Neuroimaging Methods in Child Language and Reading Acquisition

Introduction to Developmental Cognitive Neuroscience in Language Acquisition

• Evolutionary processes have provided humans with specialized neural tissue and learning capabilities that enable children to acquire language simply through exposure to parental spoken or sign languages.

• Reading acquisition often requires explicit instruction, practice, and years of effort, while relying on core brain structures and learning functions that support literacy.

• Developmental Cognitive Neuroscience: A discipline emerging at the turn of the $21\text{st}$ century (Johnson, 2005) that integrates behavioral child development with neuroscience to study the developing human mind, brain, and behavior.

• Facilitating Factors: Rapid technological advancements in noninvasive, child-friendly brain imaging allow direct investigation of infants and children during critical ages of early language and literacy development.

• Key Findings in Early Brain Development:

  • Newborn infants already demonstrate a hemispheric preference for language.

  • Early functional activation patterns in infants and preliterate children can reveal future risks for language and reading difficulties.

  • Early environmental experiences, such as bilingualism, significantly alter how language is organized in the neural architecture (Petitto, 2005).

Principles of Noninvasive Functional Neuroimaging

Functional neuroimaging measures changes in brain activity as individuals engage in cognitive and linguistic tasks. These methods fall into two primary categories based on the biological signals they record:

• Electrical Activity Methods (ERP, MEG):

  • Measure rapid, transient electrical activity generated by active neurons.

  • Temporal Resolution: High (on the order of milliseconds).

  • Spatial Resolution: Relatively poor anatomical localization (Luck, 2005).

• Hemodynamic Response Methods (fMRI, PET, fNIRS):

  • Active neurons consume energy and induce localized changes in blood flow (hemodynamic response) to supply glucose and oxygen.

  • Hemodynamic changes are indirect derivatives of neural activity (Friston, 2009).

  • Temporal Resolution: Low ($2\text{--}5\text{ seconds}$ delay).

  • Spatial Resolution: High anatomical localization (Friston, 2009).

Event-Related Potential (ERP)

• Mechanism: Electroencephalography (EEG) records continuous, spontaneous electrical activity below the scalp via surface electrodes. Event-related potentials (ERPs) represent time-locked electrical responses to specific internal or external events (Luck, 2005; Slotnick, 2010).

• Methodological Advantages:

  • Excellent millisecond temporal resolution, suitable for rapid verbal processing.

  • Acoustic silence and relative tolerance to physical movement.

  • Safe and effective for awake infants and young children without requiring active behavioral responses or focused attention.

  • Infant-friendly electrode net setups can take less than $5\text{ minutes}$ to position (Johnson et al., 2001).

• ERP Waveform Characteristics:

  • Polarity: Direction of the electrical deflection (positive- or negative-going wave).

  • Latency: Precise time delay following stimulus onset (in milliseconds).

  • Scalp Distribution: Topographical location of recording electrodes across the scalp.

  • Functional Components: Early components ($< 200\text{ ms}$) reflect sensory detection; later components ($> 200\text{ ms}$) reflect higher-order cognitive processing (e.g., sentence comprehension).

• Key ERP Components in Language Research:

  • Phonology (MMN):

  • Mismatch Negativity (MMN): Negative deflection peaking at approximately $200\text{ ms}$.

  • Triggered by categorical changes in phonological units (e.g., discriminating between /ba/ and /da/; Kuhl & Rivera-Gaxiola, 2008).

  • Emerges early in development; $3\text{-month-old}$ infants exhibit adult-like MMN amplitude with delayed latency (Cheour et al., 1998).

  • Semantics (N400):

  • N400 Component: Centro-parietal negative wave peaking at approximately $400\text{ ms}$ post-stimulus onset.

  • Induced by semantic anomalies in context (e.g., "I like my coffee with cream and sock" versus "cream and sugar", or "We bake cookies at the zoo"; Holcomb, Coffey, & Neville, 1992).

  • Children as young as $7\text{ years}$ exhibit adult-like N400 responses; latency and amplitude decrease progressively with age due to cognitive and neural maturation.

  • Observed in infants as young as $1\text{ year}$ when hearing a word that mismatches a visual picture (Friederici, 2005).

  • Syntax (E/LAN and P600):

  • Early Left Anterior Negativity (E/LAN): Peaks at $150\text{--}350\text{ ms}$; indexes automatic, online grammatical structure processing (Friederici, 2005).

