فرضیه وابستگی بازخورد حسی در اشخاصی که لکنت زبان دارند

The present study investigated the role of sensory feedback (auditory, proprioception, and tactile) at the intra- and inter-gestural levels of speech motor coordination in normal and fast speech rate conditions in two groups: (1) persons who stutter (PWS) and (2) those who do not (PNS). Feedback perturbations were carried out with the use of masking noise (auditory), tendon vibration (proprioception), and nonwords that differed in the amount of required tactile lip contact (/api/ + tactile and /awi/ − tactile). Comparisons were also made between jaw-free and jaw-immobilized (with a bite-block) task conditions. It was hypothesized that if PWS depend more strongly on sensory feedback control during speech production, they would show an increase in variability of movement coordination in the combined presence of fast speech rates and feedback perturbations, in particular, when jaw motions are blocked and adaptations in the other articulators are required to achieve the task goals. Significant feedback perturbation effects were found for both groups, but the only significant between-group effect was found at fast speech rates in the jaw-free condition, showing that control speakers were more perturbed at the intra-gestural level of coordination than PWS when simultaneous (auditory, proprioceptive, and tactile) perturbations were present. The findings do not provide support for either the feedback dependency or the sensory deficit hypotheses described in the literature to explain movement characteristics found in fluent speech production of PWS.

The “motor skill” view of stuttering (van Lieshout, Hulstijn, & Peters, 2004) posits that persons who stutter (PWS) are at the lower end of a speech-motor skill continuum relative to persons who do not stutter (PNS). According to this view, it is assumed that PNS produce speech like any other well-practiced motor task, weighted towards feedforward control that is highly automatized and based on dynamical principles of motor control (Saltzman and Munhall, 1989 and van Lieshout et al., 2004). In contrast, PWS are argued to be less skilled in speech production and are inclined to use a less automated strategy that is more dependent on sensory information for the control of speech movements (Adams et al., 1993, van Lieshout et al., 1996a, van Lieshout et al., 1996b and van Lieshout et al., 1993). Within this motor skill frame work, PWS are not considered speakers with a sensory deficit as some researchers have suggested (Archibald and De Nil, 1999, Loucks and De Nil, 2006a and Loucks and De Nil, 2006b). Rather it is proposed that their speech motor symptoms reflect limitations in efficiency and agility similar to what has been reported for less skilled performances in other types of motor tasks (Broderick and Newell, 1999 and De Nil, 1999). Others have also hypothesized that PWS might show an over-reliance on sensory feedback, but not as a compensatory strategy to prevent stuttering due to limitations in motor skill (Max, Guenther, Gracco, Ghosh, & Wallace, 2004). Rather, these authors claim that this over-reliance on sensory feedback, given the inherent delay between a motor command and its sensory consequences, may actually result in an unstable speech motor system characterized by oscillations and resets, which at a behavioral level is assumed to be reflected in disfluent speech production. It is expected that these instabilities in the speech motor system would increase with increases in speech rate due to the extra demands on temporal processing of sensory information at fast speech rates (Max et al., 2004). Although over-reliance on sensory feedback has been implicated in stuttering both as a cause and as a compensatory strategy, there is very little information in the stuttering literature on the exact type of sensory information that PWS are assumed to be over-dependent on (whether its auditory, proprioceptive, or tactile1), at what level it is being used for motor control, whether its use in PWS differs from that of PNS and so on. This paper is aimed at answering some of these questions. The role of feedback in speech production is currently a hot topic of research. Broadly stated, auditory and oro-sensory2 feedback arising from speech related activity is utilized to learn and control speech movements (Guenther, 2006, Postma, 2000, Purcell and Munhall, 2006 and Smith, 1992). For example, the importance of auditory information for the control of speech movements has been demonstrated using speech compensation and adaptation paradigms. There are studies that report of bite-block induced vowel distortions and acoustic target variability in hearing impaired (Tye, Zimmermann, & Kelso, 1983) and postlingually deaf cochlear implant users when their implant was temporarily turned off (Lane et al., 2005). Furthermore, studies have also demonstrated the significance of oro-sensory information, independent of auditory information for the control of speech movements. For example, Tremblay, Shiller, and Ostry (2003) used a robotic arm and applied perturbations to the jaw during speech and non-speech tasks that had no auditory consequences (and were undetected by the speakers). After sufficient training, they found negative after-effects in the jaw trajectories after the perturbing force was removed from the jaw, which indicates oro-sensory adaptation. Together, these studies indicate the importance of response-produced feedback in the maintenance of appropriate accuracy of speech gestures and as described by Postma (2000) may operate at three unique levels of control namely, Directive (feedback driving motor commands), Tuning (recalibration or updating of internal models), and Corrective (real-time correction of detected errors). However, the use of both auditory and oro-sensory feedback for the control of speech