Lecture | CHiLL series
The Time Course of Tone Encoding in Mandarin Speech Production and Perception: Evidence from MVPA
- Date
- Wednesday 28 October 2026
- Time
- Series
- Chinese Linguistics in Leiden (ChiLL)
- Location
-
Herta Mohr
Witte Singel 27A
2311 BG Leiden - Room
- 0.30
Abstract
In Mandarin, lexical tone and segmental information jointly distinguish word meanings, yet the temporal dynamics of these representations during speech perception and production remain incompletely understood. Across two EEG experiments (N = 24 each), we investigated passive listening and overt picture naming using a fully crossed 4 (tone) × 2 (onset consonant) × 2 (rhyme) factorial design. Combining time-resolved multivariate decoding with representational similarity analysis, we tracked the time courses and representational geometry of these linguistic features. During perception, our results showed that neural responses distinguished initals-consonant classes, talker voices, and tone categories as speech unfolded, with tone-category discriminability evolving dynamically over time. During production, decoding analyses revealed a distinct cascaded emergence of linguistic representations: animacy information appeared first, followed by tone, consonant, and rhyme, with significant decoding emerging at 160, 190, 326, and 434 ms after picture onset, respectively. Further analyses revealed that direction- or slope-related tonal representations likely drove the early neural encoding of tone during planning. Collectively, across perception and production, tonal and segmental representations emerge in a rapid, temporally overlapping cascade, with tonal information encoded earlier than previously assumed in both modalities, revealing its unique and central role in Mandarin speech processing.