Abstract
Simple sentence construction is a foundational writing skill that forms the building blocks of advanced writing. Many middle school students with disabilities struggle to consistently construct simple sentences, thereby hindering development of written expression. The present study used a single-case, multiple-baseline across participants design to investigate the effects of a multicomponent intervention on fluency of sentence construction in response to picture-word prompts. During intervention, three middle school students with disabilities completed six explicit instruction lessons, followed by 15 sessions of timed practice, referred to as sentence instruction and frequency building to a performance criterion. After the first few sentence instruction lessons, all students gradually increased the accuracy and frequency of word sequences during intervention. Tau-U values for writing sequences were large. After completing intervention, all students maintained levels similar to intervention. Results are discussed in the context of writing fluency, prior intervention studies, and implications for practice.
Keywords
Written expression is a highly complex, critical skill for middle school students with and without disabilities. Through successful engagement in writing, students experience expanded opportunities to acquire knowledge, synthesize information, and share their thoughts (Klein & Boscolo, 2016). Accordingly, educational stakeholders have placed considerable emphasis on the development and use of written expression. The Common Core State Standards adopted by the majority of states call for middle school students to produce clear and coherent composition (National Governors Association Center for Best Practices & Council of Chief State School Officers, 2010). Furthermore, survey results suggest that more than 50% of middle school teachers report that they use writing activities to strengthen and assess learning (Ray, Graham, Houston, & Harris, 2016).
Unfortunately, students with disabilities tend to struggle with handwriting, spelling, and sentence construction during the middle and high school years (Berninger, Richards, & Abbott, 2015) when teachers report spending less time instructing foundational writing skills (Graham, Capizzi, Harris, Hebert, & Morphy, 2014). These difficulties hinder overall writing development: On the most recent administration of the National Assessment of Educational Progress writing subtest, only a small portion (5%) of eighth- and 12th-grade students with disabilities scored at the proficient level, and most (60%) scored below basic level (National Center for Education Statistics, 2011). Furthermore, a meta-analysis of 53 studies (Graham, Collins, & Rigby-Wills, 2017) found that students with learning disabilities (LD) scored significantly lower than their typically developing peers on writing outcomes, which included ideation, writing output, sentence fluency, and writing conventions—effect sizes ranged from −0.81 to −1.43. Likewise, other research indicates that students with intellectual disability (ID) tend to generate incomplete sentences and produce shorter, less complex composition (Finnegan & Accardo, 2018).
One skill critical to effective written expression is sentence construction: the composition of multiple words into sentence types (e.g., simple or compound) that follow rules of syntax, semantics, spelling, and usage (Troia, Lin, Monroe, & Cohen, 2009). Research indicates that as students reach the fourth grade, the ability to construct a simple sentence of at least one subject and one verb positively affects competence in writing syntactically complex sentences (Berninger, Nagy, & Beers, 2011). Furthermore, researchers recommend that development of fluency in sentence construction, including simple sentences, can further bolster advanced writing skills (Graham et al., 2016).
Theoretical Framework for Writing Development and Intervention
For students with LD or ID, lack of fluency in constructing simple sentences may stem from one or more interrelated areas of writing. According to the simple view of writing (Berninger, Garcia, & Abbott, 2010), written expression develops across three composite areas—transcription, text generation, and executive functions—which are further subdivided into smaller, component skills. Transcription encompasses handwriting, spelling, and keyboarding/typing. Text generation refers to the writing of multiple words and phrases into sentences (i.e., simple sentence construction) or paragraphs. Executive functions are cognitive skills that encompass attention and strategies to self-regulate planning, translating, and revising—all necessary to manage an iterative process of written expression. These composite areas are interdependent and compete for a finite amount of working memory; fluency in one component (e.g., simple sentence construction) or composite area (e.g., text generation) frees up memory resources for development and use of other areas of writing. For example, proficiency in handwriting and spelling promotes the quantity and quality of text generated (Berninger et al., 2002), while fluency in simple sentence construction contributes to the acquisition of paragraph composition (Datchuk, 2016).
