Quality of language intervention provided to primary-grade students with language impairment.

G Model

JCD-5660; No. of Pages 12 Journal of Communication Disorders xxx (2014) xxx–xxx

Contents lists available at ScienceDirect

Journal of Communication Disorders

Quality of language intervention provided to primary-grade students with language impairment Tricia L. Biancone *, Kelly Farquharson, Laura M. Justice, Mary Beth Schmitt, Jessica A.R. Logan The Ohio State University, United States

A R T I C L E I N F O

A B S T R A C T

Article history: Received 4 July 2013 Received in revised form 28 February 2014 Accepted 15 March 2014 Available online xxx

This study had two aims: (a) to describe the quality of language intervention provided by school-based speech-language pathologists (SLPs) to children with language impairment in the primary grades with respect to the quality of emotional support, instructional support, and proactive management during SLP–child interactions, and (b) to determine if key characteristics of the SLPs are predictors of variance in intervention quality. Participants were 174 children nested within 40 SLPs’ caseloads from various districts in two Midwestern states involved in a larger study of speech-language therapy practices in the public schools. A total of 208 videotaped language intervention sessions were coded for emotional support, instructional support, and proactive management using the Classroom Assessment Scoring System (CLASS; Pianta, La Paro, & Hamre, 2008). The quality of language intervention varied widely and was generally mid-range to high with respect to emotional support and proactive management, and low to mid-range in terms of instructional support. The quality of interactions varied and a large percentage of the observed variance in quality was attributed to SLPs. Time pressure was a strong predictor of the quality of emotional support, instructional support, and proactive management, and job satisfaction was a significant predictor of instructional support and proactive management. This descriptive information about school-based language intervention highlights the impact of the individual SLP in terms of the quality of the interactions taking place and the potential need to ease job pressures and promote job satisfaction. Learning outcomes: Readers will be able to: (1) identify and define three aspects of SLP–child interaction quality during intervention as framed in this study using the CLASS observation tool (Pianta, La Paro, et al., 2008); (2) discuss the relevance of those three aspects of quality to children with LI; and (3) identify SLP-level factors that significantly predict SLP–student interaction quality during intervention for children with LI. ß 2014 Published by Elsevier Inc.

Keywords: Language intervention Language impairment Quality Speech-language pathologist SLP characteristics School-based

1. Introduction In the United States, there are presently more than 130,000 certified speech-language pathologists (SLPs), of which an estimated 28,000 provide services within elementary schools (ASHA, 2010; Bureau of Labor Statistics, 2012). A prevalent

* Corresponding author. Tel.: +1 614 769 4149. E-mail address: [email protected] (T.L. Biancone). http://dx.doi.org/10.1016/j.jcomdis.2014.03.001 0021-9924/ß 2014 Published by Elsevier Inc.

Please cite this article in press as: Biancone, T. L., et al. Quality of language intervention provided to primary-grade students with language impairment. Journal of Communication Disorders (2014), http://dx.doi.org/10.1016/j.jcomdis.2014.03.001

