gyrus	48, 36, 16	1.761	0.0001	587

FIGURE 4.

Between language comparations.

FIGURE 5.

Conjunction analysis of the between‐language comparation.

English‐reading children with dyslexia showed more convergent brain hypoactivation in the left ITG, and the left MOG than Chinese‐reading children (Table 3 and Figure 4). English‐reading children with dyslexia showed more convergent brain hypoactivation in the left middle frontal gyrus, the left superior parietal gyrus, the left ITG, and the left FG than German‐reading children (Table 3 and Figure 4). Conjunction analysis showed English‐reading children with dyslexia showed more convergent brain hypoactivation than Chinese‐reading children and German‐reading children in the left ITG (peak coordinate: −50, −38, −26, cluster size: 375 voxels), the left inferior occipital gyrus (peak coordinate: −32, −74, −4, cluster size: 12 voxels), and the left superior occipital gyrus (peak coordinate: −14, −90, 10, cluster size: 67 voxels) (Figure 5).

German‐reading children with dyslexia showed more convergent brain hypoactivation than Chinese‐reading children in the left anterior cingulate/paracingulate gyri, the left insula, the left MTG, the left FG, the right STG and the right MOG, and more convergent brain hyperactivation in the left IFG, the left STG, the right IFG, the right insula, and the right cerebellum (Table 3 and Figure 4). German‐reading children with dyslexia showed more convergent brain hypoactivation in the left anterior cingulate/paracingulate gyri, the left insula, the left supramarginal gyrus, the left MTG, the right STG, and the right MOG, and more convergent hyperactivation in the left IFG, the right IFG, the right insula, and the right cerebellum than English‐reading children (Table 3 and Figure 4). Conjunction analysis showed German‐reading children with dyslexia showed more convergent brain hypoactivation than English‐reading children and Chinese‐reading children in the bilateral anterior cingulate gyrus (peak coordinate: 2, 44, 10, cluster size: 295 voxels), the left insula (peak coordinate: −42, 0, 2, cluster size: 61 voxels), the left angular gyrus (peak coordinate: −54, −58, 32, cluster size: 29 voxels), the left MTG (peak coordinate: −56, −64, 14, cluster size: 52 voxels), the right STG (peak coordinate: 62, −34, 18, cluster size: 20 voxels), and the right MOG (peak coordinate: 38, −74, 12, cluster size: 233 voxels), and more convergent brain hyperactivation in the left IFG (peak coordinate: −50, 16, 18, cluster size: 595 voxels), the right insula (peak coordinate: 38, −8, 10, cluster size: 657 voxels), the right IFG (peak coordinate: 50, 38, 16, cluster size: 103 voxels) and the right cerebellum (peak coordinate: 26, −66, −32, cluster size: 1138 voxels) (Figure 5).

3.4. Developmental Changes of the Brain Functional Deficit Associated With Dyslexia in Three Languages

Meta‐regression found Chinese‐reading children with dyslexia showed increased hypoactivation associated with age in the left IFG, the left FG, the left MOG and the right middle frontal gyrus; and decreased hyperactivation associated with age in the left cerebellum and the right IFG (Table 4 and Figure 6).

TABLE 4.

Meta‐regression results in children with dyslexia.

