Task-dependent pitch auditory feedback control in cerebellar ataxia

Purpose The purpose of this study was to investigate how ataxia affects the task-dependent role of pitch auditory feedback control in speech. In previous research, individuals with ataxia produced over-corrected, hypermetric compensatory responses to unexpected pitch and formant frequency perturbations in auditory feedback in sustained vowels and single words (Houde et al., 2019; Li et al., 2019; Parrell et al., 2017). In this study, we investigated whether ataxia would also affect the task-dependent role of the auditory feedback control system, measuring whether pitch-shift responses would be mediated by speech task or semantic focus pattern as they are in neurologically healthy speakers. Methods Twenty-two adults with ataxia and 29 age- and sex-matched control participants produced sustained vowels and sentences with and without corrective focus while their auditory feedback was briefly and unexpectedly perturbed in pitch by +/−200 cents. The magnitude and latency of the reflexive pitch-shift responses were measured as a reflection of auditory feedback control. Results Individuals with ataxia produced larger reflexive pitch-shift responses in both the sustained-vowel and sentence-production tasks than the control participants. Additionally, a differential response magnitude was observed by task and sentence focus pattern for both groups. Conclusion These findings demonstrate that even though accuracy of auditory feedback control correction is affected by cerebellar damage, as evidenced by the hypermetric responses, the system still retains efficiency in utilizing the task-dependent role of auditory feedback.


Introduction
Ataxia is a neurological condition that results from damage to the cerebellum, affecting movement coordination across the body (Diener & Dichgans, 1992;Manto & Marmolino, 2009). Speech, a highly coordinative activity, is uniquely impacted in ataxia (Kent et al., 2000). Individuals with ataxia exhibit signi cant impairment in the naturalness of speech (i.e., pitch, loudness and timing) but intelligibility (i.e., how well words can be understood) is often spared or minimally affected, suggesting differential de cits in neural control for speech (Hilger et al., 2022). The goal of this study was to assess neural control of speech in ataxia, speci cally for feedback control of pitch.
A possible reason that speech naturalness is more disrupted than speech intelligibility in ataxia is because feedback control is impaired to a greater extent than feedforward control. Speech naturalness is strongly correlated with prosodic aspects of speech (i.e., fundamental frequency (f o ), vocal intensity, and timing), and is heavily reliant on feedback control (Guenther, 2016;Perkell et al., 2007;Yorkston et al., 1999). Feedback control is important for prosody to make salient distinctions in relative acoustic prominence, or the relative enhancement of phrasal stress within a phrase (Aylett & Turk, 2004;Cole, 2015;Kochanski et al., 2005;Tamburini & Caini, 2005;Turk & White, 1999). Adjusting the subtle, relative acoustic features of speech within the production of a phrase requires careful and precise self-monitoring of speech, which is known as auditory feedback control (Guenther, 2016;Tourville & Guenther, 2011). In this study, we investigate feedback control in ataxia to better understand the role of the cerebellum in pitch control and prosody. We hypothesize that due to cerebellar damage, feedback control is impaired in the adjustment of auditory targets based on sensory prediction and in generating corrective commands in response to sensory errors.
a. The role of the cerebellum in feedback control of speech The cerebellum plays an important role in predicting the sensory consequences of actions, which has important implications for adjusting production targets for speech (Davidson & Wolpert, 2005;Desmurget & Grafton, 2000;Guenther, 2016;Kawato, 1999;Larson et al., 1978;Miall & Wolpert, 1996). During speech production, cortical mechanisms send motor plans forward for production by generating auditory and somatosensory targets that are then mapped onto articulatory gestures for production (Guenther, 2016;Guenther et al., 2006;Tourville & Guenther, 2011). These targets are also used in ongoing production to compare incoming feedback with the intended targets to identify errors through mismatches in these comparisons (Tourville & Guenther, 2011).
