رمز گذاری واجی در سخنرانی صامت افرادی که لکنت زبان دارند

The purpose of the present study was to investigate the role of phonological encoding in the silent speech of persons who stutter (PWS) and persons who do not stutter (PNS). Participants were 10 PWS (M = 30.4 years, S.D. = 7.8), matched in age, gender, and handedness with 11 PNS (M = 30.1 years, S.D. = 7.8). Each participant performed five tasks: a familiarization task, an overt picture naming task, a task of self-monitoring target phonemes during concurrent silent picture naming, a task of monitoring target pure tones in aurally presented tonal sequences, and a simple motor task requiring finger button clicks in response to an auditory tone. Results indicated that PWS were significantly slower in phoneme monitoring compared to PNS. No significant between-group differences were present for response speed during the auditory monitoring, picture naming or simple motor tasks, nor did the two groups differ for percent errors in any of the experimental tasks. The findings were interpreted to suggest a specific deficiency at the level of phonological monitoring, rather than a general monitoring, reaction time or auditory monitoring deficit in PWS.

In speech production, phonological encoding can be defined as “the processes involved in retrieving or building a phonetic or articulatory plan from each lemma or word and the utterance as a whole” (Levelt, 1989, p. 12). It has been proposed that phonological encoding involves three components: (a) generation of segments that constitute words, (b) integration of sound segments with word frames, and (c) assignment of appropriate syllable stress (Levelt, 1989). This process is thought to be an interface between lexical processes, on the one hand, and speech motor production on the other (Levelt, 1989; Levelt, Roelofs, & Meyer, 1999). At least four theories have been proposed concerning the potential link between phonological encoding and stuttering. Howell (2004) in his EXPLAN theory proposed that fluency failures occur due to temporal asynchronies between execution (EX) and speech planning (PLAN). He speculated that such asynchronies are caused by difficulties associated with the planning of complex linguistic segments and fast speech rate and the resulting coping strategies adopted by the speaker. Postma and Kolk (1993) proposed the Covert Repair Hypothesis in which the primary symptoms of stuttering represent overt manifestations of covert corrections of speech plan errors that are caused by delayed phonological encoding of speech sounds. Perkins, Kent, and Curlee (1991) in their Neuropsycholinguistic theory outlined two factors as crucial elements in the causation of stuttering: (a) temporal asynchrony between linguistic, i.e., lexical and phonological, and supralinguistic planning, and (b) time pressure. Finally, Wingate (1988) in his Fault Line Hypothesis proposed that stuttering resulted from a delay in the retrieval and encoding of syllable rhyme during speech production resulting in a fault-line created at the point of integration of the syllable onset with its rhyme. These theories have motivated considerable interest in the role of phonological encoding in speech production in persons who stutter (PWS). In the next section, we will briefly review the main findings from studies investigating phonological encoding skills in PWS and persons who do not stutter (PNS). 1. Phonological encoding in PWS 1.1. Priming studies In phonological priming studies, the presentation of a target word is preceded by a prime stimulus, which shares one or more phonological segments with the target word. The underlying assumption is that the prime stimulus will increase the activation level of the shared sound segment(s), thereby increasing the probability of their selection for the target word within a network of possible phonemes. Wijnen and Boers (1994) used an implicit priming paradigm in which PWS and PNS learned sets of five word-pairs in Dutch. Within a set, all of the second words within the pairs either started with the same phoneme(s), i.e., consonant-only, e.g., lepel, lila, loeder, larie, luier, or consonant-vowel, e.g., leuven, leugen, leuze, leuning, leukerd (homogenous condition) or not, e.g., lila, pekel, kater, sable, tafel (heterogeneous condition). Upon learning the word sets, participants were presented with the first word and asked to recall and name the second word as fast as possible. The results revealed a comparable naming facilitation effect in speech initiation time for the two groups in the consonant-vowel items in the homogeneous primed condition. The consonant-only primed items, however, showed a reduced facilitation in PWS. This finding was interpreted as suggesting that PWS exhibited delayed encoding specific to the stressed vowel, and that this delay was reduced or eliminated by using a consonant-vowel prime. In a follow-up study, however, Burger and Wijnen (1999) failed to replicate their earlier findings. Melnick, Conture, and Ohde (2003) conducted one of the few investigations of segmental, phonological priming effects with children who stutter (CWS). They measured speech reaction times in 3–6-year-old CWS and age-matched children who did not stutter (CNS) during three picture naming conditions: no prime, onset-related