Brain regions	MNI coordinate	SDM‐Z	P	Voxels
Chinese (developmental increase)
TD>DD
Left fusiform gyrus	−42, ‐54, −20	2.060	0.0009	338
Left inferior frontal gyrus	−52, 16, 28	2.473	0.0001	295
Right middle frontal gyrus	46, 12, 42	1.955	0.0014	147
Left middle occipital gyrus	−30, −62, 34	2.174	0.0006	126
Left lingual gyrus	−6, −60, 0	1.677	0.0036	21
DD>TD
—	—	—	—	—
Chinese (developmental decrease)
TD>DD
—	—	—	—	—
DD>TD
Left cerebellum	−24, −68, −30	1.168	0.0011	253
Right inferior frontal gyrus	46, 24, 12	1.092	0.0024	104
Left middle temporal gyrus	−50, −22, −8	1.166	0.0012	93
Right superior frontal gyrus	22, 20, 60	1.166	0.0011	91
Right cerebellum	26, −42, −38	1.064	0.0031	46
Left middle frontal gyrus	−30,56,16	1.037	0.0040	13
English (developmental increase)
TD>DD
Left middle temporal gyrus	−52, −54, 0	1.89	0.0002	379
Cerebellum	2, −60, 0	1.511	0.0011	335
Left middle occipital gyrus	−20, −86, 2	1.996	0.0001	252
Left inferior parietal gyri	−58, −42, 46	2.135	0.0001	233
Left fusiform gyrus	−30, −70, −2	1.868	0.0003	125
Left fusiform gyrus	−38, −34, −26	1.297	0.0025	98
Right fusiform gyrus	30, −70, −2	1.652	0.0006	59
DD>TD
Left insula	−34, 28, 6	2.071	0.0001	302
Left anterior cingulate gyrus	−6, −14, 28	1.720	0.0004	247
Left inferior frontal gyrus	−40, 30, 24	1.647	0.0006	122
Left thalamus	−8, −14, 20	1.610	0.0008	73
Left insula	−34, −12, 18	1.556	0.0010	38
English (developmental decrease)
TD>DD
Left superior temporal gyrus	−56, −6, −6	2.033	0.0000	1182
Left inferior temporal gyrus	−62, −38, −24	1.108	0.0044	23
DD>TD
—	—	—	—	—
German (developmental increase)
TD>DD
Left middle frontal gyrus	−24, 42, −12	1.494	0.0002	186
Right middle frontal gyrus	26, 52, −14	1.322	0.0006	178
Left fusiform gyrus	−32, −44, −14	1.108	0.0024	40
Right middle temporal gyrus	40, −62, 2	1.199	0.0013	16
DD>TD
Left middle frontal gyrus	−38, 54, ‐8	1.868	0.0015	330
Right rolandic operculum	38, −4, 16	1.921	0.0012	208
Left insula	−36, −20, 6	1.687	0.0029	63
Right median cingulate / paracingulate gyri	12, 8, 38	1.641	0.0034	14
German (developmental decrease)
TD>DD
Right superior temporal gyrus	62, −24, 10	1.229	0.0003	2187
Left precentral gyrus	−48, 0, 40	1.393	0.0001	829
DD>TD
Right middle occipital gyrus	36, −74, 28	1.607	0.0002	465
Middle cerebellar peduncles	18, −52, −30	1.457	0.0005	363
Right postcentral gyrus	38, −20, 48	1.157	0.0031	26
Left superior occipital gyrus	−24, −92, 28	1.140	0.0034	15

FIGURE 6.

Developmental changes associated with dyslexia in different languages.

English‐reading children with dyslexia showed increased hypoactivation associated with age in the left inferior parietal gyri (excluding supramarginal and angular), the left MOG, the left FG, and the left cerebellum, decreased hypoactivation associated with age in the left STG; and increased hyperactivation associated with age in the left IFG, the left insula, and the left cingulate gyrus (Table 4 and Figure 6).

German‐reading children with dyslexia showed increased hypoactivation associated with age in the bilateral middle frontal gyrus, and decreased hypoactivation associated with age in the left precentral gyrus and the right STG; and increased hyperactivation associated with age in the left middle frontal gyrus and the right insula; and decreased hyperactivation associated with age in the right MOG and the right cerebella (Table 4 and Figure 6).

Conjunction analysis revealed that there was a convergent increase in hypoactivation with age in the left ITG (Peak coordinate: −54, −56, −6, Cluster size: 87 voxels) and the left lingual gyrus (Peak coordinate: −4, −60, 0, Cluster size: 11 voxels) observed in both Chinese‐reading children with DD and English‐reading children with DD (Figure 6).

4. Discussion

In this meta‐analysis, we identified convergent functional abnormalities in children with developmental dyslexia (DD) across three orthographically distinct languages, specifically in the left OT regions and the left inferior parietal lobule. These findings suggest that these regions may represent the core deficit of DD across different languages. Furthermore, we observed that these functional deficits are modulated by language orthographic depth, with increased deficit in the left IFG and left posterior OT regions in deep orthographic languages. Additionally, we noted distinct developmental changes for the functional deficit across different languages.