The cerebellum provides indirect mediation in the generation of the auditory target (Guenther, 2016). The auditory target is a learned motor program for a speech movement, and is adjusted by the cerebellum for production based on the current gestural state and communicative context (Tourville & Guenther, 2011).
For example, the intended loudness of the auditory target may be adjusted depending on whether the speaker is in a quiet library or a loud restaurant. When the cerebellum is damaged, it is likely that it is unable to accurately estimate the necessary adjustments for the auditory target based on sensory information. It is possible that this damage creates an inaccurate estimation of the auditory target depending on the complexity of the production target and the extent of cerebellar damage.
We theorize that this inaccurate adjustment of the auditory target has implications for feedback control in speech. The generation of production targets, such as auditory targets, is typically viewed mainly as a feedforward process that sends motor programs forward for production (Guenther, 2016;Tourville & Guenther, 2011). Feedforward impairments are observed in ataxia, providing evidence for a cerebellar role in feedforward control of auditory targets for speech (Parrell et al., 2017). However, we theorize that inaccurate adjustments of the auditory target also help account for feedback control impairments to correct for errors. As described above, the cerebellum adjusts the auditory target based on sensory prediction, and that auditory target is then used for production mechanisms. However, that same auditory target is also copied to feedback control systems to compare incoming auditory feedback with the intended auditory target (Tourville & Guenther, 2011). If the auditory target is incorrectly estimated by the cerebellum, then comparisons with incoming feedback will also be incorrectly estimated. The result is that more errors, and greater errors, may be detected in ongoing speech, triggering feedback control to make more corrections.
However, there is an additional feedback process that is impaired in ataxia. The cerebellum not only makes adjustments to the auditory target but also to the corrective movements that are generated when  (Guenther, 2016). If the cerebellum is damaged, then the cerebellar adjustments made to the corrective movements will also be inaccurate. Therefore, auditory feedback control is considerably impaired in ataxia because of disruption to the cerebellum's roles in adjusting both the auditory target for comparisons to incoming feedback as well as the corrective movements when errors are identi ed from these comparisons. assess the e ciency and accuracy of the auditory feedback control system to rapidly correct for unexpected changes in ongoing speech (Burnett et al., 1998;Larson & Robin, 2016). In these studies, participants vocalize into a microphone while their auditory feedback is unexpectedly and brie y perturbed in an acoustic dimension such as pitch or formant frequency (Bauer & Larson, 2003;Burnett et al., 1998;Purcell & Munhall, 2006). When feedback is unexpectedly perturbed, speakers rapidly produce a re exive response (Behroozmand et al., 2012;Burnett et al., 1998;Hain et al., 2000;Kim & Larson, 2019;Larson & Robin, 2016;Scheerer & Jones, 2018). This response is thought to be re exive because of the automatic and involuntary nature of the response and the inability for speakers to suppress it (Bauer & Larson, 2003;Burnett et al., 1998;Zarate & Zatorre, 2008). The magnitude and timing of the response has been measured as an indication of the accuracy of the auditory feedback control system to correct for errors in ongoing speech (Larson & Robin, 2016).
Auditory feedback control in ataxia has been investigated in three recent studies in which participants were asked to hold a sustained-vowel sound for multiple seconds while either pitch or vowel formant frequency was brie y and unexpectedly perturbed through headphone auditory feedback (Houde et al., 2019;Li et al., 2019;Parrell et al., 2017). In all three studies, individuals with ataxia produced signi cantly larger re exive responses than the control participants, indicating that the cerebellum over-corrects due to inaccurate error estimation and correction. In this current study, our goal was to further investigate the cerebellar role in auditory feedback control by studying how cerebellar damage disrupts the taskdependent role of auditory feedback control for e ciency.