segmental prime (e.g., /də/ followed by dog), and onset-unrelated segmental prime (e.g., /sə/ followed by dog). In addition, they correlated articulation mastery, as measured by the Goldman Fristoe Test of Articulation (GFTA; Goldman & Fristoe, 1986), with the speed of picture naming in their young subjects. While the extent of phonological priming in the related prime condition was comparable for both groups, the CWS exhibited higher variability in the naming reaction time. Furthermore, a significant negative correlation was observed between the reaction times and the GFTA scores in CNS, but no such correlation was observed in CWS. They attributed the lack of significant between-group differences in phonological priming to larger variability in performance within the CWS group. However, it could be argued that the finding of higher variability in picture naming and the lack of correlation between articulatory proficiency and speed of picture naming in CWS reported by Melnick et al. (2003) may partly be attributable to phonological knowledge and processing difficulties, and warrant further investigation. 1.2. Rhyme monitoring studies Rhyme monitoring involves multiple processes including (a) registration of information from the speech code in the phonological input store (through the auditory/visual models), (b) engagement of processes involved with phonemic segmentation and deletion while information is retained within the articulatory loop, (c) direct comparison of word endings, (d) execution of a decision and (e) response (Baddeley, Lewis, & Valler, 1984). Bosshardt and Fransen (1996) did not find rhyme monitoring differences between 14 adult PWS and 14 age-matched PNS during a task of silent prose reading, but the groups did differ on a semantic monitoring task, which required participants to monitor for semantic categories (e.g., fruits) while silently reading prose (e.g., At the market there is much to find. A woman gives a pear to a little girl). In a more recent study, Bosshardt, Balmer, and De Nil (2002) used a dual task paradigm with 14 PWS and 16 PNS to investigate further the phonological and semantic processing abilities of PWS. They found that PWS again performed more poorly than PNS on a semantic monitoring task while generating sentences. In contrast to Bosshardt and Fransen (1996), however, PWS also were significantly less accurate in the rhyme monitoring task compared to PNS when this task was performed concurrently with the sentence generation task. Differences were also found in the sentences generated: PWS exhibited a significantly reduced number of propositions in the generated sentences compared to the PNS. Bosshardt et al. (2002) explained their results in terms of cognitive processing limitations in PWS and suggested that the organization of the speech production system in PWS may be vulnerable to interference from concurrent attention demanding tasks (such as sentence generation). Similarly, evidence for cognitive processing limitations in conjunction with phonological tasks in PWS has also been reported by Weber-Fox, Spencer, Spruil, and Smith (2005). They used evoked response potentials (ERP) to study rhyme judgments across a variety of prime target pairs in 11 adult PWS and 11 adult PNS. ERP amplitude and latency, response time, and error data suggested that the performance of PWS and PNS was comparable across all conditions studied. The only difference found was that PWS were slower in identifying dissimilar rhymes from orthographically and visually similar targets that required additional cognitive processing (e.g., cow-own). Although the authors interpreted their findings as evidence against theories that specifically propose phonological encoding as a causal factor in stuttering, their findings did not completely rule out the role of phonological encoding in PWS. This is because PWS were found to be slower in at least some of the rhyming conditions when the task became cognitively more complex. 1.3. Other paradigms Postma, Kolk, and Povel (1990) investigated speaking rates of 19 PWS and 19 PNS using tongue twister sentences and matched control sentences spoken either silently, lipped, or overtly. Tongue twisters are words strings manipulated for phonemic similarity (e.g., the chef's sooty shoe soles). By manipulating error-inducing features, such as the extent of phonemic similarity and rate of speech, tongue twisters have been used to study phonological encoding (see Wilshire, 1999). The results reported by Postma et al. (1990) showed that PWS were slower than PNS in each of the three speaking conditions, with the largest differences observed in the overt condition and the smallest differences in the silent condition. Similarly, group differences for tongue twisters were greater than for the control sentences. These findings suggested that PWS were slower than PNS, not just in overt speech, but also in silent speech tasks involving linguistic processes, which Postma et al. regarded as involving minimal to negligible motor planning and execution. These results of course cannot be interpreted simply as indicating a phonological deficit because the speech conditions used in their study may have involved a variety of different cognitive processes, including semantic, syntactic, and phonemic encoding ( Levelt, 