4.1. Convergent Functional Deficits Across Three Languages

We found convergent functional deficits associated with DD in the left OT regions and the TP regions across various languages, which are critical components of the reading network. This aligns with the cross‐linguistic study by Paulesu et al. (2001), which demonstrated similar deficits in Italian, English, and French speakers, and is further corroborated by a cross‐cultural meta‐analysis by Yan et al. (2021), implicating these regions as the core neural underpinnings of developmental dyslexia (DD), given their pivotal roles in the reading network.

The hypothesis that DD arises primarily from dysfunction in the left OT or TP regions has garnered robust empirical support. Dębska et al. (2016) reported reduced activation in these regions among Polish‐speaking kindergarteners at familial risk for DD, suggesting that these deficits precede reading acquisition and may thus be causal rather than consequential. The primacy of TP regions in DD is further substantiated by converging evidence from functional and structural neuroimaging studies. Multiple functional neuroimaging investigations have consistently replicated findings of pre‐reading activation deficits within the TP cortex (Raschle et al. 2012; Vandermosten et al. 2020). Notably, these functional deficits are paralleled by observable structural alterations in TP regions prior to the onset of formal reading instruction (Clark et al. 2014). The convergence of both functional and structural anomalies within the TP region preceding the acquisition of reading competencies provides compelling evidence supporting the foundational role of this neural substrate in the etiology of DD. Similarly, the left OT region's involvement in DD is also well‐documented. Centanni et al. (2019) found diminished FG responses to print in at‐risk English‐speaking kindergarteners, while Pleisch et al. (2019) observed that only typical developing readers exhibited orthographic sensitivity in the left ventral OT cortex among German‐speaking children. In Chinese, a logographic writing system, OT hypoactivation has been persistently observed in individuals with DD from childhood adolescents to adulthood, which highlights the robust deficit in RD (Yan et al. 2024; Yan et al. 2026). Collectively, these findings highlight the OT region's impaired print processing as a hallmark of DD. Critically, these deficits persist beyond mere reading experience. Hoeft et al. (2007), employing a double‐control design with age‐matched (AM) and reading‐matched (RM) controls, confirmed that OT and TP dysfunction in DD is intrinsic to the disorder rather than a byproduct of insufficient exposure to print. The convergence of cross‐linguistic, longitudinal, and methodologically rigorous studies underscores the left OT and TP regions as central to DD's neural architecture. These findings support a neurobiological framework that emphasizes early identification and intervention for these deficits.

4.2. Functional Deficits of DD Are Modulated by Language Orthographic Depth

Apart from the convergent functional deficits across three different languages, the functional deficit can also be modulated by language orthographic depth. For the left OT regions, the pattern of functional deficit exhibits a posterior‐to‐anterior gradient, correlated with the orthographic depth of the language. Chinese, which has the deepest orthography, displays a convergent deficit in the posterior OT regions (Peak coordinate: −46, −56, −14). English, with a modest level of orthographic depth, shows a convergent deficit in the middle OT regions (Peak coordinate: −48, −46, −22). German, characterized by the shallowest orthography, exhibits a convergent deficit in the anterior OT regions (Peak coordinate: −40, −42, −20). Comparative group analysis revealed that English and German demonstrated greater convergent functional deficits at an anterior OT region compared to Chinese in the left OT regions. Conversely, Chinese presented a more pronounced convergent functional deficit at the posterior OT region compared to English and German, thereby confirming the posterior‐to‐anterior gradient.