b. The role of the cerebellum in task-dependent auditory feedback control Past research has demonstrated that the sensitivity of auditory feedback control to correct for perceived errors varies by vocal task, likely a tool to maintain e ciency (Chen et al., 2007;Hilger et al., 2022;Hilger et al., 2023;Natke et al., 2003). Re exive pitch-shift responses are larger in magnitude in sentence production (Chen et al., 2007) and in singing (Natke et al., 2003) than in sustained-vowel production. Additionally, semantic focus also modulates the pitch-shift response, in which sentences with corrective focus elicit larger responses than new information focus (Hilger et al., 2023). This task-dependent nature of auditory feedback control re ects task-based variation in the precision and scale of the auditory target for production e ciency. Guenther (2016) describes how the size of the auditory target is scaled by the speaking task. For example, target regions for speech sounds shrink when speakers are asked to speak more clearly, resulting in more precise articulation (Perkell et al., 2002). Guenther (2016) attributes this variable target region to a strategy employed by the speech motor system called an economy of effort: speakers minimize the amount of movement required for production while maintaining intelligibility for the listener (Lindblom, 1983(Lindblom, , 1990). Essentially, speakers tune their production in relation to the speaking context and their judgment of the listener's ability to access the information in the speech signal (Lindblom, 1990).
For example, speakers may adjust their speech differently if they are speaking with a close friend compared to a stranger, or if they are giving a presentation compared to talking in a more casual setting.
By scaling the size of the target region for production, the speech motor system can retain e ciency while maintaining intelligibility.
The economy of effort strategy used by the speech motor system helps explain the task-dependent nature of the auditory feedback control system. Larger pitch-shift responses are observed in tasks such as singing and producing a sentence than for simply holding a vowel sound because the auditory target regions for voice f o are smaller and more precise in singing and sentence-production. Singing requires matching pitch to a musical note and speaking requires the production of intonational patterns for pitch. Therefore, both production tasks require a high level of precision and less room for error. Mismatches between the auditory feedback and the auditory target will be on average greater in these tasks because the auditory target is smaller. On the contrary, sustained-vowel production requires less precision in f o so detected mismatches will on average be smaller because there is more room for error (i.e., a larger region for the f o auditory target). nding indicates that the cerebellar impairment results in over-correction in the feedback control system. In the present study, we are interested in investigating the task-dependent role of auditory feedback control in ataxia. Because the cerebellum is important for adjusting the auditory target for the gestural state and the communicative context, we hypothesized that re exive responses would show less task-dependency in ataxia because the cerebellum is less accurate in adjusting the auditory targets for the speaking context based on sensory information. Therefore, we predicted (i) that speakers with ataxia would generate larger re exive responses than the healthy control participants in both a sustained-vowel and sentence-production tasks, and (ii) a smaller differential response would be observed between these tasks in the ataxia group. Additionally, we predicted (iii) that sentence focus would not mediate the magnitude of the re exive response in ataxia as was observed in the healthy speakers in Hilger et al. (2023). These ndings would demonstrate a reduced e ciency and accuracy of the auditory feedback control system due to cerebellar damage, reinforcing the important role of the cerebellum for auditory feedback control. All participants were native speakers of American English. Participants had normal, or corrected to normal, visual acuity. Out of the twenty-seven participants with ataxia, ve participants were excluded from participating in the auditory feedback perturbation tasks in this study: two participants did not pass the pure-tone audiometric thresholds of 40dB or better in both ears at 500, 1000, 2000, and 4000 Hz; two participants exhibited severe dysarthric impairments and were not able to complete the tasks; and one participant had self-reported cognitive di culty. The remaining 22 participants all passed the hearing and cognitive screenings.
Ataxia diagnosis was con rmed through participant self-reports of neurology and/or genetic testing. Participants were recruited through local support groups, outpatient clinics of local medical/rehabilitation facilities, yers in the monthly National Ataxia Foundation newsletter (NAF, 2016), social media, word of mouth, the Communication Research Registry at Northwestern University, and the CoRDS registry (Trudeau, 2013; Coordination of Rare Diseases at Sanford). Summary characteristics of speakers with ataxia are provided in Table 1.