G Model

JCD-5660; No. of Pages 12 2

T.L. Biancone et al. / Journal of Communication Disorders xxx (2014) xxx–xxx

component of the school-based SLPs’ caseload is providing direct treatment to students with language impairment (LI; ASHA, 2010). The importance of this clinical service is well documented, as students with LI are prone not only to having problems with language skills, but also are susceptible to difficulties in literacy (Cabell et al., 2010; Catts, Fey, Tomblin, & Zhang, 2002), academic performance (Conti-Ramsden, Simkin, Knox, & Durkin, 2009), and socio-emotional functioning (Botting & Conti-Ramsden, 2000; Sundheim & Voeller, 2004). Nonetheless, there is little research to date that empirically characterizes the quality of language intervention as provided by SLPs to students in educational settings. While there is a substantial body of literature (e.g., Fey, Cleave, Long, Hughes, 1993; Fey, Finestack, Gajewski, Popescu, & Lewine, 2010; Girolametto, Weitzman, & Greenberg, 2012; Haley, Camarata, & Nelson, 1994; Leonard, Camarata, Brown, & Camarata, 2004; Petersen, Gillam, Spencer, & Gillam, 2010; Wolter & Green, 2013) describing empirically validated intervention approaches that SLPs may choose to use, particularly as tied to the framework of evidence-based practice, few studies have sought to describe the quality of SLP–child interactions during business-as-usual (BAU) intervention as it is currently received in schools by students with LI. Indeed, the field of speech-language pathology does not currently have a common framework that establishes what factors affect quality of language intervention. Considering that school-based SLPs play a central role in improving the linguistic and academic outcomes for children with LI, it is possible that specific SLP-level factors contribute to the quality of language intervention. It is the aim of the present study to examine SLP-level factors as possible predictors of quality of language intervention with a goal of advancing the field toward establishing a framework for thinking about intervention quality. This is an important omission in the literature, particularly relative to the considerably large research literature regarding BAU practices employed by other school-based professionals, including teachers along the continua of schooling (i.e., preschool to secondary school; see Guo, Piasta, Justice, & Kaderavek, 2010; Justice, Mashburn, Hamre, & Pianta, 2008; Pianta, Belsky, Houts, & Morrison, 2007; Stuhlman & Pianta, 2009). Such research has been very important for identifying specific features during the normal course of classroom instruction that contribute directly and/or indirectly to children’s language growth over time (e.g., Bowers & Vasilyeva, 2011; Christ & Wang, 2012; Mashburn et al., 2008). In recent years, a robust literature on features of classroom instruction that influence children’s learning has increasingly emphasized high-quality teaching as predominantly a function of the nature of teacher–child interactions that take place within the classroom (Mashburn et al., 2008; La Paro et al., 2009; Pianta et al., 2005; NICHD ECCRN, 2002a, 2002b, 2004). Specifically, interactions between teachers and children are considered the mechanism through which instructional experiences affect children (Paro, Pianta, & Stuhlman, 2004). Observational tools that capture the quality of teacher–child interactions are able to document significant predictive relations between teaching quality and children’s gains in language, reading, and other areas of achievement during the preschool and primary grades (Girolametto et al., 2012; La Paro et al., 2009; Mashburn et al., 2008; Pianta et al., 2005). Drawing from such literature, the purpose of the present study was to examine the quality of language intervention provided to primary-grade students with LI, focusing on BAU treatment delivered by 40 SLPs providing services to 174 children in the public schools. Intervention quality was informed by classroom-quality research that characterizes high-quality instruction as a function of the quality of teacher–child interactions (e.g., Mashburn et al., 2008; Pianta et al., 2007). A key premise of the present work is that exposure to and participation in high-quality language intervention is important to students’ outcomes relevant to therapy, although this premise has not directly been studied in the language intervention literature (but see Bowers & Vasilyeva, 2011, for an exemplar from the educational literature). In addition, we examine the extent to which SLP-level factors may help to explain variance in intervention quality. Specific aims were twofold: (a) to characterize the quality of SLP–child interactions during school-based language intervention provided to children with LI with respect to emotional support, instructional support, and proactive management, and (b) to determine the extent to which characteristics of the SLP predict variance in the quality of intervention. 1.1. Describing and measuring intervention quality Language intervention is a multidimensional construct and is often discussed in terms of such variables as treatment targets, dosage/intensity, and techniques used (Swanson, Fey, & Mills, 2005; Wolfe & Heilmann, 2010; Yoder, Fey, & Warren, 2012). For instance, treatment principles for implementation of grammar facilitation with children with LI include careful selection of treatment targets, manipulation of the treatment context to provide children with exposure to those targets, and specific techniques that may be used to evoke children’s expression of specific grammatical forms (Fey, Long, & Finestack, 2003). High-quality intervention employing grammar facilitation likely would include adherence to these principles, such that intervention quality would be largely conceptualized as fidelity of implementation (see Guo et al., 2010). A separate, albeit complementary, way to conceive of language intervention quality concerns going beyond specific treatment techniques and examining the nature of the interactions that take place within the therapy session, as these SLP– child interactions are likely a key mechanism through which therapy experiences affect the child’s language skills. As derived from studies of the quality of classroom teaching (see Mashburn et al., 2008), referenced previously, SLP–child interactions as a potential index of quality may be conceived to reflect three inter-related domains – emotional support, instructional support, and proactive management – which collectively capture the quality of the interactions between SLPs and students (each of these three domains is explained in greater detail shortly). Importantly, there is evidence to suggest that each of these interaction domains, as captured for teachers within the classroom context, has either direct or indirect effects on children’s language skills and academic growth (Downer, Sabol, & Hamre, 2010). That is, numerous studies indicate that