movements has been traditionally argued to be speech-rate dependent. Adams et al. (1993) have suggested that as movement duration and total time spent in deceleration increases the speech motor control system may shift from a strategy that is predominantly open-loop at fast movement speeds to a more closed-loop feedback control at slow movement speeds (Bullock & Grossberg, 1988; cf. Desmurget & Grafton, 2000). This would permit the use of sensory feedback to make real-time corrective adjustments as a target position is approached. However, at faster speech rates, closed-loop feedback control would be difficult, as the delay in the arrival of the sensory feedback (in particular auditory information) is considered too large to make real-time corrective adjustments to the motor commands which under these circumstances are assumed to be issued at relatively short intervals, and this situation may render the speech system unstable (Max et al., 2004). Unlike the above three levels of feedback control, the role of sensory feedback has been defined in a very different way with respect to movement coordination in the context of the dynamical systems theory (DST; e.g., van Lieshout, 2004) and the gestural linguistic model (GLM) of speech production (Browman and Goldstein, 1992, Goldstein and Fowler, 2003 and Saltzman and Byrd, 2000). With regards to the former approach, a few studies have systematically addressed the role of sensory feedback on movement coordination using tendon vibration, a powerful tool altering the activity of muscle spindles (Verschueren et al., 1999 and Verschueren et al., 2002). For example, Steyvers and colleagues (Steyvers, Verschueren, Levin, Ouamer, & Swinnen, 2001) applied unilateral tendon vibration during cyclic bimanual forearm movements, across increasing cycling frequencies and showed that (regardless of cycling frequency) vibration resulted in an increase of the phase lead of the vibrated dominant limb, along with a concomitant increase in the variability of relative phase. These results reinforce the role of proprioceptive information during bimanual coordination as suggested by others (Verschueren et al., 1999). Concerning speech, although some evidence indicates a relationship between sensory feedback manipulations and greater acoustic target variability (e.g., Hoole, 1987 and Lane et al., 2005), there is a paucity of literature involving these manipulations and variability of speech movement coordination. Speech articulators are in many ways different from limb muscles (Kent, 2004), but it is known that the masseter muscle spindles provide jaw position and velocity information during the production of speech gestures that involve jaw as the primary articulator (such as in low–high vowel productions). Additionally, given the postural support role of the jaw for other articulators such as the tongue and the lower lip, jaw proprioceptive information may be used as a reference signal for the coordination of other articulators (Loucks & De Nil, 2001). Although there is some evidence in the literature to suggest that robust tendon vibration effects can be elicited though masseter vibration for simple speech (discrete vowels) and non-speech tasks (Loucks & De Nil, 2001), masseter muscle-spindle contributions to continuous speech and inter-articulatory coordination remain unexplored up till now. However, based on several studies using cyclical uni- and bi-manual tasks (analogous to cyclic reiterative speech tasks as used in the present study) that have demonstrated robust effects of tendon vibration on complex intra- and inter-limb coordination (e.g., Steyvers et al., 2001), and given the fact that in humans, jaw closing muscles (masseter, temporalis and medial pterygoid) have muscle spindle densities similar to that of limb muscles (Dubner et al., 1978 and Rowlerson, 1990), it seems reasonable to expect similar effects of tendon vibration on cyclical reiterated speech movements. In the context of DST, it can be hypothesized that sensory feedback from speech articulators is used to stabilize the output of a coupled neural oscillatory system. Within such systems, feedback gain decreases with smaller movement amplitudes which may result in a reduction of the neural oscillator-effector coupling strength and system stability (Peper and Beek, 1998, van Lieshout et al., 2004 and Williamson, 1998). We hypothesize that this mechanism may also be the way by which tendon vibration influences system stability as its presence typically results in a decrease in movement amplitude or movement undershoot (Loucks & De Nil, 2001). The stabilization effect of movement related feedback may be all the more important when at fast speech rates the coupling strength between articulators decreases (Peper et al., 1995, Schmidt et al., 1998, van Lieshout et al., 2004 and Zanone et al., 2001). With respect to GLM, this theory allows for a differentiation of the role of feedback at very specific (intra- and inter-gestural) levels of speech motor coordination (Browman and Goldstein, 1992, Goldstein and Fowler, 2003 and Saltzman and Byrd, 2000). GLM uses the task dynamics (TD) model (Saltzman, 1991, Saltzman and Kelso, 1987 and Saltzman and Munhall, 1989) to map gestures onto simulated speech articulator movements. Within this framework, gestures are defined as abstract action goals with specific vocal tract constriction degree and location parameters (i.e., tract variables), which control functionally coupled articulatory synergies at the level of individual movements. According to the TD model, articulatory trajectories are produced by critically damped oscillatory movements of a mass-spring system and the final characteristics of the movements themselves are determined by complex and lawful mechanisms of interaction between the articulators (Browman & Goldstein, 1992). Thus, the coordination between individual