The anterior temporal lobe (ATL) is suggested as a semantic hub that may support reading inconsistent words via semantic access. Surface alexia is characterized by difficulty in reading these inconsistent words, and often co-occurs with ATL atrophy and semantic impairments. However, the role of ATL in word reading is unclear, as evidenced by cases of alexia without semantic impairments and semantic impairments without surface alexia. To test its role in reading, we stimulated the ATL using transcranial Direct Current Stimulation in neurotypical participants performing a Word Naming task including consistent and inconsistent words, and a Picture Plausibility (PP) task involving nonverbal semantic judgments. We also stimulated the left tempo-occipital cortex (TOC) during the PP task. Low-frequency inconsistent words were read more slowly than high-frequency inconsistent words in the sham condition. This interaction disappeared with ATL stimulation. We found an interaction between consistency and stimulation in the low-frequency subset, but not in the high-frequency subset. In the PP task, ATL stimulation had no effect, whereas TOC stimulation significantly influenced reaction time. These findings support the left ATL's role in reading inconsistent words, aligning with surface alexia with ATL atrophy. The results also suggest that the left lateral ATL may be involved in lexical rather than purely semantic processes, consistent with lesion studies demonstrating a dissociation between semantic impairments and surface alexia, and possibly also with graded modality-specific specialization of bilateral ATLs. These findings reconcile conflicting findings and elucidate the role of left lateral ATL in reading.
Lexical access to common words is remarkably robust to changes in surface format (e.g., font, case, color, size), supporting abstractionist models in which orthographic codes are largely invariant to perceptual details. However, recent behavioral experiments have shown that brand names, a type of lexical item that is typically encountered in a highly consistent visual format, are sensitive to modifications of surface format (Labusch, Duñabeitia, & Perea, 2024; Perea et al., 2022). This research has largely focused on behavioral measures, leaving open questions about the neural mechanisms underlying brand name recognition. Here, we used fMRI to examine whether the visual word form area (VWFA) and related regions are sensitive to font and letter case modifications in brand names in a semantic categorization task (brand related to transportation or not). Participants viewed brand names presented either in their typical visual format, or with a modified font or case. Font modifications slowed responses and elicited increased VWFA activation relative to intact logos, whereas case modifications also slowed responses but did not reliably modulate activity in the canonical VWFA ROI. This dissociation constrains strong claims of strict format invariance in the VWFA, while remaining compatible with hierarchical or interactive accounts in which some ventral occipitotemporal populations retain sensitivity to learned surface forms. Rather than supporting a categorical claim of case invariance, the present data indicate stronger sensitivity to font than to letter case when a perceptual format is consistently tied to lexical identity.
Language, as a uniquely human faculty, combines words into hierarchical phrases with diverse syntactic-semantic relations (e.g., subject-predicate and modifier-head phrases). These phrase types are theoretically generated through the labeling algorithm at the syntax-semantics interface. Nevertheless, previous research has focused predominantly on the syntactic system, whereas the neural basis and temporal dynamics of labeling remain poorly understood. To address this gap, we developed the Head-Anchored Labeling Manipulation approach. This method manipulates labeling by embedding the same Mandarin noun-verb dual-category words as heads in tightly controlled modifier-head constructions, where the head alone determines the phrase's grammatical type. The representational similarity analysis of EEG data revealed that labeling representations emerged during both the early and middle stages: 192-227, 290-318, 330-360, and 385-416 ms following the dual-category word. Labeling emerged as early as ~200 ms, concurrent with syntactic Merge, and continued into the N400 window. These results demonstrate that labeling dynamically links syntax to conceptual-intentional system and generates phrase types by determining the constituent head. Its early engagement challenges syntax-first models and supports parallel interactive accounts. Moreover, combinability representations (520-596 and 604-632 ms), together with a late event-related brain potential negativity (420-700 ms) elicited by the dual-category word in the baseline condition, reflect increased difficulty in reconciling semantic associations with atypical syntactic configurations. Together, these findings provide clear electrophysiological evidence for the temporal dynamics of labeling and elucidate the processes at the syntax-semantics interface.