Given the interdependent nature of writing development, it is important for interventions not only to emphasize fluency but also to address the multiple skills needed to construct simple sentences. Specifically, simple sentences start with a capital letter, end with an appropriate punctuation mark, and feature at least one subject and one verb that make semantic and syntactic sense. The theory of behavioral fluency (Kubina & Yurich, 2012) helps define fluency and provides a framework to develop fluency-based interventions that address both targeted skills (e.g., sentence construction) and closely related skills (e.g., capitalization/punctuation and syntax/semantics). According to the theory, the development of an academic skill or behavior follows a sequence: Skills are first developed to high accuracy through instruction, and fluency is subsequently developed through deliberate practice. Failure to first develop accuracy with the targeted and closely related skills can lead to dysfluent performance and hinder development of other learning outcomes, such as maintenance following the completion of instruction or practice (Johnson & Street, 2012). In other words, it is prudent for interventions to concurrently address both sentence construction and closely related skills (i.e., capitalization, punctuation, semantics, and syntax) to maximize the likelihood that important learning outcomes are achieved. Indeed, research has found mixed results for addressing these skills in isolation from each other, such as focusing on only syntax without clear connections to sentences or discourse (Datchuk & Kubina, 2013).
Prior Research on Sentence Construction
To teach sentence construction and the closely related skills needed to generate sentences, researchers have typically studied two approaches: sentence-combining for more complicated sentence types and explicit instruction for simple sentences (Datchuk & Kubina, 2013). For the first approach of sentence-combining, students merge one or more phrases or sentences into a complicated sentence—such as combining two simple sentences into one sentence with multiple nouns/adjectives or one compound sentence. Several studies have found positive effects for sentence-combining delivered as a peer-assisted learning strategy with fourth-grade students with and without disabilities (Saddler, Ellis-Robinson, & Asaro-Saddler, 2018). Importantly, when constructing more complicated sentence types (e.g., compound sentences), students need proficiency constructing simple sentences (i.e., simple sentences form the basis of all other sentence types, such as compound or complex sentences). As such, the second approach of explicit instruction—which has typically been used to improve the accuracy of simple sentence construction—is most relevant to the present investigation.
To improve the accuracy of simple sentences and multiple related text writing skills (i.e., capitalization, punctuation, semantics, and syntax within connected text), several studies have delivered a range of 26 to 50 explicit instruction lessons lasting approximately 30 to 50 min each (Viel-Ruma, Houchins, Jolivette, Fredrick, & Gama, 2010; Walker, Shippen, Alberto, Houchins, & Cihak, 2005; Walker, Shippen, Houchins, & Cihak, 2007). Within model–lead–test formats, instructors modeled new content, led students through guided practice, and tested for independent performance (Archer & Hughes, 2011). Students wrote sentences in response to picture-word prompts—small drawings or sketches that depicted a subject or an object engaged in an activity or an action—thought to increase the efficiency of instruction by providing a focal point for sentence construction. As students wrote responses to the same picture-word prompts, instructors quickly provided feedback and error correction.
Three of these studies (Viel-Ruma et al., 2010; Walker et al., 2005, 2007) used single-case research designs involving high school students with LD; all students gradually improved their sentence construction as measured by correct word sequences (CWS), typically scored when sentences had initial capitalization, between words with correct grammar and spelling, and for end punctuation (McMaster & Espin, 2007). In one quasi-experimental study, White, Houchins, Viel-Ruma, and Dever (2014) delivered explicit instruction to two experimental groups of middle and high school students with emotional and behavioral disabilities. Students in one group were provided an additional procedural facilitator for planning (i.e., a graphic organizer). Hierarchical growth modeling analysis revealed that both groups of students increased their mean percentage of CWS.
In a series of related single-case design studies, explicit instruction was combined with timed practice to improve both the accuracy and speed of simple sentence construction (Datchuk, 2016, 2017; Datchuk & Kubina, 2017; Datchuk, Kubina, & Mason, 2015). In these studies, students first completed three explicit instruction lessons lasting approximately 25 min each, referred to as sentence instruction (SI). After the SI lessons, students completed 10 to 15 timed practice lessons lasting approximately 15 min each, referred to as frequency building to a performance criterion (FBPC). During FBPC lessons, students wrote simple sentences in response to picture-word prompts during a series of quick practice timings. After each practice timing, students received performance feedback (i.e., a modified form of CWS and incorrect word sequences [IWS]), praise for correct responses, and error correction before self-monitoring their performance by graphing the highest score.
