Abstractness of human speech sound representations

Area of Science:

• Neuroscience
• Linguistics
• Cognitive Science

Background:

• Phonemes, the functional units of speech, are theorized to be abstract representations.
• Previous research indicated voicing underspecification in English consonants.
• Linguistic theory suggests languages may differ in underspecified features.

Purpose of the Study:

• To investigate the abstract nature of phoneme encoding in Japanese.
• To test the hypothesis of voicing underspecification in Japanese consonants.
• To compare Japanese findings with previous English speech sound research.

Main Methods:

• Examined brain responses to speech sound differences in Japanese.
• Utilized mismatch negativity (MMN) event-related potentials.
• Compared MMN asymmetries for voiceless and voiced stop consonants.

Main Results:

• Confirmed the prediction of an opposite asymmetry in Japanese MMN responses compared to English.
• Found evidence suggesting voiceless stops are underspecified in Japanese.
• Demonstrated language-specific patterns in phonemic feature underspecification.

Conclusions:

• Phoneme memory encoding is highly abstract, not directly tied to physical sound features.
• Japanese phonology exhibits underspecification for voiceless stops, contrasting with English.
• Supports a linguistic model where phonemes consist of abstract distinctive features, with variations across languages.