Speaking is a highly complicated motor behavior. Fluent speech requires not only precise coordination of many muscles, but also integration of auditory and somatosensory feedback. Auditory feedback (i.e., the sounds of our own voice), in particular, is widely acknowledged to be essential for speech production and is a central part of models of speech motor control (Hickok et al., 2011; Guenther, 2016; Parrell et al., 2019). In recent three decades, the real-time altered auditory feedback (AAF) experimental paradigm has provided plentiful evidence that auditory feedback plays a critical role in both online feedback control (e.g., Burnett et al., 1998; Hain et al., 2000) and in updating feedforward control (e.g., Houde and Jordan, 2002; Purcell and Munhall, 2006). Perturbations of auditory feedback introduce prediction errors (i.e., discrepancies) between the predicted and actual auditory feedback. When auditory feedback is perturbed inconsistently, an opposing response (compensation) that changes in the opposite direction to the perturbation direction will be elicited to reduce the perceived error within an ongoing production. When auditory feedback is perturbed consistently, speakers can gradually modify their motor plans over multiple successive utterances (adaptation) to integrate information from those errors.

By using the AAF paradigm, previous studies have demonstrated pitch compensation and adaptation across various contexts, including singing (Natke et al., 2003), nonspeech vocalization (Larson et al., 2001), and speech production (Jones and Munhall, 2002). Those studies, however, have primarily focused on non-tonal languages (predominantly English), where pitch has a less prominent suprasegmental role. In contrast, around 70% of the world’s languages use changes in pitch (i.e., tones) to alter word meaning, which serve an important linguistic function (Yip, 2002). Such a major difference could potentially affect speakers’ pitch control and determine the extent to which they adjust their pitch in response to pitch perturbations. In recent decades, to investigate the possible influence of a speaker’s entire language system on pitch control, researchers started to examine pitch compensation and adaptation in tonal language speakers. The goal of this mini review is to provide a summary of the progression of altered auditory feedback (AAF) as a method to understand pitch motor control in tonal language speakers. In total, 18 studies (see Table 1) using the AAF paradigm to investigate pitch compensation and adaptation in tonal language speakers were reviewed.

Previous studies in non-tonal language speakers have shown that participants lowered their pitch when F0 was shifted up and raised their pitch when it was shifted down (e.g., Hain et al., 2000; Jones and Munhall, 2000). Jones and Munhall (2002) first investigated pitch compensation in tonal language speakers. They found that similar to non-tonal language speakers, Mandarin speakers changed their pitch productions to the opposite direction of the manipulation when inconsistent pitch perturbations (+ 100 cents or no perturbation) were introduced to their auditory feedback during production of Mandarin word /ma/ with a high level tone (tone 1, ‘mother’). Mean response latency was 211 ms, which was longer than non-tonal language speakers producing sustained vowels (114 ms, Hain et al., 2000; 130 ms, Larson et al., 2001). The response magnitudes, however, were not reported in Jones and Munhall (2002). In a later study, compensatory responses to pitch perturbation (±100 cents or no perturbation) were examined in Mandarin speakers when they produced disyllabic Mandarin phrases (/ma ma/) with different tonal contours, High-High (H-H), High-Rising (H-R), and High-Falling (H-F) (Xu et al., 2004). They found that Mandarin speakers had larger compensatory responses (49–84 cents) when experiencing pitch perturbation to mandarin bitonal sequences, than English speakers’ compensatory responses to pitch perturbation when producing the sustained vowel /a/ (30 ~ 40 cents, Hain et al., 2000; Larson et al., 2001). The results seem to suggest that pitch feedback control in tonal languages is more sensitive than that in non-tonal languages. The same research group further investigated the compensatory responses to pitch perturbation in Mandarin speakers when producing Mandarin sentence (/mao1 mi1 mo1 ma1/, ‘kitty touches mom’) with different intonation patterns (statement v.s question) (Liu et al., 2009). Compared to meaningless disyllabic Mandarin phrases, generally smaller compensatory responses (20 ~ 30 cents) were elicited during Mandarin sentence productions.

Ning et al. (2014) conducted an initial study where they directly compared pitch compensation between speakers with and without tonal language experience. Three groups of participants (native English speakers who were never exposed to tonal languages, L2 learners of Mandarin whose native language was English, native Mandarin speakers) vocalizing the same vowel /a/ while receiving F0 perturbation with different magnitudes (±50 and ± 100 cents). No significant difference between groups were found in response amplitude or latency. In addition, the authors modeled the entire F0 contours using generalized additive models and found that F0 contours of the native English speakers and L2 learners were affected by perturbation magnitude, but Mandarin speakers were not. The authors thus suggested that instead of increasing pitch-control sensitivity, tonal language experience may contribute to the development of a more stable vocal control. In a later study, the same group of researchers further investigated whether compensatory responses induced by pitch perturbation were affected by language or vocal training experiences (Ning et al., 2015). The compensatory responses of four groups of participants (the same three language groups as in their previous study and trained vocalists with English as their native language) during production of Mandarin bitonal sequences /ma1 ma1/, /ma1 ma2/, /ma1 ma4/, and the vowel /a/ were examined. In contrast to previous findings, both native Mandarin speakers and trained vocalists showed significantly reduced response amplitudes when producing sustained vowel compared to native English speakers. During productions of Mandarin bitonal sequences, only native Mandarin speakers, but not trained vocalists, exhibited significantly smaller magnitude deviation in comparison to native English speakers. The authors concluded that tonal language speakers have more robust or stable pitch control during both linguistic (Mandarin bitonal sequences) and nonlinguistic (sustained vowel) productions and are thus less affected by auditory feedback perturbations. Liu et al. (2010) showed that pitch motor control even differed across different tonal languages. Cantonese speakers exhibited significantly smaller responses than Mandarin speakers when receiving relatively large pitch perturbations (±200, or ± 500 cents).

