Abstract
Using a multiple baseline single-subject design, this study evaluated the effects of a text structure intervention package on the ability of students with autism to comprehend traditional science texts. Three high school students with high-functioning autism and their teacher participated in this study. The intervention package included instruction in types of text structures using a text structure organization sheet before reading, and completing an analysis and summary sheet during and after reading. Results indicated that the instruction was highly effective during intervention and maintenance phase for all three participants. The first-year special education teacher was able to implement the intervention with fidelity. All participants agreed that the intervention was helpful for reading science texts. Future research and implications for classroom intervention is discussed.
Prevalence estimates suggest 1 in 68 children have an autism spectrum disorder (ASD; Centers for Disease Control and Prevention, 2014), making the disorder one of the fastest growing developmental disabilities facing educators. Of particular concern are secondary students, for whom instruction designed to improve post-secondary outcomes is a critical need (Odom, Duda, Kucharczyk, Cox, & Stabel, 2014). Like all individuals with ASD, these high school students have differences in the areas of social-communication, and restricted interests and/or repetitive behavior (American Psychology Association, 2013). These differences often manifest as challenges with social relationships, intense interests in specific subjects or topics, and challenges with organization and attention, which can negatively affect high school success, especially in academic content areas.
In high schools, the Common Core State Standards (CCSS) require teachers to provide reading instruction in content texts that continually increase in complexity over time (National Governors Association for Best Practices & Council of Chief State School Officers [NGA & CCSSO], 2010). This underlies the expectation that as students progress in their academic careers, reading is one way to learn (Israel, Maynard, & Williamson, 2013). For students to learn while reading academic content, they must be able to comprehend content-area texts. For students with ASD, both the text and the reader influence reading comprehension (Williamson, Carnahan, & Jacobs, 2012). Individuals with ASD demonstrate cognitive characteristics that may influence text comprehension and comprehension strategy learning (Carnahan & Williamson, 2010; Carnahan, Williamson, & Christman, 2011). Although differences vary, many learners with ASD need explicit instruction to comprehend the increasingly complex texts they face across their academic careers.
To date, few studies focus solely on reading comprehension strategies for students with autism (cf. Chiang & Lin, 2007; Whalon, Al Otaiba, & Delano, 2009), and even fewer target expository text comprehension in specific content areas (Carnahan & Williamson, 2013). The purpose of this article is to describe the results of a text pattern intervention designed to support expository reading comprehension, specifically science text comprehension, for three high school students with ASD.
Expository Text Comprehension for Learners With ASD
Comprehending Expository Science Text
Expository texts incorporate a variety of different structures to convey meaning. Commonly used expository text structures include description, compare–contrast, cause and effective, and sequence (Zwiers, 2010). Figure 1 contains a list of common expository text structures and the associated features. Each expository text structure has a specific purpose supported by unique language or signal words, and graphic organizers commonly used to support comprehension of the structure. For example, authors use the compare–contrast structure to tell about how two things are similar or different (Carnahan & Williamson, 2013). The structure is marked by signal words such as alike, different, and both, and the graphic organizer associated with compare–contrast is different from the organizer associated with description or cause–effect. Readers need to understand these different structures and have background knowledge, or knowledge about the passage topic for comprehension to occur.
Text structure organization guide.
At times, individuals rely more heavily on either their understanding of the text structure or on their science background knowledge for comprehension (Goldman & Rakestraw, 2000). For example, if an individual has a high level of background knowledge about a topic but limited understanding of the text structure, they may be able to comprehend the text. However, when both background and text structure knowledge are limited, comprehension diminishes. For students with ASD, knowledge of text structures, including the associated graphic organizers and signal words, may mediate common comprehension difficulties, such as integrating relevant background knowledge with the text (Carnahan & Williamson, 2013).
Why is expository text comprehension challenging?