Previous research on the gradient transition within the left OT regions has proposed a hierarchical model of function, emphasizing an increase from posterior to anterior for complex linguistic functions or psycholinguistic grain size (Vinckier et al. 2007; Zhan et al. 2023). However, other studies provided evidence for a posterior‐to‐anterior gradient that encompasses both the transition from sublexical to lexical processing (Lochy et al. 2018; Woolnough et al. 2021) and the enhancement of orthographic and phonological representation in Chinese (Zhao et al. 2017). These findings suggest a complex interplay of functions within the left OT regions. A recent study has uncovered that the left OT regions serve not only as a linguistic system but also as an attentional system, interconnected with both language and attention‐related regions (Chen et al. 2019). One potential explanation for the observed gradient pattern is that the posterior part of the left OT region is involved in visual feature extraction and has stronger connections with visual spatial attention regions, whereas the anterior part is more closely linked with the decoding language system (Lerma‐Usabiaga et al. 2018; Yablonski et al. 2024). This dual functionality may underlie the differential patterns of activation observed across languages.

The results also showed a mesial‐to‐lateral gradient pattern modulated by language orthographic depth, with shallow orthographic language (i.e., German) exhibiting a convergent deficit in a mesial OT region, while deep orthographic languages (i.e., Chinese and English) exhibiting a convergent deficit in a lateral OT region. Previous study has indicated a mesial‐to‐lateral dissociation for orthographic processing, wherein the mesial OT region is primarily involved in encoding sublexical orthographic information, whereas the lateral OT region is engaged in direct lexico‐semantic access (Bouhali et al. 2019). The nuanced dissociation aligns with the varying cognitive demands imposed by different orthographic systems. In shallow orthographic languages like German, where there is a high degree of consistency between graphemes and phonemes, the mesial OT region's role in sublexical processing is particularly critical. Conversely, in deep orthographic languages such as Chinese and English, where the relationship between graphemes and phonemes is more complex and less predictable, the lateral OT region's involvement in lexico‐semantic access becomes more pronounced.

For the left dorsal IFG/MFG, a convergent functional deficit was observed exclusively in Chinese and English, both of which are characterized by deep orthographies. In contrast, no such deficit was detected in German, a language with a shallow orthography. Comparative group analysis further demonstrated that the convergence in Chinese was more pronounced than in both English and German. Additionally, English exhibited greater convergence than German within the left dorsal IFG/MFG, thereby confirming a gradient modulation by the depth of the orthography.

These results are in alignment with previous meta‐analytic studies. One such study compared the functional deficits associated with dyslexia in deep orthographic languages with those in shallow orthographic languages, discovering that deep orthographic languages exhibited a greater convergent functional deficit in the left dorsal IFG than shallow orthographic languages (Martin et al. 2016). Another meta‐analysis revealed that Chinese readers with dyslexia presented greater convergent structural and functional deficits in the left dorsal IFG compared to readers of alphabetic languages (Yan et al. 2021).

One hypothesis suggests that the left dorsal IFG/MFG is primarily involved in phonological processing (Hancock et al. 2017; Tan et al. 2005), The increasing engagement of the left dorsal IFG/MFG is linked to the complexity of the mapping from orthography to phonology, which adds to the difficulty of decoding words through phonological assembly. The complexity arises because the deeper the orthography, the more intricate the mapping rules between written symbols and their corresponding sounds, which can complicate the decoding process. An alternative hypothesis is based on the dual processing theory, which posits the existence of two distinct pathways for word processing: a dorsal decoding pathway and a ventral lexical pathway. The dorsal pathway connects the OT regions with the temporo‐parietal (TP) regions, while the ventral pathway connects the OT regions with the inferior frontal regions. According to this theory (Fiebach et al. 2002; Friederici 2012), the complex mapping rules of deep orthography languages create greater challenges for decoding through the ventral pathway, leading to increased activation in the dorsal IFG as the brain relies much on the lexical ventral pathway.

In addition to the convergent reduced functional activation in the left IFG/MFG, we also noted an increased engagement of the left IFG/precentral gyrus associated with RD in German, which was more pronounced than in Chinese or English. This finding suggests a distinctive feature of shallow orthographic languages. Accord to Richlan (2014), the left IFG/precentral gyrus is identified as an independent reading network, presumed to be involved in articulatory processing. The overactivation in this region is likely linked to an overreliance on the compensatory articulatory processes in DD, particularly for shallow orthographies. In such languages, words can be decoded more easily by following the transparent orthography‐to‐phonology correspondence rules, even without fully comprehending the word's meaning. Consequently, it is easier for poor readers to develop a compensatory strategy of simply spelling out words.