More recently, Ning (2019, 2022b) conducted two studies in speakers of Taiwanese Southern Min to investigate the impact of pitch contour on pitch compensation. The author proposed that larger and faster compensatory responses would occur when the shifted pitch overlaps with another level tone. The results showed that when the H level tone of the HM word (/un⁵⁵-ding³³/) was downshifted by 250 cents to the M level, and when the L level tone of the LM word (/un¹¹-tong³³/) was upshifted by 150 cents to the M level, larger compensatory responses were elicited. However, no such effect was found for the MM word (/un³³-too³³/) (Ning, 2019). Specifically, upshifting the M level tone of the MM word to the H level or downshifting the M level tone of the MM word to the L level did not result in larger pitch compensation. In the later study, the author further examined how different pitch contours would affect pitch compensation and proposed that pitch perturbation to the flat contour (e.g., MM), where speakers could maintain stability more easily, would result in smaller pitch compensation compared to the rising (e.g., LM) and falling (e.g., HM) contours. The results support the author’s hypothesis, indicating that flat pitch contour exhibited a greater resistance to pitch perturbation compared to the rising or falling pitch contours. However, unlike the previous study, downshifting the rising contour or upshifting the falling contour (i.e., when the shifted pitch overlapped with the M tone) did not result in larger pitch compensation (Ning, 2022b).

By combining AAF paradigm with brain stimulation and neuroimaging techniques, a series of studies have been conducted by Liu’s group to investigate the neural basis of pitch control in tonal language speakers (e.g., Liu et al., 2020, 2023; Li et al., 2022). They have demonstrated the causal role of left dorsolateral prefrontal cortex (DLPFC) in pitch compensation by showing that inhibitory continuous theta burst stimulation (cTBS) over DLPFC elicited larger compensatory responses and smaller event-related potential (ERP) P2 responses in Mandarin speakers experiencing pitch perturbation (+200 or +500 cents), compared to sham stimulation (Liu et al., 2020). In contrast, transcranial direct current stimulation (tDCS) over DLPFC led to significantly faster but smaller compensatory responses than sham stimulation (Chang et al., 2023). The authors suggested that DLPFC is involved in top-down inhibitory control over vocal motor behavior. Altering the function of left DLPFC, either by disrupting or enhancing it, can result in impairment or enhancement of this top–down inhibitory control, which regulates the extent to which feedback perturbations influence speech production. The involvement of left inferior frontal gyrus (IFG) and supplementary motor cortex (SMA) in pitch compensation in Mandarin speakers have also been demonstrated in studies of Dai et al. (2022) and Li et al. (2023), in which enhanced or reduced compensatory responses to pitch perturbation were observed after transcranial alternating current stimulation (tACS) over left IFG or cTBS over left SMA, respectively.

Previous research has indicated that the posterior superior temporal gyrus (pSTG) plays a critical role in both speech perception and production (Okada and Hickok, 2006). The dual-stream model suggests that the pSTG, predominantly in the left hemisphere (Hickok and Poeppel, 2007), is responsible for encoding auditory-phonological representations, while the Directions Into Velocities of Articulators (DIVA) model proposes that both the left and right pSTG are involved in auditory feedback control (Guenther et al., 2006). However, it is important to note that these models are primarily based on findings from English speakers. In a recent study, a right-lateralized contribution of the pSTG to pitch feedback control has been found in tonal language speakers (Liu et al., 2023). More specifically, cTBS over right, but not left pSTG resulted in reduced compensations for pitch perturbations accompanied by smaller N1 and larger P2 responses in Mandarin speakers. In contrast, both the left and right supramarginal gyrus (SMG) has been found to causally contribute to pitch compensation in Mandarin speakers (Li et al., 2022).

By using AAF paradigm, pitch compensation and adaptation has been demonstrated in tonal language speakers. The compensation and adaptation are under the influence of various factors, such as the type of perturbation stimuli, the speaker’s language background, and their vocal training. However, the exact influence of tonal language experience on pitch control still remains unclear.

D-LT: Writing – original draft, Writing – review & editing, Writing – original draft, Writing – review & editing.