Dysarthria type and severity were assessed using the Frenchay Dysarthria Assessment (Enderby & Palmer, 2008), a standardized assessment sensitive to various severity and subtypes of dysarthria. The FDA-2 assesses level of function for speech subsystems, including respiration, articulation, phonation, resonance, and intelligibility. Dysarthria severity was assessed by comparing the level of function across the speech subsystems. This study was approved by the Northwestern University Institutional Review Board.
b. Healthy control speakers.

Page 7/30
Twenty-nine adults, with no reported history of speech, language, or neurological impairment, were recruited for this study as age-and sex-matched control participants (10 males, 19 females). All participants were native speakers of American English. Ages ranged from 24-79 years (M = 54.1, SD = 15.0). Years of education ranged from 12-22 years (M = 17.3; SD = 2.1). Participants had normal, or corrected to normal, visual acuity. Participants passed hearing and cognitive screenings.

d. Auditory Feedback Perturbation Paradigm Tasks
To investigate voice pitch auditory feedback control in ataxia, we conducted a production study to elicit both sustained-vowels and sentences in individuals with ataxia and matched control participants. In the sustained-vowel task, participants were instructed to repeatedly hold an / / sound for three seconds at a time. For the sentence-production task, a visual world paradigm was used to elicit new information and corrective semantic focus patterns. We used a repeated-measures within-and between-subjects design with four independent variables: group (ataxia or control), task (vowel or sentence), semantic focus (new information or corrective), and perturbation direction (+/-200 cents). Two dependent variables were analyzed, pitch-shift re ex magnitude (cents) and peak latency (milliseconds).

f. Design and Stimuli
For both the sustained-vowel and the sentence-production tasks, participants followed instructions from a computer monitor and were told to vocalize at a comfortable but stable pitch and loudness level. The sustained-vowel task was completed rst. At the start of each trial, words appeared on the screen instructing participants to "Say 'ahh.'" Participants vocalized the / / sound until an instruction appeared to "Stop." There was a two second inter-trial rest period before the instruction for the next trial appeared. Participants were informed before the start of the experiment that they could rest between trials for as long as they needed if they did not trigger the microphone by vocalizing. Each trial was triggered by voice onset with an intensity of 70 dB or greater. The onset of the vocalization was detected using a voice onset detector module in MIDI software. The output from the voice onset detector was used to trigger an Eventide Eclipse Harmonizer (Quadravox, Eventide) to produce two pitch perturbations of pseudorandomized magnitude (+200 cents, -200 cents, 0 cents) at random intervals during the vocalization period. The pitch-perturbed stimuli were delivered with 700-900 ms variable interstimulus intervals within each 3-sec vocalization period. The pitch perturbations had durations of 200 ms in order to elicit the re exive response that occurs after the pitch perturbation rather than a volitional response that is triggered by longer pitch perturbation stimuli (Burnett et al., 1998;Hain et al., 2000). When a pitch perturbation was presented, the f o value produced by the participant was shifted +/-200 cents for the entirety of the 200-msec perturbation region. Participants completed two blocks of 45 trials each for the sustained-vowel task with a total of 90 vocalizations collected for each participant for this task.
For the sentence-production task, participants produced instructions within a visual world paradigm modeled from Ouyang & Kaiser (2015). The results of this task for the control participants were published in Hilger et al. (2023) and are included in the current study to compare with the speakers with ataxia. The same paradigm was included in the current study to compare with the speakers with ataxia and using the same procedures as in Hilger et al. (2023). For details on the paradigm, please refer to Hilger et al. (2023). Participants were told to imagine that they were playing a game with the computer (i.e., a computerplayer), which was using the participant's verbal instructions to move the pictures on the screen accordingly. However, they were told that the computer-player would occasionally make mistakes and move the wrong picture. Color pictures were presented on the screen in circular frames with the picture names displayed underneath each picture.
