به سوی یک مدل سیستم عصبی عملکردی لکنت زبان تکاملی

This paper overviews recent developments in an ongoing program of brain imaging research on developmental stuttering that is being conducted at the University of Texas Health Science Center, San Antonio. This program has primarily used PET imaging of different speaking tasks by right-handed adult male and female persistent stutterers, recovered stutterers and controls in order to isolate the neural regions that are functionally associated with stuttered speech. The principal findings have emerged from studies using condition contrasts and performance correlation techniques. The emerging findings from these studies are reviewed and referenced to a neural model of normal speech production recently proposed by Jürgens [Neurosci. Biobehav. Rev. 26 (2002) 235]. This paper will report (1) the reconfiguration of previous findings within the Jürgens Model; (2) preliminary findings of an investigation with late recovered stutterers; (3) an investigation of neural activations during a treatment procedure designed to produce a sustained improvement in fluency; and (4) an across-studies comparison that seeks to isolate neural regions within the Jürgens Model that are consistently associated with stuttering. Two regions appear to meet this criterion: right anterior insula (activated) and anterior middle and superior temporal gyri (deactivated) mainly in right hemisphere. The implications of these findings and the direction of future imaging investigations are discussed.

In recent years positron emission tomography (PET), functional magnetic resonance imaging (fMRI), and magnetoencephalography (MEG) studies have provided converging evidence regarding the neural regions that are implicated in speech production. For the most part this research has served to verify some of the classic models of the neural regions involved in speech production—models that were largely derived from lesion studies. The origin of many models of the regional interaction system supporting speech production can be traced to Wernicke’s (1874) observations on aphasia. These observations were ultimately extended by Geschwind (1979) and formed the basis of the Wernicke–Geschwind Model—arguably, the most influential model of speech production. This model identified a sequence of brain regions that play a critical role when, for example, individuals read aloud single words [primary visual area (V1)→angular area→M1-mouth]. Inevitably, increasing knowledge about the neural regions and structures associated with speech production, especially subcortical structures, meant that this basic model had to be expanded in a number of important ways. An especially important expansion occurred when it was established that cerebral cortex links with the basal ganglia via input structures that receive direct input from the cerebral cortex, and via output structures that project back to the cerebral cortex via thalamus (Alexander & DeLong, 1985a and Alexander & DeLong, 1985b). These multiple loops, which came to be known as cortico-basal ganglia–thalamo–cortical circuits (see Alexander, DeLong, & Strick, 1986), have been found to be involved in speech production. Indeed, certain speech–motor disorders, such as dysarthria, appear to reflect a dysfunction in that loop (Crosson, 1985 and Penney & Young, 1983). Not surprisingly, therefore, there is interest in determining if other speech disorders, such as developmental stuttering, are byproducts of a fundamentally dysfunctional neural system. Major improvements to the understanding of the regions and systems that participate in speech production occurred in the mid-1980s with the arrival of PET imaging of the brain (Ter-Pogossian, Phelps, Hoffman, & Mullani, 1975; Ter-Pogossian, Raichle, & Sobel, 1980). The subsequent groundbreaking PET experiments by Petersen, Fox, Posner, and Raichle (1988) and Petersen, Fox, Posner, Mintun, and Raichle (1989) yielded the first distinctive images of neural activity during reading and during the production of single words. Other researchers using PET in conjunction with various speech tasks soon replicated and refined Petersen et al.’s findings, identifying a group of regions that were generally active during speech production (mainly single word production tasks). Across nine of these early PET studies Fiez and Petersen (1998, p. 914) reported that … the results converge to reveal a set of areas active during word reading, including left-lateralized regions in occipital and occipitotemporal cortex, the left frontal operculum, bilateral regions within the cerebellum, primary motor cortex, and the superior and middle temporal cortex, and medial regions in the supplementary motor area and anterior cingulate. The variability in regions reported across the studies reported by Fiez and Petersen, however, also prompted arguments about their validity (see Démonet, Fiez, Paulesu, Petersen, & Zatorre, 1996; Poeppel, 1996) and ignited attempts to identify the sources of this variability (Hickok, 2001). That variability was also evident in a much larger meta-analysis of imaging studies of speech production by Indefrey and Levelt (2000). Their review of the findings of 58 word production studies showed the expected variation in regions activated because of experimental task differences (Grabowski & Damasio, 2000). Nevertheless, Indefrey and Levelt provided a valuable summary of the regions that are principally associated with oral reading. Their review did not consider studies of continuous speech or continuous oral reading, but it did highlight a group of relatively broad regions that might be implicated in connected speech. A reasonable conclusion from Indefrey and Levelt’s tabulated findings (see Indefrey & Levelt, 2000, pp. 855–858) is that most regions they identified were essentially identical to those reported by Fiez and Petersen (1998). Refinements to these regions continue to occur. For instance, a recent synthesis of lesion and imaging studies by Price (2000) concluded that left anterior insula might have a much greater role in speech planning than Broca’s area. A recent meta-analysis of single word reading PET studies compared with fMRI findings by Turkeltaub, Eden, Jones, and Zeffiro (2002) highlighted the functional roles of specific areas of thalamus (left ventrolateral thalamus) and cerebellum (lobules VI and VII) during speech. This compilation of findings is gradually identifying regions that must be implicated in normal speech production—a necessary prerequisite to the understanding of the regions that must be implicated in abnormal speech production such as occurs during chronic developmental stuttering. 1. The Jürgens Model of speech production A recent model of the neural basis of speech production proposed by Jürgens (2002) has attempted to synthesize current knowledge about the neurologic foundations of speech production. This elaborate “box and arrow” model of neural regions and structures participating in speech production is derived from a careful integration of findings from lesion, invasive brain stimulation, single-unit recording and brain imaging studies. The model builds on currently known structural connections between and within neural locations. The Jürgens Model provides only partial information on the sequence with which particular regions participate in different speech tasks, but it does make it possible to locate patterns of regional innervation that characterize different speech tasks. And, of more relevance to stuttering research, it helps to focus the search for regions that are relatively inactive or overactive during the speech of individuals with developmental stuttering. Fig. 1 reproduces the Jürgens Model with some additions (e.g., specific BA regions) that were derived by analyzing the results of studies that Jürgens used to develop this model. Those additional labels have been confirmed as accurate by the author (Jürgens, 2003, personal communication).