G Model

JCD-5660; No. of Pages 12 T.L. Biancone et al. / Journal of Communication Disorders xxx (2014) xxx–xxx

3

children whose interactions with teachers (as well as other caregivers, including parents) are characterized by high levels of emotional support, instructional support, and proactive management experience greater growth in language and academic skills over time relative to children who experience low levels of support (Hamre & Pianta, 2005; Justice & Pullen, 2003; Mashburn et al., 2008; NICHD ECCRN, 2002a, 2004; Pakarinen et al., 2013; Pianta et al., 2007). The importance of high-quality interactions in language intervention is supported, in part, by the sociocultural theory of development. Lev Vygotsky’s sociocultural perspective of learning asserts that the processes of development are fundamentally social, and that children learn in a social context through interactions with adults and peers (Rogoff, 1990; Vygotsky & Cole, 1978). The focus on the specific proximal processes (Bronfenbrenner & Morris, 1969) of emotional support, instructional support, and proactive management (Hamre & Pianta, 2007) provides a framework for examining the quality of interactions between SLPs and children within the therapeutic context. 1.1.1. Emotional support Emotional support concerns the extent to which the SLP–child interactions during language intervention reflect a warm and positive relationship, embodies a supportive and comfortable atmosphere, expresses enjoyment and excitement for therapy, and facilitates appropriate levels of autonomy (Hamre & Pianta, 2005; Girolametto, Weitzman, & Greenberg, 2006). Attachment, social interactionist, and self-determination theory posit that creating positive relationships, safe environments, and a sense of autonomy for children encourages more engaged and self-reliant behaviors (Ainsworth, 1978; Bowlby, 1969) and increases their motivation to learn (Connell & Wellborn, 1991; Ryan & Deci, 2000). Studies have found that classrooms, on average, have generally high emotional support (La Paro et al., 2009; Pianta, La Paro, & Hamre, 2008) and provide evidence for the importance of high-quality, emotionally-supportive interactions with children who are typically developing as well as those with LI. Regarding the former, Mashburn and colleagues (2008) examined the quality of emotional support within 671 pre-kindergarten classrooms serving mostly high-risk children. These researchers found that children who were taught in a more emotionally supportive environment were rated as having high levels of social competence and fewer problem behaviors compared to peers in less emotionally supportive settings. In a separate study of 910 first-grade students, Hamre and Pianta (2005) found that at-risk students in classrooms in which teachers provided high emotional support had achievement scores and student–teacher relationships in proportion with their peers who were considered low-risk. Other works have similarly shown that classrooms characterized by a high level of emotional support are associated with acceleration in children’s language development (Howes et al., 2008; NICHD ECCRN, 2002b, 2004), and early reading (Hamre and Pianta, 2005; NICHD ECCRN, 2004). These potentially positive effects of high quality emotional support may be particularly relevant to students with LI, who are prone to exhibit comorbid difficulties in socio-emotional functioning (Durkin & Conti-Ramsden, 2010; Sundheim & Voeller, 2004; Vallance, Cummings, & Humphries, 1998), academic achievement (Conti-Ramsden et al., 2009), and literacy (Cabell et al., 2010; Catts, Bridges, Little, & Tomblin, 2008; McCardle, Scarborough, & Catts, 2001). Although we are not aware of specific studies that have quantified the level of emotional support observed in SLP–child interactions, studies involving children with LI and their parents indicate that emotional support is an important dimension of these interactions. McGinty and Justice (2009), for instance, examined parent–child book-reading sessions involving mothers and their preschoolers with LI, finding that the quality of these interactions (as measured by maternal emotional and instructional support) was a significant predictor of children’s emergent literacy skills. Such work suggests that emotional support may be a crucial factor to consider when examining the quality of language intervention sessions for children with LI. 1.1.2. Instructional support Instructional support concerns the extent to which the interactions between the SLP and child promotes the child’s higher-order thinking and understanding, provides feedback that expands learning and understanding, encourages participation, and features language-facilitation techniques (Hamre & Pianta, 2005). Instructional support does not specify the content or specific type of language intervention activities used during therapy, but rather the ways the SLP implements the activities and provides feedback to promote both cognitive and language development. In mediated interactions with adults, the premise is that a child’s cognitive and language development is contingent upon the opportunities adults provide to express existing skills and scaffold more complex ones (Girolametto et al., 2012; Skibbe, Behnke, & Justice, 2004; Vygotsky & Cole, 1978). Empirically, a number of studies examining the quality of instructional support provided to children in classrooms show that these are predictive of gains in academic achievement (Bransford & Schwartz, 1999; Cameron et al., 2008; Hamre & Pianta, 2005). In addition, a study of first grade classrooms by Hamre and Pianta (2007) showed that the amount of high quality instructional support provided by teachers in kindergarten consistently predicted positive effects on the academic achievement in first grade of at risk students whose parents have low levels of education. Interestingly, the instructional support domain has typically been rated in the lower range of quality across empirically examined kindergarten (La Paro et al., 2009) and first grade classrooms (Pianta, Belsky, Vandergrift, Houts, & Morrison, 2008). 1.1.3. Proactive management Proactive management concerns the SLP–child interactions with respect to how effectively the SLP manages various aspects of the language intervention session, including student behavior, productive use of the session, organization of activities, and maintenance of student’s attention. Proactive management can be characterized as efficiently managing instructional time by using an appropriate pace and routines to align with children’s needs and progress. Researchers in the