Based on how the pictures moved and how the computer player responded, the participants were cued to produce either new information focus or corrective focus on the target word, which was the moved object in the visual display. Figure 1B represents the production of a new information focus pattern. The pictures presented on the screen within Figures 1A and 1B are new within the discourse context because they are within a new set of pictures. Therefore, when the instruction is produced, the name of the picture within the instruction cannot be inferred from the previous discourse context. Figures 1C and 1D both represent productions using corrective focus.
The carrier phrase used in this task was, "Lay/not your OBJECT by your LOCATION." This phrase was chosen because voicing is continuous across the production of the phrase (apart from the break in voicing for the /b/ sound in "by" and the /t/ sound in "not"). Continuous or near-continous voicing was essential to implement pitch perturbations within the phrase and to measure a pitch-shift response, both of which use pitch tracking analyses that require modal voicing. The target word that was manipulated in this task was always the word in the OBJECT position. Words in the OBJECT position occur in the middle of the phrase where modal voicing is frequently used. On the contrary, words in the LOCATION position occur at the end of the phrase, which is a position highly vulnerable to creaky voice (Kreiman, 1982). By manipulating words in the OBJECT position, we hoped to elicit new information and corrective focus for our target words with modal voicing. Target words were chosen from the MultiPic pictures that were monosyllabic and contained all voiced sounds. Participants produced a total of 250 instructive phrases that were subdivided into ve blocks of 50 trials each. Within each block, there were around 20 trials each of new information and corrective phrases, depending on the ordering of the pictures. Around 10 trials per block were produced with target words that were neither new information nor bearing corrective focus, but which referred to items that were accessible from the prior context, having been explicitly mentioned in the prior instruction.
To study the effect of semantic focus on pitch auditory feedback control, brief pitch perturbations were applied in random trials of sentence production. The same perturbation magnitude as the sustainedvowel task was used in the sentence-production task, +/-200 cents. Pitch perturbations were applied 50 ms after voice onset on the rst word in the phrase (i.e., lay or not) on random trials. Perturbations were 200 ms in duration before auditory feedback was switched back to normal (i.e., unperturbed). We chose to apply the perturbation on the rst word in the phrase for two reasons: (1) there is evidence that auditory feedback control is more sensitive at the start of the phrase, possibly because the acoustic features at the start of the phrase are used as a reference to calibrate the relative acoustic production of the rest of the phrase (Hilger et al., 2020;Liu et al., 2009), and (2) we were interested in how auditory feedback control is utilized at the start of the phrase to prepare for anticipatory phrasal stress for semantic focus. This second goal was the topic of a future analysis and was not addressed in this current paper. Figure 2 displays an example production of the phrase, "Lay your well by your van" with well as the target word. Both the pitch perturbation and the pitch-shift response occur well before the onset of the stressed word (i.e., the pitch accent). At least thirty trials of each perturbation condition (i.e., +200 cents, -200 cents, and 0 cents) were included for each sentence focus type (i.e., new vs. corrective focus), which has been shown to be su cient for the signal averaging technique used in the pitch-shift response analysis (Bauer & Larson, 2003).

g. Acoustic Analysis
Acoustic data from the voice recordings for each trial were rst analyzed using autocorrelation-based pitch tracking in Praat software to transform the raw data into time-course measures of pitch (Praat To isolate the PSR from the pitch movement due to phrasal intonation, we completed a difference wave analysis. Without the difference wave analysis, we would not be able to determine if a change in pitch was due to the pitch perturbation or from natural changes in intonation. The difference wave analysis was accomplished by subtracting out the average intonation contour per participant and focus pattern from each individual experimental trial. First, the control trials per participant per focus condition were averaged together to calculate the average intonation contour each participant produced. Then, the average intonation contour was subtracted from the individual experimental trials (i.e., trials with perturbations) for that participant and focus condition. The resulting pitch contours re ected changes in pitch from the pitch perturbation. By completing this analysis for each individual trial, we were able to subtract out variability in intonation that may occur trial-by-trial. This analysis technique has been successfully utilized to analyze the PSR in phrase production for a variety of intonation patterns ( Mixed-effects models were chosen because of their ability to e ciently handle random by-participant variation (Guilley et al., 1999).
