feature of speech disorder rather than typical development (Holm et al., 2007). ...... continuum and that inconsistency indicates an excessive amount of variability.
INCONSISTENCY OF SPEECH IN CHILDREN WITH CHILDHOOD APRAXIA OF SPEECH, PHONOLOGICAL DISORDERS, AND TYPICAL SPEECH
Jenya Iuzzini
Submitted to the faculty of the University Graduate School in partial fulfillment of the requirements for the degree Doctor of Philosophy In the Department of Speech and Hearing Sciences, Indiana University February 2012
ii
Accepted by the Graduate Faculty, Indiana University, in partial fulfillment of the requirements for the degree of Doctor of Philosophy.
Doctoral Committee
Karen Forrest, Ph.D.
Tessa Bent, Ph.D.
Judith A. Gierut, Ph.D.
David Koceja, Ph.D.
David Pisoni, Ph.D.
January 9, 2012
iii ACKNOWLEDGEMENTS First, thank you to my outstanding committee who provided feedback and motivation for various facets of the current work. Judith Gierut, Ph.D. provided elegant examples of theoretically-based clinically-relevant research. In addition, our conversations on other important matters (e.g., food, fashion) provided much needed breaks throughout the years. David Pisoni, Ph.D. encouraged my interest in cognitive factors associated with Childhood Apraxia of Speech, and reminded me how great science sounds in a Brooklyn accent. Tessa Bent, Ph.D. contributed a useful perspective throughout this process and has been a model new investigator. Thanks also to David Koceja, Ph.D. for feedback and support throughout my doctorate; my kinesiology training was an adventure which I look forward to continuing in kinematic research. My greatest gratitude is directed to Karen Forrest, Ph.D.—she has been a mentor in the truest and best sense of the word. She has served as a role model for all things academic and otherwise, and I appreciate her guidance and support whether it pertained to the dissertation, applications, or matters related to life in general. It was through my early experiences in her lab that I realized this was the best path for me; I cannot imagine this process without her. Thank you to the many graduate students who worked in the lab: Casey Wepsic, Kelly Campbell, Mariana Millan, Kaitlyn Clendenen, Becky Jessmer, and Libby Buck contributed what seemed like endless hours of data collection, transcription, and analysis—and so enthusiastically I might add. I enjoyed working closely with you all—this lab felt like home, in part, because of each of you. Thanks also to CJ Seong, Ph.D. and Yi-Fang Chiu for providing help with Praat training and script development. Thank you to the participants and their families, and to the IU Health Bedford Hospital, St. Vincent de Paul Catholic School, the IU Speech and Hearing Research Joint Database, and CASANA for help with recruitment. Specifically, thanks to Angie Stroud and Kacy Winegar for being advocates for this research, and wonderful friends.
iv Many friends supported me in this process. Jean Sokolowski, thank you for helping me to get to this point. Our daily updates undoubtedly made the difference between a good idea for a study and a completed dissertation. Thanks also to my fellow doctoral students—I am very proud to be part of this cohort. Great appreciation also goes to Tepanta Fossett, Ph.D.—our tea breaks saved me on more than occasion. Enormous thanks go to my parents, Jane and Edward Iuzzini, for providing support my whole life and who didn’t miss a beat when I majored in Printmaking and then decided to earn a doctorate in Speech Science. It was quite a road to get to this point and my ability to get here was in large part due to them. They also provided support and sustenance during the final days of dissertating—coming home to their cooking each night surely added years back on to my life. Cousin, Megan Phelan, and brother, Jonathan Iuzzini, were loving and honest sounding boards and friends throughout this process. Finally, thanks to Daniel Kutz who helped me a hundred different ways; especially, he helped me to smile and laugh daily and to remember the world beyond this dissertation. This research was supported by NIH-DC04575 (P.I. Karen Forrest), T32 DC 12-32 (P.I. David Pisoni), the ASHF New Century Scholars Doctoral Scholarship, and the Lion’s Club McKinney Research Award.
v Jenya Iuzzini INCONSISTENCY OF SPEECH IN CHILDREN WITH CHILDHOOD APRAXIA OF SPEECH, PHONOLOGICAL DISORDERS, AND TYPICAL SPEECH There is a lack of agreement on the features used to differentiate Childhood Apraxia of Speech (CAS) from Phonological Disorders (PD). One criterion which has gained consensus is lexical inconsistency of speech (ASHA, 2007); however, no accepted measure of this feature has been defined. Although lexical assessment provides information about consistency of an item across repeated trials, it may not capture the magnitude of inconsistency within an item. In contrast, segmental analysis provides more extensive information about consistency of phoneme usage across multiple contexts and word-positions. The current research compared segmental and lexical inconsistency metrics in preschool-aged children with PD, CAS, and typical development (TD) to determine how inconsistency varies with age in typical and disordered speakers, and whether CAS and PD were differentiated equally well by both assessment levels. Whereas lexical and segmental analyses may be influenced by listener characteristics or speaker intelligibility, the acoustic signal is less vulnerable to these factors. In addition, the acoustic signal may reveal information which is not evident in the perceptual signal. A second focus of the current research was motivated by Blumstein et al.’s (1980) classic study on voice onset time (VOT) in adults with acquired apraxia of speech (AOS) which demonstrated a motor impairment underlying AOS. In the current study, VOT analyses were conducted to determine the relationship between age and group with the voicing distribution for bilabial and alveolar plosives. Findings revealed that 3-year-olds evidenced significantly higher inconsistency than 5year-olds; segmental inconsistency approached 0% in 5-year-olds with TD, whereas it persisted
vi in children with PD and CAS suggesting that for child in this age-range, inconsistency is a feature of speech disorder rather than typical development (Holm et al., 2007). Likewise, whereas segmental and lexical inconsistency were moderately-highly correlated, even the most highly-related segmental and lexical measures agreed on only 76% of classifications (i.e., to CAS and PD). Finally, VOT analyses revealed that CAS utilized a distinct distribution pattern relative to PD and TD. Discussion frames the current findings within a profile of CAS and provides a validated list of criteria for the differential diagnosis of CAS and PD.