Three of these studies included elementary and middle school students with and without disabilities (Datchuk, 2016, 2017; Datchuk & Kubina, 2017). One study included only students with disabilities (Datchuk et al., 2015). Across these experiments, nearly all students showed overall greater fluency in construction of simple sentences as indicated by increases in CWS and decreases in IWS on 1-min sentence construction probes. However, there were mixed results for some students (Datchuk, 2016; Datchuk & Kubina, 2017). An error analysis revealed persistent errors in skills not addressed by the intervention, leading authors to suggest expanding the focus of intervention to encompass additional skills of capitalization, punctuation, and verb use.
Purpose of the Current Study and Research Questions
Given the limited opportunities for middle school students with disabilities to develop proficiency in sentence construction (Graham et al., 2014) and the small number of studies that have focused exclusively on middle school students with disabilities, the current investigation addresses a considerable need for efficient and effective intervention, extending prior research on SI and FBPC intervention procedures in two ways. First, instruction was modified to address persistent errors noted in prior studies (Datchuk, 2016; Datchuk & Kubina, 2017). Specifically, three additional SI lessons were incorporated to teach the related skills of capitalization/punctuation (i.e., capitalization of proper nouns and capitalization/punctuation of titles, such as “Mrs.”) and grammar (i.e., use of past tense verbs and subject–verb agreement). Second, SI and FBPC lessons were delivered as a tertiary-level intervention to a population of middle school students who had LD or ID and goals in written expression. Prior studies have mainly investigated the effects of SI and FBPC as a secondary-level intervention for middle and elementary school students with and without disabilities (Datchuk, 2016, 2017). The current study posed two research questions. First, what are the effects of SI and FBPC on simple sentence construction of middle school students with disabilities? Second, what are the maintenance effects following the completion of intervention?
Method
Students and Setting
The study included three participants, Jason, Milan, and Parker—all middle school students with disabilities who had individualized education plans that included goals and services for written expression. Table 1 shows characteristics, achievement scores reported by the participating school, and screening outcomes for each student. Within pullout, resource classrooms, the students received specially designed instruction in various areas, including (a) language arts, reading, math, science, and living skills (for Jason); (b) language arts and reading (for Milan); and (c) language arts, reading, and math (for Parker). All three students attended the school where the study took place—an urban, Midwestern middle school serving approximately 550 students who were 59% White, 21% Black, 10% Hispanic, 7% multiracial, 1% Asian, and 1% American Indian/Alaska Native. Students from low-income families, as defined by enrollment in the free or reduced lunch program, comprised 69% of the school population.
Student Characteristics.
Note. M = male; W = White; B = Black; ASD = autism spectrum disorder; ID = intellectual disability; LD = learning disabilities; NPR = national percentile ranking; CWS = correct word sequences; IWS = incorrect word sequences; CLPM = correct letters per min; WRCPM = words read correctly per 1 min.
The first author, a full-time doctoral student in special education who was unaffiliated with the school, delivered one-on-one intervention seated across from the student in quiet, unoccupied areas of the school (e.g., school offices or empty classrooms). She had a master’s degree in special education, K–12 teacher licensure, and 12 years of special education teaching experience.
Dependent Variable
The dependent variable was the number of CWS and IWS. A CWS was scored when initial words in sentences were capitalized, between words that made semantic and syntactic sense, and for end punctuation. An IWS was scored inversely, when initial words in sentences were not capitalized, between words that were syntactically incorrect or semantically confusing, and for missing end punctuation. Dependent variable scoring and administration differed from research on writing curriculum-based measurement (McMaster & Espin, 2007) in two main ways. First, to more closely reflect skills targeted by the independent variable that did not address spelling, a CWS was counted for incorrectly spelled words that had clear intention (e.g., realy instead of really) or were phonologically similar to the correct word. Second, probes were conducted for 1 min instead of 3 min to minimize handwriting fatigue that may have resulted from multiple practice timings during intervention.
Dependent measurement materials and procedures
To measure CWS and IWS, a 1-min sentence construction probe containing 10 researcher-developed picture-word prompts used in prior research (e.g., Datchuk, 2016) was administered during each baseline and maintenance session and at the end of every intervention session. Across all study phases, a total of 36 probes, which each contained 10 unique picture-word prompts, were used to assess the students’ writing performance under conditions with no feedback or encouragement. Each picture-word prompt was unique and did not overlap with intervention materials; it featured a 2.5 cm by 5 cm image, two to three words similar in level to the oral reading fluency probe used to screen students for adequate decoding skills, and three writing lines. The interventionist began each probe administration by saying, When I say start, write as many sentences as you can in 1 min. Work as quickly and accurately as you can. You may use the words given if you want. Do you have any questions about the pictures or words?
