TL;DR
This study uses ERPs to investigate how deaf individuals process ambiguous words in Chinese. They found that deaf readers process words with equally common meanings differently than words with one dominant meaning, suggesting that meaning frequency and context play distinct roles in their reading.
What Problem Does This Paper Address?
Chinese, unlike alphabetic languages, has a high prevalence of polysemous characters, meaning many characters have multiple meanings. This creates "ambiguous morphemes" within compound words, where a single morpheme can contribute different meanings to a word. Understanding how readers process these ambiguities is crucial for language acquisition and reading comprehension. While research has explored this in hearing populations, it remains largely unexamined in deaf individuals. This paper aims to fill this gap by investigating how deaf readers process ambiguous morphemes, specifically looking at the influence of meaning frequency and contextual information on their reading.
Methods
The researchers employed a masked priming lexical decision paradigm combined with event-related potentials (ERPs). This means participants were briefly shown a "prime" word (which could be related or unrelated to the target word) before seeing the "target" word and had to decide if it was a real word. ERPs measure brain activity with high temporal resolution, allowing researchers to see when different cognitive processes occur.
They specifically looked at two ERP components:
- N250: Associated with early, sublexical processing (like recognizing the form of morphemes).
- N400: Associated with semantic processing (understanding the meaning).
They tested two types of ambiguous morphemes:
- Balanced: Meanings occur with similar frequency.
- Biased: One meaning is significantly more frequent (dominant) than the other (subordinate).
Datasets
Not explicitly detailed as a "dataset" in the traditional machine learning sense. The study utilized a carefully curated set of Chinese compound words containing ambiguous morphemes. These materials were adapted from a previously established ambiguous morpheme database and further validated through semantic association questionnaires and relatedness ratings from undergraduate participants.
Key Insights and Ideas
- Deaf readers do access morphemic semantics: Despite potential challenges related to auditory input, deaf individuals actively process the meaning of morphemes within words.
- Two distinct processing pathways: The way deaf readers handle ambiguous morphemes depends on the frequency of their meanings.
- Balanced morphemes (equal meaning frequency): These seem to involve a "meaning competition" phase. Early processing (N250) is influenced by shared form, but semantic integration (N400) only occurs when the prime and target meanings are contextually congruent. This suggests a reliance on context to resolve ambiguity.
- Biased morphemes (dominant meaning): These show a "meaning frequency" advantage. The dominant meaning is activated early and consistently, even if the prime is not perfectly aligned. Subordinate meanings can be activated, but they require contextual support.
- Visual compensation is key: The findings suggest that deaf readers may rely more heavily on visual cues and form-meaning mapping, potentially compensating for the lack of auditory input.
- Implications for language instruction: Understanding these processing differences can inform how ambiguous words are taught to deaf learners, perhaps by structuring instruction around meaning dominance and providing strong contextual support.
Key Results
- N250 (early processing): For balanced morphemes, any orthographically similar prime (regardless of meaning) elicited priming. For biased morphemes, both dominant and subordinate primes showed some early facilitation, with the dominant meaning showing a consistent advantage.
- N400 (semantic processing): For balanced morphemes, only contextually congruent meanings were activated. For biased morphemes, the dominant meaning was activated regardless of context, while the subordinate meaning required contextual congruence.
- Behavioral data: While behavioral results showed some trends, they were not consistently significant, highlighting the power of ERPs in revealing subtle temporal processing differences.
Limitations and Future Work
- Whole-word semantic relatedness: The authors acknowledge that it was challenging to fully control for whole-word semantic relatedness, which could potentially influence priming effects. However, they argue that their findings are unlikely to be solely driven by this confound, citing previous research and the specific patterns observed.
- Visual stimuli only: The study relied solely on visually presented stimuli. In real-world communication, deaf individuals utilize multiple channels (sign language, lip-reading, etc.). Future research should explore multimodal input paradigms to better reflect naturalistic communication.
- Individual differences: The study notes that reliance on multimodal cues might vary among deaf individuals based on their oral language experience and sign language proficiency. Future research should account for these individual differences.
Summary updated