بررسی عمل حرکتی گفتار و یادگیری در افرادی که لکنت زبان دارند

In this exploratory study, we investigated whether or not people who stutter (PWS) show motor practice and learning changes similar to those of people who do not stutter (PNS). To this end, five PWS and five PNS repeated a set of non-words at two different rates (normal and fast) across three test sessions (T1, T2 on the same day and T3 on a separate day, at least 1 week apart). The results indicated that PWS and PNS may resemble each other on a number of performance variables (such as movement amplitude and duration), but they differ in terms of practice and learning on variables that relate to movement stability and strength of coordination patterns. These findings are interpreted in support of recent claims about speech motor skill limitations in PWS. Educational objectives: The reader will be able to: (1) define oral articulatory changes associated with motor practice and learning and their measurement; (2) summarize findings from previous studies examining motor practice and learning in PWS; and (3) discuss hypotheses that could account for the present findings that suggest PWS and PNS differ in their speech motor learning abilities.

In the past decade, various authors have proposed the presence of motor learning/skill limitations in people who stutter (PWS; e.g., Ludlow, Siren, & Zikria, 1997; Smits-Bandstra, De Nil, & Saint-Cyr, 2006; Van Lieshout et al., 1996a, Van Lieshout et al., 1996b and Van Lieshout et al., 2004). Primarily, the evidence comes from speech and non-speech motor learning studies. These studies have implied that motor learning and automatization of new speech skills may be limited in PWS and that these limitations may have implications for treatment strategies that focus heavily on learning a new set of speech motor skills (Smits-Bandstra, De Nil, & Rochon, 2006). The results of these specific studies relating to motor learning in PWS and some of their limitations will be reviewed first, followed by a description of the present investigation. 1.1. Motor practice and learning Typically, practicing a new motor task over repeated trials results in decreases in response latency, shorter execution time (as inferred by increases in speed), higher accuracy and more consistency/lower variability of newly practiced movements (Hamstra-Bletze & Blote, 1990; Meulenbroek & Van Galen, 1988; Moore & Marteniuk, 1986; Salthouse, 1986; West & Sabban, 1982; Zesiger, Mounoud, & Hauert, 1993). The nature of these immediate changes subsequent to practice are typically reflected in so-called performance curves and represent motor practice effects (Schmidt & Wrisberg, 2004). Motor learning on the other hand, is not directly observed but rather inferred by how well these practiced tasks are transferred to other similar types of tasks that were not practiced, and by the extent to which they are retained after a period of rest. Further, motor learning is inferred when these practice-related changes become a relatively permanent feature of the demonstrated behaviors (Schmidt & Wrisberg, 2004). 1.2. Assessment of motor learning The issue of practice versus learning becomes more complicated in the context of different theoretical views. Traditional motor control models (e.g., the motor schema theory of Schmidt, 2003; Shea & Wulf, 2005) would predict that decreases in movement amplitude (Meulenbroek & Van Galen, 1988; Salthouse, 1986) and variability (Hamstra-Bletze & Blote, 1990; Zesiger et al., 1993) and increases in movement speed (Moore & Marteniuk, 1986; West & Sabban, 1982) would be associated with practice (cf. Schulz, Dingwall, & Ludlow, 1999). However, many of the more traditional kinematic measures show substantial individual variability over time (Alfonso & Van Lieshout, 1997; Van Lieshout et al., 2004). For example, Alfonso and Van Lieshout (1997) found high intra-subject variability in PNS for discrete kinematic parameters and movement sequence patterns (peak velocity timing of upper lip, lower lip and jaw sequence) including some complete pattern reversals across three test sessions 2 weeks apart. These findings suggest that these more traditional discrete spatial and temporal order measurements in speech, including movement amplitude and duration, can be seen as emergent features that do not necessarily reflect organizational principles of the underlying control mechanism (Kelso, 1995; Smith, Johnson, McGillem, & Goffman, 2000; Van Lieshout, 2004). Different approaches have surfaced in recent years to reflect principal mechanisms of motor control. These include linear and non-linear indices of movement pattern stability (e.g., Lucero, Munhall, Gracco, & Ramsay, 1997; Smith, 1997; Wohlert & Smith, 1998), relative phase measures