The link between brain development and language outcomes in children with DLD is unclear, and this lack of connection is apparent when reviewing the DLD neuroimaging literature. Over the past 50 years, there have been fewer than 60 neuroimaging studies (excluding EEG studies) with children diagnosed with DLD. The majority of these studies have focused on structural brain differences when compared to language-unimpaired (neurotypical) children or children with other neurodevelopmental language disorders, such as children diagnosed with ASD and concomitant language impairment. Though there are some consistencies that will be discussed below, it is important to note that the picture portrayed here is tenuous at best due to the limited number of studies that confirm these consistent findings as compared to the larger number of studies that have contrasting results.
2. Structural Brain Differences
Across development, the human brain undergoes a wide variety of structural changes in order to support increasing cognitive demands and the acquisition of new skills
[13]. The emergence of white matter pathways in the brain begins in utero following formation of the neural tube and production and migration of neurons
[14][15]. The brain then continues to differentiate and refine following birth. Infancy and early childhood are periods of rapid brain development, where the myelination of axons and synaptic reorganization and pruning are abundant in order to strengthen neuronal populations that frequently fire together as well as support those neuronal networks associated with new skills
[14][15]. As a result of these changes, the gross structure of the brain continues to visibly change within the first few years of life, with subtle decreases in gray matter volume and increases in white matter volume up until early adulthood
[16]. Thus, changes in individual neurons as well as neural networks impact gross brain structure across development.
Any perturbations to the tightly orchestrated processes that contribute to brain development can contribute to a range of developmental disorders
[17]. In fact, across different neurodevelopmental disorders, there is widespread evidence of volumetric brain differences compared to neurotypical peers
[18][19][20][21][22][23]. For example, individuals with ASD have been shown to have whole and regional brain volume differences when compared to age-matched control children. Lange et al. (2015) found that young children with ASD had larger overall brain volumes than their typically developing peers but as they aged, they showed atypical regional volume decreases
[24]. Findings such as these underscore the importance of investigating brain differences between typically and atypically developing populations, so that we can begin to uncover how structural (and functional) alterations contribute to language impairment. In the sections that follow, the researchers provide an overview of structural brain differences in DLD starting with global brain volume, moving to a discussion of gray and white matter volume and integrity. It should be noted that the structure of this overview should not be viewed as an annotated summary of findings, but instead presents thematically related information based on the broader categories that were just described.
2.1. Global Brain Volume
Studies investigating measurements of global brain volume (i.e., the whole brain, the hemispheres, and the cerebral lobes) in children with DLD have found differences when compared to neurotypically developing (TD) children, suggesting that like other neurodivergent populations, there are abnormalities in brain structure that are associated with the disorder. In typical development, total cerebrum volume and surface area increases from birth until reaching its peak at around 11–12 years of age
[25]. The few studies that have reported on these metrics in DLD point to children with DLD having smaller overall brain volumes, hemispheres, and cerebral lobes than typically developing children
[26][27][28][29]. However, other studies, such as Herbert et al. (2003), have found contrasting results (e.g., larger brain volumes compared to TD children)
[30]. These discrepant findings may be due to differences in methodological choices (e.g., voxel-based vs. semi-automatic brain morphometry). Another possibility to account for differences reported here, as well as throughout this research, is that children with DLD may have different brain volumes relative to their TD peers at different stages across development, since DLD is indeed a developmental disorder. Therefore, studies aimed at measuring longitudinal changes in DLD (similarly to Lange et al.’s 2015
[24] approach with ASD) are needed as they will help elucidate differences in measures of global brain volume across development.
