نظارت واج در نامگذاری ساکت و ادراک در بزرگسالانی که لکنت زبان دارند

The present study investigated phonological encoding skills in persons who stutter (PWS). Participants were 10 PWS (M = 31.8 years, S.D. = 5.9) matched for age, gender, and handedness with 12 persons who do not stutter (PNS) (M = 24.3 years, S.D. = 4.3). The groups were compared in a phoneme monitoring task performed during silent picture naming. The phonological complexity of the target items in the task was varied such that participants monitored either compound words or noun phrases. Performance in this task was compared to phoneme monitoring performed on aurally presented target words to investigate whether any differences observed in silent naming were also evident in perception. Analysis of the response time data, in milliseconds, indicated that PWS were significantly slower as compared to PNS in phoneme monitoring during silent naming; group differences were not obtained in the perception task. The groups were also comparable in the response time to phoneme monitoring within compound words and noun phrases in both silent naming and perception. The findings suggested that PWS were slower in the encoding of segmental, phonological units during silent naming. Furthermore, absence of such differences in perception ruled out a general monitoring deficit in PWS. Findings are interpreted within the context of the psycholinguistic theories of stuttering that postulate phonological encoding and/or monitoring as a causal variable in stuttering. Educational objectives: As a result of this activity, the participant should: (1) describe relevant literature on phonological encoding skills in children and adults who stutter, (2) identify paradigms that can be used to investigate phonological processing in PWS, and (3) discuss the role of phonological encoding in speech production.

Phonological processes in speech production are a set of retrieval and encoding processes involved in word form or phonological encoding (Schiller, 2000). Levelt and Wheeldon (1994) described three such processes, namely segmental spell-out, metric spell-out, and syllabification. Segmental spell-out involves the generation of individual sound segments that constitute words; metric spell-out involves generation of word stress; and, syllabification involves assignment of syllable boundaries during speech production. Wheeldon and Levelt (1995) proposed that the process of phonological encoding is available for editing through the prearticulatory monitoring process, which enables experimental investigation of the time course of this process. Recent data ( Schiller, Bles, & Jansma, 2003) have suggested that during speech planning, speakers encode segmental and metric information in a parallel fashion, while within each of these levels, the segmental and metric units are encoded from left to right in a serial incremental manner thereby resulting in a just-in-time incremental speech production process. Consequently, the time course of phonological encoding is crucial to the timely generation of the phonetic code for speech motor planning and execution. Starting with Brown's research in 1945, several factors at the phonological level have been linked to the presence and maintenance of stuttering (Louko, Edwards, & Conture, 1990; Paden, Ambrose, & Yairi, 2002; Paden & Yairi, 1996; Wolk, Conture, & Edwards, 1990). A review of the literature reveals several paradigms that have been used to investigate phonological encoding skills in persons who stutter (PWS) including phonological priming, nonword tasks, phoneme monitoring, and experimental manipulations of phonological complexity and its effect in speech production. The following is an overview of the findings and some of the limitations of the paradigms that have been used in the past to investigate phonological encoding in PWS. The priming paradigm has been used in children (CWS) and adults who stutter (AWS) to study the organization and activation of units in the phonological output lexicon ( Byrd, Conture, & Odhe, in press; Melnick, Conture, & Odhe, 2003; Wijnen & Boers, 1994). For instance, in an implicit priming paradigm involving the production of words, Wijnen and Boers (1994) reported a larger facilitatory priming effect for consonant–vowel (CV) primes (e.g., leuven, leugen, leuze, leaning, leukerd) than C primes (e.g., lepel, lila, loeder, larie, luier) in AWS as compared to adults who did not stutter (ANS). They suggested that AWS experienced difficulties in the encoding of the stress-bearing nucleus of a syllable, which was overcome by facilitating the encoding of such units with CV primes. However, in a study designed to replicate the findings from Wijnen and Boers (1994), Burger and Wijnen (1999) failed to find reduced facilitatory priming effects in AWS. Using a priming task with CWS, Byrd et al. (in press) reported that picture naming was facilitated by end-related phonological primes (e.g., /ed/-bed) as compared to onset-related primes (e.g., /bə/-bed); a reverse pattern was observed in children who did not stutter (CNS). Byrd et al. suggested that the phonological system in CWS may be less well developed or efficient compared to CNS. In contrast, Melnick et al. (2003) failed to find