to index the nature and stability of movement coordination (Kelso & Tuller, 1987; Van Lieshout, Hulstijn, Alfonso, & Peters, 1997; Ward, 1997), and cross-spectral coherence, as a measure of the strength of the frequency coupling between articulators (Kay, 1988; Morrison & Newell, 1996; Van Lieshout, 2001 and Van Lieshout, 2004). More details on these measures are provided in Section 2. One approach to measure individual articulator variability is the spatio-temporal index (STI) which captures movement stability over time (Smith, 1997). The assumption is that an adult speaker's speech is highly practiced and stable. The index is a measure of how well amplitude and time normalized movement trajectories of repeated utterances converge upon a single core template. The lower the STI the lesser the deviation from a single template and vice versa (cf. Smith, 1997; Wohlert & Smith, 1998). Interestingly, the pattern of interest or template could be either defined in linguistic terms in the form of short meaningful phrases (Smith, 1997; Wohlert & Smith, 1998) or in more basic motor terms, in the form of individual movement cycles, in which case it is referred to as cyclic STI (cSTI; Van Lieshout & Moussa, 2000). It is claimed that the cSTI measure is less susceptible to long-term non-linear influences relating to local temporal variations in linguistic structures (cf. Lucero et al., 1997) and more indicative of basic motor control stability (Van Lieshout & Moussa, 2000), and hence could be an appropriate tool to study changes subsequent to motor practice. cSTI does not address motor coordination between individual articulators during speech production. Motor coordination and learning in speech can be addressed by coordination dynamics theory, which describes lawful relationships for the coupling of limb and speech structures over time (Kelso, 1995). Combining this theory with concepts from the Articulatory Phonology (AP) model, motor coordination can be defined at two distinct levels, namely within (intra) and between (inter) gestures (Browman & Goldstein, 1992; Goldstein, Byrd, & Saltzman, 2006; Saltzman & Byrd, 2000; Saltzman & Munhall, 1989; Van Lieshout, Bose, Square, & Steele, 2007). The nature and variability of coordination (or coupling) both within and between gestures can be assessed using a measure of relative timing that is not confounded with variations in movement duration, called relative phase (for a recent review of this approach and related theories, see Van Lieshout, 2004). Following this combined approach, it can be argued that the phasing (or coupling) between the individual articulators involved for a particular gesture (e.g., between upper and lower lip during bilabial closure) is not directly dependent on sensory feedback and primarily reflects lower level physiological and biomechanical influences (e.g., differences in mass) on the intrinsic coupling between the individual articulators which act as a functional synergy (Saltzman & Munhall, 1989). On the other hand, inter-gestural levels of coordination refer to the phasing between two independent gestures (e.g., between bilabial closing and tongue body during the production of syllable /ba/). It has been suggested that sensory input is deemed relevant at this level (Saltzman, Löfqvist, Kay, Kinsella-Shaw, & Rubin, 1998). According to the coordination dynamics approach (Kelso, 1995), practice stabilizes the to-be-learned coordination pattern (Vereijken, Van Emmerik, Bongaardt, Beek, & Newell, 1997; Zanone & Kelso, 1997). This stabilization is evidenced as a decrease in the variability (S.D.) of relative phase (during practice). Vereijken et al. (1997) suggest that stability may be established very early in practice. Such early stabilization according to Vereijken et al. (1997) may be followed by ongoing modulation of movements that enables improvement of performance and promotes transfer to new tasks and further improvement of performance (i.e., brings back flexibility in the system). Whereas relative phase measures provide insight into the nature and stability of coordination, other measures can be used to index the relative strength of the coupling between articulators or gestures. This information can be acquired through cross-spectral coherence analysis (Kay, 1988; Morrison and Newell, 1996 and Morrison and Newell, 2000; Timmer, Lauk, Pfleger, & Deuschl, 1998). The basis for cross-spectral coherence as applied to speech motor control is that each articulator's movement trajectory can be treated as a complex wave, which when decomposed (e.g., using Fast Fourier Transformation) reveals that it is a composite output of a number of underlying singular waves or frequencies. The underlying frequency components of the complex movement