Rapid Orthographic and Delayed Phonological Processing: ERP and Oscillatory Evidence from Masked Priming in Korean

Using EEG and masked priming in Korean, this study demonstrates that orthographic syllable processing occurs rapidly at pre-lexical stages and cascades into lexical access, whereas phonological effects emerge later, providing neural evidence for a sequential rather than parallel coordination of orthographic and phonological codes.

The Gist

The Big Question: How Do We Read?

Imagine your brain is a busy airport control tower. When a plane (a word) lands, the tower needs to figure out what it is and where it belongs. But there's a debate among the controllers:

1. The "Parallel" Theory: The tower checks the plane's shape (what it looks like) and its sound (what it sounds like) at the exact same time, instantly.
2. The "Sequential" Theory: The tower checks the shape first. Only after it recognizes the shape does it bother to check the sound.

This study, conducted by researchers in Korea, wanted to settle this debate using the Korean writing system (Hangul) as a unique testing ground.

The Secret Weapon: Korean Hangul

Most writing systems (like English) are tricky because the shape of a word and its sound are usually tangled together. If you change the letters, the sound usually changes too.

But Korean is special. It uses "blocks." Think of Korean letters like LEGO bricks. You snap them together to make a syllable block.

• The Magic Trick: Because of how Korean spelling works, the researchers could create a situation where two words look identical but sound different, or sound identical but look different.

The Experiment:
They showed participants a "prime" word (a hint) very quickly (so fast you barely see it) before showing a "target" word.

• Scenario A (Same Shape, Different Sound): The hint looks exactly like the first part of the target, but the sound is slightly different due to Korean grammar rules.
• Scenario B (Same Sound, Different Shape): The hint sounds exactly like the first part of the target, but the letters look completely different.
• Scenario C (Unrelated): The hint has nothing to do with the target.

They measured brain waves (EEG) to see what happened inside the brain milliseconds after seeing the hint.

The Results: A Race Between Shape and Sound

The study found that Shape wins the race, and Sound is a latecomer.

1. The Shape Effect (The "Fast Lane")

When the brain saw a word with the same shape as the hint:

• The Reaction: It was like a green light turning on instantly. The brain recognized the visual pattern immediately.
• The Brain Waves: Within 150–250 milliseconds (faster than a blink), the brain showed a "P200" spike. This is like a "Hello! I know this shape!" signal.
• The Oscillation: Simultaneously, the brain's "upper beta" waves synchronized. Think of this as the brain's internal team high-fiving to say, "We've got this visual pattern locked in!"
• The Result: People pressed the "Yes, that's a word" button faster.

Analogy: It's like recognizing a friend's face from a distance. You don't need to hear their voice to know who it is; the visual shape is enough to trigger a quick, happy recognition.

2. The Sound Effect (The "Slow Lane")

When the brain saw a word that sounded like the hint but looked different:

• The Reaction: The brain was confused. It didn't get the "green light" early.
• The Brain Waves: There was no early spike. The brain didn't react until much later (around 300–590 milliseconds).
• The Oscillation: The brain showed "lower beta" waves. This is like the brain working hard, grinding its gears, and trying to assemble the sound from the visual letters. It's a "computational effort" signal.
• The Result: People did not get faster. In fact, because the shape didn't match, the sound hint didn't help them at all.

Analogy: It's like hearing a voice in a crowd but not seeing the person. You have to squint, turn your head, and search the crowd to figure out who it is. It takes more time and effort, and it doesn't help you find the person any faster if you were already looking for someone else.

The "N400" Moment: Making Sense of It All

Later in the process (around 350–550 ms), the brain has to connect the word to its meaning (like "dog" = a furry animal).

• With the Shape Match: The brain was relaxed. The "N400" wave (which usually gets big when the brain is struggling to understand) was smaller. The brain said, "Oh, I already knew this shape, so getting the meaning was easy."
• With the Sound Match: The brain was still working hard. The "N400" wave was large, and "theta" waves (associated with mental effort) were high. The brain was saying, "I'm still trying to figure this out."

The Big Conclusion

The study proves that for Korean readers, reading is a two-step process, not a one-step magic trick.

1. Step 1 (Pre-lexical): The brain looks at the visual shape of the syllable block first. This happens super fast and is very efficient.
2. Step 2 (Lexical): Only after the shape is recognized does the brain bother to assemble the sound and check for meaning.

Why does this matter?
It challenges the idea that our brains process sound and sight perfectly simultaneously. Instead, it suggests that our brains are smart enough to pick the most efficient route. In Korean, the visual "blocks" are so clear that the brain uses them as the main entry door to the dictionary, leaving the sound processing for later.

In a nutshell:
If you want to know what a word is, look at it first. The sound will catch up later, but the shape is the one that gets you to the finish line first.

Technical Summary

1. Problem Statement and Research Questions

The study addresses a central debate in visual word recognition: whether orthographic (visual) and phonological (sound) codes are processed sequentially (as predicted by Dual-Route Cascaded models) or in parallel (as predicted by Interactive Activation models like BIAM).

While previous research in alphabetic languages (e.g., English, Spanish) has struggled to dissociate these two codes because syllable structure is confounded in spelling, the authors utilize Korean Hangul to solve this. Hangul is an alpha-syllabic system where morphophonemic spelling principles allow for the independent manipulation of orthographic and phonological syllable overlap.