The challenge of understanding expository science texts relates to both the nature of the content and text structure (Carnahan & Williamson, 2015). Academic content frequently includes complex, abstract topics. Spiral content curricula provide students deeper understandings through multiple experiences with the same concept. For some students with ASD, however, emerging evidence suggests that their cognitive reading profile may limit the extent to which they access or integrate background knowledge (Wahlberg & Magliano, 2004; Williamson et al., 2012). Thus, although students with ASD may have previous exposure to content, exposure may not facilitate comprehension. Both the abstract nature of the science curriculum and challenges integrating previously learned knowledge make comprehending expository text challenging.
Another compounding issue is that most science texts are structurally dense. Unlike narrative texts that have consistent text structures (i.e., story grammars), expository texts typically contain several different text patterns in close proximity. Signal words (e.g., because signals cause and effect), or language clues, point to text patterns, which clarify the author’s purpose (i.e., what does the author want me to learn about this concept; Meyer & Poon, 2001; Zwiers, 2010). To be skilled at expository science texts, students must identify key words, especially when they have limited background knowledge. These signal words support expository text comprehension by helping readers “mentally examine how ideas in text are interrelated” (Meyer & Ray, 2011, p. 128).
Explicit text pattern instruction has been effective for improving reading comprehension of students with learning disabilities (Berkeley, Scruggs, & Mastropieri, 2010). Additional research has shown the effectiveness of cause and effect text pattern instruction for second-grade struggling readers (Williams et al., 2007). Compare–contrast pattern instruction has been shown effective for second-grade struggling readers (Williams, Stafford, Lauer, Hall, & Pollini, 2009), and for improving expository science text comprehension for students with ASD (Carnahan & Williamson, 2013).
Reading Comprehension in ASD
Individuals with ASD demonstrate cognitive processing differences that influence academic reading. Given the cognitively intensive nature of reading, these differences influence reading comprehension (Brown, Oram-Cardy, & Johnson, 2013; Carnahan et al., 2011; Williamson et al., 2012) and academic achievement (Charman et al., 2011). Two studies provided important insight into characteristics that may influence reading comprehension for learners with ASD. Brown and her colleagues (2012) conducted a meta-analysis of 47 studies related to comprehension for learners with ASD. The analysis suggested that while variability exists, having autism alone did not predict reading comprehension challenges. Rather, it was an intersection of the influence of both individual characteristics and task demands. Semantic knowledge (i.e., word/vocabulary knowledge and the concepts associated with these words) and decoding skills best predicted reading comprehension, but other factors such as performance IQ and social knowledge also influenced comprehension. Brown et al. (2013) hypothesized that semantic knowledge and decoding skills might be related to language differences, a hallmark of ASD.
Williamson and her colleagues (2012) used grounded theory to understand how individuals with ASD made meaning from text. This resulted in the description of three reading comprehension profiles that may help explain the variability across the autism spectrum. The profiles suggested a continuum of characteristics between text bound and imaginative, with strategic comprehenders in the middle. Individuals in the text-bound (i.e., concrete) profile had difficulty moving beyond the words on the page to integrate prior knowledge. Text-bound learners exhibited syntactic and semantic language differences, underdeveloped conceptual knowledge, over selective attention, and challenges with social understanding. Similar to learners in the text-bound profile, imaginative comprehenders demonstrated semantic language differences and underdeveloped conceptual knowledge. Unlike learners in the text-bound profile, learners in the imaginative group frequently answered comprehension questions based solely on experience. The imaginative profile was characterized by overreliance on unrelated background knowledge and previous experiences. However, text type (narrative or expository) did not seem to influence comprehension for learners in either profile. Finally, although learners in the strategic profile demonstrated language differences such as unique word usage, these differences rarely altered text meaning. In addition, these learners had well-developed conceptual knowledge and social understanding, and frequently applied strategies, such as questioning the text, to support comprehension.
Influence of genre on comprehension among individuals with ASD
Recent research suggests that students with ASD differ in their abilities to comprehend narrative and expository texts. Williamson and her colleagues (2012) found that some individuals comprehended expository text better than narrative, while others comprehended narrative text better than expository. Brown and her colleagues (2012) found that for individuals with limited social knowledge, text requiring social understanding was more challenging. Thus, these studies suggest that while variability exists, individuals with ASD struggle with reading comprehension. Specific individual differences that influence reading comprehension include language differences (e.g., vocabulary), conceptual knowledge (e.g., schema related to the text), and the ability to integrate relevant background knowledge (e.g., combine existing knowledge with new information contained in the text), social knowledge, and executive function (e.g., selection of relevant existing knowledge; generating the gist of the story in spite of the details).