facilitatory priming effects for onset related primes in a picture naming task in CWS. Arnold, Conture, and Ohde (2005) reported comparable performances for CWS and CNS in the naming of target pictures with sparse and dense phonological neighborhood. Arnold et al. (2005) suggested that phonological processes contribute minimally to the difficulties experienced by CWS in producing fluent speech. Such equivocal support for a phonological encoding deficit in PWS from priming studies may, in part, be related to methodological differences. For instance, phonological encoding in CWS has been investigated using both segment (e.g., Melnick et al., 2003) and rhyme primes (e.g., Byrd et al., in press). These tasks differ in the nature of the primes and the word form structures they facilitate. Nevertheless, contrary findings clearly indicate the need for further detailed investigation of the organization and activation of the phonological output lexicon in PWS. Nonword tasks have also been used to investigate phonological encoding skills in PWS. Performance in these tasks is a predictor of learning of the phonology of new words in children ( Gathercole & Baddeley, 1990). Previous studies correlating nonword repetition skills with speech motor tasks have reported negligible correlations between these tasks ( Edwards & Lahey, 1998), suggesting that the performance in the nonword task may be independent of the motor difficulties experienced frequently by PWS. Hakim and Bernstein Ratner (2004) investigated nonword repetition skills in eight CWS and eight CNS. The target stimuli were nonwords varying in length (two to five syllables) that had stress on the first syllable. On the four-syllable nonwords, the stress pattern was altered to study the effect of stress position on task performance. CWS on average exhibited more errors compared to CNS in nonword repetition, although this difference was significant only at the three-syllable level. Furthermore, both groups exhibited higher error percentages in the four- and five-syllable nonwords as compared to the two- and three-syllable nonwords. The findings also indicated that the unfamiliar stress pattern used in the study might have partly challenged the phonological encoding process thereby contributing to a higher phonemic error percentage in the four-syllable nonwords in CWS. Keeping in mind the relatively small sample size (N = 8), their findings supported the hypothesized link between stuttering and phonological encoding, and highlighted the need to study word stress and its influence on phonological encoding in PWS. It could be argued, however, that factors other than phonological encoding difficulties may have resulted in a higher percentage of errors in CWS as compared to CNS, such as reactive behaviors to the unfamiliar nature of the target stimuli in CWS. In a more recent investigation of phonological encoding skills in adults who stutter, Sasisekaran, De Nil, Smyth, and Johnson (2006) studied phonological encoding in PWS and persons who do not stutter (PNS) using phoneme monitoring during silent picture naming. This task presumably requires encoding of the sound segments during silent naming in order to provide a phoneme monitoring response ( Wheeldon & Levelt, 1995; Wheeldon & Morgan, 2002). Performance in the phoneme monitoring task was compared to performance in tone monitoring, picture naming, and simple motor response tasks. PWS were slower compared to PNS in phoneme monitoring while being comparable in auditory monitoring, picture naming, and simple motor responses. The findings from Sasisekaran et al. (2006) were interpreted to suggest that PWS were delayed in the encoding of sound segments during silent naming. Their findings, however, need to be interpreted with caution as the phoneme monitoring task involves several subprocesses including speed of lexical retrieval, phonological encoding, general monitoring skills, and motor responses. Although the groups were comparable in general monitoring and manual response speed, further testing is warranted to investigate the extent to which initial lexical retrieval speed may have contributed to the group differences observed in phoneme monitoring. Finally, phonological complexity and its effect on task performance in PWS has also been investigated in several studies (Logan & Conture, 1995; Prins, Main, & Wampler, 1997; Watkins & Yairi, 1997), although the findings have been equivocal. For instance, Howell, Au-Yeung, and Sackin (2000) investigated the effect of phonologically complex sounds such as late-emerging consonants (LEC) and consonant clusters (CC) on percentage disfluencies in the conversational samples of 51 participants divided into three age groups (3–11 years, 12–18 years, >18 years). Results indicated that the frequency of stuttering was higher on LEC and CC in function words than content words in the younger age groups as compared to the older age groups. Additionally, the older age group exhibited a higher percentage of disfluencies when both LEC and CC co-occurred in word-initial position compared to other positions. Howell et al. (2000) suggested that PWS might exhibit difficulties in the planning and motor execution of complex phonological categories. There is also some