2.2. Total Gray Matter Volume
Across the human lifespan, total gray matter volume increases in utero until its peak around 6 years of age followed by a non-uniform decrease from late childhood to late adulthood
[25]. Like with overall brain volume, studies investigating total gray matter volume in children with DLD have inconsistent findings because they employ different methods (e.g., volume-based vs. surface-based analysis) and their participant groups (for both DLD and TD children) rarely overlap in age, demographics, and inclusionary/exclusionary criteria (e.g., prior neurological history, assessment scores, etc.). As the researchers focus this research on converging findings, one common pattern is that reductions in overall gray matter volume (compared to TD children) are evident in younger children with DLD but diminish as they age. Soriano-Mas et al. (2009) used voxel-based morphometry (VBM) to examine gray matter differences in two groups of children with DLD, younger children under 12 years of age (
n = 19), and older children over 12 years of age (
n = 17)
[31]. They found that when looking at the DLD group as a whole, they had greater global gray matter volumes than TD age-matched controls, but when they separated participants by age group, the younger children with DLD had greater gray matter volumes than control children of the same age, while the older children with DLD showed no differences in gray matter volume compared to their neurotypical peers. Consistent with Soriano Mas et al.’s findings that morphological differences in DLD diminish with age, Badcock et al. (2012) found no differences in overall gray matter volume between older TD children and age-matched children with DLD (mean age: 13.5)
[32]. Typically, total cerebrum volume, surface volume, and subcortical volume reach their peaks around twelve years of age, with total gray matter volume peaking earlier, followed by a slow decline
[25]. It may be the case that the proportion of gray matter in TD children after six years of age decreases enough to match that of children with DLD at twelve years of age, while overall brain volume differences remain detectable between groups across late childhood and adolescence. These findings align well with the literature from other neurodevelopmental disorders, such as ASD, that report amplified volumetric differences in younger age groups that eventually diminish around 9–11.5 years of age
[33].
As described above, maturational differences underscore the importance of carefully considering the age range of participants included, but other methodological differences, such as not accounting for overall brain size within participants, can also impact results. Larger brains have more brain tissue and can give the appearance of increased gray matter, when in reality, the amount of gray matter may be proportional to overall brain size. To account for proportional differences in gray matter volume, Girbau-Massana et al. (2014) included intracranial volume (ICV) as a covariate in their analyses
[27]. Unlike Soriano-Mas et al.’s (2009) study
[31], which found that younger children with DLD had larger gray matter volumes, Girbau-Massana et al. (2014)
[27] found that the young children with DLD had
lower overall gray matter volumes than their TD peers. The discrepant findings again may be related to accounting for ICV or they could be due to the fact that 6 of the 10 children with DLD in their study were also diagnosed with a concomitant reading disorder. Regardless of the inconsistent findings, there is limited evidence indicating that children with DLD as a whole have abnormalities in gray matter volume. What remains unclear is the source of gray matter differences; that is, does it represent a difference in the overall surface area or is it due to differences in cortical thickness? Both cortical thickness and surface area follow different developmental trajectories, suggesting that they are driven by partially distinct processes
[34]. In the only large-scale study investigating these metrics in DLD, Bahar and colleagues (2023) found that children with DLD (ages 10–16 years old) had lower measures of surface area and, to a lesser extent, volume, but not cortical thickness, when compared to their TD peers
[35]. Changes in the surface area of the brain are associated with gyrification (or the folding of brain tissue), which is related to more efficient neural processing
[36]. Based on these findings, it is possible that differences in gray matter volume may be related to less efficient neural networks (as indexed by decreased surface area) within language-related brain regions in children with DLD.
In the next section of this research, the researchers summarize regional gray matter volume differences for areas linked to language function to explore how differences in gray matter volume within language-related brain regions can impact language abilities
[37][38][39][40]. The researchers also discuss differences in regional asymmetries as they are a common feature of typically developing brains that support growing functional specializations between the two hemispheres.
3. Regional Brain Differences
One of the most consistent findings in children with DLD is anomalous gray matter volume and symmetry within the perisylvian language zone. Though across the DLD literature a number of regions within this zone have been discussed
[41][42], this section highlights three specific regions which have consistently been shown to have different characteristics in children with DLD as compared to TD children, namely the planum temporale and inferior frontal gyrus (
Figure 1a). In addition to these standard language regions, the other region that will be discussed is the caudate nucleus, as it has been theorized to support speech and language processes (
Figure 1b).