The objective of the rst analysis was to separately assess whether the magnitude and peak latency of the pitch-shift response were predicted by group (ataxia vs. control), task (sustained-vowel or sentenceproduction), sentence focus (new information vs. corrective), perturbation direction (+/-200 cents), and response direction (opposing vs. following the perturbation direction) while controlling for by-participant variance. After checking for and meeting assumptions of normality, separate models for response magnitude and peak latency were run. For each model, response magnitude and peak latency were separately used as dependent variables. In the rst model, group, task, perturbation direction, and response direction were used as xed effects, and participant was included as a random effect. In the second model, only the sentence-production task was analyzed to determine the effect of sentence focus on response magnitude and latency by group. Signi cant effects from the mixed-effects models were assessed by pairwise comparisons of Least Squares Means from the R package "lsmeans" (Lenth, 2017). Cohen's D was calculated for each signi cant effect using the "lme.dscore" function from the R package "EMAtools" (Kleiman, 2017).
A nal exploratory analysis was performed to compare the number of opposing and following responses by group, task, perturbation direction, and sentence focus pattern. Response direction was converted to a binary variable and used as a dependent variable in a logistic regression with group, task, perturbation direction, and sentence focuses included as xed effects using the "glm" function in R.

a. Production of Phrasal Stress by Sentence Focus
The motivation for this rst analysis was to determine whether the production task successfully elicited Refer to Table 2 for pitch-shift response magnitudes by task and group. Overall, response magnitudes were signi cantly larger in phrase production, for opposing responses, and for the ataxia group.  e. Response Direction The nal analysis was exploratory to compare the number of opposing and following responses by group, task, perturbation direction, and sentence focus pattern to determine if an experimental condition elicited more opposing or following responses. Additionally, we asked the exploratory question of whether cerebellar disease would elicit more following responses, potentially due to inaccurate estimation in the direction of the error and/or the direction of the correction. There was a total of 204 potential averaged responses for the sustained-vowel production task (51 participants X 2 perturbation directions X 2 response directions) and 408 averaged responses for the phrase-production task (51 participants X 2 perturbation directions X 2 response directions X 2 sentence focus patterns). After analysis, 566 averaged responses were measured out of a total of 612 potential responses. Table 3 lists the number of responses by experimental condition. Not all participants opposed and followed every perturbation direction for each experimental condition, which accounts for the number of "non-responses" indicated in the table.
There were no signi cant differences in the count of responses for any condition (p > 0.05), indicating that both groups of speakers opposed and followed the perturbation for all perturbation direction, task, and sentence focus conditions.

Discussion
The purpose of this study was to investigate the task-dependent role of pitch auditory feedback control in ataxia to measure the effect of cerebellar damage on the e ciency and accuracy of the auditory feedback control system. Impaired auditory feedback control has been measured in a number of studies study, we were interested if cerebellar damage would also affect the e ciency of auditory feedback control for scaling auditory targets appropriately for the speaking context. According to Lindblom (1983), the speech motor control system employs a strategy of economy of effort in order to reduce the magnitude and velocity of articulator movement required in speech production while maintaining intelligible speech. This strategy can be accomplished by scaling the size of the auditory target for production so that the system retains e ciency while maintaining intelligibility. Previous research has shown that healthy speakers produce larger pitch-shift responses in more complex tasks, such as sentence-production or singing, than for more simple production tasks (Chen et al., 2007;Natke et al., 2003). For example, people produce large responses in sentences with more salient semantic focus patterns, such as corrective focus compared to new information focus (Hilger et al., 2023). We predicted that individuals with ataxia would produce larger responses than the control group, similar to prior research, but that they would not have differentially larger responses in more complex tasks. In other words, we predicted large responses for the ataxia group across all tasks, regardless of task or focus pattern. This result would re ect cerebellar damage for accuracy of auditory feedback correction (i.e., responses are too large) as well as e ciency of correction (i.e., responses are not scaled by vocal task).