The interventionist then answered any questions and started a countdown timer. After 1 min, the interventionist said, “Pencil up,” noted the last written response, and withheld performance feedback, praise, or error correction.
Interobserver agreement
Given the inductive nature of the study and the need to make timely, data-based instructional decisions, the first author (i.e., the interventionist) completed primary scoring of the 1-min sentence construction probes and graphed data after each session. To measure potential observer drift or systematic bias in the primary scoring, the first author trained a secondary observer, a doctoral student uninvolved with the study, on scoring procedures. The secondary observer practiced scoring 20 sample sentences until meeting 100% agreement with the first author. Per suggested guidelines of single-case research (Ledford & Gast, 2018), the secondary observer scored 33% of all probes selected randomly across students and experimental phases, but was uninformed of the sequence of probes or research questions.
Point-by-point agreement was calculated by dividing the number of agreements by the number of agreements plus disagreements, multiplying by 100, and then calculating the average across probes (Ledford & Gast, 2018). The mean exact agreement was 92% for word sequences: 84% during baseline (range = 71%-96%), 95% during SI and FBPC (range = 94%-96%), and 97% during maintenance (range = 92%-100%). While in baseline, students tended to produce small numbers of word sequences (e.g., four CWS). As such, slight differences in scores between observers resulted in lower exact agreement during the baseline phase compared with the other phases. For example, scoring that differed by one writing sequence (e.g., three CWS vs. four CWS) resulted in 75% agreement.
Experimental Design and Data Analysis
The current study used a single-case, multiple-baseline design across students. For this design, behavior is repeatedly measured across phases, intervention is applied in a staggered fashion across students, and data are visually analyzed for level, trend, and variability (Ledford & Gast, 2018). A multiple-baseline design is ideal for investigating likely irreversible academic behaviors such as sentence construction and allows for multiple opportunities to detect experimental effects between independent and dependent variables (Horner et al., 2005).
Two decision rules guided the systematic manipulation of the SI and FBPC phase. First, students needed at least three baseline data points that showed stable (i.e., flat CWS and IWS) or worsening (i.e., increasing IWS) performance trends. Second, preceding students needed to demonstrate at least 90% accuracy on the first three SI lessons. Although there were at least three attempts to demonstrate an effect at three different points in time, a collection of just three baseline data points only meets What Works Clearinghouse Pilot Single-Case Design Standards With Reservations (Kratochwill et al., 2010) and potentially limits robust conclusions about baseline performance. The study authors visually inspected data after each session to judge any effects of intervention (i.e., to compare the CWS and IWS data paths) and decide upon phase changes. Descriptive statistics were calculated to reveal overall performance levels. An online tool (Vannest, Parker, Gonen, & Adiguzel, 2016) was used to calculate an overall Tau-U value, an effect size metric of nonoverlap between baseline and intervention phases, which accounts for undesirable baseline trends. Baseline trends for Jason’s CWS and Milan’s IWS were corrected due to Tau-U values that exceeded .20 (Vannest & Ninci, 2015).
Procedures
Screening
Participant selection followed three steps used in prior studies (e.g., Datchuk, 2016). First, teachers identified students with disabilities who had goals in written expression. Second, students completed a 1-min sentence construction probe, which was scored identically to dependent measurement procedures. Third, students who had low production (<20 CWS) or low accuracy (<90%) on the 1-min sentence construction probe (i.e., well below the performance criterion used in the study) were screened using a 1-min handwriting copy task and a 1-min oral reading probe of sentences drawn from intervention materials. To confirm adequate handwriting and decoding skills to engage with intervention materials, students needed to show at least 80% accuracy on the handwriting and reading probes to qualify for participation.
Baseline and concurrent intervention
Concurrent to all study phases, the participating students received 40 min of specially designed language arts instruction from a special education teacher who had a master’s in special education and was uninvolved with study sessions. She delivered mini-lessons on editing strategies, sentence and paragraph composition, and commonly seen grammar/usage errors on everyday assignments. The special education teacher did not use timed practice procedures for typical daily instruction. During all baseline sessions, the interventionist (i.e., the first author) administered one 1-min sentence construction probe.
