🧠 neuroscience

Neural processing of natural speech by adults with and without dyslexia: Evidence for atypical cortical decoding of speech information in the delta and theta EEG bands

This study demonstrates that adults with dyslexia exhibit atypical neural processing of natural speech, characterized by reduced delta- and theta-band cortical decoding accuracy and cerebro-acoustic coherence, alongside increased delta-band power, particularly over the right temporal region.

Original authors: Keshavarzi, M., Moore, B. C. J., Goswami, U.

Published 2026-02-19

📖 5 min read🧠 Deep dive

CC BY 4.0

Original authors: Keshavarzi, M., Moore, B. C. J., Goswami, U.

Original paper licensed under CC BY 4.0 (https://creativecommons.org/licenses/by/4.0/). ⚕️ This is an AI-generated explanation of a preprint that has not been peer-reviewed. It is not medical advice. Do not make health decisions based on this content. Read full disclaimer

The Big Picture: Listening to a Story with a Broken Radio

Imagine your brain is a high-tech radio station trying to tune into a live broadcast of a story (the speech). To understand the story, your brain needs to sync up perfectly with the rhythm of the speaker's voice.

The "Delta" rhythm is the slow, heavy beat of the story (like the pauses between sentences or the rise and fall of a voice).
The "Theta" rhythm is the faster beat (like the syllables: ba-na-na).

For most people, this radio tunes in perfectly. The brain waves lock onto the voice, making the story clear and easy to understand. This paper asks: What happens inside the brain of an adult with dyslexia when they listen to a story? Do their radios still tune in, or is there static?

The Experiment: A 16-Minute Story

The researchers gathered 48 adults (24 with dyslexia, 24 without) and had them listen to a 16-minute audiobook of The Iron Man. While they listened, the researchers put sensors on their heads (EEG) to record the brain's electrical signals.

They didn't just ask, "Did you understand the story?" Instead, they used a clever computer trick called "Neural Decoding."

The Analogy: Imagine the brain is a translator. The computer tries to take the brain's electrical signals and "reconstruct" the original story from them.

If the computer can rebuild the story perfectly from the brain signals, the brain is doing a great job.
If the computer builds a garbled, fuzzy version of the story, the brain's representation of the speech is "noisy" or imprecise.

What They Found: The "Static" in the System

The study revealed three main things about how adults with dyslexia process speech:

1. The "Pattern" is Different (Between-Group Analysis)

When the researchers compared the average brain pattern of the dyslexic group against the control group, they found the dyslexic brains were struggling to lock onto both the slow beat (Delta) and the fast beat (Theta).

Analogy: It's like trying to dance to a song where the music is slightly out of sync with the beat. The dancers (the brain) are moving, but they aren't stepping on the right beat as precisely as the control group.

2. The "Individual" Struggle (Within-Participant Analysis)

When they looked at each person individually, the results were interesting:

Delta (Slow beat): The dyslexic adults were actually doing just fine here. Their brains could still track the slow, big-picture rhythm of the story.
Theta (Fast beat): This is where the trouble was. The dyslexic adults had a much harder time tracking the syllable-by-syllable rhythm.
The Twist: In previous studies with children with dyslexia, this Theta problem wasn't as obvious. The researchers suggest that as dyslexic adults grow up and read less (because reading is hard for them), their brains lose some of the "muscle memory" for syncing up with the fast rhythm of speech. It's like a musician who stops practicing a specific instrument; eventually, their fingers get a bit clumsy with it.

3. The "Volume" is Too Loud (Brain Power)

The researchers also measured how "loud" the brain activity was in different areas.

They found that the dyslexic group had more electrical activity (power) in the Delta band, specifically in the right side of the brain (near the ear).
Analogy: Imagine a lightbulb that is flickering and buzzing too brightly. The brain is working harder to process the slow rhythm, perhaps trying to compensate for the difficulty. It's like an engine revving high while stuck in traffic—it's using more fuel (energy) but not necessarily moving faster.

What They Did Not Find

The researchers also checked for "Cross-Frequency Coupling."