Karen Forrest, Ph.D.
Tessa Bent, Ph.D.
Judith A. Gierut, Ph.D.
David Koceja, Ph.D.
David Pisoni, Ph.D.
vii
Table of Contents ACKNOWLEDGEMENTS ................................................................................................... iii LIST OF TABLES................................................................................................................. x LIST OF FIGURES ............................................................................................................. xi INTRODUCTION ....................................................................................................................... 1 LITERATURE REVIEW .............................................................................................................11 Diagnostic Features of CAS ..................................................................................................13 Inconsistency and Variability .................................................................................................36 Measures of speech variability and inconsistency .................................................................40 Lexical measures of inconsistency ........................................................................................41 Segmental analysis of inconsistency .....................................................................................47 Acoustic analysis ...................................................................................................................52 Research Questions ..............................................................................................................60 Segmental and Lexical Comparisons .................................................................................62 Acoustic Analysis of VOT ...................................................................................................62 METHODS ................................................................................................................................64 Participants ...........................................................................................................................64 Stimuli ...................................................................................................................................67 Procedures ............................................................................................................................69 Intra- and Inter-rater Reliability ..............................................................................................75 Segmental/Lexical level .....................................................................................................75 Acoustical level ..................................................................................................................76 Statistical Analyses................................................................................................................76 Group equivalence .............................................................................................................76 Acoustical Data ..................................................................................................................77 Segmental Data .................................................................................................................78 Lexical Data .......................................................................................................................79 Group Classification ...........................................................................................................79 Summary ...............................................................................................................................80 RESULTS .................................................................................................................................81 Group equivalence ................................................................................................................82 Group Equivalence for participants in Protocol 1 ................................................................82 Group Equivalence for participants in Protocol 2 ................................................................83
viii Segmental Level....................................................................................................................86 Relationship between age and segmental level inconsistency in TD, PD, and CAS ...........86 Comparison of different segmental measures ....................................................................90 Group classification by segmental measures .....................................................................90 Lexical level ...........................................................................................................................91 Relationship between age and lexical inconsistency in TD, PD, and CAS ..........................92 Comparison of different lexical measures...........................................................................94 Group Classification based on Segmental and Lexical Level Inconsistency Measures ..........95 Group assignment across measures ..................................................................................96 Summary ...............................................................................................................................97 Voice onset time ....................................................................................................................98 Age- and group-related changes to VOT ............................................................................98 Age-related changes to VOT ..............................................................................................99 Group related differences in VOT .....................................................................................105 Voicing distribution patterns by group ..............................................................................108 Relation between VOT and inconsistency ...........................................................................114 Segmental level comparisons ..........................................................................................114 VOT vs. CSIP ..................................................................................................................114 VOT vs. ISP .....................................................................................................................114 VOT vs. ISP-A .................................................................................................................115 VOT vs. ISP/PCC.............................................................................................................115 Lexical level comparisons ................................................................................................115 VOT vs. WIS ....................................................................................................................115 VOT vs. WIS-R ................................................................................................................115 VOT vs. WIS-A.................................................................................................................116 VOT vs. WIS-AR ..............................................................................................................116 Summary .........................................................................................................................116 Group Classification ............................................................................................................124 Summary .........................................................................................................................130 Additional Comparisons .......................................................................................................130 DISCUSSION..........................................................................................................................131 Overview .............................................................................................................................131 Inconsistency of Speech ......................................................................................................132 Relationship between age and inconsistency ......................................................................132