trajectory are referred to as motion primitives (Van Lieshout, 2004). The motion primitives themselves are assumed to originate from the coupling between an effector (e.g., lips, jaw, tongue, etc.) and an underlying neural oscillator system (Beek, Peper, & Daffertshofer, 2002; Van Lieshout, 2004 and Williamson, 1998). The cross-spectral coherence is then a measure of the association between pairs of such individual frequency components across articulators (intra-gestural) or across independent gestures (inter-gestural). High cross-spectral coherence ratios indicate a strong frequency coupling or entrainment between two paired components (between articulators or gestures) and vice versa (Van Lieshout, 2004 and Van Lieshout et al., 2007). Measurement of inter-gestural cross-spectral coherence is particularly relevant from the AP model perspective, as recent evidence has shown that different gestures could be coordinated in time by coupling their corresponding nonlinear neural oscillators (Goldstein, Pouplier, Chen, Saltzman, & Byrd, 2007; Saltzman et al., 1998). Further, stability in relative timing between the two gestures is achieved when these coupled oscillators settle towards a stable relationship (Nam, 2007). In this context, cross-spectral coherence can be interpreted as the strength of bonding or coupling between the outputs of the neural oscillators driving the independent gestures (more details under Section 2.6.2). 1.3. Speech and non-speech motor learning in PWS Several studies have demonstrated that PWS take a longer time to demonstrate improvements in movement speed subsequent to practice. Studies on speech motor learning in PWS have shown slower rates of pseudoword learning (Ludlow et al., 1997) and slower acquisition of sequential finger tapping and syllable reading tasks in PWS compared to PNS (Smits-Bandstra, De Nil, & Saint-Cyr, 2006). The first study compared PWS and PNS on their ability to learn two pseudowords. It was found that PWS were less accurate and needed more trials to learn these pseudowords than PNS (Ludlow et al., 1997). The authors attributed the difficulty of PWS in learning new sound sequences to a limitation in learning new phonological patterns (i.e., a problem at the phonological level of processing), but obviously, nonsense sequences of the type used in that study (also) impose challenges in terms of the timing and execution of speech gestures compared to over-learned existing word patterns, so it remains unclear what aspects of learning are reflected in this type of task. One alternative possibility is that, since the nonwords or phoneme sequences do not already exist in the lexicon they need to be created by assembling the sequence of previously known motor gestures or functional synergies that make up the nonword (cf. Smits-Bandstra, De Nil, & Saint-Cyr, 2006). Specifically, in such a task, the coupling or phasing between the individual gestures (or inter-gestural coupling) in the nonword both between and across syllable boundaries has to be learnt (Nam, 2007). Thus, in a way what is being “learnt” during multiple repetitions of a nonword may parallel sequence skill learning tasks in speech and limb control literature (Schmidt & Wrisberg, 2004; Smits-Bandstra, De Nil, & Saint-Cyr, 2006). In the study by Smits-Bandstra, De Nil, and Saint-Cyr (2006), learning of novel finger tapping and nonsense syllable sequences was investigated in PWS and PNS. Their study design included an initial practice period (about 30 repetitions) which was then followed by transfer (to unpracticed novel sequences) and retention (following a 40-min rest period) tests of the newly learnt skills. The learning outcomes were measured using performance accuracy, sequence duration and reaction time (RT). The results of their study indicated that although PWS were able to maintain accuracy levels of finger tapping and syllable sequence productions similar to those of PNS, the groups differed on a number of task conditions and variables. RT data for the finger tapping following practice task showed that PWS were not as fast as PNS and portrayed a lesser degree of transfer and retention abilities. For nonsense syllable sequence tasks, it was found that the RT data for PNS, unlike PWS, showed a fast decrease in reaction time within the first five practice trials, suggesting group differences in the rate of skill acquisition. Furthermore, the authors observed a trend for a stronger decrease in sequence duration over practice in PNS compared to PWS. 1.4. Limitations of the current motor learning studies with PWS First, Smits-Bandstra, De Nil, and Saint-Cyr (2006) and others (Peters, Hulstijn, & Van Lieshout, 2000) have pointed out that performance measures of accuracy and sequencing, as well as