We found that, overall, the ataxia group produced larger pitch-shift responses than the control group for both the sustained-vowel and the sentence-production tasks. These results support the ndings from previous studies showing that ataxia results in hypermetric compensatory responses to unexpected changes in auditory feedback. From the current study, over-corrected responses were observed across speaking tasks for both simple tasks (i.e., sustained-vowel production) and more complex tasks (i.e., sentence-production) in ataxia. This over-corrected response demonstrates the inaccuracy of the feedback control system for appropriately scaling the corrective movement for error.
Next, we assessed the e ciency of auditory feedback control in ataxia by measuring whether taskdependent scaling of auditory targets was retained. Contrary to our hypothesis, we observed a taskdependency in auditory feedback control in ataxia, nding that participants with ataxia were able to produce a differential response magnitude both by task and by semantic focus pattern. Similar to the control speakers, speakers with ataxia produced larger pitch-shift responses in the sentence-production task than the sustained-vowel task, and for sentences with corrective focus than for new information focus. Overall, this nding demonstrates that despite the cerebellar disruption, the motor speech system is still able to employ the economy of effort strategy by scaling the auditory target by speaking context.
For example, the larger responses observed in the sentence-production task for both groups demonstrates that the auditory target was scaled down for precision in this task, resulting in a greater detected mismatch. It is possible that e ciency is retained in the speech motor control system in ataxia to scale the auditory target by speaking task, even though accuracy is reduced from cerebellar damage when correcting for errors. Another interpretation is that attentional demands for sentence production as well as for corrective focus could in uence feedback correction, resulting in larger responses. This interpretation is not mutually exclusive with the theory of economy of effort and also aligns with the neural mechanisms studied in this paper. Higher-level cortical mechanisms are typically preserved in ataxia, which are the mechanisms responsible for attention. In this case, these higher-level cortical mechanisms may be responsible for scaling the auditory target based on attentional demands as well as economy of effort.
These ndings have implications for both our understanding of the role of the cerebellum in feedback control as well as our clinical approach to treating speech naturalness in ataxic dysarthria. For the role of the cerebellum, individuals with cerebellar damage exhibit a hypermetric response to auditory feedback perturbations across speaking tasks, but the task-dependent role of auditory feedback control remains intact. This demonstrates that the auditory target was effectively scaled to the demands of the speaking task, despite the inaccuracy of the corrective response. A potential explanation for this nding is that there may be several variables that are used to scale the auditory target, some of which might involve more cortical processes rather than cerebellar processes. For example, scaling the target may include factors such as the speaking context, the current gestural state, task complexity, prosodic targets, and fatigue. Any combination of these factors may scale the production target greater or smaller in size depending on whether greater precision is required for the speech production task. Since we found that the pitch-shift response was effectively scaled by production task and semantic focus pattern in ataxia, it is likely that there are higher-level cortical processes that are involved with scaling production of targets for these factors. Overall, these ndings show that cerebellar damage disrupts the accuracy in estimating corrective responses to unexpected auditory feedback perturbations, but e ciency for scaling the response by task and sentence focus remains intact.
Clinical techniques can take advantage of the intact task-dependency in auditory feedback control in cerebellar patients by addressing the inaccurate, over-correction in pitch control. As found in Hilger et al.
(2022), speech naturalness is more signi cantly impaired in ataxic dysarthria than speech intelligibility, re ecting the patterns in disrupted prosody. The prosodic impairments are likely in-part due to overcorrection in f o , and probably intensity, in online speech production. Because the original auditory target is inaccurately estimated by cerebellar processes, more errors may be detected, triggering more corrective movements, which are also likely to be inaccurately estimated. We theorize that this feedback impairment results in a pattern of frequent and over-estimated corrections in pitch, and likely loudness as well. This impairment would help explain the variable pitch and loudness characteristics observed in ataxic dysarthria. To apply this work clinically, a treatment study could test the use of other feedback mechanisms, such as visual and kinesthetic feedback, to improve prosodic control by strengthening the internal models of the auditory targets. Since the task-dependent role of auditory feedback is intact, the production target itself is not wholly compromised in ataxia and has potential for responding to rehabilitative techniques.