ix Comparison of different segmental inconsistency measures ...............................................134 Comparison of different lexical inconsistency measures ......................................................138 Agreement between Segmental and Lexical inconsistency measures .................................140 Reconsidering the cutoff score for the WIS-A ......................................................................143 VOT .....................................................................................................................................143 Relation between VOT and age ...........................................................................................143 VOT performance in CAS and PD .......................................................................................146 Variability in VOT productions for typical and disordered speakers......................................149 Language in Children with CAS and PD ..............................................................................151 Profile of CAS ......................................................................................................................154 Clinical application ...............................................................................................................159 Limitations of the current research and recommendations for future investigation ...............162 REFERENCES .......................................................................................................................165 APPENDIX..............................................................................................................................173 APPENDIX A. Child Case History Questionnaire ............................................................173 APPENDIX B. Place-voice-manner template ..................................................................176 APPENDIX C. Consonant Articulation Long Probe Stimuli ...............................................177 APPENDIX D. Word Inconsistency Assessment (Dodd et al., 2006) ................................179 APPENDIX E. Supplemental Word List ............................................................................181 APPENDIX F. Acoustic Stimuli.........................................................................................182 APPENDIX G. Voicing Distribution Histograms by Group and Distribution Pattern...........183 APPENDIX H: VITA .........................................................................................................212
x LIST OF TABLES Table 1. Means and SD of group characteristics from participants in Protocol 1. .....................84 Table 2. Means and SD of group characteristics from participants in Protocol 2 .......................85 Table 3. Voicing distribution patterns as evidenced by each group. . .....................................113 Table 4. Means and SD for measures associated with VOT for PD and CAS as classified by CSIP. ...............................................................................................................................117 Table 5. Means and SD for measures associated with VOT for PD and CAS as classified by ISP. ..................................................................................................................................118 Table 6. Means and SD for measures associated with VOT for PD and CAS as classified by ISP-A. ..............................................................................................................................119 Table 7. Means and SD for measures associated with VOT for PD and CAS as classified by ISP/PCC. .........................................................................................................................120 Table 8. Means and SD of VOT measures for CAS and PD as differentiated by WIS. ............121 Table 9. Means and SD of VOT measures for CAS and PD as differentiated by WIS-R..........122 Table 10. Means and SD of VOT measures for CAS and PD as differentiated by WIS-A ........123 Table 11. Means and SD of VOT measures for CAS and PD as differentiated by WIS-AR. ....124 Table 12. Means and SD of speech and language measures for PD and CAS as specified by the two-step cluster analysis ............................................................................................128 Table 13. Means and SD for select language, inconsistency, and BRIEF subtests for CAS and PD as specified by the cluster analysis ............................................................................128 Table 14. Means and SD for VOT measures for PD and CAS as specified by cluster analysis ........................................................................................................................................129
xi LIST OF FIGURES Figure 1. Age-by-group changes to segmental inconsistency....................................................89 Figure 2. Age-by-group changes to lexical inconsistency ..........................................................94 Figure 3. Scatterplots depicting relationships between select segmental and lexical inconsistency measures for all participants in Protocol 2 (i.e., CAS, PD, and TD) .....................97 Figure 4. Age-by-group comparisons of VOT measures. ........................................................105 Figure 5. VOT measures by group; CAS and PD as classified by CSIP cutoff score. .............108 Figure 6. Examples of voicing distribution patterns evidenced by participants in the current research ..................................................................................................................................109 Figure 7. Speech severity compared across participants sorted by voicing distribution patterns ...............................................................................................................................................111 Figure 8. Mean lexical and segmental inconsistency performance for participants sorted according to their voicing distribution pattern for bilabial plosives ............................................111 Figure 9. Mean lexical and segmental inconsistency performance for participants sorted according to their voicing distribution pattern for alveolar plosives. .........................................112
1
INTRODUCTION The choice of the frame of reference for determining variability in motor control is sometimes arbitrary—that is, it merely follows a tradition that has been initiated in a certain experimental protocol. The use of different frames of reference can lead to different descriptions and interpretations of the role of variability of the sensorimotor system. The determination of the frame of reference for measurement should be motivated a priori by a theoretical perspective and not merely the operational traditions in motor control (Newell & Corcos, 1993, p.3). According to the American Speech-Language-Hearing Association’s (ASHA) technical statement on Childhood Apraxia of Speech (2007), “Childhood apraxia of speech (CAS) is a neurological childhood (pediatric) speech sound disorder in which the precision and consistency of movements underlying speech are impaired in the absence of neuromuscular deficits” (pp. 34). Although this technical statement reported speech and prosodic features which have gained some consensus among speech-language pathologists, the paper did not include diagnostic criteria or procedures to ascertain a diagnosis of CAS. This omission may be caused by the controversy surrounding the disorder and the limited research that has been conducted to define diagnostic criteria. Historically, the list of features which contribute to a diagnosis of CAS has been long and lacking in specificity. In a survey conducted by Forrest (2003) 75 speech-language pathologists who had experience working with the pediatric population were asked to report the top three criteria they use to make a diagnosis of CAS. Fifty different features across a variety of speech and oral behaviors were reported ranging from poor response to traditional speech treatment to drooling. The feature identified most commonly was inconsistency of errors which accounted for 14.1% of responses. As the ASHA technical statement suggests, there is increasing consensus that ‘inconsistency’ is central to the CAS diagnosis; however, this feature continues to be ill-defined and there is no operational definition of inconsistency relative to