4.3. Different Developmental Trajectories of Functional Deficit Across Various Languages

The findings also elucidate potential divergent developmental patterns of functional alterations associated with dyslexia across different languages. In deep orthographic languages, such as Chinese and English, age‐related increases in convergent functional alteration were evident within the left occipitoparietal cortex. This may tentatively indicate a tendency toward greater ventral pathway engagement as reading proficiency develops in these languages. Conversely, in alphabetic languages, English readers with dyslexia exhibited a reduced functional deficit in the left STG, while German readers with dyslexia showed a decreased functional deficit in the right STG. Collectively, these findings imply a decreasing reliance on the dorsal pathway during reading development in alphabetic languages. These observations are consistent with previous studies on typical reading development. A meta‐analysis comparing children and adults (Martin et al. 2015) found that children demonstrated higher convergence in the left STG than adults, whereas adults showed greater convergence in the left occipitoparietal regions. This implicates a shift from dorsal to ventral pathway involvement as readers mature. Another study employing a longitudinal design and effective connectivity analysis (Wise Younger et al. 2017) directly tested this hypothesis. It revealed that improvements in reading ability were associated with an initial increase in connectivity within the dorsal reading pathway, which is involved in phonological decoding. The increase in connectivity was followed by a decrease in this stream's connectivity over time. This pattern indicates that reading development initially depends on the dorsal reading pathway for phonological decoding, which is then gradually replaced by a greater reliance on the ventral pathway. In summary, the results underscore the dynamic and language‐specific nature of reading development and the neural mechanisms underlying dyslexia. The putative developmental transition from a dorsal pathway to a ventral pathway appears to be a common feature of reading development, but the timing and extent of this transition may vary depending on the orthographic depth of the language, as well as the broader linguistic and sociocultural contexts.

5. Limitations

Despite the promising findings of this meta‐analysis, several limitations warrant consideration. First, the included studies exhibited substantial heterogeneity arising from differences in neuroimaging tasks, demographic profiles of participants, diagnostic criteria for dyslexia, experimental imaging paradigms, and MRI scanning protocols. Second, although the present meta‐regression analyses revealed age‐associated neural patterns linked to reading development, these findings are inherently based on cross‐sectional comparisons across independent studies rather than genuine longitudinal trajectories. Future longitudinal investigations are therefore required to validate the developmental inferences reported here. Third, relatively liberal statistical thresholds were adopted in the current analyses, which may increase the possibility of false‐positive activations and should be interpreted with appropriate caution. Finally, the scope of this meta‐analysis was constrained by available comparable datasets, limiting inclusion to only three languages. Future cross‐linguistic meta‐analyses should incorporate a broader range of languages with varying orthographic complexity to improve generalizability and better isolate typological effects.

6. Conclusion

In this cross‐linguistic meta‐analysis, we identified a consistent pattern of functional deficits across various languages within the left reading networks, including the left occipitoparietal and the left TP regions. These deficits appear to be related to the core symptoms of dyslexia. Additionally, we observed that the functional deficits associated with dyslexia, as well as its developmental changes, are modulated by the depth of the orthography. In languages with deep orthographies, there are more pronounced convergent deficits in the left IFG and the left posterior OT regions, along with an increased reliance on the ventral pathway as age increases. Conversely, in alphabetic languages, there is a decreased reliance on the dorsal pathway as age increases. These findings demonstrate the complex interplay between language characteristics and the neural correlates of dyslexia.

Author Contributions

Xiaohui Yan: conceptualization, methodology, investigation, writing – original draft, visualization, formal analysis, supervision, writing – review and editing. Ziyi Wang: validation, conceptualization, investigation, methodology, data curation, writing – original draft, writing – review and editing, formal analysis.

Funding

The authors have nothing to report.

Conflicts of Interest

The authors report no competing interests.

Declaration of AI Usage

No AI tools were used for the preparation of the manuscript.

Data Availability Statement

The data that support the findings of this study are openly available in OSF at https://osf.io/rpkyu/overview?view_only=240a4645bd3b486f9ada8ece5fe5bd70.