G Model



field of psychology have theorized that organizational structure supports the development of self-regulatory skills (Blair & Razza, 2007; Paris & Paris, 2001; Raver, 2004) through high-quality management of time, behavior, and attention. Research on management of students’ behavior in classrooms indicates that teachers who provide clear, consistent behavioral expectations, positive, proactive strategies (Emmer & Stough, 2001) and have a more organized structure (Cameron et al., 2008) have less oppositional behavior and higher levels of engagement among their students. Large scale studies have shown that proactive management quality is generally moderate to high in general education classrooms (La Paro et al., 2009) and indicate that children in well-managed classrooms are more engaged and have higher literacy gains compared to children in poorly managed classrooms (McClelland, Acock, & Morrison, 2006; McClelland, Morrison, & Holmes, 2000; Ponitz, RimmKaufman, Grimm, & Curby, 2009). 1.2. Predictors of Intervention Quality As is the case with classrooms (Bransford & Schwartz, 1999; Cameron et al., 2008; Curby et al., 2011; Hamre & Pianta, 2005; McClelland et al., 2000, 2006; NICHD ECCRN, 2002b; Ponitz et al., 2009), we anticipate that language intervention provided to children may vary substantially with respect to the quality of emotional support, instructional support, and proactive management. For instance, some intervention sessions may be characterized by very high levels of instructional support, whereas others may be less so; research on classrooms has shown that such variability is the norm, rather than the exception (see Justice et al., 2008). Given this natural variability among intervention sessions, it is possible to explore whether variance in intervention quality might be associated with clinician characteristics. In other fields such as childcare, teaching, clinical psychology, and medicine, linkages between caregiver characteristics and the quality of intervention have been well documented (Guo et al., 2010; Mohr, Young, Meterko, Stolzmann, & White, 2011; Ponitz et al., 2009; Wampold, Lichtenberg, & Waehler, 2005). For instance, in the medical field, a systematic review of literature (Choudhry, Fletcher, & Soumerai, 2005) found that the number of years in practice among physicians is negatively associated with the quality of care received by patients. In education, teachers’ beliefs about the challenges they experience (Stuhlman & Pianta, 2009; Yoon, 2002), as well as their feelings of self-efficacy (Brandel and Frome, 2011; Guo et al., 2010), positively predict the quality of teacher–student relationships and the quality of instruction provided. However, in the SLP literature, few studies have examined the association between specific clinician characteristics, such as job satisfaction or stress, and intervention quality (but see Brandel and Frome, 2011; Ebert & Kohnert, 2010; Ratner, 2006). Drawing from literature on predictors of intervention quality from other fields, in the present study we investigated three SLP-level factors for their potential role in predicting intervention quality: self-efficacy, job satisfaction, and time pressure. 2. The current study This descriptive study seeks to provide a common framework for thinking about language intervention quality in an effort to advance the field of speech-language pathology with regards to SLP-level factors that can impact quality. In total, two aims were addressed: (a) to describe quality of SLP–child interactions during school-based language intervention provided to children with LI in the primary grades with respect to emotional support, instructional support, and proactive management, and (b) to determine if key SLP-level factors are predictors of variance in intervention quality. It is important to note that the examination of the quality of emotional support, instructional support, and proactive management provided during language intervention in this study is focused on all SLP–child interactions during a therapy session; put differently, this examination of quality concerns the global characteristics of a session irrespective of the skills targeted within therapy or the specific intervention techniques used by the SLPs during the session. 2.1. Method 2.1.1. Participants Participants were 40 SLPs and 174 students from 16 public school districts in the Midwestern and Mid-southern regions of the United States. The participants were from the first two cohorts of a three-cohort study (Speech-Therapy Experiences in Public Schools – STEPS), which was designed to examine characteristics of school-based language intervention for primary grade students with LI. The SLPs were recruited to participate in STEPS prior to the start of the academic year, and were included in the study based on (a) their consent to participate, (b) their anticipated caseloads (which needed to include a minimum of five students with LI in kindergarten and/or first grade, and (c) approval of the school district. The 40 SLPs had a minimum of a master’s degree and a state license in speech-language pathology, and 88% held a national certification from the American Speech Language Hearing Association (ASHA). Table 1 provides a demographic overview of the SLP sample. Students were recruited into the study following enrollment of SLPs. Specifically, SLPs identified up to five kindergarten and/or first grade students on their caseloads who currently had an individualized education program (IEP) for LI, and provided consent forms for participation to the student’s legal guardian. For those children for whom consent was provided, SLPs completed a brief questionnaire to provide the research team with information about the child, including age, gender, home language, primary and comorbid diagnoses, hearing screening information, and current level of functioning. Study investigators then used the information from the questionnaires to purposefully select up to five students per SLP to