RT and movement duration data may not be sensitive enough to capture group differences (if any) in motor control strategies related to practice and learning that may be more evident in specific changes of speech movement kinematics and coordination. For example, reaction time data encompasses the collective influence of a number of higher level cognitive-linguistic and lower level motor processes and hence can only provide relatively crude inferences regarding the processes that underlie the delay in RT (for further details see Peters et al., 2000, Van Lieshout et al., 1996a and Van Lieshout et al., 1996b). Second, motor practice and learning literature on stuttering has not specifically included speech rate as an independent variable. Apart from the fact that speech rate is an important factor that allows one to addresses the question of whether or not motor learning can be evidenced independent of articulation rate in terms of its effects on variability of individual movements and coordination, speech rate also provides a means to study potential underlying motor control differences between PWS and PNS (Van Lieshout et al., 2004). For example, as argued by other researchers (Smits-Bandstra, De Nil, & Saint-Cyr, 2006), in healthy motor control systems the strength of the coupling between individual synergies (e.g., as in inter-gestural coupling) within a speech sequence that has to be learnt improves with practice. However, PWS may differ from PNS in the acquisition, processing or use of sensory information (Archibald & De Nil, 1999; De Nil & Abbs, 1991; Loucks & De Nil, 2006; Max, 2004; Van Lieshout, Peters, Starkweather, & Hulstijn, 1993) and given that sensory information may be relevant for inter-gestural coupling (Saltzman et al., 1998), PWS may experience difficulties at this level. It is assumed that such difficulties would increase for PWS relative to PNS under fast rates of speech in comparison to normal speech rates as the ability to use sensory information for movement control would be more limited in the former than the latter (Adams, Weismer, & Kent, 1993; Baum, 1999). Third, one important aspect of motor learning, namely the long-term retention of the learned tasks, has not been addressed in previous motor learning studies on stuttering (Ludlow et al., 1997). This aspect of motor learning is important given that the neurophysiological processes underlying learning continue to evolve after practice or training has ended (Karni et al., 1998). That is, consolidation of experience dependent changes occurs not during, but rather a minimum of 4–8 h after practice, and continues to develop over the course of several days (Karni et al., 1998). The study by Smits-Bandstra, De Nil, and Saint-Cyr (2006) tested for retention of the practiced motor task within a single session (after a 40 min rest period). Testing retention and transfer of practiced (or learned) motor tasks after a short rest period may be too limited. Further, it is well known from the non-speech motor control literature that learning different tasks, if not separated by several hours, can interfere with each other (Sage, 1984; Wolpert, Ghahramani, & Flanagan, 2001). This implies that motor learning requires a period of consolidation during which memory is susceptible to disruptions (cf. Wolpert et al., 2001). Interestingly, observation of the graphed finger tapping RT data from Smits-Bandstra, De Nil, and Saint-Cyr's (2006) study did show some evidence (although non-significant) for such a negative interference effect. Thus, in order to demonstrate that changes subsequent to motor practice are not simply performance effects but are relatively permanent learning changes, and to control for negative interference effects from performing the same or a similar task again after a relatively short period of time, the experimental design would require a test for practice effects on the same day and a test for retention a week later (Wolpert et al., 2001). 1.5. Present investigation The purpose of the present study was to specifically investigate whether or not any differences in motor practice and/or learning of novel speech utterances exist between PWS and PNS using an assessment of short and long term changes in different motor variables. Based on the review of the literature (as presented above) the focus of this investigation will be on variables that have been shown to be influenced by practice and/or learning. For individual movements this includes movement amplitude, duration, as well as movement stability. For coordination, the variables of interest are mean coherence (as an index for frequency coupling) and the variability of relative phase (as an index of the stability of the coupling) for inter-gestural coordination, as changes due to learning are more likely to occur at this level (see Section 1.3). It is important to note that in the present investigation, given the nature of the task, no specific task goal or feedback regarding motor performance related to the dependent variables could be given to the subject. Likewise, we could not provide any specific instruction on what aspect of performance the subject should try to optimize. In this sense, learning was considered to be implicit with respect to gestural phasing and movement execution as detailed above. To this end, the following hypotheses were formulated. 