Interventions to Support Expository Text Comprehension
A review of the literature on expository text instruction including content enhancement strategies (e.g., graphic organizers) and cognitive strategy instruction (e.g., text structure) for individuals with learning disabilities in Grades 4 through 12, found generally large effects on comprehension (Gajria, Jitendra, Sood, & Sacks, 2007). Teaching students, including those with learning disabilities, to comprehend text structure is an effective intervention beginning in the primary grades (Meyer & Ray, 2011; Williams, 2005; Williams et al., 2009). Furthermore, Jitendra, Burgess, and Gajria (2011) noted that text structure instruction is an evidence-based cognitive strategy for learners with learning disabilities. Meyer and Ray (2011) described critical components of text structure instruction, especially for younger children or students with less sophisticated reading comprehension abilities. These critical components included (a) careful matching texts with learner reading levels, (b) support and feedback during skill acquisition, and (c) gradual increase of text complexity.
Reading comprehension interventions for learners with ASD
Two literature reviews describing reading interventions for students with ASD were published in 2007 and 2009 (Chiang & Lin, 2007; Whalon et al., 2009). Effective reading comprehension strategies included anaphoric cueing (e.g., proving nouns to match pronouns found in the text; O’Connor & Klein, 2004) and reciprocal questioning (i.e., generating and answering questions about the text with teacher-like roles; Whalon & Hanline, 2008). Additional studies have emerged supporting specific reading comprehension strategies such as repeated reading and question generation (Hua et al., 2012), story mapping (Stringfield, Luscre, & Gast, 2011), and teaching compare–contrast text structures (Carnahan & Williamson, 2013). Of these, only one (Carnahan & Williamson, 2013) specifically addressed content area (i.e., science) expository text comprehension.
Carnahan and Williamson (2013) conducted a study with three individuals with ASD and their teacher. The intervention included explicitly teaching students to recognize and comprehend the compare/contrast text pattern in science texts. The intervention proved effective or highly effective and was deemed socially valid. Although important, the Carnahan and Williamson study was limited to one text pattern in tightly controlled passages. Thus, the purpose of the current study was to determine whether a text pattern intervention could facilitate reading comprehension of naturalistic (i.e., textbooks or off-the-shelf) science books. Specifically, the current study sought to address the following research question:
Method
Participants and Setting
A university review board approved this study, and three male students with ASD and their teacher participated. Although all three students were described by school personnel as having “high-functioning” autism, they each had a medical diagnosis of autism provided by the local children’s hospital and received school services under the Individuals With Disabilities Education Improvement Act (IDEA; 2004) educational identification of autism. In addition, the students all comprehended text at the fifth-grade level or higher, and provided verbal or written responses to questions. In addition to receiving pull out services, two of the students participated in general education courses, including science and social studies/history. All of the students had exposure to expository text; the two students included for academics read expository text daily as part of their content-area instruction. Neither the special education teacher nor content-area teachers provided explicit expository text instruction or strategies for increasing comprehension of expository text.
This study took place in the spring semester of the school year during a portion of the students’ pull out language arts instruction. Before beginning the study, the Qualitative Reading Inventory–5 (QRI-5; Leslie & Caldwell, 2010), including the look back procedure, was used to assess each student’s reading levels, and students’ records were reviewed to gather information about standard language scores and IQ. What follows is a description of the student participants, Andy, Ryan, and Sam, based on available information.