Figure 1. Commonly reported regions of interest in studies investigating structural gray matter differences in DLD. (
a) The inferior frontal gyrus and the planum temporale. Note that the transparency of the planum temporal is meant to indicate that it is not visible on the lateral surface of the brain. (
b) The caudate nucleus located subcortically. Note that it can be further subdivided into the head, body, and tail, but will be discussed as a whole in the text. Figure adapted from Hugh Guiney, CC BY-SA 3.0, via Wikimedia Commons
https://en.m.wikipedia.org/wiki/File:Human-brain.SVG (accessed on 12 September 2023).
3.1. Planum Temporale
Perception and processing of speech sounds requires the ability to attend to rapidly changing auditory stimuli. The planum temporale is thought to play a key role in this process and has been posited to be a factor in the left hemisphere’s dominance for language
[43]. Prior research has indicated that the left planum temporale is comprised of densely packed cortical columns that facilitate processing at higher temporal resolutions
[44]. As a result of this left hemisphere specialization, the planum temporale is generally larger on the left than the right in typically developing populations
[45][46][47]. However, children with DLD, as well as children from other language-impaired populations, such as those with ASD with language impairment
[48] and dyslexia
[49], do not exhibit this pattern, suggesting that there is a connection between language impairment and alterations in volume of the planum temporale.
In children with DLD, the exact nature of these differences is unclear as some studies have demonstrated that they have larger planum temporale in the right hemisphere
[16][26][50], while others have found that when compared to controls, the planum temporale size is reduced in the left hemisphere
[51][52][53], or equal in size
[54]. Differences in methodological approaches in the segmentation of the planum temporale may have contributed to the conflicting outcomes. For example, in a study by De Fossé et al. (2004), the authors noted that they included broader regions (the horizontal and posterior ascending portions) of the planum temporale in their measurements (delineated by a trained technician using anatomical landmarks), whereas other studies only included surface volume measurements based on automatic parcellation methods
[55]. Though the studies referenced here contrast in terms of the directionality of differences (likely due to methodological differences), both indicate a potential link between abnormal planum temporale asymmetry and language impairment.
3.2. Inferior Frontal Gyrus
Another commonly implicated region in children with DLD is the inferior frontal gyrus (IFG). The IFG is comprised of three functional regions, the pars triangularis, the pars opercularis, and the pars orbitalis, though the pars opercularis is sometimes further subdivided into dorsal and ventral regions (
Figure 2)
[40]. This distinction is important as it can impact structural and functional comparisons across studies, yet studies often report results for the entire IFG or for the region known as Broca’s area (pars opercularis and pars triangularis) rather than the individual components. The functional role of left IFG regions has been a long-standing source of debate amongst researchers (see summary by Rogalsky et al., 2008)
[56][57][58]. Though theories differ, they seem to converge on Broca’s area being a language support region and the left pars orbitalis playing a role in covert articulation and semantic processing
[59].
Figure 2. The three subdivisions of the inferior frontal gyrus shown in the left hemisphere.
