Elsevier

Brain and Language

Volume 220, September 2021, 104983
Brain and Language

Poor reading is characterized by a more connected network with wrong hubs

https://doi.org/10.1016/j.bandl.2021.104983Get rights and content

Highlights

  • Reduced clustering in the left IFG in children with RD in the visual task.

  • Stronger connections with the bilateral STG in children with RD in the visual task.

  • More hubs in the right hemisphere in children with RD in both tasks.

  • Greater global efficiency in children with RD in both tasks.

Abstract

Using graph theory, we examined topological organization of the language network in Chinese children with poor reading during an auditory rhyming task and a visual spelling task, compared to reading-matched controls and age-matched controls. First, poor readers (PR) showed reduced clustering coefficient in the left inferior frontal gyrus (IFG) and higher nodal efficiency in the bilateral superior temporal gyri (STG) during the visual task, indicating a less functionally specialized cluster around the left IFG and stronger functional links between bilateral STGs and other regions. Furthermore, PR adopted additional right-hemispheric hubs in both tasks, which may explain increased global efficiency across both tasks and lower normalized characteristic shortest path length in the visual task for the PR. These results underscore deficits in the left IFG during visual word processing and conform previous findings about compensation in the right hemisphere in children with poor reading.

Introduction

Reading difficulty (RD) is a learning disorder that is characterized by impaired reading skills despite appropriate intelligence, learning motivation, and sufficient education (Ferrer, Shaywitz, Holahan, Marchione, & Shaywitz, 2010). Divergent hypotheses have been proposed to explain the occurrence of RD. For instance, the phonological deficit theory argues that individuals with RD fail to represent speech sounds accurately and to map them to the corresponding written letter strings (Ramus et al., 2003, Shaywitz and Shaywitz, 2005), which hinders their subsequent reading development. Consistent with this hypothesis, behavioral evidence has shown deficient phonological processing (Breier et al., 2001, Manis et al., 1997, Pennington et al., 1990, Snowling, 1995, for review; Wilson & Lesaux, 2001) and deficient mapping between orthography and phonology (Bailey et al., 2004, Booth et al., 2002, Hultquist, 1997, Lesch and Martin, 1998, Snowling, 1980, Tree, 2008) in individuals with RD. Neuroimaging studies have also documented findings suggesting poor phonological processing in the brain. For example, deficient phonological processing can be the result of reduced brain activation in the temporo-parietal area (Breier et al., 2003, Cao et al., 2006, Hoeft et al., 2006, Paulesu et al., 2014, Pugh et al., 2001, for review; Richlan, 2020, for a review; Rumsey et al., 1997, Shaywitz et al., 2002, Simos et al., 2000, Temple et al., 2001). Brain activation at the left temporo-parietal area is essential for phonological representation (Kuo et al., 2004, Paulesu et al., 1996, Tanaka et al., 2011) and for the conversion from orthography to phonology (Bitan et al., 2007, Booth et al., 2004, Simos et al., 2001). Abnormal brain activities have also been observed in the left inferior frontal gyrus (IFG), which is related to phonological memory (Kambara et al., 2018, Nixon et al., 2004, Strand et al., 2008), phonological retrieval (Georgiewa et al., 1999, Nixon et al., 2004), phonological manipulation (Booth et al., 2002, Fiez et al., 1996, Pugh et al., 1996), and phonological rehearsal (Kambara et al., 2018, Strand et al., 2008). However, some studies reported increased activation (Hoeft et al., 2007, MacSweeney et al., 2009, Shaywitz et al., 1998), while other studies reported decreased activation in this region (Cao et al., 2006, Cao et al., 2017, Georgiewa et al., 1999, Liu et al., 2012) in individuals with RD compared to controls, depending on the difficulty of the task and the age of the participants. In summary, the phonological deficit hypothesis has been a very influential theory in the literature of RD.

On the other hand, a visuo-orthographic theory has been proposed to explain the emergence of RD. It has been found that prolonged orthographic processing speed (Bowers and Wolf, 1993, Corcos and Willows, 1993, Serrano and Defior, 2008, Sunseth and Bowers, 2002), letter confusion errors (Manis et al., 1993, O’Brien et al., 2011, Valdois et al., 2012), letter reversal (O’Brien et al., 2011, Treiman, 1997), failure of irregular spelling-sound mapping (Bruck, 1993, Kemp et al., 2009, Manis et al., 1993, Treiman, 1997), and phonologically inaccurate spellings (Caravolas and Volín, 2001, Treiman, 1997) are associated with RD in some individuals. Supporting the visuo-orthographic deficit theory, reduced brain activation in the visuo-orthographic brain areas including the left occipito-temporal areas has been consistently reported in the literature (Cao et al., 2018, Hu et al., 2010, Maisog et al., 2008, Paulesu et al., 2001, Richlan et al., 2011). However, it is important to note that there is no single theory that can explain all cases of RD due to the heterogeneity of this condition.