Intervention
A modified form of researcher-developed intervention and assessment materials and procedures used in prior studies (e.g., Datchuk, 2017) included six SI lessons that lasted approximately 20 min each, followed by up to 15 FBPC lessons that lasted approximately 10 min each. An online appendix contains specific details about instructional foci and examples of student tasks used across the intervention. The SI and FBPC intervention lessons were provided supplementary to students’ typical, daily specialized instruction delivered by the special education teacher. At the end of each intervention session, students completed a 1-min sentence construction probe with no performance feedback, praise, or error correction.
Sentence instruction
An explicit instruction framework using model–lead–test formats (Archer & Hughes, 2011) was used to teach SI lessons: The interventionist modeled new content by writing and thinking aloud accurate responses, led students through practice by eliciting frequent vocal and written responses, praised correct responses, and immediately demonstrated correct responses for errors. Throughout SI lessons, a complete, simple sentence was defined as having (a) a part that names someone or something (i.e., subjects/nouns) and a part that tells more (i.e., verbs/predicates), (b) an initial capitalized word, and (c) appropriate end punctuation. The first three SI lessons focused on identifying and writing subjects/nouns and verbs/predicates within sentences, changing present tense verbs into past tense, discriminating between complete sentences and sentence fragments, and writing sentences in response to picture-word prompts.
To address persistent errors observed in prior research (e.g., Datchuk, 2017), the last three SI lessons taught additional skills in capitalization/punctuation and grammar. These lessons involved (a) revising errors in beginning capitalization, capitalizing proper nouns/titles, and end punctuation; (b) changing irregular, present tense verbs into past tense; and (c) using appropriate subject–verb agreement within sentences. On a test for independence following the third and sixth SI lessons, students needed to meet a 90% accuracy criterion on a subset of tasks from previous SI lessons. All students met this criterion, meaning that no student needed to repeat any SI lessons. After completing six SI lessons, students started the FBPC component.
Frequency building to a performance criterion
Each FBPC lesson featured three copies of a practice sheet with a unique set of 10 picture-word prompts, similar in format to the 1-min sentence construction probes. The interventionist first verbally stated the performance criterion—30 CWS and zero to three IWS—drawn from a prior study (Datchuk et al., 2015) that had set a local norm based on the performance of students who had scored proficient on a statewide writing assessment. Next, the interventionist noted previous high scores on a bar chart and highlighted a specific skill focus based on persistent error patterns in sentence construction (e.g., missing capital letters). After that, the interventionist and students previewed the picture-word prompts for that lesson and brainstormed five additional word prompts to support idea generation during practice timings. If students showed difficulty coming up with appropriate words, the interventionist provided suggestions or alternatives.
Then, during three consecutive 1-min practice timings, students wrote as many complete, simple sentences as possible to the same set of 10 picture-word prompts. After each timing, the interventionist tallied CWS and IWS, praised instances of CWS, and modeled corrections for any IWS. Last, students graphed their highest score on a bar chart. It should be noted that timed practice scores for CWS and IWS were not reported as study data. Students completed the FBPC component of intervention after either (a) achieving the performance criterion (i.e., 30 or more CWS and 0-3 IWS) on the majority of practice timings for three lessons in a row or (b) finishing the 15th FBPC lesson—whichever occurred first. No student reached the performance criterion.
Maintenance
Starting from 1 to 3 days following the end of the SI and FBPC phase, each student completed one 1-min sentence construction probe across five maintenance sessions. Maintenance phase ranged from 7 (for Parker) to 15 (for Jason and Milan) days. The 1-min sentence construction probes were conducted under no-feedback conditions.
Treatment Integrity
The second author used demonstration and rehearsal to train the first author (i.e., the interventionist) on study procedures during two 2-hr sessions. For all study phases, sessions were recorded with a digital audio-video camera. Only the first author provided intervention using an implementation checklist and an intervention script with recommended language and level of assistance for model–lead–test formats. The first author trained a secondary observer, a master’s student in special education, on study procedures. The secondary observer viewed 33% of all videos, randomly selected across study phases and students and checked for accuracy of implementation by marking occurrences and nonoccurrences of the interventionist’s behaviors during the steps for model–lead–test and timed practice. Fidelity across phases was 97% in baseline, 98% in SI and FBPC, and 100% in maintenance. Treatment integrity during baseline and maintenance phase was only conducted for the 1-min sentence construction probe administration.