In typically developing individuals, the left IFG tends to be larger in the left hemisphere than in the right, which is thought again to reflect left hemisphere language dominance
[55]. In children with DLD, the regions within the left IFG have been shown to lack this pattern with some studies reporting that they have more gray matter in this region
[32][60][61] and other studies reporting that the left IFG is smaller in children with DLD compared to controls
[26][55][62]. Interestingly, of the studies reporting overall smaller left IFG volumes in DLD, some pointed to a group correlation between a smaller left pars triangularis and worse observed language outcomes
[26][62]. However, as has been emphasized throughout this research, the interpretation of findings should be approached with caution as results are often affected by research design. For example, Plante et al. (1991) found that the size and symmetry pattern of the IFG in children with DLD did not differ from age-matched peers as a group, but when looking at
individual patterns of symmetry within the IFG, results varied from child to child, with some demonstrating reversed asymmetry while others did not
[50]. Plante et al.’s results may be due to maturational effects, given that the participants varied in age from 4 years to almost 10 years of age. Another possibility is that there may be a tradeoff between abnormal development of the left IFG early in life and compensation of other structures in the brain. In support of the latter possibility, Lee et al. (2020) reported individual differences in the size of the left IFG with some children with DLD in their study having increased volume in the right pars orbitalis, which may indicate compensation of the right IFG in those who had abnormal left IFG volumes
[60]. More research is clearly needed to understand differences in the left IFG volume and symmetry in children with DLD as well as the functional and structural connection between left and right homologue regions. As pointed out in the planum temporal section above, results converge on a similar theme; abnormal brain patterns in DLD can be linked to observed language impairment.
3.3. Caudate Nucleus
Other subcortical regions outside of the cortex, such as the basal ganglia and thalamus, have also received attention in children with DLD, as they connect to other language regions and are thought to play a supportive role in certain aspects of language such as complex syntactic processing, semantic processing, phonological processing, word generation, and more generally, attentional resource allocation
[63][64][65][66][67]. Focusing in on one of the structures within the basal ganglia, the caudate nucleus (also referred to as the caudate), some studies have found that it tends to be larger in the left hemisphere than in the right in typically developing individuals and it has projections to other cortical regions involved in verbal working memory, language comprehension and production, and procedural learning
[65][68][69]. However, in children with DLD, studies have reported a bilateral reduction in gray matter volume of the caudate, with the exception of Soriano-Mas et al. (2008), who reported an increase in caudate volume in younger children that was not present in older children
[29][30][31][32].
The caudate has also been implicated in the neuropathology of the famous KE family, known for their genetic mutation of the FOXP2 gene which, it has been argued, has rendered the language/speech of over half the family members agrammatic (i.e., lacking grammatical features) and unintelligible. In a study specifically looking at the relationship between language impairment and the caudate in this family, Watkins et al. (2002) found that the affected family members had smaller caudate volumes than unaffected family members or age-matched controls, and that the size correlated with nonword repetition abilities
[61]. Interestingly, in a study by Krishnan et al. (2022) investigating the macromolecular content of gray matter in the caudate in DLD, they found a reduction in myelin content in the caudate nucleus (as well as the inferior frontal gyrus) when compared to controls and this reduction was related to lower language proficiency
[70]. During childhood and adolescence, myelin content increases across cortical gray matter regions, while subcortical regions see less pronounced changes
[71]. Further, brain regions associated with higher-order cognitive functions, such as language, require longer periods of myelination and are impacted by not just genes but also the environment
[72]. It may be the case that the corticostriatal circuit, connecting regions of the caudate to the frontal lobe, associated with learning, is aberrant in DLD. Though more studies are needed, novel findings about the underlying architecture of commonly implicated brain regions in DLD may reveal important insights into DLD pathology. Given that brain regions do not function in isolation, the researchers now turn the discussion to the white matter pathways that support structural connections between these commonly implicated gray matter regions.
4. White Matter Pathways
Historically, investigations into brain function have focused on the size and engagement of cortical and subcortical gray matter regions of the brain. However, a missing piece of those investigations are the connections that allow gray matter regions to coordinate activity. Since we know brain regions do not operate in silos, recent technological advances have provided scientists a way to measure the important connections that exist between gray matter regions, known as white matter, which comprises the structural wiring of the brain (
Figure 3). Due to the early development of the brain, white matter pathways are already present by 30 weeks of gestation
[73]. However, between birth and two years of age, children undergo a period of rapid brain development, highly influenced by genes and the environment, that helps further shape the neural architecture of the brain
[14][74]. As children continue to develop, white matter volume continues to increase until around the fourth decade of life to support improvements in cognitive skills
[13][75][76].