The phonological theory was originally developed for alphabetic languages in which reading is based on grapheme-phoneme correspondence. Therefore, when the phoneme representation is deficient, it influences the correspondence with graphemes. In contrast, Chinese is a morpho-syllabic language in which each character represents a morpheme and a syllable. Written forms do not represent sound but meaning. No part of a character corresponds to a phoneme, but the whole character maps to the whole syllable. There are many homophones at the character level. Therefore, character-syllable mapping emphasizes the critical role of visuo-orthographic processing and the direct mapping from orthography to semantics in Chinese reading. Recent behavioral and neuroimaging research on Chinese RD revealed impairment in visual-orthographic processing (Cao et al., 2018, Ho et al., 2002, Meng et al., 2011, Siok et al., 2008, Tan and Liu, 2018, Wang et al., 2014, Woo and Hoosain, 1984). For example, some Chinese children with RD display orthographic deficit in tasks tapping into character orientation, character structure, and radical position (Ho et al., 2004, Ho et al., 2002). Chinese children with RD also show reduced activation in the orthographic regions such as the left occipito-temporal areas in semantic decisions on written words (Hu et al., 2010), rhyming judgment and semantic association judgment tasks (Liu et al., 2012), visual spelling judgment (Cao et al., 2018) and in the right lingual gyrus (LG) for a character size judgement task (Siok, Spinks, Jin, & Tan, 2009), suggesting their deficient orthographic processing.

As for phonological impairment, previous behavioral studies have found that Chinese children with RD had worse performance in phonological awareness (Ho et al., 2004, Ho et al., 2000, Huang and Zhang, 1997, Shu and Li, 2012) and phonological memory (Ho et al., 2000, Zhang et al., 1998) than controls. Neuroimaging evidence shows that Chinese children with RD have reduced activation in phonological related regions such as the left IFG, and middle frontal gyrus (MFG) (Cao et al., 2017, Liu et al., 2012, Siok et al., 2004). However, Chinese research has not shown that poor phonological processing in RD is associated with reduced activation in the left temporo-parietal areas in Chinese (Cao et al., 2017, Hu et al., 2010, Liu et al., 2013, Siok et al., 2008, Siok et al., 2009, Zhou et al., 2015), probably because Chinese is non-alphabetic and the conversion between orthography and phonology is different than that in English. In normal adult readers, it was found that the left temporo-parietal areas are more involved in English than in Chinese while the MFG is more involved in Chinese than in English (Bolger et al., 2005, Tan et al., 2005). Taken together, there are orthographic and phonological deficits associated with RD in Chinese as well as in other languages, even though the brain regions that are affected might be different. However, it is important to investigate phonological deficits and orthographic deficits in both spoken and written language processing and to understand the relationship between orthographic deficits and phonological deficits in RD. For example, is one deficit more severe than the other? Is one deficit more related to causal risks of RD than the other? How do orthographic and phonological deficits manifest in written word and spoken word processing? We planned to address these questions in the current study by directly comparing deficits in an auditory rhyming task and a visual spelling task in a group of children with poor reading in comparison to a group of age-matched control children and a group of reading matched control children.