The non-selective processing hypothesis for bilinguals posits that during the processing of the target language, the lexical information of the non-target language is concurrently activated. Nonetheless, the lexical representation of bilinguals in both languages and the temporal dynamics of lexical information representation remain unclear. Here, we utilized fMRI in Experiment 1 and EEG in Experiment 2, combined with representational similarity analysis (RSA), to explore the neural representation of lexical access in both the target and non-target languages during L2 word reading and the modulatory effects of processing demands. Results of two Experiments jointly revealed that the lexical information of L2 was represented widely and earlier during L2 lexical processing, whereas the lexical information of L1 was represented widely and earlier during L2 semantic processing. These findings provide spatiotemporally integrated evidence for the bilingual non-selective access hypothesis, indicating that non-selective processing occurs only under conditions of high semantic processing demands.
Writing systems are cultural inventions that differ in how they represent spoken language. We tested how the brain learns to map arbitrary visual symbols to sound and meaning by comparing neural activity for artificial writing systems that were either alphabetic (systematic symbol-sound mappings) or logographic (arbitrary symbol-sound mappings). Twenty-four adults learned to read aloud and comprehend novel words written in each system. After 2 weeks of training, functional magnetic resonance imaging during reading comprehension revealed that the dorsal pathway (inferior parietal gyrus, inferior frontal gyrus), which is involved in mapping from print-to-sound, was more active for the alphabetic system, whereas the ventral pathway (anterior fusiform gyrus, middle temporal gyrus), which is involved in mapping from print-to-meaning, was more active for the logographic system. Combined with performance differences, these findings indicate that systematic symbol-sound mappings allowed the brain to bridge the interface between vision and meaning via sound, whereas an absence of systematicity made it more efficient to link vision directly to meaning. Thus, the same brain finds different solutions to the problem of reading that capitalise on the statistical properties of culturally invented writing systems.
Semantic memory retrieval is essential for language, thought, and adaptive behavior. Although both the prefrontal cortex (PFC) and the cerebellum have been implicated in this function, the role of the PFC remains poorly understood and the contribution of the cerebellum largely overlooked in current neurocognitive models. To address these gaps, we conducted a double-blind, randomized, placebo-controlled experiment in which healthy adults received anodal transcranial direct current stimulation (tDCS) targeting the left lateral PFC, the right posterior cerebellum, or sham stimulation. Participants completed a novel process-sensitive paradigm comprising lexical decision, free-associative (FA; automatic) retrieval, dissociative (DA; controlled) retrieval, and intrusion monitoring, while manipulating response predictability and rule switching. Cerebellar tDCS selectively impaired FA performance, particularly for cues evoking predictable responses, supporting its role in automatic access to overlearned associations. In contrast, prefrontal tDCS disrupted DA performance and increased associative intrusions, implicating the PFC in retrieval inhibition. Importantly, mediation analysis showed that the reduction in DA fluency was largely explained by higher probability of intrusions, indicating a perturbation of proactive inhibitory control that normally prevents irrelevant memory activations from entering working memory. Further exploratory analyses ruled out several alternative accounts of these stimulation effects, underscoring their process specificity. Together, these findings advance models of semantic cognition by demonstrating complementary contributions of the cerebellum to automatic retrieval and of the PFC to inhibitory control over intrusions.
Classroom-based language learning has typically taken place in relatively static bodypositions, but research suggests that embodied learning through sensorimotor engagement andtechnical immersion, using virtual realities, can significantly enhance learning outcomes.Recent research has linked differences in the learning context to different cortical structureswithin the language learning network. In this study, we investigated the effect of technicalimmersion and sensorimotor engagement on performance in behavior and gray matter volumein the brain after a single 20-min language learning task. We tested two learning environments:a low-embodied desktop-based virtual environment (dVE) using a computer screen and a high-embodied immersive virtual reality (iVR) environment using a head-mounted display, as wellas a no training group. We assessed morphological brain changes using magnetic resonanceimaging at 7 Tesla before and after training. Participants with less sensorimotor engagement,compared to those with high, performed significantly better and showed higher gray mattervolume in the left angular gyrus, a key hub region for vocabulary training within the languagenetwork, as well as in the left middle temporal gyrus, a region associated with lexical semanticprocessing. However, we could not identify a difference between the dVE and iVR groups. Ourresults suggest that both virtual platforms, although different in the level of immersion andwhole-body involvement, rely on similar cortical structures within the language learningnetwork. Furthermore, sensorimotor engagement might have a stronger influence onperformance and related brain changes than the learning context itself.