Figure 3. Example of white matter pathways connecting gray matter regions in the left hemisphere.
There is a strong relationship between neural activity associated with new skills and the formation of efficient, myelinated white matter pathways that connect gray matter regions throughout the brain
[13]. As a result of this critical interaction, any deviation in the typical formation of white matter pathways will likely contribute to functional impairments
[14]. Prior research has demonstrated a link between white matter alterations and language impairment in children with a range of neurodevelopmental disorders, including autism spectrum disorder, dyslexia and other reading disorders, and epilepsy
[77][78][79][80]. While research on white matter connectivity in children with DLD is limited, like other neurodevelopmental populations, there seems to be a connection between language impairment and altered white matter volume and diffusivity (movement of water molecules along white matter pathways). Here, the researchers provide an overview of these findings starting with white matter volume changes in DLD as compared to TD and then the researchers turn attention to the diffusivity of white matter tracts involved in language processing within dorsal and ventral regions.
4.1. White Matter Volume
The findings from studies investigating overall white matter volume in children with DLD parallel the findings reported earlier in this research for gray matter volume, in that results are varied due to differences in methodology and age of participants. However, there is some consistent evidence for children with DLD having overall increased white matter volume
[28][30][31]. Similar to the gray matter pattern discovered by Soriano-Mas and colleagues (2009), these researchers discovered that an increase in white matter may be mediated by age. In their 2009 paper, Soriano-Mas and colleagues found that younger children (<12 years of age) with DLD showed an increase in white matter volume, while older children (>12 years of age) did not
[27][31]. In contrast to these results, Girbau-Massana et al. (2014) did not find significant differences in white matter volume from their sample of younger children (mean age: 9.4 years) when compared to controls
[27]. These difference in findings may be attributed to different methodological approaches. Unlike Soriano-Mas (2009)
[31], Girbau-Massana et al. (2014)
[27] included measures of intracranial volume in their analysis, which as noted previously, is an important consideration when calculating measures of brain volume. At first glance, the results from Soriano-Mas that link an increase in white matter volume to language impairment in DLD are surprising, since increases in white matter volume have been interpreted as improved cognitive skills across development
[76]. In this case, a different interpretation is warranted. Findings of abnormally increased white matter volume in children with DLD may indicate that the underlying microstructure is impacted, which could result in poor connectivity across networks. Thus, in the next section, the researchers turn to the literature investigating white matter integrity using diffusion MRI.
4.2. White Matter Diffusivity in Child Language-Impaired Populations
If the transmission of signals between different neural regions is impaired or slowed in any way, it could lead to disruptions of the tightly orchestrated processes needed for successful language processing and production, such as early auditory processing or the production of appropriate grammatical morphemes. Further, it can impact the strength and efficiency of language networks, which could lead to changes in connectivity as well as white and gray matter volume. Diffusion MRI (dMRl) is an imaging method that can be used to reveal details about the integrity of white matter pathways by measuring the diffusion of water molecules in brain tissue. This method has been commonly used to track neurodevelopmental changes attributed to normal brain maturation, as well as assess the underlying integrity of these connections in children with neurodevelopmental disorders.
As children mature and new skills are mastered, the brain develops more efficient structural and functional neuronal connections. These changes can be characterized by different diffusion tensor imaging (DTI) indices, such as fractional anisotropy (FA), mean diffusivity (MD), axial diffusivity (AD), and radial diffusivity (RD), that reflect movement and direction of water molecules along fiber bundles over a period of time (see
Appendix B for a description of these indices)
[75]. In typically developing children, the general expectation is that across development axons become more densely packed and myelinated
[78]. This is represented by
increases in FA and AD along the primary axis and
decreases in MD and RD along the orthogonal axes. Additionally, these indices are expected to exhibit asymmetries across the two hemispheres as different brain regions develop at different rates due to experience. However, in populations with neurodevelopmental language disorders, such as ASD, researchers have found altered patterns of white matter diffusivity in left hemisphere language tracts, which may contribute to language impairment
[81][82]. In children with DLD, differences in white matter development may underlie the language impairments observed. The next section of this research explores dMRI findings along dorsal and ventral language pathways in children with DLD.