In addition to the deficient phonological and orthographic network, another neurological abnormality associated with RD is reduced left-lateralization and/or compensation in the right hemisphere (e.g. Hynd et al., 1991, Qi et al., 2016). Normal readers usually develop left-hemispheric dominance in language processing (Richlan, 2012, for review). However, dyslexic readers have reduced left-asymmetry in both brain structures and brain functions. Previous studies found reduced gray matter in the left fusiform gyrus (FG) (Linkersdörfer, Lonnemann, Lindberg, Hasselhorn, & Fiebach, 2012), left IFG (Eckert et al., 2016, Jednoróg et al., 2014), left MFG (Siok et al., 2008), left superior temporal sulcus (Eckert et al., 2016, Richlan, 2020), and increased bilateral symmetry in the physical size of planum temporale for participants with RD (Bloom et al., 2013, Larsen et al., 1990, Morgan and Hynd, 1998, for review). Structural research on white matter suggested reduced left-lateralization in the structural integrity of inferior fronto-occipital fasciculus in children with RD, as well as increased right-lateralization in the second branch of the superior longitudinal fasciculus (Zhao, de Schotten, Altarelli, Dubois, & Ramus, 2016). Along with the structural findings, functional studies also found reduced left-lateral activations in the left FG (Linkersdörfer et al., 2012), and in the left-hemispheric reading network including occipito-temporal, temporal-parietal, and inferior frontal cortex (Richlan, 2020, for a review). Individuals with RD also tend to show greater involvement in the right hemisphere during language tasks. For example, stronger activation in the right prefrontal and bilateral posterior parietal cortices was found in children with RD during an N-back phonological judgement task (Xu, Yang, Siok, & Tan, 2015), suggesting potential compensation in the right hemisphere. A meta-analysis study (Maisog et al., 2008) also found that hyperactivities in the right thalamus and anterior insula were associated with RD, implying compensation in the right-hemispheric areas across various reading tasks. In summary, previous research suggest that RD is also associated with reduced left-lateralization and increased involvement of the right hemisphere.

Holistic network analysis provides novel information about the connectome, compared to brain activation at separate regions and/or functional connectivity between several brain regions. Graph theory is a mathematical method to conceptualize and quantify the topological structure of complex networks (Bullmore & Sporns, 2009). Graphs are mathematical representations of networks which define members of the network as vertices (also called nodes) and the relationships between nodes as edges (also called links) (Rubinov & Sporns, 2010). In studies of language processing, graph theory assesses the comprehensive pattern of information flow, concerning both global organization of the network and specific roles of individual regions (Beharelle & Small, 2016). Recently, a few studies have used graph theory to analyze aberrant neural organization in RD (Dimitriadis et al., 2013, González et al., 2016, Vourkas et al., 2011). In an EEG study conducted by González et al. (2016), a less integrated resting-state network in the theta-band was observed in children with RD compared to controls, and the theta-band was relevant to linguistic and cognitive abilities in reading. Several MEG studies consistently found that children with RD had lower global efficiency in both resting-state networks (Dimitriadis et al., 2013) and networks involved in pseudoword reading and letter sound naming (Vourkas et al., 2011). These studies showed abnormal long-range connectivity (e.g. global efficiency) in readers with RD.

Recent MRI research has also used graph theory to investigate abnormalities in brain structures in individuals with RD (Hosseini et al., 2013, Liu et al., 2015, Qi et al., 2016). In structural networks defined by cortical surface area and/or cortical thickness, dyslexic children’s networks had higher betweenness and nodal degree in several right-hemispheric regions such as the right inferior frontal gyrus and right lingual gyrus (Qi et al., 2016, Hosseini et al., 2013), but reduced nodal degree and betweenness in the left regions, such as the left precentral gyrus (Hosseini et al., 2013). This is consistent with the finding of increased involvement of the right hemisphere and decreased involvement of the left hemisphere in previous studies. On the other hand, research of structural networks defined by gray matter volume in children with RD (Liu et al., 2015) found sparse long-range connections and dense short-range connection around the left precentral gyrus in RD. In summary, these MRI studies defined structural networks based on anatomical properties such as cortical thickness, surface area, and gray matter volume, which may face a challenge when explaining the meaning of the connection between regions (Avena-Koenigsberger, Misic, & Sporns, 2018).

To our knowledge, there are three published fMRI studies that have examined the functional abnormalities in individuals with RD using graph theory. One study focused on narrative comprehension in RD (Horowitz-Kraus, Buck, & Dorrmann, 2016), and they found higher global efficiency in children with RD than control children indicated by lower average path length between nodes. In consistent, a study by Edwards and colleagues (2018) found that global efficiency negatively predicted performance in a visual-auditory rhyming task in children with RD, while the functional segregation indices positively predicted performance on the visual-auditory rhyming task. The third study found that the visual network in RD showed higher clustering coefficient and greater shortest path length than controls during an orthographic component judgement task, along with lower global efficiency (Yang & Tan, 2019), indicating that children with RD may have more segregated functional clusters and adopt different mechanisms than peers in orthographic processing. In summary, findings in the three studies are not consistent at all and that the topological alteration may vary according to tasks. Therefore, it is important to include both visual and auditory tasks and then to directly compare alternations in written and spoken language processing in order to have a comprehensive understanding of deficits associated with RD.

