Parallel pitch processing in speech and melody: A study of the interference of musical melody on lexical pitch perception in speakers of Mandarin

Abstract

Music and language have long been considered two distinct cognitive faculties governed by domain-specific cognitive and neural mechanisms. Recent work into the domain-specificity of pitch processing in both domains appears to suggest pitch processing to be governed by shared neural mechanisms. The current study aimed to explore the domain-specificity of pitch processing by simultaneously presenting pitch contours in speech and music to speakers of a tonal language, and measuring behavioral response and event-related potentials (ERPs). Native speakers of Mandarin were exposed to concurrent pitch contours in melody and speech. Contours in melody emulated those in speech were either congruent or incongruent with the pitch contour of the lexical tone (i.e., rising or falling). Component magnitudes of the N2b and N400 were used as indices of lexical processing. We found that the N2b was modulated by melodic pitch; incongruent item evoked significantly stronger amplitude. There was a trend of N400 to be modulated in the same way. Interestingly, these effects were present only on rising tones. Amplitude and time-course of the N2b and N400 may suggest an interference of melodic pitch contours with both early and late stages of phonological and semantic processing.

Fig 1. Examples of stimuli and the order of events for each experimental trial.

(A) provides an example of stimuli with a rising lexical tone (C^R and I^R). Each row contains speech (written out in pinyin) and the music notation. The dashed rectangle highlights the critical item. The arrow above the last syllable indicates the direction of the lexical tone. In (B) the order of events for each trial with time shown in milliseconds (see Procedure below).

Fig 2. Correct response rate for the lexical verification task.

Behavioral scores are displayed in percentages (errors bars indicate 95% coincidence intervals). The dashed line indicates chance level. C^R (rising speech with rising melodic contour), C^F (falling speech with falling melodic contour), I^R (rising speech with falling melodic contour), I^F (falling speech with rising melodic contour).

Fig 3. Grand average waveforms and scalp maps.

(A) shows the grand average ERP waveform for all participants at electrode AFz. The left graph contrasts the waveforms for congruent and incongruent items with rising speech contours. The right graph contrasts that with falling contours. Dashed boxes demarcate latency windows used in the statistical analyses for the N2b (150–225 ms), P3b (250–360 ms) and N400 (400–500 ms). The y-axes denote amplitude in microvolts (-3 μV to 3 μV). The x-axes time in milliseconds (-200 ms to 600 ms). (B) contains topographical maps that display the scalp distribution averaged over the latency windows used in the analyses.