Rapidly processed magnocellular (M) information may facilitate visual object recognition but its role in reading is unclear. A previous study with Chinese characters and masked foveal primes did not find a unique role of the M system as compared to the parvocellular (P) system in mediating repetition effects. As M cells are better represented in the parafoveal visual field, the present study tested whether the M and P systems contribute differentially to parafoveal processing during reading. We combined EEG recordings and eye tracking to measure parafoveal preview effects in fixation-related potentials, using the boundary paradigm. In two experiments, we contrasted high versus low spatial frequency previews and luminance versus color contrast previews and also included standard previews as a manipulation check. As expected, the N250 component was diminished after valid as compared to invalid normal previews, especially over the left hemisphere. We also obtained left-lateralized preview effects for the N250 component for both M- and P-biased previews in both experiments. In the experiment involving a spatial frequency manipulation, P-biased preview effects tended to be larger than M-biased preview effects over the left hemisphere, but not over the right hemisphere. No interactions with preview validity were found for the luminance contrast manipulation. This null effect was supported by a Bayesian analysis. Taken together, these results indicate that the M pathway does not exclusively mediate the preview effect, even for stimuli presented in the parafovea. Instead, both M- and P-based information appear to contribute to early, left-lateralized neural processes underlying visual word recognition.
Successful coordination of the functional networks underlying reading is highly dependent on the underlying white matter tracts that connect these regions. Continuous theta burst stimulation (cTBS) can temporarily inhibit brain activity in targeted brain networks and has been shown to modulate reading ability. It was hypothesized that measures of diffusivity would predict change in reading efficiency following stimulation to the left temporal-parietal junction (TPJ), a targeted node of the dorsal stream of the reading network. Fifty-three adults between the ages of 18 and 50 years (M = 22.79, SD = 5.40; 34 female) with a range in reading ability completed sight word and pseudoword fluency measures prior to and immediately following cTBS to either the left or right TPJ or a control site. Participants also completed an MRI session including anatomical and diffusion weighted imaging sequences. Regression analyses were conducted to predict change in reading fluency following cTBS. Tracts associated with the reading network including the arcuate fasciculus, inferior longitudinal fasciculus, and portions of the superior longitudinal fasciculus, as well as the corpus callosum significantly predicted reliable change in a test of word reading efficiency-sight word efficiency subtest following stimulation of the left TPJ. These findings suggest that increased diffusivity of white matter tracts associated with the left hemisphere reading network and their right hemisphere homologues may support the impact of cTBS following stimulation to a targeted node of the reading network. Individual differences in white matter diffusivity may underlie differences in behavioral outcomes following neuromodulation.
Public speaking is a fundamental form of communication across a wide range of domains; however, the neural mechanisms underlying audience engagement during different speeches remain poorly understood. In particular, it is unclear which functional brain networks support the dynamic fluctuations of audience engagement and what neurobiological processes underlie these effects. In this study, we used naturalistic fMRI combined with intersubject correlation (ISC) analysis to examine how carefully selected and matched speeches, with varying levels of audience engagement, influence neural activity. Our results revealed that the more engaging speech elicited significantly greater interbrain neural synchronization, as indexed by ISC, across a broad range of brain regions. Notably, these engagement-related effects were most prominent in networks associated with language processing and theory of mind, highlighting their critical roles in facilitating shared audience experiences during compelling public communication. A sliding-window analysis further revealed substantial temporal fluctuations in interbrain synchronization throughout the speech. Additionally, neurobiological annotation analyses identified strong associations between engagement-related ISC effects and molecular pathways involved in trans-synaptic signaling, suggesting that intrabrain neuronal communication may contribute to modulating interbrain synchronization. By integrating naturalistic fMRI with ISC analyses, this study offers a promising framework for investigating dynamic neural synchronization among audience members. These findings have broad implications for fields such as education and leadership development, where a deeper understanding of the neural basis of audience engagement could inform strategies to enhance public speaking and communication effectiveness.