Treatment Acceptability
Following the last maintenance session, a school counselor who was uninvolved in study implementation administered a treatment acceptability questionnaire. Students wrote responses to four items. First, what do you think is the purpose of learning to write complete sentences? Second, is there anything you did not like about the instruction or practice? Third, is there anything you would change? Fourth, how do you feel about your writing after the lessons?
Results
Students completed three to eight baseline sessions over a range of 4 to 14 calendar days, 21 SI and FBPC lessons over a range of 34 to 42 calendar days, and five maintenance sessions over a range of 7 to 15 days. For Jason, no data were collected on Calendar Day 36 due to a procedural error. Data collection did not occur during a 5-day Thanksgiving break (i.e., Calendar Days 45-54). The overall Tau-U across students was .87 for CWS (SD = 0.17, p < .001, 95% confidence interval [95% CI] = [.53, 1.0]) and .89 for IWS (SD = 0.17, p < .001, 95% CI = [.54, 1.0]), indicating large effect sizes for intervention (Parker, Vannest, & Davis, 2011). Table 2 displays descriptive statistics across phases. Figure 1 shows the frequency of CWS and IWS. Dots represent CWS, and Xs represent IWS.
Means and Standard Deviations for CWS and IWS Across Experimental Phases.
Note. CWS = correct word sequences; IWS = incorrect word sequences; SI = sentence instruction; FBPC = frequency building to a performance criterion.
The frequency of word sequences per 1-min across experimental phases.
Jason
Jason only completed three baseline sessions; however, he showed very low accuracy. He consistently wrote fewer CWS than IWS, averaging 5.0 CWS and 11.0 IWS. After introduction of the SI and FBPC intervention, Jason’s frequencies of CWS and IWS initially remained unchanged from baseline. His data paths separated after the sixth SI lesson, when CWS consistently outnumbered IWS as he gradually improved in the accuracy and speed. Jason’s CWS increased in trend and showed variability, while IWS decreased in trend and stabilized. Across the SI and FBPC phase, his average CWS increased to 12.6, and IWS reduced to 3.0. Jason showed more stable performance during maintenance, when he achieved four of his highest frequencies of CWS and very low frequencies of IWS. While Jason’s average CWS further increased to 21.8 CWS, and IWS reduced to 0.8, there was a slight decline in performance during his last maintenance session.
Milan
During baseline, half of Milan’s sentence construction tended to be inaccurate. He had similar frequencies of CWS and IWS for most sessions, producing on average 8.2 CWS and 9.2 IWS. For the last baseline session, Milan’s accuracy decreased markedly. During the SI and FBPC phase, Milan’s sentence construction gradually increased in the accuracy and speed. His data paths diverged from baseline following the second SI lesson, as CWS trended upward and remained variable, and IWS trended downward and stabilized. Milan consistently wrote more CWS than IWS within sentences across the entire SI and FBPC phase, averaging 15.2 CWS and 2.2 IWS. During maintenance, a clear separation of data continued, with overall levels similar to SI and FBPC phase. Although Milan’s average CWS increased slightly to 16.4, and IWS reduced slightly to 1.6, his CWS trended slightly downward, while IWS trended slightly upward.
Parker
Parker’s data paths for CWS and IWS were initially separated (i.e., higher CWS than IWS) for three baseline sessions, but merged as his accuracy diminished toward the end of baseline. In baseline, he averaged 10.6 CWS and 8.5 IWS, reflecting numerous errors in his writing. During the SI and FBPC phase, Parker gradually improved his accuracy and speed of sentence construction. His data paths separated after the third SI lesson, at which point he had higher frequencies of CWS than IWS for all sessions. Across the SI and FBPC phase, Parker’s CWS data remained variable and trended upward, while IWS data trended downward and stabilized; his average CWS reached 20.7, and IWS reduced to 2.5. During maintenance, Parker’s data paths remained separated, and his performance levels were similar to the SI and FBPC phase. Although his average CWS increased to 23.0, and IWS decreased to 0.8, his CWS trended sharply downward during the maintenance phase.
Treatment Acceptability
Students wrote responses to four questions after the last maintenance session. When asked about the purpose of writing complete sentences, Jason replied, “to be good at completing sentences.” Milan indicated, “to help us learn.” Parker wrote, “keeping up with punctuation and making sure that there is a capital letter.” When asked whether there was anything to dislike about instruction or practice, Jason only replied, “yes.” Milan and Parker replied, “no.” When asked whether there was anything they would change, each student responded “no.” To answer how they felt about their writing after lessons, all students wrote “good” or “pretty good.”