Andy was 16 years old at the time of the study and received academic instruction in a self-contained classroom setting. Related services included speech, occupational, and physical therapy. He received a medical diagnosis of autism at the age of 4 from a diagnostic team from a large children’s hospital in the region. According to the Stanford–Binet Intelligence Scale, 5th edition, administered in 2009, his full scale IQ was 76. Using the Clinical Evaluation of Language Fundamentals (CELF-4; Semel, Wiig, & Secord, 2003), his standard score was 91 (27th%) on the Core Language measure and 80 (9th%) on the Expressive Language measure. Based on the QRI-5 (Leslie & Caldwell, 2010), he independently comprehended text at the fifth-grade level, but only decoded text at a fourth-grade level.
Ryan was 15 years old at the time of the study. He received academic instruction in three general education classes (science, social studies, and math) and one self-contained/resource room class (language arts). Ryan was diagnosed with autism in the fourth grade at a large children’s hospital in the region. Although no formal IQ scores were available, using the CELF-4 (Semel et al., 2003), Ryan’s Core Language score was 60 (.4%), and his expressive language score was 47 (<.1%). He was able to decode text at the upper middle school level, but comprehended text independently only at the fifth-grade level based on the QRI-5 (Leslie & Caldwell, 2010). Ryan received weekly services from a speech language pathologist, occupational therapist, and physical therapist.
Sam was 16 years old at the time of the study. He received a medical diagnosis of autism when he was 3 years old from a diagnostic team at a large children’s hospital in the region. Sam was included for science and social studies, but received math and language arts instruction in a self-contained/resource room setting. Based on a record review, no recent formal IQ scores were available. However, his standard score was 75 (5th%) on the Core Language measure and 51 (.10th%) on the Expressive Language measure of the CELF-4 (Semel et al., 2003). Using the QRI-5 (Leslie & Caldwell, 2010), he comprehended and decoded text at the upper middle school level. He received weekly services from a speech language pathologist, occupational therapist, and physical therapist.
Materials
Two different types of science text were used in this study. Sam, who comprehended at an upper middle level, read a high school science text used in his grade level. Andy and Ryan, who independently comprehended fifth-grade text, read science trade books written at the late sixth-grade level. One passage from each text was selected for each session in the study. Passage length, number of text patterns, and reading level were consistent across all passages. Passages were three to four pages in length, including figures and tables, and contained no more than four different text patterns. Readings addressed topics students had not read about or discussed during the 2011–2012 school year. Table 1 contains a list of the science topics addressed in the passages used.
Science Topics Addressed.
To maintain consistency across all sessions in the study, a series of 10 comprehension questions was developed for each passage. The question types included 5 explicit questions and 5 inferential questions. Of the explicit questions, 2 were definition questions (e.g., What are the two properties of matter?), and 3 questions were based on the text patterns used in the passages (e.g., describe two properties of ionic compounds.) The inferential questions included a main idea question (e.g., What is one main idea you read about?), two drawing conclusions questions (e.g., Why is an object’s weight the same everywhere on earth?), and two application questions (e.g., Why can’t you blow a square bubble?). After one author wrote questions for each passage, another author reviewed the questions to ensure they followed the question type outline to ensure consistency across all phases of the study.
In addition to the passages and questions used during baseline, the intervention stage included a text pattern organization guide, and a text analysis and summary sheet. Figures 1 and 2 contain the text structure organization guide, and the text analysis and summary sheet. The second author adapted the text structure organization guide from Zwiers (2010) to include only the four text patterns found in the passages (i.e., description, cause/effect, compare/contrast, and sequence), the purpose of the text patterns (i.e., to explain an idea or thing), signal words associated with the text patterns (e.g., is, are, consists of, for instance), and a picture of a graphic organizer associated with each pattern. The text analysis form was developed by the second author and contained five questions. The first question was a prediction, followed by three questions related to the text pattern, and the final question was a summary question that required students to use the text pattern graphic organizers. Questions were developed to draw students’ attention to both the text pattern and, importantly, the content contained in the passages (i.e., prediction and summary questions).
Text analysis and summary form.