In the current study, we combined data from two published studies, one with an auditory rhyming task (Cao et al., 2017), and one with a visual spelling task (Cao et al., 2018). We examined the topology of functional networks involved in written language and oral language processing in Chinese children with poor reading (PR), in comparison to both an age-matched control group (AC) and a reading-matched control group (RC). Having two control groups would help distinguish possible causes and consequences of PR. When AC and RC are similar, both of whom are different from PR, the results indicate a possible cause of RD; on the other hand, when RC and PR are similar, both of whom are different from AC, the results indicate a consequence of PR, because they represent a performance effect. Having both an auditory task and a visual task allows us to directly compare deficits in these two tasks and to determine whether there are more deficits in written language processing than in oral language processing or vice versa. We were also interested in determining whether deficits in one task tend to be causal while deficits in the other tend to be consequences of RD.

Section snippets

Participants

Forty-two Chinese children in fifth-grade and third-grade were recruited from 8 different public elementary schools in Beijing, China. There were 17 fifth-grade children in the PR group (11 males, 6 females), 14 fifth-grade children in the AC group (7 males, 7 females), and 11 third-grade children in the RC group (6 males, 5 females). The exact age information is reported in Table 1. The criteria for PR in this study were: (1) the standard score on Raven was above 80, and (2) the score fell 1

Behavioral results

Table 2 presents behavioral performance in the AR and VS tasks for each group. We calculated the adjusted reaction time (RT) by taking accuracy into consideration in order to correct the speed-accuracy trade-off effect. A repeated-measure ANOVA of group (PR, AC, and RC) by task (AR and VS) was conducted for lexical trials.

A marginally significant group effect was found in accuracy for lexical trials (F(2, 39) = 3.09, p = .057). Post-hoc analysis showed that accuracy for lexical trials was

Discussion

In the present study, we examined topological characteristics of the brain network in Chinese poor readers during written word and spoken word processing, compared with age-matched and reading-matched controls. First, we found a lower clustering coefficient in the opercular part of the left IFG for PR than both control groups during the VS task, but not the AR task, suggesting deficient functional segregation in the left IFG in PR in the VS task. The PR group showed higher nodal efficiency than

Conclusions

In the current study, we examined the neural topological organization in Chinese poor readers in an auditory rhyming judgment and a visual spelling judgment task. Results indicate aberrant topological organization in PR in the visual spelling task as the left IFG forms a less specialized functioning cluster and bilateral STG show higher nodal efficiency. Furthermore, PR adopted additional right-hemispheric hubs in both the auditory and visual tasks, which may explain the higher global

Declaration of Competing Interest

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

Acknowledgements

Funding: This work was supported by the “Fundamental Research Funds for the Central Universities” (31650130) awarded to Dr. Fan Cao; by “Guangdong Planning Office of Philosophy and Social Science” (GD19CXL05) awarded to Dr. Fan Cao; and by “Science and Technology Program of Guangzhou, China, Key Area Research and Development Program (202007030011)”.

References (169)

  • F. Cao et al.

    Development of brain networks involved in spoken word processing of Mandarin Chinese

    NeuroImage

    (2011)
  • F. Cao et al.

    Neural signatures of phonological deficits in Chinese developmental dyslexia

    NeuroImage

    (2017)
  • L. Cohen et al.

    Distinct unimodal and multimodal regions for word processing in the left temporal cortex

    NeuroImage

    (2004)
  • N.E. Cone et al.

    Developmental changes in brain regions involved in phonological and orthographic processing during spoken language processing

    NeuroImage

    (2008)
  • M. Dhar et al.

    Reduced interhemispheric coherence in dyslexic adults

    Cortex

    (2010)
  • G.M. Di Liberto et al.

    Atypical cortical entrainment to speech in the right hemisphere underpins phonemic deficits in dyslexia

    Neuroimage

    (2018)
  • S.I. Dimitriadis et al.

    Altered temporal correlations in resting-state connectivity fluctuations in children with reading difficulties detected via MEG

    Neuroimage

    (2013)
  • Y. Dong et al.

    Neural mechanisms underlying the processing of Chinese words: An fMRI study

    Neuroscience Research

    (2005)
  • A.A. González-Garrido et al.

    The supramarginal and angular gyri underlie orthographic competence in Spanish language

    Brain and Language

    (2017)
  • C.S.H. Ho et al.

    Cognitive profiling and preliminary subtyping in Chinese developmental dyslexia

    Cognition

    (2004)
  • S.H. Hosseini et al.

    Topological properties of large-scale structural brain networks in children with familial risk for reading difficulties

    Neuroimage

    (2013)
  • J. Kronschnabel et al.