Age-related declines in cognitive function are often accompanied by changes in brain activity and network organization. This study investigated the relationship between resting state brain activity and age-related differences in speech production. We hypothesized that older adults would exhibit altered functional connectivity and activation intensity, correlating with reduced speech quality. Resting state functional MRI data were collected and a composite measure of speech complexity and fluency was calculated from younger and older adults. Results revealed significantly worse speech performance in older adults, accompanied by less segregated whole-brain networks, reduced amplitude of low-frequency fluctuations, and more heterogeneous brain states. Univariate regression analyses indicated stronger brain-behavior relationships in younger adults, while multivariate regression analyses revealed that age-related differences in resting state brain state patterns critically relate to speech production differences. Notably, the language network remained relatively stable with age, whereas whole-brain status became very important for speech performance in older adults. These findings suggest that resting state brain activity, particularly whole brain network characteristics, may serve as a stable biomarker of age-related changes in speech production.
Developmental stuttering is a complex neurodevelopmental condition associated with structural and functional anomalies in the basal ganglia-thalamo-cortical (BGTC) circuits that support speech planning and execution. In this study, we examined hypothesized impairments in the planning and motor circuits of the speech network in children who stutter (CWS), compared to children who do not stutter (CNS), using person-specific functional connectivity maps derived from resting-state functional magnetic resonance imaging (rsfMRI) data. RsfMRI data were acquired from 73 CWS and 74 CNS, aged 3 to 10 years. Twelve regions of interest within the speech motor networks were extracted. Functional connectivity was assessed using confirmatory subgrouping group iterative multiple model estimation (CS-GIMME), which estimates group-, subgroup-, and individual-level connections. Subgroup-level functional connectivity patterns revealed altered connections among CWS in both planning and motor loops, including reduced within-network connectivity, compared to CNS. CWS showed connectivity between the left posterior inferior frontal sulcus and left ventral lateral thalamus that was not observed in CNS. Furthermore, centrality of the left ventral lateral thalamus and right ventral premotor cortex were increased in CWS relative to CNS. Significant differences between CWS and CNS in within-network connectivity highlight early developmental alterations that affect the BGTC circuitry, pointing toward inefficiencies in the neural network that supports the programming, planning and timing of speech motor sequences.
Speech development requires precise timing and sensorimotor integration, supported by neural oscillations that synchronize activity across auditory, motor, and cognitive circuits. Among speech-relevant frequency bands, beta oscillations (13-30 Hz) are critical for timing and coordination, supporting sensorimotor processing and speech preparation. Beta desynchronization (power decreases) is typically observed prior to movement, reflecting motor planning, and beta activity also supports cognitive functions such as attention and anticipation. Although age-related changes in beta power have been documented, its developmental trajectory during speech processing remains underexplored. Here, we compared beta power dynamics in 28 adults (mean age = 27.8 yr) and 50 children (mean age = 10.3 yr) during speech perception and production tasks using EEG. On each trial, participants received a visual cue indicating the condition ("Say," "Hear," or "See"), followed by rhythmic tones and a warning cue presented as a picture with its name (e.g., cat) in Say and Hear, but only as a picture in See. A Go cue then prompted participants to speak, listen, or maintain fixation. Beta power was analyzed in three time windows: postwarning cue (P1), pre-Go cue (P2), and post-Go cue (P3). Adults exhibited significant beta power decreases across all time windows, particularly in Say, indicating mature sensorimotor and cognitive integration. In contrast, children showed no significant condition effects and minimal beta reduction in P3. Beta modulation was negatively correlated with age in children, suggesting ongoing maturation of beta oscillations. These findings highlight key developmental differences in beta oscillations relevant to speech processing.