Dependent Variable
The dependent variable in this study was students’ ability to comprehend the science texts measured by answers to the questions written for each section of text. Following procedures outlined in the QRI-5 (Leslie & Caldwell, 2010) and in other reading comprehension research for students with ASD (Carnahan & Williamson, 2013; Smith-Myles et al., 2002), students were given the opportunity to look back in the text to correct any incorrect responses. That is, after students responded to the questions, the adult identified any incorrect responses, read the question to the student, and then asked the student to look back in the text before answering. Allowing students to look back in the text mediates issues related to memory, thus allowing a true measure of understanding as opposed to measuring students’ recall abilities. In addition, allowing students to look back in a text often lessens the role of prior knowledge (Ozuru, Best, Bell, Witherspoon, & McNamara, 2007). That is, looking back decreases reliance on prior knowledge, which according to Williamson and her colleagues (2012) is challenging for many individuals with ASD because of either over or under reliance on relevant background knowledge. Thus, allowing students to look back in a text may increase their ability to apply text structure understanding and reading comprehension strategies to increase their ability to answer comprehension questions about text. In addition, the use of look backs approximates other testing procedures that allow access to text during question answering (e.g., standardized state assessments). The percentage of questions reported correct is based on responses after look backs.
Independent Variable
A multicomponent text structure intervention designed to teach students to understand text patterns was the independent variable in this study. The text structure intervention included instruction in the different types of text patterns using the text structure organization sheet before reading, and completing the text analysis and summary sheet during and after reading.
Design
A multiple baseline design was used to evaluate the effects of the text structure intervention package on students’ comprehension of science passages. Student order was randomly assigned by drawing names, and phase changes were dictated by three data points of improvement in the treatment phase compared with no baseline upward trend from the remaining untreated students. Follow-up data were collected 2 weeks after the end of the intervention.
Procedures
Extant literature suggests texts for learning should be matched to students’ reading levels (see Meyer & Ray, 2011). We used two science texts written at different levels, and the teacher taught students in two different groups (i.e., high school and middle school). Andy and Ryan both comprehended text at the late fifth-grade level and were placed in a group together. Sam was placed in his own group because his reading comprehension level (i.e., late middle school) was above the other students. Students read texts slightly above their independent levels during all phases of the study. However, baseline and intervention procedures were consistent for all students. The teacher handed the text to students and reviewed the written activity schedule as the first two steps in all study phases.
Baseline
After reviewing the activity schedule, the teacher asked students to review the title of the passage and then read aloud from the text during baseline. Sam read the entire passage aloud because there were no other students in his group. Andy and Ryan alternated pages when reading the passages aloud. The teacher asked the students to verbally summarize the content contained in the text at the end of each page before asking the other student to read. At the end of the passage, she asked the students to summarize the entire passage.
After the summaries, students individually answered the series of 10 questions about the passages verbally while the teacher or researcher scribed their responses. Each set of questions was printed using 14-point font. Students received a copy of the questions to view as either a teacher or researcher read the questions aloud; one teacher or researcher worked with one student during the question/answer portion of each session. At the end of the question/answer segment, students were given an opportunity to look back in the text to correct any incorrect or unanswered questions (Leslie & Caldwell, 2010). The researchers scored the questions at the end of each session, marking the total number of questions correct after look backs.
Intervention
Each student received a text structure organization guide, and text pattern analysis and summary form, in addition to the reading passage. Similar to baseline, the teacher started the lesson by reading the title and reviewing the activity schedule. Table 2 contains the specific intervention procedures. After reviewing the schedule, the teacher directed students’ attention to the text pattern organization guide. Students verbally reviewed each of the text patterns by discussing the purpose, features, and signal words for each pattern. Then, students and the teacher discussed the corresponding graphic organizer, including how the graphic organizer could be used as a thinking tool.
Task Analysis of Text Pattern Intervention Lesson.