G Model


5

Table 1 SLP demographics. Characteristic

n

%

Years of Experience Age Ethnicity White/Caucasian Black/African America Hispanic Gender Male Female Caseload size 21–40 41–60 61+

40 40

– –

34 5 1

85 12.5 2.5

2 38

5 95

6 21 13

15 52.5 32.5

M

SD

Min

Max

16.43 41.7

12.27 12.66

0 23

36 63

Table 2 Primary SLP-level factors of interest.

Self-efficacy Job Satisfaction Time pressure

M

SD

Min

Max

3.48 3.88 2.66

.53 .35 .70

2.47 2.92 1.60

4.63 4.75 4.20

participate in ongoing assessments and language intervention documentation, prioritizing enrollment of English-speaking children with a primary disorder of language and no significant comorbidities (e.g., hearing loss, autism). Consequently, children with primary LI (and no comorbidities) are over-represented in the sample (77% of participants). Students with LI and comorbid difficulties included those with phonological disorder (n = 2), autism (n = 12), hearing loss (n = 2), epilepsy (n = 1), attention deficit disorder (n = 10), traumatic brain injury in infancy (n = 1), possible drug exposure in utero (n = 1), and cognitive impairment (n = 8). Per these recruitment procedures, all of the 174 students (63% male) were receiving school-based speech-language services by a participating SLP during the academic year in a public school; students were five to eight years of age, with 43% of the students in kindergarten and 57% in the first grade. The students’ race/ethnicity varied, and was represented by caregivers as Caucasian (52.9%, n = 69), African American (14.9%, n = 26), Hispanic (4.6%, n = 8), Native American, Japanese, Chinese, (each .6%, n = 1) or other (4%, n = 7); this information was unavailable for 36 students. In the fall of the academic year, following consent and selection procedures, students were administered select subtests of the Clinical Evaluation of Language Fundamentals-4th Edition (CELF:4; Semel, Wiig, & Secord, 2003), namely Concepts and Following Directions, Word Structure, Recalling Sentences, and Formulated Sentences. The mean Core Language Composite for these students was 68.1 (SD = 17.11, range = 40–115). 2.1.2. General procedures Two principle procedures conducted as part of the STEPS study are of relevance to the present study: (a) collection of questionnaires from SLPs, and (b) collection and analyses of videos of students’ language intervention sessions. 2.1.2.1. SLP questionnaire. Prior to the start of the school year, the SLPs completed a questionnaire that captured a variety of information about their demographic background, self-efficacy, and two work-specific matters, namely job satisfaction and time pressures. Of relevance to this study were three sets of items we examined as primary SLP-level factors (see Table 2). First, SLPs completed a 19-item self-efficacy questionnaire, which sought the SLPs’ perspective on the amount of influence they felt that they had in their professional work; the questionnaire is adapted from Bandura’s (1997) self-efficacy scale used widely with school-based personnel (e.g., Guo et al., 2010). Adaptations included elimination of items not directly relevant to SLPs (e.g., self-efficacy questions regarding reducing student absenteeism) and rewording some questions so the focus was on SLPs instead of teachers. SLPs respond to a series of questions relating to self-efficacy (e.g., How much can you do to get through to the most difficult students?) that were rated on a scale of 1 (i.e., having no feeling of influence) to 5 (i.e., having a very strong feeling of influence). To arrive at a score for each SLP with respect to self-efficacy, we calculated the average response per SLP to the 19 questions. Second, SLPs completed a 17-item questionnaire related to job satisfaction and stress, based on an adaptation of a teacher satisfaction questionnaire used by the National Center for Education Statistics (NCES) School and Staffing Survey (NCES, 1994). Adaptations included rewording some questions so the focus was on SLPs instead of teachers. SLPs responded to a series of statements (e.g., I receive appropriate recognition for my efforts) on a scale of 1 (i.e., strongly disagree with statement) to 5 (i.e., strongly agree with statement).