Functional imaging and clinical cases implicate the left thalamus in object naming, yet the prevalence of naming impairment after focal thalamic damage is low with variable impact and often rapid resolution. This suggests that compensatory mechanisms, within or beyond the thalamus, may support recovery. We hypothesized that thalamic damage would (a) not cause chronic anomia if other naming-related regions remain intact but (b) exacerbate anomia when co-occurring with damage to non-thalamic naming regions. To test these hypotheses, we retrospectively assessed naming ability in 550 left hemisphere chronic stroke survivors (52% with anomia). Lesion sites included focal thalamic lesions (n = 14), combined thalamic and non-thalamic lesions (n = 271), and lesions sparing the thalamus (n = 265). Whole-brain lesion-symptom mapping (LSM), using multivariate support vector regression, identified brain regions where damage was significantly related to naming ability. Contributions of different thalamic subregions to naming were assessed using ridge regression. Focal thalamic lesions were not associated with chronic anomia. LSM identified two naming-related clusters: a temporoparietal region of interest (ROI-TP) and a subcortical-insular region of interest (ROI-SC) including the lateral thalamus. However, lesion load in the lateral thalamus did not independently contribute to naming performance when controlling for damage to other parts of the ROI-SC, nor did any thalamic nuclei show additive effects beyond the ROI-TP and the non-thalamic ROI-SC. These findings suggest that thalamic damage in the dominant hemisphere does not cause long-term anomia in chronic stroke. Future research therefore needs longitudinal designs to track the trajectory of transient thalamic effects from the acute to chronic phases and to investigate whether naming impairments after thalamic lesions are (a) lesion specific but context dependent, emerging under increased cognitive load, or (b) attributable to non-lesion-site-dependent post-stroke factors such as fatigue.
The neural origins of the Left Anterior Negativity (LAN) component of the Event-Related brain Potentials (ERP) have never been directly probed although this information is of the highest interest for a comprehensive view of the neural foundation of language. The LAN emerges specifically after morphosyntactic violations and is affected by both linguistic and extralinguistic, non-syntactic information. Here, we explored the neural sources of the LAN by analyzing data from three previously published ERP data sets obtained from canonical morphosyntactic violation conditions. The neuroelectric source analyses were based on LAN data from N = 76 participants and comprised two distributed source algorithms: sLORETA (standardized low-resolution brain electromagnetic tomography), and CLARA (classical LORETA analysis recursively applied), and a discrete dipole model (BESA, brain electrical source analysis). The results indicate that the most acceptable candidate as primary neural source of the LAN is the left frontal operculum (LFO), though the right FO might also be implicated. Considering its location, functions and connections, we speculate that the FO may be monitoring articulatory (morphological or morphophonological) predictions during language comprehension. The direct links between the FO and the anterior temporal pole might also account for nonlinguistic influences on the LAN.
Infant-directed speech (IDS) is highly rhythmic, and in European languages, it is dominated by patterns of amplitude modulation (AM) peaking at ∼2 Hz (reflecting prosody) and ∼5 Hz (reflecting individual syllables). The rhythm structure of spoken Japanese is thought to differ from European stress-timed and syllable-timed languages, depending on moraic units (∼10 Hz) comprising any onset phoneme and vowel phonemes within a syllable, PA-N-DA. As the infant brain must be prepared to acquire any human language, initial speech encoding is likely to utilize language-universal physical acoustic structures in speech. These physical structures are, however, probabilistic and may thereby simultaneously accommodate language-specific structures like morae. Here, a language-blind computational model of linguistic rhythm based on the amplitude envelope (AE) is used to compute the physical acoustic stimulus characteristics for Japanese. Using ∼18,000 samples of natural IDS and child-directed speech (CDS) recorded longitudinally over the ages 0-5 years, the data show that the temporal modulation patterns that characterize the AE of Japanese are similar to those found for stress-timed and syllable-timed European languages. However, the AM band corresponding to the syllabic level in CDS/IDS in European languages (∼2-12 Hz) was elongated in Japanese (2.5-17 Hz), possibly accommodating the faster modulation peaks reflecting morae. Furthermore, the phase synchronization ratios between the two slowest AM bands were as likely to be 1:3 as 1:2, differing from European languages where 1:2 ratios (delivering the perceptual experience of a temporally regular beat) are dominant. Accordingly, the amplitude-driven physical acoustic structures important for cortical speech tracking flexibly accommodate both universality and specificity.
The visual word form area (VWFA) has been consistently identified as a crucial structure in word reading, and its function differs across subregions. Nevertheless, the functional roles of its subregions and their functional origins remain controversial. Here, we adopted multimodal neuroimaging techniques (i.e., task-state fMRI, resting-state fMRI, and diffusion MRI) combined with representational similarity analysis to investigate the functional role of VWFA subregions and the brain circuitry supporting their function in two experiments. Results revealed respective roles of the posterior and anterior VWFA subregions in visual and semantic processing, which is consistent with their respective connectivity to orthographic and semantic networks. In addition, processing demands modulated the neural representations of high-level linguistic information in the VWFAs. These convergent findings elucidated the local neural computations in the VWFAs and their cooperative mechanism with distant brain regions related to language processing, jointly providing multimodal neuroimaging evidence for the connectivity-biased hypothesis.