After reviewing the text pattern organization form, participants read the title of the passage and completed the first part of the text structure analysis and summary form shown in Figure 2. The teacher asked the students, “After looking at the title and any pictures, what do you think this passage will be about?” The students made a prediction, and the teacher then directed them to begin reading. Similar to baseline, Sam was in a group by himself and read the entire passage aloud to the teacher. Ryan entered intervention after Sam. He also read the entire passage aloud to the teacher until Andy entered intervention. When both Andy and Ryan were in intervention, they took turns reading a section of the text aloud. At the end of each section, students identified signal words used in the passage and completed the second part of the text analysis and summary form. If students provided an incorrect answer for either of the first two questions, the teacher asked them to again look at the signal words. Finally, to complete the summary, the teacher quickly drew the corresponding graphic organizer, as students provided verbal information for her to write. After completing these procedures for each passage section, students reviewed the graphic organizers, verbally summarizing the ideas presented in the text. After completing the intervention, each student answered 10 comprehension questions using the same procedures described for baseline.
Maintenance
The maintenance phase of this study was designed to assess whether the intervention continued to influence student comprehension as measured by their responses to a series of 10 comprehension questions. It took place 2 weeks after the final intervention session. Maintenance strictly followed the procedures outlined for the intervention phases.
Measures
Student outcome measures
Student responses to a series of 10 comprehension questions (see “Materials” section for details) were the primary measure of intervention effectiveness in this study. The teacher and researchers reviewed all student question responses, agreeing on the total number correct for each participant. The researchers recorded the total number of questions answered correctly after look backs at the top of question pages at the end of each session.
Reliability
Point-by-point reliability was used to calculate interobserver agreement for the quizzes. A second observer was trained prior to the study to independently collect data on participant responses. Reliability data for quizzes were taken at least once per condition, per participant, for a minimum of 20% of sessions. For quiz reliability recording, the rater sat beside the teacher and both independently recorded the quiz responses. Percentage agreement was calculated by dividing the number correct by the number correct plus incorrect multiplied by 100% (Ayres & Gast, 2010). Reliability for quizzes was 100% across all participants and all conditions, eliminating the threat of instrumentation.
Intervention fidelity
One researcher collected intervention fidelity data on 100% of all sessions, and a second researcher was present for at least 40% of these sessions. Data were collected using checklists during baseline, intervention, and maintenance. Checklists addressed the specific steps of each phase from reviewing the schedule and reading the title to the verbal summary, which was the last step of the intervention, and question answering. The teacher implemented the intervention with 100% accuracy, and the interobserver agreement for intervention fidelity was 100%.
Social validity measures
A social validity questionnaire, adapted from Teacher Post-Intervention Acceptability and Importance of Effects Survey (Lane & Beebe-Frankenberger, 2004), was created for the student and teacher participants. The student questionnaire included four questions in which students rated the importance of the skills taught, the role of the intervention in their understanding of what they read, their ability to discuss the topic, and their ability to answer questions about the topic on a Likert-type scale. The survey also included two open-ended questions in which students were asked what was beneficial or what they would change regarding the intervention. The teacher’s questionnaire included six questions rated on a Likert-type scale and two open-ended questions. The teacher questions addressed issues of importance, appropriateness, outcome, and future use.
Data Analysis
Visual analysis, including attention to trend, level, and stability, was used to evaluate the effectiveness of the text structure intervention on reading comprehension (Gast & Spriggs, 2010; Horner et al., 2005). To supplement the visual analysis, the Conservative Dual Criterion (CDC) method was used to help affirm visual analysis decisions by considering both trend and level (Fisher, Kelley, & Lomas, 2003; Swoboda, Kratochwill, & Levin, 2010). Finally, the analysis addressed immediacy of effect, calculated by determining the magnitude of change between baseline and intervention, and calculated percentage of non-overlapping data (PND) using guidelines outlined by Scruggs, Mastropieri, and Casto (1987).
Results
Comprehension Questions
Data in Figure 3 demonstrate the results for all three students across baseline, treatment, and maintenance phases. During baseline, Sam correctly answered between four and five questions for each passage, with an average comprehension level of 42%. Andy and Ryan correctly answered between four and six questions, with Ryan averaging 54% and Andy averaging 51%. All of the students demonstrated increased levels of comprehension during the intervention phase, and maintained the increase during follow-up.
Comprehension question scores.
Sam was the first student to enter intervention. Sam answered an average of 88% of questions during intervention and 95% during follow-up, with an average of 89% across both intervention conditions. After a stable baseline with no trend, an immediate change in level occurred when the intervention was introduced, with an increase from 42% to 88% questions answered correctly on average. Upon introduction of the intervention, the stability of the data continued and a slight upward trend emerged.