4.3. Dorsal and Ventral Language Pathways
There is an abundance of evidence supporting a dual-stream (dorsal and ventral;
Figure 4) model of language processing in the brain. Dorsal stream white matter pathways are thought to map sound to distinct linguistic units, and ventral stream white matter pathways are thought to map sound to meaning
[83][84][85]. As described below, some studies have reported differences along these pathways in children with DLD when compared to age-matched, typically developing children.
Figure 4. White matter pathways within the dorsal (blue) and ventral (pink) streams. Dorsal stream paths include the superior longitudinal fasciculus (SLF) and arcuate fasciculus (AF). Ventral stream paths include the uncinate fasciculus (UF), inferior frontal occipital fasciculus (IFOF), middle longitudinal fasciculus (MdLF), and inferior longitudinal fasciculus (ILF).
4.4. Dorsal Pathway Findings in DLD
From the few studies that have investigated microstructural brain differences in children with DLD, one of the common findings in the dorsal language pathway is decreased FA in the superior longitudinal fasciculus (SLF) and an increase in either MD or RD in the arcuate fasciculus (AF)
[60][86][87]. This pattern is in contrast to what occurs in typically developing children in which as language skills develop, myelin content increases to create more efficient white matter pathways, which is indexed by an increase in FA and a decrease in MD and RD
[87]. This lack of change in children with DLD indicates abnormal development of dorsal stream tracts involved in language processing.
4.5. Ventral Pathway Findings in DLD
Longitudinal studies of ventral white matter development in typically developing children have indicated that the inferior fronto-occipital fasciculus (IFOF) and inferior longitudinal fasciculus (ILF) develop early on with increases in FA and RD and decreases in AD and MD values throughout childhood and adolescence until they peak in young adulthood. Interestingly, the uncinate fasciculus (UF) does not reach peak values until around 30–40 years of age
[75][88]. Even fewer studies have investigated ventral language tracts in children with DLD, but the most consistent findings are decreased FA in the left hemisphere IFOF, UF, and ILF, with some limited evidence of decreased FA in the right hemisphere as well
[60][87][89]. Vydrova et al. (2015) also found increases in MD and RD in the left hemisphere IFOF and ILF and a bilateral increase in RD for the UF
[87]. This pattern of decreased FA and increased MD and RD in ventral stream language tracts of children with DLD differs from the typical pattern of white matter development. Additionally, in line with reported asymmetry and volume differences in gray matter brain regions, there also seems to be a lack of leftward asymmetry and an increase in volume in the ILF and IFOF
[87][89].
In sum, there is converging evidence that suggests that children with DLD have altered micro- and macro-structures within language-related white matter pathways, but the correlation with observed language abilities remains elusive as studies have used a limited range of standardized language assessments, if at all, and as such, only a few have found correlations between language abilities and DTI indices. Importantly, differences in white matter architecture point to a possible contributor to language impairments in DLD. While there is a dearth of evidence for DLD, we can look to findings with other language-impaired populations to direct future investigations of white matter pathway anomalies and their effect on language abilities. Alterations in the development of the architecture supporting different higher-order functions such as language and learning systems could lead to alterations in function of different brain regions, which may in turn contribute to neurodevelopmental disorders. In fact, research with infants at risk for developing autism and dyslexia has revealed that early alterations in white matter pathways can impact developing language abilities among others, as the coordination and function of cortical networks is constrained by the architecture of white matter pathways
[78][90][91]. Given the structural alterations reported in gray and white matter regions of the brain in children with DLD, the investigation of functional brain activity during language tasks is warranted. In the next section of this research, the researchers explore studies investigating functional brain activation patterns in those diagnosed with DLD.