Another factor we analyzed in this study was response direction, or whether the speaker opposed or followed the direction of the perturbation. Similar to the results from the control speakers in Hilger et al.
(2023), there were no signi cant differences in the number of opposing or following responses for any task or group in this study. A potential hypothesis was that the cerebellar damage in ataxia would result in more following responses because the estimated corrective movement is inaccurate. However, this hypothesis was not supported because the speakers with ataxia opposed and followed the perturbations to the same extent as the control speakers. A simple explanation for when the response follows the perturbation is that it is possible that the direction of the response occurs due to chance.
Limitations in this study include both the nature of the task as well as the number of perturbation variables that were analyzed. Although we attempted to elicit more natural productions using a visualworld paradigm, the task still involved a rather unnatural-sounding carrier phrase because of the constraint for voiced sounds with the words in the phrase. Additionally, the sentence-production task did not involve the same level of complexity as spontaneous speech or conversation entails, which may demonstrate interesting ndings in relation to task complexity. Lastly, only one pitch perturbation magnitude was tested in this study because of an attempt to limit the number of trials needed. However, because we only studied one perturbation magnitude, we could not measure a differential response to different perturbation magnitudes in the production tasks. Next steps in this research should include various perturbation magnitudes and various tasks of differing complexities. Despite these limitations, clear effects were measured by group and task that shed light on the role of the cerebellum in auditory feedback control.

Conclusion
The purpose of this study was to measure the task-dependent role of auditory feedback control in individuals with ataxia to investigate how accuracy and e ciency are affected by cerebellar damage. We found that the ataxia group produced larger pitch-shift responses than the control group for both the sustained-vowel and sentence-production tasks. Additionally, a differential response magnitude was observed by task and by semantic focus pattern in both the ataxia group and the control group. These ndings demonstrate that even though accuracy is affected by cerebellar damage, as evidenced by the hypermetric responses, the system still retains e ciency in utilizing the task-dependent role of auditory feedback.

Declarations Ethical Approval
This study was approved by the Northwestern University Institutional Review Board. Figure 1 Sample display of the task. The rst screen presented is 1A in which four pictures are presented with a cue to wait. In 1B, an arrow appears between knee and net, cueing the participant to produce the instruction (new focus), "Lay your knee by your net." In 1C, an incorrect picture is moved, and the participant is cued to provide a corrective statement, "Not your WHALE by your net." In 1D, the participant is cued to repeat the original instruction with corrective emphasis, "Lay your KNEE by your net." In 1E, the correct picture is moved. Figure 2 Example timing of the pitch perturbation in the phrase, "Lay your well by your van." The speech waveform (top) and the spectrogram (middle) are segmented by words and phonemes (bottom). The pitch track is displayed as a red line within the spectrogram. In this example, the target word, well, is the stressed word, termed "pitch accent." The pitch perturbation occurs on the word lay, indicated by the red box and arrow at the bottom, and the pitch-shift response occurs shortly after the onset of the perturbation, indicated by the dashed box. Production of the stressed word by sentence focus for the ataxia group. 3A displays the mean and standard error for mean f o of the stressed word for corrective focus (black circle) and new focus (gold triangle). 3B displays the same mean and standard error for mean intensity, and 3C for vowel duration.
Statistical signi cance is indicated by stars corresponding to p-values (* p < 0.05; ** p < 0.01; *** p < .001).  Sentence-production grand averaged pitch-shift responses for all participants by downward perturbations (5A) and upward perturbations (5B). The duration of the perturbation is indicated by the grey bar. Solid lines indicate opposing responses and dashed lines indicate following responses. The ataxia group is represented by the black lines and the control group is indicated by the gold lines. New information focus