Ryan entered intervention second. He averaged 90% correct responses during intervention and 95% during follow-up, with a cumulative average of 91% correct responses across all intervention conditions. Similar to Sam, after a stable baseline with no trend, there was an abrupt change at intervention from 54% to 90% correct responses on average. Data were mostly stable during the intervention phase, and a very slight upward trend was observed.
The third participant to enter intervention was Andy. He averaged 97% correct on the comprehension questions during intervention and 95% during follow-up, with a cumulative average of 96% across intervention conditions. Andy also demonstrated an abrupt change from baseline to intervention after a stable baseline with no trend, with an increase from 51% to 97% correct responses on average. The data were very stable during intervention and showed no trend.
Figure 4 shows the results with visual aids added from the CDC method for the intervention phase. For the data to support a significant treatment effect on test scores, a significant portion of the treatment data points should exceed both lines drawn on the graph, representing conservatively adjusted level and trend lines. Specifically, for Sam, Ryan, and Andy, 7, 6, and 3 of the intervention data points should exceed both lines, respectively. These criteria were met in this analysis, supporting the assertions of the visual analysis that student performance was considerably higher in the treatment phases than in the baseline.
Comprehension question scores with CDC.
As a final affirmation of the results, the PND was calculated and was 100% for all of the students. This percentage indicates the intervention was very effective and corroborates the results from visual analysis and the CDC method.
Social Validity
We measured intervention social validity using a teacher and student questionnaire. Both the teacher and the students rated the questions high (agree or strongly agree on all questions by all participants) indicating the intervention met the social validity criteria (i.e., acceptable, feasible, effective, and likely to be continued) outlined by Horner et al. (2005). When asked about changes they would make to the intervention, all student participants indicated they would “just leave it the same.” The teacher described the intervention as extremely helpful, and suggested it helped her learn a great deal about her individual students’ reading comprehension skills.
Discussion
At a time when the need to expand the pipeline of students entering professions in science, technology, engineering, and math (STEM) is clear (National Science Foundation, 2011), students with disabilities of all kinds, including students with ASD, underperform in their STEM coursework (AccessSTEM, 2007). One possible explanation for this underperformance may be the lack of embedded reading instruction during science instruction in content-area classrooms (Israel et al., 2013). Reading is one important way to learn science content.
The purpose of this study was to extend the work of Carnahan and Williamson (2013) by evaluating the effectiveness of a text structure intervention package using typical science texts, rather than modified, exemplar passages. In the original study, Carnahan and Williamson followed recommendations from the literature by developing “highly controlled passages” that presented one text pattern. Students were taught to identify key words signaling a comparison and complete a Venn diagram to summarize content. Like the Carnahan and Williamson study, students in this study were able to access science texts with high levels of comprehension. Unlike the Carnahan and Williamson study, and perhaps most importantly as it relates to STEM literacy, students in this study demonstrated comprehension, as measured through their responses to the comprehension questions, of complex science texts (i.e., high school, upper middle school) written one or two levels above their instructional reading levels. Furthermore, a first-year special education teacher was able to implement the intervention with a high degree of integrity (i.e., 100%).
Limitations
As with all studies, limitations exist. First, there were only three students in this study and the study occurred in a special education resource room, which limits generalizability. Although student participants were randomly assigned participation order, randomization occurred informally rather than through a randomization table. Second, although measure (i.e., using structured question stems and a review by a literacy expert) was taken to ensure consistency across questions, only one author reviewed the questions. In the future, review by two authors would be beneficial. Third, maintenance sessions occurred only 2 weeks after the conclusion of the intervention. Although this was necessary because of the school calendar, it limited opportunities to address treatment effectiveness over time. Finally, although the three participants had not previously been taught about the topics addressed in the study during the school year in which the study occurred, we did not assess their background knowledge before reading each passage.