evidence suggesting that the loci of stutter events are determined by the phonological complexity of utterances (Howell & Au-Yeung, 1995; Howell, Au-Yeung, & Sackin, 1999; Throneburg, Yairi, & Paden, 1994; Wolk, Blomgren, & Smith, 2000). Van Lieshout, Hulstijn, and Peters (1996) studied speech response time to picture naming and word naming tasks in PWS and PNS groups, while also manipulating lexical complexity using words of varying length and number of phonemes. Although Van Lieshout et al. (1996) reported that PWS were generally slower in both tasks, neither word length nor number of phonemes influenced task performance. They interpreted the findings to suggest that PWS do not differ from PNS in the stages of word processing leading up to motor planning and execution. However, caution is warranted in interpreting the findings from locus studies in stuttering as the exact nature of the complexity of the units studied, that is, phonological versus phonetic, is unclear. This question is crucial, as the study of phonological complexity (depending on the type of manipulation) might relate to the acquisition and representation of phonological rules, the activation and retrieval of such units from the lexicon, as well as the study of phonetic complexity to phonetic/motor performance. In PWS, given the evidence for the presence of both phonological and phonetic difficulties, studies that do not make a clear distinction between both factors are likely to misinterpret the nature of the deficit under investigation. Several theories of stuttering attribute a causal role to phonological encoding in stuttering (Howell, 2004; Perkins, Kent, & Curlee, 1991; Postma & Kolk, 1993; Wingate, 1988). These theories are mostly based on the model of speech production by Levelt (1989) and Levelt, Roelofs, and Meyers (1999). Of these theories, the Covert Repair Hypothesis (CRH; Postma & Kolk, 1993), which has received considerable research interest, proposes that stutter events are the overt manifestation of covert corrections of speech errors. More recently, Vasic and Wijnen (2005) proposed that PWS exhibit a deficit not in phonological encoding per se, but in their speech monitoring skills, the Vicious Circle Hypothesis (VCH). According to the VCH, PWS exhibit rigid monitoring of the speech produced by self and others. In general, however, studies investigating phonological encoding skills in PWS fail to offer unequivocal support for a mechanism specific deficit in the encoding of phonological units during speech production in PWS. The gaps identified in the literature raise certain outstanding issues that need to be addressed in the investigation of phonological encoding skills in PWS. From a theoretical perspective, it is important to understand whether a deficit in phonological encoding does exist in PWS. This would require a systematic investigation of the nature and extent of phonological encoding difficulties, if any, in PWS. From a methodological perspective, it would be crucial to choose an appropriate paradigm and baseline tasks that would enable isolation of the process of interest, i.e., phonological encoding, from processing differences between PWS and PNS at other levels of speech production such as, differences in lexical retrieval and phonetic encoding. 1.1. The paradigm: phoneme monitoring in silent picture naming With these considerations in mind, the present study investigated phonological encoding in AWS using phoneme monitoring in silent naming paradigm. Phoneme monitoring has been used to study phonological encoding in PNS (Costa, Sebastian-Galles, Pallier, & Colomé, 2001; Wheeldon & Levelt, 1995; Wheeldon & Morgan, 2002) and more recently in PWS (Sasisekaran et al., 2006). The assumption behind the paradigm is that participants rely on prearticulatory monitoring of the output of phonological encoding to provide a phoneme monitoring response (Levelt, 1989; Wheeldon & Levelt, 1995). In the present investigation, the phoneme monitoring paradigm was chosen to study phonological encoding in PWS for several reasons. First, past studies using phoneme monitoring in silent speech have indicated that the time course of phoneme monitoring parallels the time course of phonological encoding (Costa et al., 2001; Wheeldon & Levelt, 1995; Wheeldon & Morgan, 2002), supporting a left to right incremental encoding process (Levelt, 1989 and Levelt et al., 1999; Sevald & Dell, 1994). Second, the ability to study performance in this task during silent speech, without overt speech requirements, is useful in PWS where overt speech motor processes may otherwise render interpretation difficult. Third, by varying phonological complexity of target items and investigating the time course of phoneme monitoring across different complexity categories, this paradigm can also be used to study the effect of complexity on phonological encoding in PWS and PNS. Fourth, previous findings of differences in monitoring patterns in silent speech and perception offer an opportunity to study specific monitoring skills in PWS and PNS. In the present study, it was hypothesized that if group differences are observed only in silent naming and not in perception, then such differences may be attributable to phonological encoding in