Analogy: Think of this as a conversation between the slow beat (Delta) and the fast beat (Theta). Usually, the slow beat tells the fast beat when to start and stop.
Result: Surprisingly, the conversation between these two rhythms was normal in the dyslexic group. The problem wasn't that the rhythms were talking to each other wrong; the problem was that the rhythms themselves were struggling to lock onto the voice.

The Takeaway: Why This Matters

This study supports a theory called Temporal Sampling (TS) Theory. This theory suggests that dyslexia starts early in life because the brain has trouble syncing with the rhythm of speech.

The Good News: The brain can still understand speech; it's just doing it differently and with more effort.
The Bad News: The "noise" in the system (especially the fast Theta rhythm) gets worse over time if the person doesn't get enough practice reading and listening.
The Future: If we can help the brain "tune the radio" better—perhaps through rhythm training or special brain stimulation—we might be able to help people with dyslexia process speech more clearly, which could help them read better.

In short: Adults with dyslexia aren't "broken" listeners. Their brains are just trying to tune into a complex radio station, and they are working extra hard to catch the fast beats of the story, leading to a bit of static in their internal processing.

1. Problem Statement

Dyslexia is a developmental disorder characterized by reading difficulties linked to phonological processing deficits. The Temporal Sampling (TS) theory posits that dyslexia arises from atypical neural entrainment to slow temporal modulations in speech (specifically in the delta [0.5–4 Hz] and theta [4–8 Hz] bands), which are crucial for tracking prosodic and syllabic rhythms.

While previous studies have documented these deficits in children, research on adults with dyslexia has been limited. Existing adult studies often:

Focused on the gamma band (phoneme-level processing) rather than low-frequency bands.
Used segmented or brief speech stimuli rather than continuous natural speech.
Rarely measured delta-band cortical tracking or combined neural decoding (reconstruction) with cerebro-acoustic coherence.

The study aims to determine if atypical low-frequency neural tracking persists into adulthood during naturalistic listening and to characterize the nature of these deficits using advanced decoding and coherence metrics.

2. Methodology

Participants:

N = 48 native English-speaking adults (24 with dyslexia, 24 controls).
Groups were matched for age and nonverbal IQ.
Dyslexic participants had formal clinical diagnoses and scored significantly lower on reading fluency (TOWRE-SWE) and phonemic decoding (TOWRE-PDE) tasks.
All participants had normal hearing.

Stimuli and Procedure:

Participants listened to a 16-minute continuous spoken narrative ("The Iron Man" by Ted Hughes) via insert earphones.
EEG was recorded using a 128-channel HydroCel Geodesic Sensor Net (1 kHz sampling rate).
Participants fixated on a cross to minimize eye-movement artifacts.

Data Pre-processing:

Data were band-pass filtered (0.2–42 Hz), downsampled to 200 Hz, and cleaned using Independent Component Analysis (ICA) to remove artifacts.
Specific bands (Delta: 0.5–4 Hz; Theta: 4–8 Hz) were extracted for analysis.

Analytical Framework:
The study employed four primary analytical approaches:

Backward Multivariate Temporal Response Function (mTRF) Decoding:
- Within-participant analysis: A separate TRF model was trained for each individual to reconstruct the speech envelope from their EEG. Accuracy was measured via Pearson correlation between the reconstructed and actual envelope.
- Between-group analysis: A "normative" model was trained on a subset of control data and tested on remaining controls and the dyslexic group to assess how well dyslexic neural patterns align with typical representations.
Cerebro-Acoustic Coherence: Computed using the mscohere function to measure the synchrony between neural activity and the speech envelope in the 0.5–8 Hz range.
Band Power Analysis: Power Spectral Density (PSD) was calculated for Delta, Theta, Beta, and Low-Gamma bands. Topographic maps were generated to identify spatial differences (specifically in the right temporal region).
Cross-Frequency Coupling:
- Phase-Amplitude Coupling (PAC): Measured using the Modulation Index (MI).
- Phase-Phase Coupling (PPC): Measured using Phase-Locking Value (PLV).
- Pairs analyzed included Delta-Theta, Delta-Beta, Theta-Beta, Delta-Low Gamma, and Theta-Low Gamma.