2 speech sound production. For example, studies of speech inconsistency have been conducted at acoustical (Whiteside, Dobbin, & Henry, 2003), segmental (Iuzzini & Forrest, 2010; Tyler, Lewis, & Welch, 2003), and lexical (ASHA, 2007; Dodd & McCormack, 1995; Schumacher, McNeil, Vetter, & Yoder, 1986) levels; however, investigations comparing CAS performance across these levels have yet to be conducted so the relation between these levels of performance is unclear. It is also unknown which level of analysis contributes the most information toward determining the underlying deficit of CAS. A further confusion in discussions of inconsistency is its distinction from variability. Many studies report that children with typical speech development exhibit high variability in their phonemic, acoustic, and kinematic output relative to the adult system prior to age 7 (Goffman & Smith, 1999; Sharkey & Folkins, 1985; Terband, Maassen, van Lieshout, & Nijland, 2010) or even age 11 (Whiteside, et al., 2003), although Stathopoulos (1995) did not find that children’s speech is characterized by high variability. This presents a clinical challenge in that CAS, characterized by inconsistent production, may be diagnosed during the time window when children are expected to exhibit variability as a part of normal learning and development. Little is known about how to differentiate variability, which is fundamental to and reflective of motoric and cognitive growth, from inconsistency, now frequently considered the basis for diagnosis of CAS. The word variability, per the Oxford English Dictionary (OED; 1989), is defined as “tendency towards, capacity for variation or change . . . in amount, magnitude, or value”. Variability, therefore, may be assessed in terms of the quantity of different productions as well as the quality of those alternations (Touwen, 1993). In this context, variability is considered a basic characteristic of human performance and learning, and may reflect a state of transitional knowledge (Chen, Siegler, & Daehler, 2000) in various domains (e.g., cognitive, motor, neural). Although intra-individual variability may reflect noise in the system (Newell & Corcos, 1993), it also relates to the flexibility of a system to adapt to new
3 environments or contexts (Thelen & Smith, 1994) and will often be evidenced by expert performers (Muller & Sternad, 2009). By contrast, very stable (e.g., stereotypy) productions might suggest a behavior that is further from transition (Wilkinson, 1982). Variability, as a basic characteristic of human performance and learning, has been studied across numerous systems including cognitive (Breckinridge Church & Goldin-Meadow, 1986; Siegler, 2007), developmental (Thelen & Smith, 1994), and motor (Newell & Corcos, 1993). These studies indicate an association between variability and readiness for learning and may reflect the emergence of a new skill (e.g., Eguchi & Hirsch, 1969; Forrest, Weismer, Elbert, & Dinnsen, 1994; Kent, 1976). This is important as it may impact clinical prognosis and anticipated treatment outcomes. Breckinridge Church and Goldin-Meadow (1986) found that children with greater response-variability on a cognitive task evidenced greater learning following training compared with children who exhibited consistent inaccurate productions prior to treatment (Church & Goldin-Meadow, 1986). In this case, variability may simply reflect the use of different strategies across responses. This pattern has also been demonstrated in speech development (e.g., Forrest et al., 1994). Forrest and colleagues (1994) conducted a spectral analysis of /t/ and /k/ in 10 children (ranging in age from 3;6 to 6;6) with TD or PD (the term PD is used here and throughout this document to describe speakers with an idiopathic, not otherwise specified, speech sound disorder (e.g., phonological disorder). The PD children were assigned to one of two groups on the basis of their /t/ and /k/ productions such that one group (PDA) produced both sounds accurately across all word positions and the other group (PDI) produced /t/ and /k/ in word-initial position only. Forrest and colleagues found that children in the PDI group exhibited less variability across repeated productions than children with a greater knowledge of the contrast. Here variability reflects greater knowledge and flexibility whereas the stability evidenced by the children with limited knowledge indicates the use of a stereotypy or ballistic pattern.
4 Whereas variability is associated with growth and learning, inconsistency is associated with disorder (ASHA, 2007; Dodd & McCormack, 1995; Forrest, 2003). The OED defines inconsistency as a “discrepancy between principles and practice, or between one action and another” (OED online, para 3). If this definition is applied to speech production one might expect unpredictable sound use or sounds that are not in keeping with the trajectory of typical speech acquisition. Another possibility is that variability and inconsistency are two behaviors on the same continuum and that inconsistency indicates an excessive amount of variability. Holm et al. (2007) operationally defined speech inconsistency as “speech characterized by a high proportion of differing repeated productions with multiple error types (unpredictable variation between a relatively large number of phones and/or structural changes (Bradford & Dodd, 1996) that cannot be attributed to factors responsible for normal variability” (p. 468). By this definition, it is only the quantity of different productions that provides the distinction between variability and inconsistency. To date limited research has been conducted which investigates the similarities and differences between variability and inconsistency in children with typical development or those with disordered speech. Two of the definitions commonly used in the literature on inconsistent speech production in children with CAS include: different productions of the same exemplar across multiple trials (ASHA, 2007; Dodd, 1995; Marquardt, Jacks, & Davis, 2004) (e.g., producing /kata/, /gai/, /kæt/ for the target /kæt/); and multiple substitutes for phonemes within and across all positions (Forrest, Dinnsen, & Elbert, 1997; Iuzzini & Forrest, 2010; Tyler, et al., 2003), (e.g., producing [d, t, g, ʃ] for /k/ targets independent of positional constraints). The first definition reduces inconsistency due to an external source of variability (i.e., context) but provides a limited amount of information about the child’s inventory. The second definition includes assessment of the inconsistency of productions across different positional, word, and stress contexts thereby
5 providing information about the participant’s entire inventory. The impact of these different definitions on our understanding of inconsistency will be discussed below. As the definitions of inconsistent speech disorders indicate, the level of analysis influences how diagnosis would be undertaken. Variability and inconsistency have been explored at the lexical level in studies of adults with AOS (e.g., Johns & Darley, 1970; Miller, 1992; Shuster & Wambaugh, 2008) in an effort to characterize the features of AOS as well as to differentiate them from speakers with other speech (e.g., dysarthria) or language disorders (e.g., aphasia). Unfortunately, although speakers with AOS do exhibit variability/inconsistency in their responses, differentiation from other disorder types (e.g., fluent aphasics exhibiting phonemic paraphasias) on this basis alone has not been successful (Miller, 1992). The research investigating lexical level inconsistency in adults with AOS is limited and has yielded mixed results (McNeil, Odell, Miller, & Hunter, 1995; McNeil, Robin, & Schmidt, 1997). Empirical investigation of lexical level inconsistency in the CAS population also is limited. Dodd and McCormack (1995) studied inconsistency at the lexical level by asking children to repeat a series of 25 words. Children who evidenced more than 40% inconsistent productions across 2 of 3 productions of the same target word were considered to have inconsistent speech disorders. The main argument for assessing inconsistency at the lexical level is that it helps to reduce to the influence of contextual variation. However, one of the challenges of assessing inconsistency at the lexical level is that there are numerous patterns exhibited by children during typical speech development which could result in variability. Examples of these were outlined by Davis, Jakielski, and Marquardt (1998) and include harmony, stopping of a consonant, simplification of a more complex consonant or word shape, and syllable deletion. In addition, a child may use complex phonological rules such as using all fricatives interchangeably (Dodd, 1995) which may appear highly inconsistent until investigated at a deeper level. As lexical level analysis is based on rating responses as “same/different,” a
6 deeper segmental or acoustical analysis may still be needed in order to learn about a child’s full knowledge and usage of speech sounds across contexts. Segmental analysis of error substitutes provides information about the specific consonant errors as well as the inconsistency of their usage.