silent production as opposed to a generalized phoneme monitoring deficit in both production and perception. Therefore, the phoneme monitoring paradigm enables a theoretically based, systematic investigation of phonological encoding deficits, if any, in PWS. 1.2. Purpose of the present study In the present study, participants performed phoneme monitoring in silent picture naming, monitoring target phonemes in pictures that were pre-assigned to target words. This task was used to investigate whether PWS differ from PNS in the encoding of sound segments during silent speech. Participants also performed phoneme monitoring during perception in which they monitored target phonemes in the auditory presentations of target words. This task was used to investigate whether the differences in phoneme monitoring, if any, between PWS and PNS are related to general monitoring skills. For both these tasks, phoneme monitoring was studied in two target positions across two different phonological complexity levels, that is, compound words and noun phrases, to study the effect of complexity on segmental phonological encoding (for further description of complexity categories see, ‘Stimuli’). It was hypothesized that if PWS differed from PNS in phonological encoding, the groups would differ only in phoneme monitoring during the silent naming task while being comparable in monitoring during the auditory perception task. This would support the psycholinguistic theories of stuttering, which postulate a phonological encoding deficit in PWS ( Howell, 2004 and Perkins et al., 1991; Postma & Kolk, 1993; Wingate, 1988). Contrarily, if PWS experienced a generalized linguistic monitoring deficit, as evidenced by differences in both the silent naming and the perception tasks, this would offer preliminary support for models, such as the VCD hypothesis ( Vasic & Wijnen, 2005) that proposes a monitoring deficit as a causal variable in stuttering. In addition, it was hypothesized that phonological complexity would adversely influence monitoring performance in PWS.

The present study provides qualified support for the presence of a phonological processing deficit in PWS. In interpreting these results a number of limitations of the present study need to be kept in mind. First, the type of complexity manipulation used in the present study was not successful in effectively challenging phonological encoding skills, i.e., the expected within-group differences between compound words and noun phrases were not obtained in the present study. In addition, the complexity manipulation was not restricted specifically to the phonological level. The compound words differed from noun phrases not just in stress location, but also in the number of lexical units and grammatical categories. Future investigations need to study whether the present findings are comparable to monitoring performance using other types of phonological complexity manipulations that are specific to the phonemic level (e.g., word frequency, number and type of syllables, and stress manipulation). Second, although a delay in phonological encoding may be one factor contributing to the observed group differences in phoneme monitoring, other factors may have contributed as well: (a) slower monitoring in PWS may be a learned behavior attributable to a life long history of stuttering (e.g., James, 1981 and La Croix, 1973); and (b) it could be argued that the between-group differences in phoneme monitoring are due to differences in the encoding of the phonetic code, rather than the phonological code (e.g., McGuire et al., 1996 and Shergill et al., 2002). Experiments targeted at investigating these factors while studying phoneme monitoring performance in PWS are required to rule out these and other possible interpretations. Third, the present study tested phoneme monitoring skills in a relatively small group of participants with the stuttering severity of the participants in the PWS group being biased towards mild and very mild, and the participant groups were not matched systematically for age although efforts were made to retain homogeneity across age range and years of education. Although it seems unlikely that the relatively small difference in age between the two groups could account for the observed differences, future studies investigating phonological encoding using phoneme monitoring need to test further the effect of such variables in task performance. In conclusion, the present findings indicated that PWS are significantly slower than PNS in performing phoneme monitoring in a silent naming task while being comparable to PNS in performing phoneme monitoring in a perception task. These findings offer support for the psycholinguistic theories of stuttering that have implicated phonological encoding in stuttering. Due to its strategic location within the speech production system (see Levelt, 1989 and Levelt et al., 1999), delays in phonological encoding are likely to challenge fluent speech production in PWS. Keeping in mind that the present findings of delayed phonological encoding in PWS are specific to a silent naming paradigm, it remains to be seen whether a bidirectional influence exists between phonological encoding and overt speech motor control processes in PWS.