1989), as well as speech motor planning ( Yetkin et al., 1995). Au-Yeung and Howell (2002) reinterpreted an earlier study by Packman, Onslow, Coombes, and Goodwin (2001) in which the latter reported that stuttering speakers showed a significantly higher percentage of stuttering when reading a nonsense passage compared to a meaningful passage. Packman et al. interpreted their findings as indicative of a lexical retrieval deficit. However, Au-Yeung et al. argued that the nonsense passage used in the original study engages the phonological encoding processes without lexical access. Instead of lexical retrieval, they suggested that the higher percentage of disfluencies in the nonsense passage might reflect phonological encoding difficulties in PWS. In conclusion, although the research reported to date is not conclusive, there is at least partial evidence to suggest that PWS may exhibit a phonological encoding deficit or delay which, in turn, may be linked to speech fluency breakdown. As discussed before, the findings reported by Melnick et al. (2003) may partly be attributable to phonological encoding difficulties. Similarly, the findings from Weber-Fox et al. (2005) cannot completely rule out a phonological processing deficit in PWS. Furthermore, the fact that Melnick et al. (2003) did not find differences between CWS and CNS on a segmental priming task and that Weber-Fox et al. (2005) failed to find differences in adult PWS and PNS in a rhyme monitoring task, may have reflected age-related processing differences, rather than the absence of differences in phonological encoding between these groups, as suggested by studies on fluent children and adults by Brooks and MacWhinney (2000), Bonte and Blomert (2004), and Jusczyk (1993). An additional problem with many previous studies is that they have involved spoken responses. This raises the possibility that any observed group differences especially in older children, adolescents, and adults may, at least in part, represent reactive behaviors to the presence or the anticipation of stuttering.1 The lack of a conclusive interpretation of the data motivates the need for further investigation of phonological encoding skills in PWS, but these studies are complicated by several factors. First, phonological encoding is a covert operation unavailable for direct experimental manipulation and interpretation (Coles, Smid, Scheffers, & Otten, 1995); hence experimental paradigms available to study phonological encoding in production are limited (for a discussion, see Meyer, 1992). Second, findings based on production tasks are open to other interpretations due to methodological limitations in isolating phonological encoding from additional on-going linguistic and motor processes involved in speech planning and production. In PWS, overt speech obviously can lead to disfluencies, potentially masking the very phonological encoding effects researchers wish to isolate. For instance, while it is possible that the higher percentage of stuttering reported in phonologically complex items (e.g., Hakim & Bernstein-Ratner, 2004; Prins, Hubbard, & Kraus, 1991) is due to a phonological encoding deficit, it is also possible that it reflects speech motor timing and articulatory coordination difficulties in PWS (e.g., Howell, Au-Yeung, & Sackin, 2000). Given the need for a better understanding of the role of phonological encoding in stuttering, and with the need for careful experimental control in mind, the present study was intended to investigate phonological encoding skills during silent speech in PWS and PNS using a phoneme-monitoring paradigm. This paradigm was adapted from an earlier study by Wheeldon and Levelt (1995) in which they used a translation task to investigate phoneme monitoring. In the current study, a silent picture naming task was used instead of a language translation task. Some researchers have claimed that both translation and silent picture naming involve similar subprocesses, namely lexical retrieval, followed by phonological encoding and monitoring of the phonological code, which then lead to a motor response (Levelt et al., 1999; Wheeldon & Levelt, 1995). While the translation and picture naming tasks are not identical (for instance, translation requires competence in two languages and possibly a wider lexical search strategy), both can be expected to reveal deficits in phonological encoding, as both tasks require participants to phonologically encode segments to make a phoneme-monitoring decision. To control for the possibility that any group differences could be due to processes other than phonological encoding, the present study also included an overt picture naming task, an auditory tone-monitoring task, and a simple motor task. These tasks were aimed at comparing the two participant groups in lexical access, general monitoring skills, and speed of motor response. Specifically, the following research questions were addressed in the present study: (a) Do PWS and PNS differ in the time taken to detect the presence or absence of target phonemes in the phoneme-monitoring task? (b) Do PWS and PNS differ in the percent of error judgments in phoneme monitoring? and (c) Do PWS and PNS differ in their response time and percent of error judgments in the overt picture naming, auditory, or simple motor tasks?