Betz and Stoel-Gammon (2005)
compared error consistency at the lexical level in children with CAS and PD. Results revealed a greater proportion of errors for children with CAS but no difference between CAS and PD on the consistency measures. As such, only severity differentiated CAS from PD in this study. One possible reason for this was the metrics used to measure consistency (i.e., (1- (# of different error types/# erred productions)*100)). That is, inconsistency was measured relative to the number of erred productions and as the children in the CAS group evidenced a greater proportion of errors, they would have needed to exhibit a greater number of inconsistent productions in order to achieve the same level inconsistency as those with fewer errors. In addition, as severity is not a diagnostic feature of CAS it is imperative that the pathognomonic value of inconsistency measures is tested by comparing children with CAS and PD who are matched for severity. Tyler, Lewis, and Welch (2003) considered inconsistency of errors as a possible factor to predict phonological change resultant from treatment. They assessed inconsistency at the phonemic level in 20 children with PD and morphosyntactic impairments, ranging in age from 3;0 to 5;11. Tyler and colleagues used a raw score metric which reported the total number of different substitutes across all phonemes and all word positions. Their results indicated that the more inconsistent inventories were associated with the greatest PCC change. Tyler et al. reported that this was an unexpected finding. It is possible that Tyler and colleagues had actually investigated variability (associated with learning) rather than inconsistency (associated with disorder). A limit of this study is that Tyler and colleagues did not include a control group (e.g., children with TD or a different disorder type such as CAS); therefore it is unclear if their
7 data can differentiate typical speech from disordered, or if it can differentiate between disorder subtypes. Iuzzini and Forrest (2008) incorporated some aspects of the previous formulas in a metric designed to capture inconsistency at the segmental level. The consonant substitute inconsistency percentage (CSIP) reports the number of different error substitutes a participant uses divided by the total number of erred productions across the entire inventory. This measure was originally based on data collected from a 200-word probe which provided 340 opportunities to produce all American English consonants, multiple times, across all word positions. Analysis of 16 children with phonological/articulatory disorders (PAD) revealed two clusters, one more stable group with CSIP scores below 21% and a more inconsistent group with CSIP scores greater than 24%. These two clusters are believed to represent children with PD and CAS respectively. Although the CSIP is useful in differentially diagnosing children with PD and CAS, the metric does not reflect within-subject changes over time including post-treatment assessment. That is, CSIP often remains the same as the numerator and denominator change in the same direction and, therefore, masks improvement. This limitation was addressed in the development of the inconsistency severity percentage (ISP; Iuzzini & Forrest, 2010) which reports the number of different substitutes summed across all consonants, divided by the total number of consonant production opportunities. The stable denominator allows comparison within and across children and captures intrasubject change better than the CSIP. This segmental measure, which is similar to the total token variability measure developed by Marquardt et al. (2004), was originally applied to the speech of children with PAD (i.e., children with either PD or CAS) based on a speech probe (340 consonant opportunities). In a later study, the ISP was calculated from the Goldman-Fristoe Test of Articulation-2 which provides 106 singleton-consonant production opportunities. Comparisons of 21 children with PAD ranging in age from 3;1 to 6;9 and 21 age-
8 matched TD controls (Artman, 2010), revealed that the TD children produced ISPs lower than 7.5%. One group of PAD children had ISPs ranging from 8.4-13%, and a more inconsistent group had ISPs greater than 18% which may represent children with PD and CAS respectively (Jean, 2010).This provides preliminary evidence that the ISP may be used to differentiate children with TD, PD and CAS. A limit of perceptual analyses, such as those discussed above, is that they may be impacted by listener characteristics. Listener strategies, including the use of top down processing, may result in a speaker being given credit for a sound which is not actually present in the acoustic signal (Remez, 1981). Research also suggests that inter-rater reliability for perceptual measures is inversely related to speaker intelligibility which may be relevant in analyzing the speech of children with severe speech disorders. Whereas lexical and segmental analysis may be influenced by listener knowledge, acoustical analysis can provide objective information that links perception and production.