G Model



Third, for our purposes, four items were extracted from these 17 (a = .80) to represent time pressure; these items examined perceptions regarding overall workload, participation in non-therapy tasks, and other demands on time (e.g., My workload as a SLP is too heavy, My job includes too many non-therapy tasks). Responses to these statements were averaged to create a measure of time pressure. The remaining 13 statements were averaged to create a measure of job satisfaction. 2.1.2.2. Language intervention session video collection and coding. SLPs submitted one to five video-taped intervention sessions across the academic year, capturing BAU intervention for participating children. SLPs were asked to capture representative, BAU intervention sessions; therefore, participating children could have been captured in the same intervention session. For the purposes of this study, we were interested in intervention sessions from two time points: 1 in the fall and 1 in the spring. A total of 208 sessions were collected from these two time points, representing an average of 5 videos (SD = 1.8) per SLP. SLPs were provided all materials to do so (e.g., video camera, tripod, recording media) and participated in an orientation to learn how to operate these effectively. Each session was coded using an observational coding system, the Classroom Assessment Scoring System (CLASS; Pianta, La Paro, et al., 2008), used extensively in the educational literature for documenting emotional support, instructional support, and proactive management (e.g., Burchinal et al., 2008; Guo et al., 2010; Hamre & Pianta, 2005; La Paro et al., 2009; Mashburn et al., 2008; Pianta, Belsky, et al., 2008). Assessment of classrooms using CLASS focuses exclusively on the nature of interactions between teachers and students or, in this case, of the nature of interactions between the SLP and student(s). Although the CLASS was designed for use in classroom settings, it has been shown to be valid in a variety of activity settings, group-size dynamics, and content areas (Pianta & Hamre, 2009; Pianta, Belsky, et al., 2008). It was thus seen as applicable to the intervention sessions observed, which included one-on-one pull-out therapy sessions (29%), small-group pull-out therapy sessions (70%), and large-group settings (1%). The CLASS consists of ten separate dimensions designed to measure three global domains of quality (emotional support, instructional support, and classroom organization, the latter of which we refer to as proactive management given that not all intervention sessions took place in a classroom setting). Dimensions reflecting the domain of emotional support include positive climate, sensitivity, and regard for student perspectives; the domain of instructional support includes dimensions of concept development, quality of feedback, and language modeling; and proactive management includes the dimensions of behavior management, productivity, and instructional learning formats. Dimensions scores, which span a continuum of quality that encompasses low (1, 2 points), medium (3, 4, 5 points), and high (6, 7 points) levels of quality, are averaged to create the three domain composites. It is important to note that CLASS codes, which are quantitative in nature, do correspond to these qualitative reference points of low, medium, and high levels of quality which are largely theoretically driven as they correspond to descriptions of teacher–child interactions that are observed as low quality (e.g., Low quality interactions in the emotional support domain reflect anger and