articulators within the constraints of a particular gesture (i.e., intra-gestural) such as between upper and lower lip during production of a bilabial /p/ sound, is said to emerge as an implicit consequence of gesture-specific dynamical parameters. This provides a mechanism for independent reorganization of individual articulatory contributions in support of motor equivalence demands without the need for explicit trajectory planning and/or reparameterization in order to compensate for oral-articulatory perturbations (Saltzman & Munhall, 1989). With regards to the role of sensory feedback at the intra-gestural level, the GLM assumes that this information is implicitly available in the form of the articulator state information (e.g., position and velocity of the upper lip and lower lips) and is used to derive the gestural state of the system (e.g., bilabial closure size and closing velocity). This mapping also forms the basis for functional constraints or gesture-specific coupling functions among the individual articulators3. However, an explicit articulator state-estimation process has not been incorporated into the GLM nor has the type and role of sensory information at this level been empirically demonstrated using actual speech data. On the other hand, sensory information in the form of peripheral articulator state is considered important within the GLM for maintaining temporal relationships between independent gestures (i.e., inter-gestural level) as for example, between bilabial closing gestures and tongue or laryngeal gestures (Saltzman, Löfqvist, Kay, Kinsella-Shaw, & Rubin, 1998). Evidence for this statement is based on an experiment wherein unanticipated lip perturbation during discrete and repetitive production of the syllable /pa/ resulted in phase shifts in the relative timing between the two independent gestures (lip closure and laryngeal closure) both within the phoneme /p/ and between successive /pa/ syllables. These findings suggest that the timing at the inter-gestural level is not rigidly specified in a pure feedforward manner over an entire utterance, but is coupled bidirectionally to the feedback arising from the peripheral articulators (Saltzman et al., 1998; see also Beek, Peper, & Daffertshofer, 2002 for a similar approach in limb control). Thus, based on the findings from Saltzman et al. (1998), the assumption can be made that sensory feedback is critical for timing at the inter-gestural level of speech motor coordination, but there is not sufficient evidence to indicate if and how this information is used at the intra-gestural level. Studies from the stuttering literature suggest that this process may be different for PWS, as it has been argued that they differ from PNS in the acquisition, processing or use of sensory information at the intra-gestural level of coordination (Archibald and De Nil, 1999, De Nil and Abbs, 1991, Loucks and De Nil, 2006a, Loucks and De Nil, 2006b, van Lieshout et al., 1993, van Lieshout et al., 1996a and van Lieshout et al., 1996b). Particularly, as pointed out earlier, limited speech motor skills in PWS may make them more inclined to use a strategy that is more dependent on sensory information for the control of individual speech movements, such as those used within the context of a particular gesture (van Lieshout et al., 1993, van Lieshout et al., 1996a and van Lieshout et al., 1996b). It is argued that a feedback driven motor control strategy would typically lead to a stronger than normal temporal asynchrony of individual movements than usually seen for articulators involved in a given speech gesture (van Lieshout et al., 1994, van Lieshout et al., 1996a and van Lieshout et al., 1996b). Support for this claim was found in studies showing systematic lags at the intra-gestural level for PWS as indexed by the relative timing (of peak velocities) of upper lip, lower lip, and jaw movements (Caruso et al., 1988 and van Lieshout et al., 1994). However, these findings have to be dealt with cautiously since these measures of extrinsic timing are often confounded with specific changes in movement characteristics (duration, amplitude) as a function of variations in speech rate, and other studies in the past have not been able to find systematic differences in relative timing of discrete kinematic events between PWS and controls using the same measures (for a review see Alfonso, 1991). This is one reason why some researchers have promoted the use of relative phase measures that are independent of absolute changes in movement amplitude and duration due to rate variations (see also Kelso and Tuller, 1987, Saltzman et al., 1998, van Lieshout et al., 1997 and Ward, 1997). Thus, by measuring changes (M and SD) in relative phase at both the intra- and inter-gestural levels it is possible to discern the role of response-produced feedback on the nature and stability of speech motor coordination. 1.1. The present investigation The aim of this investigation is to test the notion of feedback dependence in the control of speech movement coordination in PWS, specifically as relating to intra- and inter-gestural coordination. To this end, we temporarily perturbed auditory (using masking noise), and proprioceptive (tendon vibration) feedback each on their own and in combination, with or without a bite-block in place, and at two different speech rates (normal and fast). In addition, we studied the role of tactile feedback in speech coordination by using speech stimuli that differ in the amount of tactile lip contact. The use of a bite-block condition is incorporated in this study since researchers have suggested that sensory information may be utilized (and more relevant) when overcoming (developmental or experimentally induced) peripheral changes in oral morphology (Lane et al., 2005). Thus, it was expected that the effects of sensory feedback perturbations would be