Acoustical analysis is less impacted by
intelligibility and is less vulnerable than perceptual measures to errors related to severity of disorder. In addition, although acoustical and perceptual measures often correlate with one another, the acoustic analysis may reveal information which is not evident in the perceptual signal. That is, a speaker may display an acoustical covert contrast that is not perceived by adult listeners, thus revealing information about the speaker’s knowledge and speech sound emergence which would not otherwise be apparent (Macken & Barton, 1980). Although there is substantial information to be gained from acoustical analysis, to date limited research (Iuzzini & Forrest, 2008; Shriberg, Green, Campbell, Mcsweeny, & Scheer, 2003) has been conducted at this level that can help distinguish PD from CAS. A similar problem of differentiating linguistic and motoric disorders is seen in the population of adults with Apraxia of Speech (AOS), presumed to be the adult counterpart of CAS (Morley, Court, & Miller, 1954). As AOS rarely exists in isolation, challenges exist in differentiating apraxic errors from
9 those caused by aphasia, particularly from phonemic paraphasias. Perceptual analysis of the speech of people with AOS suggests that they make phonemic substitutions which would reflect a linguistic etiology of the disorder. Acoustical evidence suggests that the errors could be actually severe phonetic distortions which would reflect a motor basis; that is, the responses are so distorted they present as responses from a different phonemic category (Blumstein, Cooper, Goodglass, Statlender, & Gottlieb, 1980). In their classic study, Blumstein et al. (1980) helped define the motor consequences of AOS through a voice onset time (VOT) analysis, (i.e., the duration between the release of the burst of a stop consonant and the onset of periodicity in the voicing of the adjacent vowel), of 13 participants with aphasia, 1 with dysarthria, and 4 healthy control subjects. Blumstein and colleagues found that the speakers with Broca’s aphasia and AOS evidenced significantly more phonetic errors than the Wernicke’s speakers. That is, speakers with Wernicke’s aphasia maintained discrete phonemic categories for voiced and voiceless cognates whereas participants with Broca’s evidenced more VOT productions in between the expected voiced and voiceless cognate categories (supporting a motor etiology). This study was the first to demonstrate a motor impairment underlying AOS and helped lay the framework for its differential diagnosis. Although VOT has been investigated in TD (Tyler & Watterson, 1991; Whiteside, et al., 2003) and PD (Forrest & Rockman, 1988; Forrest, Weismer, Hodge, Dinnsen, & Elbert, 1990; Gierut & Dinnsen, 1986), no studies have compared VOT in these populations to that of children with CAS. On the basis of Blumstein et al.’s findings, one might hypothesize that VOT in speakers with CAS is characterized by a continuous unimodal distribution wherein productions will be produced close to categorical boundaries. As noted previously, more information about the underlying deficit of CAS may be garnered from perceptual and acoustic analysis at the segmental level as opposed to lexical level analysis. The CSIP and ISPs are two measures of segmental inconsistency. Perceptual
10 analysis can be supplemented by acoustic analysis as a means to differentiate variability and inconsistency for diagnostic purposes. Following the above line of reasoning, this dissertation will investigate intrasubject inconsistency across multiple levels of production. Specifically it will compare different methods of inconsistency assessment to determine how well each tool differentiates children with typical developing speech, disordered inconsistency, and disordered consistent speech. This question will be addressed by comparing performance and group assignment based on acoustic, segmental, and lexical level measures. Investigation will include acoustical comparison of VOT performance, segmental comparison as specified by the inconsistency severity percentage (ISP; Iuzzini & Forrest, 2010) as well as a lexical comparison based on performance on the Inconsistency Assessment from the Diagnostic Evaluation of Articulation and Phonology (DEAP) (Dodd, Hua, Crosbie, Holm, & Ozanne, 2006).
11
LITERATURE REVIEW “There is debate in the speech pathology literature regarding definition, terminology, symptomatology, diagnostic features, therapy techniques, and the very existence of developmental dyspraxia” (McCabe, Rosenthal, & McLeod, 1998, p. 105). Childhood apraxia of speech (CAS) is a pediatric speech sound disorder that is entrenched in controversy. The controversy spans all aspects of the disorder beginning with its name. The term “apraxia,” as it is used today, was coined by Steinthal in 1871 to describe the behavior of patients who had lost motor ability secondary to neurological injury (Brown, 1988). Liepmann (as cited in Brown, 1988) expanded on the profile of this disorder in his work focusing on the localization of functions within the brain. In addition to specifying three types of what he termed motor apraxia (i.e., limb-kinetic, ideational, and ideomotor), Liepmann also described motor aphasia, a preliminary term for apraxia of speech. He considered this to be “apraxia of the glossolabiopharyngeal musculature: [in which] the nonparalyzed muscles cannot be innervated to bring forth the sounds of speech” (Brown, 1988, p. 34). Liepmann defined apraxia as a loss of purposeful movements that occurs even though the articulator or limb is intact. In addition, Liepmann specified that apraxia would not necessarily affect all movements, but rather, might only affect specific movements or types of movements (e.g., a person with limb apraxia might only have impaired intransitive movements; Brown, 1988). In the case of apraxia of speech, speech production may be impaired while nonspeech oral movements (e.g., licking the lips or raising and lowering the tongue) may be spared. The term articulatory dyspraxia was applied to children’s speech by Morley, Court, and Miller (1954) to describe six children who exhibited appropriate oral musculature for voluntary and involuntary non-speech movements, but who became “clumsy and awkward” (p. 9) during speech. Morley, Court, Miller, and Garside (1955) expanded the description of developmental articulatory apraxia to specify that the oral movements “are inadequate for the complex and