  • P600 Component: Centro-parietal positive deflection peaking at approximately $600\text{ ms}$; indexes syntactic reanalysis and structural revision (e.g., in response to ungrammatical sentences like "My uncle watched about a movie my family"; Friederici, 2005).

  • Children show P600 responses by $2\text{ years}$ of age, characterized by larger amplitude, longer latency, and broader scalp distribution compared to adults (Friederici, 2005; Kuhl & Rivera-Gaxiola, 2008).

Magnetoencephalography (MEG)

• Mechanism: Measures weak magnetic fields generated by neural intracellular electrical currents (dipoles) using Superconducting Quantum Interference Devices (SQUIDs) embedded within a sensor helmet.

• Comparison with ERP:

  • Shares millisecond temporal resolution, safety, silent operation, and noninvasiveness.

  • Provides superior spatial localization compared to ERP because magnetic fields pass through skull tissue without being distorted or dispersed (Luck, 2005).

  • Requires child-sized helmets to keep SQUID sensors close to smaller skulls (Tesan et al., 2010).

• Magnetic Source Imaging (MSI): Overlaying functional MEG localization data onto individual anatomical MRI scans or standard pediatric brain templates.

• Technical Limitations:

  • Highly sensitive equipment requires installation inside a specialized magnetically shielded room, as neural magnetic fields are approximately $10\text{ million}$ times weaker than Earth's magnetic field.

  • High purchase and operational costs; non-portable hardware.

• Source Modeling and N400m Findings:

  • Uses dynamic source modeling to track neural activity progression from early sensory stages ($< 200\text{ ms}$) to cognitive processing (Halgren et al., 2002).

  • Detects the N400m magnetic counterpart to the N400 ERP component at approximately $400\text{ ms}$ post-stimulus (Halgren et al., 2002).

  • Equivalent Current Dipole (ECD) Model: Localizes N400m generation to the left superior temporal sulcus (STS).

  • Distributed Source Model: Indicates semantic anomaly processing originates in the left temporal lobe (Wernicke's area) at $250\text{ ms}$ and spreads to the frontal lobe (Broca's area) by $370\text{ ms}$.

• Applications: Investigating typical development, language learning impairments, developmental dyslexia, and autism spectrum disorders (Salmelin, 2007).

Functional Magnetic Resonance Imaging (fMRI)

• Mechanism: Measures changes in blood oxygenation using Blood Oxygen Level Dependence (BOLD) contrast (Huettel, Song, & McCarthy, 2008).

  • Neuronal activity consumes oxygen, triggering an influx of oxygenated blood.

  • Transports oxygen via hemoglobin: oxy-hemoglobin (oxygen-bound) and deoxy-hemoglobin (unbound).

  • Differential magnetic susceptibility of deoxy-hemoglobin is detected using static and transient magnetic fields.

• Resolution Profile:

  • Spatial Resolution: Superior (localized to precise millimeters).

  • Temporal Resolution: Poor ($5\text{-second}$ hemodynamic delay following neural activity).

• Linguistic Activation Patterns: Phonological tasks (e.g., rhyming judgments on word pairs like "cat-hat") recruit left inferior frontal, superior temporal, and parietal regions.

• Pediatric and Methodological Adaptations:

  • Infants and Toddlers ($< 3\text{ years}$): Scanned while sedated or during natural sleep (Freund, 2008).

  • Older Children: Environment acclimation using mock scanners (simulating scanner shape and sound) to train head motion restraint (Raschle et al., 2009).

  • Hardware Adaptations: Child-sized head coils increase signal-to-noise ratio and limit head movement.

  • Image Processing: Uses child-specific anatomical templates (Freund, 2008) and advanced coregistration software (Ghosh et al., 2010).

  • Motion Artifact Correction: Post-processing tools such as the Artifact Detection Toolbox (developed by Susan Whitfield-Gabrieli).

  • Auditory Task Optimization: "Silent" fMRI designs pause scanner gradient noise during auditory stimulus delivery, or utilize quieter scanning sequences (Freund, 2008).

Functional Near Infrared Spectroscopy (fNIRS)

• Mechanism: Noninvasively monitors hemodynamic changes by emitting and detecting near-infrared light through scalp optodes (Plate 4B; Shalinsky et al., 2009).