greater in the presence of a bite-block. Although bite-blocks are sometimes referred to as perturbations (Namasivayam, van Lieshout, & De Nil, 2008) for the sake of clarity in the current study the presence or absence of a bite-block will be referred to as a task “condition”, while auditory (masking noise) and proprioceptive (tendon vibration) feedback alterations will be referred to as “perturbations”. Given the GLM framework used for this study (Goldstein et al., 2007 and Saltzman and Munhall, 1989), we expect to see the effects of feedback perturbation at the inter-gestural level of coordination for both groups. However, if PWS as suggested in the literature have a stronger dependency on feedback at the intra-gestural level we expect them to show more variability than PNS in the coordination between the lips for bilabial closure during sensory feedback perturbations. Thus, we expect significant feedback perturbation effects at both levels of coordination only for the former group. However, if we do see effects of sensory feedback perturbations (such as demonstrated by an increase in variability or changes in coordination) at the intra-gestural level for PNS it would warrant a more explicit role for sensory information at this level. The present study is unique in that it is the first study to be designed to specifically investigate the role of auditory, proprioceptive, and tactile feedback at different (intra- and inter-gestural) levels of speech motor coordination across variations in speech rate in both PWS and PNS. 1.2. Study hypotheses In the current study it was hypothesized that: (1) For both groups of speakers, the combined effects of tactile, proprioceptive, and auditory feedback perturbations would be greater than each of them on their own. (2) For PNS, one could expect an increase in the variability of movement coordination in the combined presence of a bite-block and feedback perturbation only at (a) normal speech rates and (b) at the inter-gestural level of coordination. (3) If PWS depend more strongly on sensory feedback control during speech production, they would show an increase in the variability of movement coordination during feedback perturbation conditions, in particular when jaw motions are blocked and adaptations in the other articulators are required. These feedback perturbation effects would be evident at (a) both speech rates (normal and fast, but greater at fast speech rates given the additional demands on temporal processing of sensory information – Adams et al., 1993 and Max et al., 2004) and (b) both levels of coordination (intra- and inter-gestural).

In sum, these findings do not suggest that PWS are overly dependent on sensory feedback (auditory, proprioceptive, or tactile) for speech motor coordination, as we did not see an increase in variability of relative phase in PWS (relative to PNS) as result of sensory feedback perturbations at either normal or fast speech rates (with or without a bite-block). The parallel increases in variability of relative phase across both groups in the presence of masking noise does not seem to validate the notion of significant sensory deficits (Loucks and De Nil, 2006a and Loucks and De Nil, 2006b) nor the assumption that PWS are overly dependent on auditory feedback information (Tourville, Ghosh, Reilly, & Guenther, 2008). For example, if there were any such sensory deficits, presence of masking noise should have increased their difficulties in movement coordination relative to PNS, especially when they had to rely exclusively on oro-sensory feedback (Loucks & De Nil, 2006a). Based on the above findings, it is likely that PWS selectively use motor control strategies (e.g., use larger upper lip movements) to avoid instabilities in movement coordination, which may be an effective strategy in the light of their proposed speech motor skill limitations (van Lieshout et al., 2004). The results of the study have two important implications for theories of speech motor control. First, there is an indication of a stabilizing role of (auditory, proprioceptive, and tactile) sensory feedback at both intra- and inter-gestural levels of speech motor coordination, especially at fast speech rates when the coupling strength is assumed to be reduced (see Steyvers et al., 2001 for a similar approach in limb control). The presence of perturbation effects at both the intra- and inter-gestural levels of coordination are also in line with the state-dependent control scheme of the GLM (Saltzman and Byrd, 2000 and Saltzman and Munhall, 1989). Second, the finding that (auditory, proprioceptive, and tactile) sensory feedback perturbation effects (in terms of expected increase in variability) are stronger for intra- than inter-gestural coordination highlights the need to explicitly specify (rather than implicitly assume) the role of sensory feedback information at the intra-gestural level of coordination within the GLM (Browman and Goldstein, 1992, Goldstein and Fowler, 2003, Saltzman and Byrd, 2000 and Saltzman and Munhall, 1989). Furthermore, the current results are similar to those found in some postural reaction studies (e.g., Cordo & Nashner, 1982) and in general, compatible with models that allow for rapid adjustments and stabilization of ongoing limb movements and speech gestures by response-produced feedback (Desmurget and Grafton, 2000, Peper and Beek, 1998, Shaiman and Gracco, 2002, Steyvers et al., 2001 and van Lieshout et al., 2004). Finally, results and conclusions of the present study have to be interpreted cautiously. Lower statistical power and higher probability of type II errors may be an issue in studies with small sample size such as this. Small group size may also limit generality of the findings of the study to a larger population. For these reasons, the findings of the study should be considered preliminary and subject to further research.