12 rapid movements used for articulation and the reproduction of the sequences of sounds used in speech [and that] it may occur with a minimal motor disability or as an isolated disorder of articulate speech” (p. 217-218). Whereas the original definition of apraxia specified a loss of skilled movement, children with idiopathic CAS, never have produced speech movements correctly. In this case, apraxia does not describe the loss of speech movements, but rather, a disorder which has disturbed the acquisition of speech movements. Currently, ASHA’s Ad Hoc Committee on Apraxia of Speech in Children defines CAS as: …A neurological childhood (pediatric) speech sound disorder in which the precision and consistency of movements underlying speech are impaired in the absence of neuromuscular deficits (e.g., abnormal reflexes, abnormal tone). CAS may occur as a result of known neurological impairment, in association with complex neurobehavioral disorders of known or unknown origin, or as an idiopathic neurogenic speech sound disorder (ASHA, 2007, p. 3). The ASHA technical statement on CAS addresses the numerous terminological iterations of this disorder since Morley et al.’s (1954) description. These terms include: apraxic dysarthria, cortical dysarthria, developmental apraxia of speech, developmental verbal apraxia, and developmental verbal dyspraxia. In addition, the term “inconsistent speech sound disorder” may or may not refer to the population of children with apraxia. Dodd and McCormack (1995) suggest that a CAS diagnosis requires concomitant oral-motor apraxia whereas inconsistent speech disorder does not. Without a validated list of criteria, it is unclear whether all these terms refer to distinct disorders, the same disorder, subtypes of the same disorder, or perhaps different symptoms within a syndrome. The term Childhood Apraxia of Speech was chosen as a preferred alternative to Developmental Apraxia of Speech by the ASHA Ad Hoc Committee on Apraxia of Speech in Children (2007) for the following reasons: the term developmental could be interpreted by insurance companies as a disorder which a child might outgrow without treatment or following treatment provided sufficiently by the schools. In addition, the term childhood was selected to include all children who exhibit the disorder including those with neo- or perinatal causes (e.g.,
13 galactosemia), those who have acquired the disorder during childhood secondary to an illness or neurological event (e.g., head injury or stroke), as well as those who exhibit CAS due to an unknown etiology (i.e., idiopathic CAS). The term apraxia of speech was chosen to include those with segmental and/or suprasegmental manifestations.
Diagnostic Features of CAS The controversy surrounding CAS extends beyond the name to the very features which are used to make the diagnosis; Guyette and Diedrich (1981) referred to CAS as a disorder “in search of a population” (p.39). Over the past 55 years, numerous case studies (e.g., Marquardt, et al., 2004; Morley, et al., 1954; Morley, et al., 1955; Stackhouse & Snowling, 1992), retrospective case reviews (e.g., Davis, et al., 1998; McCabe, et al., 1998; Rosenbek & Wertz, 1972; Stackhouse, 1992), and fewer group comparisons (e.g., Bradford & Dodd, 1996; Dodd, 1995; Dodd & McCormack, 1995; Shriberg, Aram, & Kwiatkowski, 1997b, 1997c; Shriberg, et al., 2003; Williams, Ingham, & Rosenthal, 1981; Yoss & Darley, 1974) or surveys (e.g., Forrest, 2003; Williams, Packman, Ingham, & Rosenthal, 1980) have been conducted which describe the possible nosological features of CAS. Unfortunately, much of the extant research is limited by one of the following issues: a- inclusion of a broad age range of participants spanning both development and residual disorders (Crary, 1984; Rosenbek & Wertz, 1972; Shriberg, et al., 1997b); b- participants who have experienced various quantities and types of therapy (e.g., Crary et al., 1984); c- inclusion of participants with concomitant deficits such as language disorder or mental retardation (e.g., Rosenbek & Wertz, 1972); d- use of the same factors for participant selection and dependent variables (e.g., Yoss & Darley, 1974; Dodd & McCormack, 1995); and e- no control group or the included control group contains typical developing speakers (TD) rather than children with a different type of speech disorder. The issue of control group selection is critical to defining CAS in that we have valid methods for determining that a
14 child is not TD, but differentiation of children with CAS from those with other speech acquisition disorders (e.g., functional articulation disorders) remains challenging. To date, there is not a validated list of pathognomonic features which serves to differentially diagnosis CAS from other pediatric speech sound disorders. ASHA’s Ad Hoc Committee on Apraxia of Speech in Children reviewed the literature between 1995 and 2007 and reported that “the core impairment [of CAS] in planning and/or programming spatiotemporal parameters of movement sequences results in errors in speech sound production and prosody (ASHA, 2007, p. 3). As the following review will highlight, this definition does little to enhance diagnostic accuracy of children with CAS. The earliest description of CAS features was made in Hadden’s (1891) report of three children with articulation disorder, ranging in age from 4 to 11 years. Features reported included unintelligible speech, inconsistent speech errors, appropriate mental capacity, and poor response to treatment. Hadden also described a prosodic disturbance following treatment in one participant whose attempts at correct articulation led to “a slow, staccato character and the words were mouthed” (p. 99). Although Hadden noted that prosody was altered in one child, he ascribed this alteration to a compensatory strategy implemented to increase intelligibility and in response to treatment. This is in contrast to the current view that prosodic impairment might be an underlying impairment of CAS (ASHA, 2007). Sixty years later Morley, Court, and Miller (1954) described articulatory dyspraxia in a paper focused on developmental dysarthria. Morley and colleagues stated that whereas dysarthria would be characterized by abnormal appearance and movement of the articulators in speech and nonspeech movements, speakers with CAS would accurately produce nonspeech oral movements but would evidence disruption in producing the rapid sound sequences required for speech. Rosenbek and McNeil (1991) elaborated on the distinction between dysarthria and
15 apraxia in adult speakers and specified that dysarthric speakers evidence disturbances of strength, range of motion, speed, and precision of oral movements whereas these are preserved in apraxic speakers. Morley et al. (1954) specified that the sound-sequencing, speech disturbance in CAS occurs in the absence of neuromuscular impairment and is functionspecific. This specificity of function is an important distinction that often is overlooked in that CAS currently is defined by a range of impairments across many different speech and oral functions that affect segmental features (e.g., Yoss & Darley, 1974; Iuzzini & Forrest, 2010), prosody (Shriberg, et al., 1997c; Shriberg, et al., 2003) oral movements (Yoss & Darley, 1974; Bradford & Dodd, 1996); and even full-body motor control (Crary, 1984). In an early attempt to refine the definition of CAS, Rosenbek and Wertz (1972) used a retrospective analysis to describe the neurological and speech/language histories of 50 children previously diagnosed with CAS. Rosenbek and Wertz aimed to describe the features of idiopathic CAS and illustrate the overlaps and differences compared with acquired apraxia of speech (AOS) and other speech disorders. Because of the preliminary nature of the study, and the retrospective design, participants ranged in age from 2;9 (years; months) to 14 years and included children with language disorder, mental retardation, dysarthria, and aphasia. Children with a history of a known neurological event such as head injury were excluded in order to focus on congenital CAS compared with acquired AOS. Standardized test scores were not available for 45 of the 50 subjects and diagnosis of CAS was based on clinical reports from the referring clinicians. The remaining five participants had been tested at the University of Colorado Speech and Hearing Clinic using a variety of standardized testing measures. Rosenbek and Wertz found that there were 13 speech and language features that were common across all participants providing a constellation of symptoms to fuel future investigations. The 13 features which were associated with idiopathic CAS are similar to those used to diagnose acquired apraxia of speech:
16 1.