    The level of audiovisual print-speech integration deficits in dyslexia

    Neuropsychologia

    (2014)
  • W.J. Kuo et al.

    Orthographic and phonological processing of Chinese characters: An fMRI study

    Neuroimage

    (2004)
  • J.P. Larsen et al.

    MRI evaluation of the size and symmetry of the planum temporale in adolescents with developmental dyslexia

    Brain and Language

    (1990)
  • J. Liebig et al.

    Neurofunctionally dissecting the reading system in children

    Developmental Cognitive Neuroscience

    (2017)
  • R. Lindenberg et al.

    Supramodal language comprehension: Role of the left temporal lobe for listening and reading

    Neuropsychologia

    (2007)
  • K. Liu et al.

    Altered topological organization of brain structural network in Chinese children with developmental dyslexia

    Neuroscience Letters

    (2015)
  • L. Liu et al.

    Chinese dyslexics show neural differences in morphological processing

    Developmental Cognitive Neuroscience

    (2013)
  • L. Liu et al.

    Similar alterations in brain function for phonological and semantic processing to visual characters in Chinese dyslexia

    Neuropsychologia

    (2012)
  • F.R. Manis et al.

    Development of phonological and orthographic skill: A 2-year longitudinal study of dyslexic children

    Journal of Experimental Child Psychology

    (1993)
  • F.R. Manis et al.

    Are speech perception deficits associated with developmental dyslexia?

    Journal of Experimental Child Psychology

    (1997)
  • K. Murphy et al.

    The impact of global signal regression on resting state correlations: Are anti-correlated networks introduced?

    Neuroimage

    (2009)
  • A. Oron et al.

    Cross-modal comparisons of stimulus specificity and commonality in phonological processing

    Brain and Language

    (2016)
  • S. Achard et al.

    A resilient, low-frequency, small-world human brain functional network with highly connected association cortical hubs

    Journal of Neuroscience

    (2006)
  • A. Avena-Koenigsberger et al.

    Communication dynamics in complex brain networks

    Nature Reviews Neuroscience

    (2018)
  • J.R. Binder et al.

    Human temporal lobe activation by speech and nonspeech sounds

    Cerebral Cortex

    (2000)
  • T. Bitan et al.

    The interaction between orthographic and phonological information in children: An fMRI study

    Human Brain Mapping

    (2007)
  • D.J. Bolger et al.

    Neural correlates of orthographic and phonological consistency effects in children

    Human Brain Mapping

    (2008)
  • D.J. Bolger et al.

    Cross-cultural effect on the brain revisited: Universal structures plus writing system variation

    Human Brain Mapping

    (2005)
  • J.R. Booth et al.

    Development of brain mechanisms for processing orthographic and phonologic representations

    Journal of Cognitive Neuroscience

    (2004)
  • P.G. Bowers et al.

    Theoretical links among naming speed, precise timing mechanisms and orthographic skill in dyslexia

    Reading and Writing

    (1993)
  • L. Bradley et al.

    Categorizing sounds and learning to read—a causal connection

    Nature

    (1983)
  • M.M. Bradley et al.

    Activation of the visual cortex in motivated attention

    Behavioral Neuroscience

    (2003)
  • J.I. Breier et al.

    Abnormal activation of temporoparietal language areas during phonetic analysis in children with dyslexia

    Neuropsychology

    (2003)
  • M. Bruck

    Component spelling skills of college students with childhood diagnoses of dyslexia

    Learning Disability Quarterly

    (1993)
  • E. Bullmore et al.

    Complex brain networks: Graph theoretical analysis of structural and functional systems

    Nature Reviews Neuroscience

    (2009)
  • F. Cao et al.

    Deficient orthographic and phonological representations in children with dyslexia revealed by brain activation patterns

    Journal of Child Psychology and Psychiatry

    (2006)
  • F. Cao et al.

    The brain adapts to orthography with experience: Evidence from English and Chinese

    Developmental Science

    (2015)
  • F. Cao et al.

    Brain mechanisms underlying visuo-orthographic deficits in children with developmental dyslexia

    Frontiers in Human Neuroscience

    (2018)
  • F. Cao et al.

    Writing affects the brain network of reading in Chinese: A functional magnetic resonance imaging study

    Human Brain Mapping

    (2013)
  • Cited by (7)

    • The Brain Connectome for Chinese Reading

      2022, Neuroscience Bulletin
    View all citing articles on Scopus
    View full text