1.5.1. Study predictions with respect to motor practice First, PWS in comparison to PNS were expected to demonstrate a lesser amount of improvement in terms of faster, more stable movement kinematics and increases in the strength and stability of inter-gestural coordination with practice. Within a given day, this divergence in performance (greater improvements in PNS compared to PWS) following practice would be more evident for fast rates of speech. That is, for PWS it was expected that faster rates of speech would limit improvements in the strength of coordination and may even lead to greater instability when compared to PNS. 1.5.2. Study predictions with respect to motor learning PWS were expected to demonstrate a lesser amount of retention of the practiced tasks (in terms of faster, more stable movement kinematics and increases in the strength and stability of coordination patterns) in comparison to PNS, especially at faster rates of speech (across a period of 1 week).

Overall, the present findings imply that, although PWS and PNS may resemble each other on a number of performance variables (such as movement amplitude and duration), they do differ in terms of practice and learning on variables that reflect organizational aspects of speech motor control in terms of movement stability (S.D. of inter-gestural relative phase) and strength of coordination patterns (inter-gestural mean coherence). In our first hypothesis related to practice, we predicted that PWS (relative to PNS) would demonstrate a lesser amount of improvement in terms of faster, more stable movement kinematics and increases in the strength and stability of inter-gestural coordination patterns, and that these group differences would be more evident for fast rates of speech. The results from the present study revealed that practice effects (within a given day), in terms of decreases in variability (inter-gestural S.D. of relative phase) of coordination patterns and increases in the strength of frequency coupling (mean coherence) between gestures, are present to a greater degree in PNS (relative to PWS) in the fast and normal speech rates respectively. In our second hypothesis, we predicted that PWS would demonstrate a lesser amount of retention (motor learning) of the practiced tasks across days in comparison to PNS, especially at faster rates of speech. The results of the present study revealed significant increases in the strength of inter-gestural frequency coupling (mean coherence) for PNS in comparison to PWS at normal speech rates. This may imply that motor learning of sequences may be limited in PWS even at habitual speech rates. In general, these findings support the motor skill approach proposed by Van Lieshout and coworkers (Hulstijn & Van Lieshout, 1998; Van Lieshout, 1995 and Van Lieshout et al., 2004). According to this approach, speech is a motor skill acquired by practice and is characterized by highly self-organized, adaptive, energy-effective and purposeful coordination patterns. Within this framework, PWS are hypothesized to be at the lower end of the (speech) motor skill continuum. Although there are different views as to how practice improves speech motor skills (Guenther, 1995; Smith & Goffman, 1998), it is currently accepted that speech movements are not innate and that they require a certain amount (years) of practice to become accurate and stable (Green, Moore, Higashikawa, & Steeve, 2000; Green, Moore, & Reilly, 2002; Smith & Goffman, 1998). Thus, if practice leading to motor learning is a critical factor in the process of acquiring speech motor skills, then the results of the present study indicate that PWS may indeed have limited speech motor skills as evidenced by the differences in motor practice and learning changes in the variables linked to stability and strength of movement coordination. Additionally, some of the data (e.g., larger UL amplitude in PWS) from the present study may suggest that PWS use a speech motor control strategy that promotes stability of production in the presence of these motor skill limitations (see also Van Lieshout et al., 1996a, Van Lieshout et al., 1996b and Van Lieshout et al., 2004).