Poor comprehenders (PCs) have typical word reading skill and intelligence but poorer than expected reading comprehension. While the prevalence of PCs is similar to that of poor decoders (individuals who struggle fluently converting written text into spoken language), less is known about the neurobiological substrates of poor comprehension. Extant studies have found small differences in gray matter volume between PCs and poor or typically reading peers. However, a detailed quantification of cortical morphometric features and white matter integrity remains unexplored. Data from 2,100 children (1,200 with imaging data), aged 8-16 years were analyzed to determine if there is a distinct neuroanatomy associated with poor reading comprehension across three common methods for classifying PCs. We computed gray matter volume, cortical thickness, and surface area, as well as white matter measures (mean diffusivity, fractional anisotropy, neurite orientation, and neurite density) for PCs and compared these measures with those of poor decoders and typical readers. Results revealed small but widespread white matter differences, but no gray matter differences between PCs and other readers. PCs showed decreased white matter integrity (increased mean diffusivity, decreased neurite density) in tracts previously associated with reading, including the superior longitudinal fasciculus and inferior longitudinal fasciculus, and in tracts that have been associated with cognitive performance such as the uncinate fasciculus. These results suggest that diffuse structural connectivity differences may underlie reading comprehension weaknesses in the face of intact decoding skills. This is consistent with the behavioral profile of PCs who exhibit a broad pattern of subclinical impairments in language and integrative cognitive processes.
Recent behavioral and neural research on reading shows that humans can extract syntactic structure from short sentences within a fraction of a second-faster than many estimates for recognizing the meaning of a single word. This challenges a core assumption of many language processing models-that combinatory operations depend on prior lexical access. Furthermore, studies using parallel presentation of full sentences have revealed electrophysiological responses remarkably similar to those well established for single words. This raises the question of whether words, phrases, and sentences all move through the same processing stages, regardless of syntactic complexity. Using magnetoencephalography, we examined how single words, phrases, and sentences are processed when all visual information is available at once. Across all three levels, we observed highly similar waveform dynamics, with early responses reflecting bottom-up detection of form, followed by activity in the left anterior and posterior temporal cortices and ventromedial prefrontal cortex consistent with combinatory processing. Of these regions, the left anterior temporal lobe showed effects of bigram frequency suggestive of serial left-to-right dynamics. Together, these results support a Global-to-Sequential Assembly model in which the brain first detects the global form of the stimulus in a snapshot-like manner and then probes its combinatory properties through partially serial processes.
Oral reading relies on lexical and sublexical processes with distinct neural mechanisms. Damage within the sublexical system causes phonological alexia, a blanket diagnosis describing acquired deficits in reading unfamiliar words. Improving the precision of alexia diagnosis requires understanding the neurocognitive basis of specific reading subprocesses. This study investigated the neural correlates of sublexical reading in 64 adults with chronic left-hemisphere stroke (LHS), focusing on lesions that impair the use of learned orthography-to-phonology (OP) mappings to read new words. Participants read aloud real words and three types of pseudowords varying in the number of plausible OP mappings at the level of the orthographic body: zero mappings (0M), one mapping (1M), and multiple mappings (MM). LHS participants exhibited phonological reading deficits with an exaggerated lexicality effect compared to 71 neurotypical controls. Across both groups, pseudowords with learned OP mappings were read more accurately than those without. Voxelwise and connectome-based lesion-symptom mapping revealed that relative lexical reading deficits were associated with lateral temporal lesions, while sublexical reading deficits were associated with lesions or disconnections of the left inferior frontal (IFG), supramarginal, and pre/postcentral gyri. Applying learned OP mappings relied on anterior IFG and frontoparietal connections, while resolving multiple plausible OP mappings relied on intraparietal connections. These results underscore the role of learned mutigraphemic OP mappings in sublexical reading, and demonstrate that disruptions of different sublexical reading subprocesses result in subtly different deficit patterns. Dissecting the neurocognitive basis of reading subprocesses may improve the precision of alexia diagnosis and point to new treatments.