irritation) compared to high quality (e.g., high quality emotional support reflects warm, responsive, and respectful interactions). There have been some efforts to empirically assess the validity of these qualitative categories (Burchinal, Vandergrift, Pianta, & Mashburn, 2010). Recent studies have suggested that in order for children to experience academic growth related to the quality of instructional support provided, there may be a threshold that must be met requiring quality scores above 3.25, with greater gains achieved with higher quality scores (Burchinal et al., 2010). Similarly, positive effects from emotional support that include positive social outcomes, may not be achieved if the quality of emotional supportive interactions are below the medium-high to high range with CLASS scores of 5–7. To conduct CLASS scoring, an individual must complete a comprehensive training program (usually two full days) that involves working toward achieving a reliability criterion across all domains. For the present study, all coders completed a training course on the CLASS system delivered by a CLASS-certified trainer and passed the standard benchmark to be considered a CLASS-reliable coder, namely 80% scoring agreement with five master-coded observations (calculated as total agreement across all 10 constructs across all five observations) and agreement on any particular CLASS dimension on at least two out of the five master-coded observations. Agreement is defined as a score within one point of the master code. CLASSreliable coders subsequently participated in bi-monthly drift meetings. The purpose of the drift meetings was to ensure ongoing discussions regarding coding, and to reduce potential threats to inter-rater reliability. Reliability of the coding procedures was adequate, as based on a randomly selected double-coding of 10% of the SLP videos; overall inter-rater agreement was 87%. 2.2. Results 2.2.1. Quality of language intervention The first aim of this study was descriptive, such that we sought to describe the quality of language intervention as experienced by students with LI. A total of 208 intervention sessions were coded using the CLASS (Pianta, Belsky, et al., 2008), which provides an index of quality that is strongly substantiated in the literature as predictive of students’ growth over time (Burchinal et al., 2008; Guo et al., 2010; Hamre & Pianta, 2005; La Paro et al., 2009; Mashburn et al., 2008). As noted previously, each SLP had an average of five sessions coded across two observation points. CLASS scores are based on a scale from 1 to 7, with scores of 1–2 indicating low quality; 3, 4, or 5 indicating mid-range of quality; and 6 or 7 indicating high quality. Table 3 illustrates that the quality of intervention was generally moderate to high with respect to emotional support (M = 5.42, SD = .56, range = 3.75–6.75) and proactive management (M = 5.3; SD = .72, range = 3.00–7.00), and low to moderate


G Model


7

Table 3 CLASS domain quality descriptives (n = 208). CLASS Domain Emotional support Low quality (1–2) Medium quality (3–5) High quality (6–7) Instructional support Low quality (1–2) Medium quality (3–5) High quality (6–7) Proactive management Low quality (1–2) Medium quality (3–5) High quality (6–7)

n

%

0 94 115

0 45 55

35 166 7

17 80 3

0 127 81

0 61 39

M

SD

Min

Max

3.75

6.75

5.42

.56

3.49

1.03

1

6.00

5.29

.72

3

7.00

Note. For percentage of low, medium, and high quality scores of 1–2.4 were included in low scores of 2.5–5.4 were included in medium quality, and scores of 5.5–7 were included in high quality.