Statistical Analysis:

Group differences were assessed using t-tests, Wilcoxon rank-sum tests, and Bayesian t-tests (calculating Bayes Factors, BF10).
Significance was determined against chance levels derived from null models (permuted data).

3. Key Contributions

First Adult Study with Natural Speech: This is the first study to apply backward mTRF decoding to continuous, naturalistic speech in adults with dyslexia, moving beyond segmented sentences or non-speech stimuli.
Dual-Analysis Approach: It uniquely combines neural decoding (assessing the fidelity of the neural representation) with coherence (assessing synchrony strength) to distinguish between "noisy" representations and "misaligned" ones.
Delta-Band Focus: It explicitly investigates the delta band in adults, a frequency range often overlooked in adult dyslexia research compared to the gamma band.
Developmental Perspective: By comparing these adult results with previous child data (using the same paradigm), the study offers insights into how neural tracking deficits evolve with reading experience.

4. Key Results

A. Neural Decoding (mTRF):

Between-Group Analysis: Dyslexic adults showed significantly lower decoding accuracy than controls in both Delta ( $p < 0.001$ ) and Theta ( $p < 0.001$ ) bands. This indicates that the pattern of neural representation in dyslexic adults is fundamentally different from the normative model.
Within-Participant Analysis:
- Delta Band: No significant difference between groups ( $p = 0.46$ ). Individual dyslexic brains could reconstruct the envelope as accurately as controls.
- Theta Band: Dyslexic adults showed significantly lower decoding accuracy than controls ( $p = 0.03$ ). This suggests a specific deficit in the precision of syllabic-rate encoding in adults, contrasting with child studies where this difference was not observed.

B. Cerebro-Acoustic Coherence:

Dyslexic adults exhibited significantly reduced coherence compared to controls across the 2–6 Hz range (spanning both delta and theta bands).
This indicates a persistent deficit in the temporal synchrony between neural oscillations and the speech envelope in adults.

C. Band Power:

Delta Band: Dyslexic adults showed significantly increased delta-band power compared to controls ( $p = 0.038$ ).
Topography: This power increase was localized specifically to the right temporal region ( $p = 0.028$ ).
Other Bands: No significant group differences were found for Theta, Beta, or Low-Gamma power.

D. Cross-Frequency Coupling (PAC & PPC):

No significant differences were found between groups for any Phase-Amplitude (PAC) or Phase-Phase (PPC) coupling measures across the tested frequency pairs. This suggests that the core deficit lies in the low-frequency tracking itself, not in the coupling between low and high frequencies.

5. Significance and Discussion

Persistence of Deficits: The study confirms that atypical low-frequency neural tracking (delta and theta) persists into adulthood. However, the nature of the deficit changes: while children show broad deficits, adults show a specific reduction in theta-band decoding precision and increased delta-band power, likely influenced by cumulative reading experience.
Support for TS Theory: The findings strongly support the Temporal Sampling theory, demonstrating that dyslexia involves a fundamental disruption in the brain's ability to entrain to the slow temporal structure of speech, affecting both prosodic (delta) and syllabic (theta) processing.
Neural Markers:
- Reduced Theta Decoding: Suggests that the precision of syllabic unit extraction is compromised in adults.
- Increased Right Temporal Delta Power: Aligns with the Asymmetric Sampling in Time (AST) theory, suggesting atypical right-hemisphere dominance for longer temporal integration windows in dyslexia.
Clinical Implications: The findings suggest that remediation strategies (e.g., rhythmic training or neurostimulation) targeting low-frequency neural synchrony could be beneficial. The fact that theta deficits persist into adulthood implies that reading experience alone does not fully compensate for these underlying neural timing issues.
Methodological Advance: The study demonstrates that combining decoding accuracy with coherence measures provides a more nuanced view of neural deficits than coherence alone, distinguishing between "weaker coupling" and "altered representation."

In conclusion, Keshavarzi et al. provide robust evidence that adults with dyslexia exhibit atypical cortical decoding of natural speech, characterized by reduced theta-band precision and altered delta-band power, supporting a developmental model where low-frequency entrainment deficits persist and evolve with age.