Discussion
Middle school students encounter high expectations to use written expression to demonstrate knowledge in content areas (Ray et al., 2016), but have fewer opportunities to improve foundational writing skills (Graham et al., 2014). While fluency in simple sentence construction contributes to development of advanced writing (Berninger et al., 2011), many students with disabilities struggle to construct complete, simple sentences (Graham et al., 2017). Therefore, middle school students with disabilities stand to benefit from a targeted, supplemental intervention on sentence construction. The current study investigated the effects of a supplemental writing fluency intervention, SI and FBPC, on the simple sentence construction of middle school students with disabilities.
A visual analysis of the results suggests three experimental effects between the SI and FBPC intervention and performance on the 1-min sentence construction probes. Upon introduction of intervention, no student showed immediate changes in performance. However, after completing two to six SI lessons, each student showed gradual improvement, as CWS trended upward, and IWS trended downward. Across the SI and FBPC phase, all students increased their accuracy and speed of writing sequences per 1 min, producing approximately 7 to 10 more CWS and 6 to 7 fewer IWS. This divergence of data paths indicates moderate improvements in both the accuracy and speed of simple sentence construction.
The current investigation is the first to apply a modified form of the SI and FBPC intervention to middle school students with disabilities. The overall positive findings extend the research regarding sentence construction fluency, explicit instruction, and timed practice. However, more research is needed to establish the generality of effects. Subsumed as a component of text generation, sentence construction refers to the composition of multiple words into sentence types that make semantic and syntactic sense to readers (Berninger et al., 2010). The theory of behavioral fluency suggests that fluent sentence construction should result in (a) immediate increases in the accuracy and speed of writing and (b) little to no decrease in performance (Kubina & Yurich, 2012). In the current study, intervention did not lead to immediate improvement, but gradual increases in the accuracy and speed were observed across the SI and FBPC phase. Fluency-based procedures were instituted after the sixth SI lesson, at which time performance had shown a distinct change from baseline. Following the completion of SI and FBPC lessons, all students demonstrated performance levels similar to intervention, with average increases of 1.2 to 9.2 CWS and decreases of 0.6 to 2.3 IWS. However, there was some decrement in performance, possibly attributable to Thanksgiving break, after which Jason’s and Milan’s CWS shifted slightly lower, and Parker’s CWS showed a sharp, downward trend.
The current study also extends prior research on using explicit instruction and picture-word prompts to teach accurate sentence construction. Past studies have applied instruction over a range of 26 to 50 sessions that lasted approximately 30 to 50 min each (Viel-Ruma et al., 2010; Walker et al., 2005, 2007; White et al., 2014). The current study followed model–lead–test formats across a total of 21 SI and FBPC lessons that lasted approximately 10 to 20 min each. Picture-word prompts may have provided an efficient focal point for instruction, allowing the interventionist to quickly model sentence construction skills, elicit frequent responses, and provide error correction. Positive outcomes from the current study suggest the possible benefits of a relatively short intervention of 3.5 to 7 hr over approximately 6 weeks. When considering the efficiency of instruction to improve sentence construction for middle school students with disabilities, a shorter duration intervention may be more acceptable to teachers. Specifically, middle school students with disabilities in the present study increased the accuracy and speed of their sentence construction by an average of seven to 10 writing sequences (i.e., increased their CWS and decreased their IWS) through delivery of a supplemental writing intervention. In practical terms, students were able to increase their sentence construction accuracy on a per-minute basis, making important but small improvements across multiple sentences (e.g., ensuring correct capitalization/punctuation across four sentences) or attempting more total sentences (e.g., writing three to four sentences instead of one to two sentences).
The results also extend procedures from prior studies (Datchuk, 2016, 2017; Datchuk & Kubina, 2017; Datchuk et al., 2015) that used SI and FBPC to promote fluency of simple sentence construction. First, based on mixed results and persistent error patterns observed in past research (Datchuk, 2016; Datchuk & Kubina, 2017), the current study increased the scope of sentence construction skills to include verb use/tense, subject–verb agreement, and capitalization of common/proper nouns and titles. As a result of this expanded scope, the number of SI lessons was increased from three to six (i.e., all prior studies only delivered three SI lessons), and these persistent errors were not found on the 1-min sentence construction probes.