in terms of instructional support (M = 3.50; SD = 3.49, range = 1.00–6.00). Of further note is that for each domain, considerable variability was observed. Nearly the full range of scores was seen for the instructional support domain, which is similar to findings in large scale studies of teacher–child interactions in kindergarten and first grade classrooms (La Paro et al., 2009; Pianta, Belsky, et al., 2008). Some children participated in sessions coded as 1 (low quality) for instructional support, whereas others participated in sessions coded as 6 (high quality). The range observed for the other two domains (emotional support, proactive management) also varied from relatively low to relatively high, indicating that children’s experiences in therapy, at least with respect to SLP–child interactions, was variable. Table 3 presents data concerning the number and percentage of sessions that may be characterized as low, medium, and high quality across the domains. 2.2.2. Predictors of quality As each SLP was observed multiple times, it is likely that some variance in the repeated observations of CLASS scores was SLP-specific. Hierarchical linear modeling (HLM) was used to partition the variance within each CLASS domain into that attributable to differences between SLPs (SLP specific variance), as opposed to differences specific to a particular session. SLP specific variance (or between-SLP variance) refers to characteristics of a particular SLP that influence scores on the three CLASS domains. This is as opposed to within-SLP variance (session specific variance), which may thus reflect the time of day, number of children, or location of the therapy session. The HLM results varied depending on the three CLASS domains assessing intervention quality. For emotional support, 37% of the observed variance in observations was SLP-specific, leaving 63% attributable to within-SLP or session-specific factors. For instructional support, 45% of the variance between intervention sessions was SLP-specific, leaving 55% due to session-specific factors. Finally, proactive management showed 18% of the variance was due to SLP (82% session-specific). The comparably large SLP-specific correlations for emotional and instruction support suggest that regardless of the therapy session, individual SLPs were providing a similar level of emotional and instructional support to all students irrespective of session-specific factors. The opposite is true for proactive management: 82% of the variance was due to session-specific factors such as the size of the group, time of day, who the children were in the group, or the topic being covered that day. 2.2.2.1. SLP-level factors. Pursuant to the second aim of this work, we sought to further explain the variability in intervention quality by determining whether and to what extent three SLP characteristics may predict intervention quality. To do so, three SLP-level factors were added to the HLM models: self-efficacy; job satisfaction; and time pressure (see Table 4). SLP selfefficacy was not found to be a significant predictor for any of the quality domains. However, this was not the case for job satisfaction and time pressure. For job satisfaction, there was a significant and positive relationship with intervention quality specific to instructional support (Estimate = .70, p = .03) and proactive management (Estimate = .42, p = .02). This suggests that SLPs who reported higher levels of job satisfaction provided higher quality instructional support and higher levels of proactive management. For time pressure, there was a significant and negative relationship with intervention quality for all three domains, indicating that SLPs who reported higher levels of stress related to time pressure provided lower quality emotional support (Estimate = .32, p =

Dual language intervention for bilinguals at risk for language impairment.

Characteristics of children with specific language impairment attending language units.

No population bias to left-hemisphere language in 4-year-olds with language impairment.

Specific Language Impairment Across Languages.

Language impairment in cerebellar ataxia.

Association of mandated language access programming and quality of care provided by mental health agencies.

Impacts of parent-implemented early-literacy intervention for Spanish-speaking children with language impairment.

Stability of Language and Literacy Profiles of Children With Language Impairment in the Public Schools.

The relationship between mathematics and language: academic implications for children with specific language impairment and English language learners.

Executive functions in mono- and bilingual children with language impairment - issues for speech-language pathology.

Language barriers: challenges to quality healthcare.

Language impairment and dyslexia genes influence language skills in children with autism spectrum disorders.

Balanced intervention for adolescents and adults with language impairment: a clinical framework.

Deficits in narrative abilities in child British Sign Language users with specific language impairment.

Music Perception Influences Language Acquisition: Melodic and Rhythmic-Melodic Perception in Children with Specific Language Impairment.

Examining the Language Phenotype in Children With Typical Development, Specific Language Impairment, and Fragile X Syndrome.

Language Proficiency and Sustained Attention in Monolingual and Bilingual Children with and without Language Impairment.

How Language Is Embodied in Bilinguals and Children with Specific Language Impairment.

Pronominalization in the language of deaf students.

Associations of HLA alleles with specific language impairment.

Language impairment from 4 to 12 years: prediction and etiology.

In search of a language intervention lexicon.

Processing Binding Relations in Specific Language Impairment.

Language impairment in bilingual children: from theory to practice.