The study aims to answer two specific questions:

1. Representational Basis: Do syllabic priming effects in Korean reflect orthographic or phonological overlap?
2. Processing Stage: Do these effects arise at the pre-lexical stage (form encoding) or the lexical stage (competitive activation)?

2. Methodology

• Participants: 29 healthy, native Korean speakers (after excluding one for low accuracy).
• Task: A masked priming lexical decision task using high-density EEG (64 channels).
  • Stimuli: 120 disyllabic Korean words.
  • Conditions:
    1. Orthographically Identical: Prime and target share the same visual syllable block but differ in pronunciation due to resyllabification (e.g., Prime: sik /식/ → Target: singlang /식량/).
    2. Phonologically Overlapping: Prime and target share the same sound but differ in visual spelling (e.g., Prime: sing /싱/ → Target: singlang /식량/).
    3. Unrelated: No overlap.
  • Procedure: Forward mask (500ms) → Prime (50ms) → Target (until response).
• Data Analysis:
  • Behavioral: Linear Mixed-Effects Models (LMM) for Reaction Times (RT) and Generalized LMM for accuracy.
  • ERP Analysis: Focused on P200 (150–250 ms) and N400 (350–550 ms) components using Region of Interest (ROI) analysis and Cluster-Based Permutation Tests (CBPT).
  • Time-Frequency Analysis (TFR): Examined oscillatory dynamics in Theta (4–8 Hz), Lower Beta (13–20 Hz), and Upper Beta (20–30 Hz) bands.

3. Key Results

A. Behavioral Results

• Orthographic Priming: Produced significant facilitation (faster RTs) compared to unrelated and phonological conditions.
• Phonological Priming: Produced no significant behavioral facilitation compared to the unrelated baseline.

B. ERP Results (Event-Related Potentials)

• P200 (150–250 ms):
  • Orthographic: Showed significant enhancement (larger amplitude) over fronto-central sites.
  • Phonological: No modulation compared to unrelated.
  • Interpretation: Early visual form encoding is driven by orthography, not phonology.
• N400 (350–550 ms):
  • Orthographic: Showed significant reduction (reduced negativity) over centro-parietal sites, indicating facilitated lexical-semantic integration.
  • Phonological: No significant modulation.
  • Interpretation: Orthographic overlap cascades into efficient lexical access; phonological overlap alone does not.

C. Time-Frequency Results (Oscillatory Dynamics)

• Upper Beta (20–30 Hz):
  • Orthographic: Early synchronization (30–290 ms) over frontal regions, temporally overlapping with the P200. Suggests rapid top-down binding of visual features.
• Lower Beta (13–20 Hz):
  • Phonological: Sustained increase in power (310–590 ms) over central regions.
  • Interpretation: Reflects effortful processing, phonological assembly, and lexical competition, occurring after orthographic processing.
• Theta (4–8 Hz):
  • Orthographic vs. Phonological: Orthographic priming showed suppression of theta power (400–730 ms) relative to phonological priming.
  • Interpretation: Reduced cognitive control/conflict demands for orthographic pairs, indicating efficient semantic integration.

4. Key Contributions

1. Methodological Innovation: Successfully dissociated orthographic and phonological syllable effects in a single experimental design using the unique properties of Korean Hangul, a feat impossible in standard alphabetic scripts.
2. Neural Dissociation: Provided converging evidence (ERP + TFR) that orthographic and phonological processing follow distinct spatiotemporal trajectories:
   • Orthography: Pre-lexical, rapid, and facilitative (P200/Upper Beta → N400/Theta suppression).
   • Phonology: Lexical, delayed, and effortful (Lower Beta increase), with no early facilitation.
3. Theoretical Resolution: The data strongly support sequential/cascaded models (e.g., DRC) over strong parallel activation models (e.g., BIAM). The absence of early phonological modulation (N/P150 or N250) challenges the notion that phonological codes are activated simultaneously with orthographic codes in this script.

5. Significance and Implications

• Script-Specific Processing: The findings suggest that reading systems adapt to the statistical regularities of their orthography. In Korean, the compact, visually bounded syllable blocks allow for a "visual-first" strategy, where orthographic parsing serves as the primary and most efficient entry route to the lexicon.
• Re-evaluating Parallel Activation: The study places empirical constraints on the "Parallel Activation" hypothesis, demonstrating that in Hangul, phonological assembly is a downstream process that only engages after orthographic analysis is complete.
• Neural Mechanisms: The study links specific oscillatory bands to cognitive operations:
  • Upper Beta: Rapid perceptual grouping and orthographic stabilization.
  • Lower Beta: Sustained phonological assembly and conflict resolution.
  • Theta: Cognitive control and retrieval effort (suppressed when orthographic priming is effective).
• Future Directions: The authors note that while phonological effects were delayed, they were not absent (evidenced by lower beta), suggesting phonological processing is active but secondary. Future work should explore cross-modal priming and morphological complexity to further refine these models.

In conclusion, the paper provides robust neurophysiological evidence that in Korean visual word recognition, orthographic processing precedes and facilitates phonological processing, supporting a sequential, cascaded architecture of lexical access.