may occur alone or with concomitant disorders such as dysarthria;
2.
delayed/deviant speech;
3.
receptive language is superior to expressive language;
4.
possible concomitant oral apraxia;
5.
phonemic errors including omissions, substitutions, distortions, additions, repetitions,
and prolongations; 6.
metathetic errors;
7.
increased errors with increased word length;
8.
although sounds and words may be produced intelligibly in isolation, breakdown occurs
when these are sequenced at the connected speech level; 9.
more errors on motorically complex sounds including fricatives and affricates;
10.
may affect vowels and/or consonants;
11.
inconsistent errors;
12.
prosodic disturbances of rate, stress, and spacing which they suggested could be as a compensatory strategy; and
13.
oral groping may precede or interrupt speech production.
These characteristics continue to dominate the features reported in the literature and serve as experimental variables for recent studies. Although 22 of the 36 children in Rosenbek and Wertz’s study (1972) who participated in a pediatric neurological evaluation scored within normal limits except for generalized or oral
17 apraxia (n=12) or orofacial apraxia (n=7), the remaining 14 children exhibited neurological deficits including but not limited to muscle weakness and drooling; likewise, feeding difficulty was also frequently reported. Of the 26 children who participated in the electroencephalography (EEG) portion of the study, 15 displayed focal or diffuse neurological abnormalities of areas including the motor strips, Sylvian and parietal regions, as well as lesions to the right hemisphere. Rosenbek and Wertz suggested that these diffuse neurological abnormalities, compared to more focal lesions in acquired AOS, indicate that there are less localized praxic centers in children than in adults. As such, there may be more areas (bilaterally) in the pediatric brain where a lesion could result in apraxia. This information should be considered with caution as the extent to which the generalized lesions found in this study were associated with apraxia versus the other neurological deficits or confounding factors is unknown (e.g., mental retardation or dysarthria). Whereas Rosenbek and Wertz’s study (1972) served to highlight features associated with idiopathic CAS, it lacked a control group; as such, it is unclear how many of the characteristics reported might overlap with TD or other types of speech or language disorder. Yoss and Darley (1974) compared children with speech disorders (SpD) including CAS and functional articulation disorders (PD) to children with TD. They investigated the features of 30 children with moderate to severe articulation disorder ranging in age from 5;1-9;10 and 30 ageand sex-matched TD controls. Children were included in the study if they evidenced normal cognition (IQ of 90 or above), language-age no lower than 6 months below chronological-age; no obvious organic etiology for the speech impairment; and a speech impairment severe enough to impact clinical, social, or academic performance, and therefore to require intervention. Yoss and Darley compared the SpD group and TD on oral movement performance including isolated and sequenced volitional oral movements (IVOM and SVOM, respectively),
18 diadochokinesis (DDK) rates, auditory perception and discrimination on the Denver Auditory Phonemic Sequencing Test (DAPST; Aten, 1973), and phoneme production on real and nonsense words. The nonsense words included 42 items produced in repetition during a short story activity and in spontaneous speech during a story retell task. A sign test revealed significant group differences on IVOM, SVOM, and on the DAPST. In addition, t-test comparison revealed that children in the SpD group evidenced lower DDK rate than the TD group; however, the youngest subgroups (5-year-olds) of SpD and TD did not differ on mean number of repetitions per second. The SpD group was then subdivided into two groups based on their performance on IVOM; children with low IVOM scores were assigned to the CAS group and those with IVOMs above the median score were assigned to the PD group. As group division was based on disturbance of oral movements rather than a metric of speech it is possible that comparison was between children with and without oral apraxia rather than apraxia of speech. The authors calculated difference scores between each disordered child and his/her TD match as a means of assessing the magnitude of the oral motor disturbance in PD and CAS children. The group-wise difference scores were then compared for the CAS and PD groups by t-tests. The authors reported that the groups differed statistically (p