Implications for Practice
Previous research demonstrated that text pattern instruction is an effective strategy for increasing comprehension in students without disabilities (Williams et al., 2009), students with learning disabilities (Berkeley et al., 2010), and to a limited extent, students with ASD (Carnahan & Williamson, 2013). In this study, we extended the research on text pattern instruction to high school students with ASD, providing important insight for both practice and future research.
For practice, Table 2 outlines the specific procedures for this study, including the language the teacher used to support students in completing the text structure analysis form and the graphic organizer. The teacher used analytical questions (e.g., What is the author’s purpose? How do you know?) to guide students’ thinking. As illustrated in Table 2, these questions were systematically faded from the beginning to the end of the study, which was critical given that without explicit direction toward the gist of the text (i.e., prediction and summary), students with ASD may over attend to text pattern (e.g., locating key words) or over rely on existing background knowledge and experiences. Such teacher language is perhaps the first critical step in developing students’ abilities to independently access texts with increasing complexity (Culatta, Blank, & Black, 2010; Culatta, Hall-Kenyon, & Black, 2010).
Implications for Research
Israel and her colleagues (2013) argued for the need to embed authentic literacy into STEM instruction for all learners. Although a first-year special education teacher delivered this intervention in a resource room, both special and general educators could use this strategy to effectively teach groups of students with and without ASD about text structures in a variety of settings. However, given the critical role of language and fading played in the intervention, teachers may need additional information beyond a task analysis to apply this strategy in their own classrooms to meet the needs of individual students. Future research should address the effectiveness of the text structure instruction in larger populations of students with ASD, and in a variety of settings. Moreover, it will be important for researchers to evaluate the level of support educators need to implement the strategy with fidelity with students at different levels.
Despite the fading of teacher support, students never completed the graphic organizer with complete independence. Although teacher support is critical in developing students’ ability to apply text structure knowledge to learn from expository text, it also raises issues for research and practice. In practice, teachers will need to find ways to systematically increase and maintain students’ independent use of the text structure intervention package until such time as the student can comprehend increasingly complex text without it. Thus, a critical next step for researchers is to implement not only a maintenance phase as in the current study but also a generalization phase, in which students independently use the intervention materials while reading science passages, and, ultimately, a follow-up phase in which students demonstrate text comprehension without needing the package. For example, an alternate design, including a post-treatment/intervention phase, generalization probes, and increasing the time between intervention and maintenance, would provide important insight into intervention effectiveness.
Similarly, research is needed to evaluate other issues of generalization. One important question addresses the issue of whether or not students can independently generalize their use of the text structure intervention to other content areas given explicit instruction in only one area. For example, if students are taught the text structure intervention in science, can they independently apply it in social studies texts? Given the interconnectedness of reading and writing (Lipson & Wixson, 2013), it would also be valuable to assess the influence of this text structure instruction on writing. Using writing probes throughout the intervention, researchers might evaluate the effects, if any, on changes to expository text writing without explicit instruction in writing. Specifically, research is needed to evaluate the level of support necessary for students with ASD to apply new text structure knowledge to their writing.
Given the small number of students and the methodology in this study, replication with large populations of students with ASD is necessary before deeming it an evidence-based practice ready for dissemination to wide numbers of teachers and their students. However, the intervention package effectively increased both reading comprehension and content-area learning as measured by responses to a series of comprehension questions, two critical requirements of the current CCSS, for all participants, suggesting that these replication and extension efforts are worthwhile.
Conclusion
This study extends Carnahan and Williamson’s (2013) findings on the use of systematic text pattern interventions to support academic reading comprehension as measured by responses to comprehension questions for students with ASD. When presented with a systematic text structure intervention, three students with ASD demonstrated increased comprehension of traditional science texts. The results of the present study confirm Whalon and Hart’s (2011) assertion that given systematic instruction, individuals with ASD benefit from academic reading comprehension interventions. Perhaps most important, the intervention was embedded in traditional science text, allowing learners to access academic content commensurate with their designated grade levels.
Footnotes
Declaration of Conflicting Interests
The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.
Funding
The author(s) received no financial support for the research, authorship, and/or publication of this article.