  • Light paths form a curved "banana-shaped" trajectory between paired emitters and detectors, penetrating up to $3\text{ cm}$ into the cerebral cortex (Plate 4C).

  • Uses dual wavelengths of near-infrared light (e.g., $690\text{ nm}$ and $830\text{ nm}$ in the Hitachi ETG-4000 system) to measure concentration changes of oxy-hemoglobin and deoxy-hemoglobin independently.

• Resolution Profile:

  • Temporal Resolution: $\text{Higher than } 10\text{ samples per second}$ (compared to fMRI at approximately $1\text{ sample per } 2\text{ seconds}$).

  • Spatial Resolution: $2\text{--}3\text{ cm}$ (greater spatial accuracy than ERP, but lower than fMRI).

• Experimental Advantages:

  • Completely silent, lightweight, and tolerant of modest motion.

  • Highly portable: operational in schools, home environments, and neonatal intensive care units.

  • Accommodates naturalistic language paradigms in awake infants (seated on parent lap), children, speaking adults, and bimodal deaf or hearing signers utilizing signing space (Huppert et al., 2009; Kovelman et al., 2009; Petitto et al., in press; Shalinsky et al., 2009).

• Methodological Controls:

  • Reclining seating and footrests minimize physiological interference (e.g., Mayer waves; Shalinsky et al., 2009).

  • Optode placement guided by international 10-20 EEG conventions, $3\text{D}$ digitizing position trackers, and fMRI coregistration with vitamin E markers.

  • Video recording synchronized with imaging tracks head movements and behavioral responses.

Multimodal and Structural Anatomical Imaging

• Multimodal Fusion Techniques: Combines complementary neuroimaging techniques to capture high resolution in both temporal ("when") and spatial ("where") dimensions (Friston, 2009).

  • Example: Simultaneous recording of EEG and fMRI, or combined fNIRS and ERP testing within a single experimental session.

  • Analytical Tools: Publicly available fNIRS time-series and mapping software include HomER (Huppert et al., 2009) and NIRS-SPM (Statistical Parametric Mapping; Ye et al., 2009).

• Structural Anatomical Imaging:

  • Structural MRI: Measures structural volumes, cortical thickness, and gray matter tissue density.

  • Diffusion Tensor Imaging (DTI): Maps white matter fiber tracts to quantify anatomical connectivity between brain regions.

Neuroimaging Discoveries in Early Language Acquisition

• Phonetic Unit Processing:

  • Phonemes serve as fundamental building blocks of linguistic competence (e.g., distinguishing /b/ and /d/).

  • Infants younger than $8\text{ months}$ act as "universal" perceivers, distinguishing phonetic boundaries across all human languages (spoken and signed; Petitto, 2005).

  • After $8\text{ months}$, nonnative phonetic sensitivity declines, completing a transition to "native" category restriction by $12\text{ months}$ (Werker & Tees, 1992).

  • Requires direct, in-person social interaction; passive video exposure (e.g., television) fails to induce phonetic learning (Kuhl & Rivera-Gaxiola, 2008).

• fNIRS Infant Phoneme Discrimination Study (Petitto et al., in press):

  • Paradigm: Oddball detection task presenting standard phonemes $80\text{\%}$ of the time (e.g., /ba/) and deviant phonemes $20\text{\%}$ of the time (e.g., /da/).

  • Sample: Monolingual and bilingual infants grouped by age into $3\text{--}4\text{ months}$ ("young") and $10\text{--}12\text{ months}$ ("old").

  • Superior Temporal Gyrus (STG): Both age groups and language groups showed equivalent bilateral STG activation to native and nonnative phonetic contrasts.

  • Left Inferior Frontal Gyrus (left IFG / Broca's area): Only region exhibiting age-related developmental changes. Posterior left IFG activated to phonetic boundary shifts (oddballs).

  • Monolingual infants: Posterior left IFG activation increased from $3\text{--}4\text{ months}$ to $10\text{--}12\text{ months}$.

  • Bilingual infants: Posterior left IFG activation decreased across the same developmental period.

• Rhythmic Processing and Babbling:

  • The infant brain possesses dedicated tissue in the left STG optimized for slow-rhythmic linguistic modulations at approximately $1.5\text{ Hz}$ (Petitto, 2005).