Second, although each student gained accuracy and speed, no student reached the performance criterion of 30 CWS and zero to three IWS on the majority of practice timings for three lessons in a row. This criterion was drawn from prior research that set a local norm based on high-performing elementary school students (Datchuk et al., 2015) and may have been unsuitable for middle school students with disabilities. Despite not reaching criterion, all students still showed gradual increases in CWS and decreases in IWS across the intervention. Prior studies had similar outcomes (Datchuk, 2016, 2017), suggesting that fluency practice procedures may result in improved sentence construction, even when an outcome of fluency, as defined by a predetermined performance criterion, is not achieved.
Third, prior studies have delivered SI and FBPC as a secondary-tier intervention for elementary/middle school students with and without disabilities (Datchuk, 2016, 2017; Datchuk & Kubina, 2017). The current study is the first to apply SI and FBPC procedures as a tertiary-tier intervention to a population of middle school students with disabilities. The gradual improvement in sentence construction suggests that the SI and FBPC intervention was suitable as a tertiary-level intervention for these middle school students with disabilities. However, given such preliminary findings, more research is needed to explore generality to this population.
Overall findings suggest that SI and FBPC positively affected sentence construction for the participating students. Furthermore, all students responded with favorable but brief comments regarding the treatment procedures and outcomes. The SI and FBPC lessons required relatively small amounts of time out of the school day, suggesting an efficient and effective means of improving the simple sentence construction of middle school students with disabilities.
Limitations and Future Directions
There are five main limitations and future directions. First, SI lessons were broadened to target additional sentence construction skills, but not expanded to address advanced writing skills such as paragraph composition. Future studies can include additional lessons and assessment procedures that incorporate longer composition such as paragraph or story writing. Second, scoring procedures were modified from prior research on writing measurement (e.g., McMaster & Espin, 2007)—misspelled but phonologically similar words with clear intention were counted as correct. Since the SI and FBPC lessons did not address spelling, modified scoring allowed for a more sensitive detection of intervention effects. Future research can test the generality of SI and FBPC to impact traditionally scored word sequences that account for spelling.
Third, the 1-min sentence construction probes were administered each session. Although minimal to no improvement was observed during baseline, repeated administration may have resulted in practice effects over time. Moreover, conclusions about baseline performance were potentially restricted by the small number of baseline data points for one student (Jason). Future studies can use a multiple-probe design (i.e., collect intermittent data across a larger range of calendar days) to reduce the potential threat of practice effects and lengthen the baseline phase to help verify any effects of intervention. Fourth, no student reached the performance criterion of 30 CWS and zero to three IWS. Future research can tailor performance criteria to better reflect local norms or transcription rates of individual students. Fifth, students produced only brief responses on a handwritten treatment acceptability survey, suggesting that future research should elicit oral responses, thus lessening any transcription difficulties.
Implications for Practice
There are several recommendations stemming from the results of the current study. First, practitioners of middle school students with disabilities who struggle with simple sentence construction can use an explicit instruction framework: model correct responses, lead students through guided practice, and test for independence. Explicit instruction should continue until students achieve 90% accuracy or higher on independent responses. Instruction should focus on skills needed to construct simple sentences, including capitalization, punctuation, and grammar/usage. Second, to promote the speed of simple sentence construction, practitioners can engage students in short practice timings of constructing simple sentences to picture-word prompts. After each timing, practitioners should deliver performance feedback on the numbers of CWS and IWS, praise correct responses, model the correct response for any errors, and encourage students to improve their score. Then, after completing several practice timings, students should graph their best score. Third, practitioners can monitor progress in simple sentence construction by examining scores and documenting error patterns during practice timings. Practically speaking, practitioners may see increased speed of sentence construction as students incorporate more semantically appropriate words within syntactically correct sentences and decrease the number of words containing errors in capitalization or subject–verb agreement.
Supplemental Material
Online_appendix_sentence_construction_RASE_7-24-19 – Supplemental material for Constructing Simple Sentences: Effects of a Writing Fluency Intervention for Middle School Students With Disabilities
Supplemental material, Online_appendix_sentence_construction_RASE_7-24-19 for Constructing Simple Sentences: Effects of a Writing Fluency Intervention for Middle School Students With Disabilities by Kristin L. Panos and Shawn M. Datchuk in Remedial and Special Education
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.
Supplemental Material
Supplemental material for this article is available online.
References
Supplementary Material
Please find the following supplemental material available below.
For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.
For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.