  • Babbling Milestone: At approximately $5\text{ months}$ to $12\text{ months}$, infants produce rhythmically alternating syllabic units at $1.5\text{ Hz}$ (vocal babbling in speech-exposed infants, manual babbling at $1.5\text{ Hz}$ in sign-exposed infants; Petitto, 2005).

  • Bilingual Neural Plasticity: Left IFG alterations at $12\text{ months}$ coincide with first word emergence, mapping phonology to semantics (Blumstein, 2009). Reduced left IFG activation in bilinguals reflects a prolonging of the sensitive period, maintaining heightened openness to language patterns (Kovelman, Baker, & Petitto, 2008; Ramon-Casas et al., 2009).

Neuroimaging Discoveries in Reading Acquisition and Dyslexia

• Phonological Awareness and Literacy:

  • Phonological awareness—the explicit manipulation of speech sound structures (e.g., rhyming "cat-hat", segmenting "ti-cket" into "ti" and "cket")—predicts successful reading acquisition across languages (Wolf, 2007).

  • Deficits in phonological awareness constitute the core etiology of developmental dyslexia (Kovelman, Christodoulou, & Gabrieli, in press; Wolf, 2007).

• Rhythmic Modulation Theory of Dyslexia:

  • Phonological awareness depends on sensitivity to low-frequency amplitude modulations in speech streams ($1.5\text{--}2\text{ Hz}$) that demarcate syllable and rime boundaries (Thomson & Goswami, 2008).

  • Children with developmental dyslexia exhibit impairments in perceiving low-frequency amplitude modulations and motor/auditory rhythms at $1.5\text{--}2\text{ Hz}$ (Thomson & Goswami, 2008).

• Neural Substrates of Rhythmic Frequency:

  • Pilot neuroimaging research (Kovelman et al., 2010) reveals that typically developing children recruit the left STG significantly more when listening to speech-like frequencies of $1.5\text{ Hz}$, compared to non-linguistic frequencies of $0.5\text{ Hz}$ or $3\text{ Hz}$.

  • Children with dyslexia consistently demonstrate under-recruitment of the left STG during phonological tasks, linking early rhythmic auditory perception to reading failure (Kovelman, Christodoulou, & Gabrieli, in press).

Key Neuroimaging Terminology and Definitions

• Blood-Oxygen-Level Dependence (BOLD): Signal dependent on blood oxygen concentration. BOLD changes correlate with blood flow alterations, serving as an indirect measure of underlying neuronal activation.

• Deoxyhemoglobin: Hemoglobin molecules without bound oxygen. Deoxy-hemoglobin possesses paramagnetic properties utilized in fMRI BOLD signal detection.

• Diffusion Tensor Imaging (DTI): A specialized MRI technique that tracks the diffusion of water molecules to map structural white matter tracts and anatomical connectivity between regions.

• Electroencephalogram (EEG): Noninvasive recording technology measuring continuous electrical potentials produced by brain activity via scalp electrodes.

• Event-Related Potential (ERP): Time-locked electrical voltage fluctuations in the EEG signal triggered by discrete sensory, motor, or cognitive events.

• Functional Magnetic Resonance Imaging (fMRI): Neuroimaging procedure measuring localized hemodynamic changes via the BOLD contrast mechanism inside a magnetic scanner.

• Functional Near Infrared Spectroscopy (fNIRS): Optical neuroimaging technique quantifying cortical oxy- and deoxy-hemoglobin changes by measuring tissue absorption of near-infrared light.

• Hemodynamic Response: Circulatory changes in blood flow, volume, and oxygen distribution supporting active neural tissue.

• Hemoglobin: Iron-containing oxygen-transport metalloprotein present in red blood cells.

• Magnetic Resonance Imaging (MRI): Noninvasive imaging technology using strong static and transient magnetic fields to align nuclear magnetization of hydrogen atoms in body tissue.

• Magnetoencephalography (MEG): Noninvasive technology measuring magnetic fields produced by neuronal electrical activity.

• Motion Artifacts: Signal noise or distortion introduced into brain imaging data by head or body movements.

• Multimodal Imaging: Simultaneous or integrated acquisition of data from two or more neuroimaging modalities (e.g., EEG-fMRI) to combine spatial and temporal precision.

• Oxy-hemoglobin: Hemoglobin molecules bound to oxygen.

• Superconducting Quantum Interference Device (SQUID): Highly sensitive magnetic sensor integrated into MEG systems to detect minute magnetic fields generated by the human brain.