Abstract
This study compared the writing motivation of students with language impairments (LI) and their typically developing (TD) peers. For the study 272 students (33 students with language impairments, 242 TD peers) aged 8–10 years, in 11 elementary schools, were sampled. The two groups completed self-report measures of writing motivation and 20 grade-appropriate spelling words from a standardized measure. Multiple regression analyses were used to examine the hypothesis that perceived writing competence – a self-evaluative judgment of writing competence – mediates the relationship between LI status (LI or TD) and intrinsic forms of writing motivation. Statistical analyses indicated that (1) students with specific language impairments reported lower levels of perceived writing competence and intrinsic writing motivation; (2) LI status was a significant predictor of perceived competence after spelling, grade and gender were controlled; and (3) when spelling, grade and gender were controlled, perceived writing competence was a significant predictor of intrinsic writing motivation, but LI status was not. Future research is necessary to further describe external influences on the writing motivation of students with LI and appropriate motivational supports.
Keywords
I Introduction
Students with specific learning difficulties have considerably more difficulty than their peers in their approach to writing and their knowledge of writing, resulting in significantly lower outcomes (Graham and Harris, 2003; Graham and Perrin, 2007; Salahu-Din et al., 2008). Without consistent participation during writing tasks and writing interventions, students with specific learning difficulties are unlikely to close the gap with their peers and develop adequate literacy skills (Guthrie, 2004; Morgan and Fuchs, 2007). For this reason, building motivation to write should be an essential instructional focus for speech language pathologists, classroom teachers and other special education service providers.
Schunk et al. (2008) define motivation as ‘the process whereby goal-directed activity is instigated and sustained’ (p. 4). To date, most research investigating the skills and motivation of struggling writers has focused on students who are broadly identified as ‘learning disabled’ or as having ‘specific learning difficulties’ (Nelson et al., 2004; Scott, 2002; Singer and Bashir, 2004). Students with specific learning difficulties, however, are a heterogeneous group who present difficulties related to a number of primary and co-existing challenges in underlying processes. This study focused on a subset of students with specific learning difficulties, namely students with language impairments (LI). These students have a disability related to their use of language to express ideas and to comprehend messages.
Because language use is an essential part of the writing process, difficulty with language is likely to negatively affect writing outcomes and attitudes towards writing (Dockrell et al., 2009; Scott, 2002). For example, writers with smaller vocabularies and decreased syntactic maturity will be at a disadvantage when attempting to express themselves through writing. In addition, students with LI will be less proficient at retrieving linguistic knowledge for text generation from long-term memory. This is further complicated by their difficulty in using language to acquire an adequate base of topic knowledge. Because of these unique challenges, writing can be an extraordinarily demanding cognitive task for students with LI. Thus, despite the heterogeneous nature of language impairments, they are likely to have detrimental effects on a written language tasks such as text production (e.g. quantity of ideas, sentence complexity), text revision and text quality (e.g. word choice, cohesion of ideas, sentence grammar) (Dockrell et al., 2007; Mackie and Dockrell, 2004; Nelson et al., 2004; Scott, 2002; Singer and Bashir, 2004). For this reason, it seems realistic to hypothesize that these unique challenges will affect the perceived writing competence, or the belief that one has the capacity to complete a specific task (Bandura, 1997; Ryan and Deci, 2000), of students with LI. From a number of motivational perspectives, it is expected that a consistent lack of academic success will decrease perceived competence and motivation for academic tasks (Bandura, 1997; Deci et al., 1992; Grolnick and Ryan, 1990; Guay et al., 2008). It would follow that students with LI, because they have significant difficulties with the cognitive-linguistic challenges of the writing process, are at-risk for lower levels of writing motivation. While there is certainly a bi-directional relationship between perceived writing competence and writing outcomes, a large body of research suggests that an individual’s writing motivation is determined more by their competence beliefs than objective measures of competence (for a review, see Pajares, 2003). As asserted by Bandura (1997), ‘people’s level of motivation, affective states, and actions are based more on what they believe than on what is objectively true’ (p. 2). Students who have higher levels of perceived competence are more likely to choose difficult activities and persist during difficulty, which is an important consideration for speech language pathologists (SLPs) attempting to provide frequent, intense intervention.
In addition to using language for text generation and revision, proficient writers must also possess adequate decoding and phonics knowledge to transcribe language into print. While some students with language impairments have deficits in these areas, some do not (e.g. Bishop and Snowling, 2004; van Weerdenburg et al., 2010). Deficits in decoding and spelling will impede the writing process in different ways than oral language difficulties. For example, at a very basic level, spelling ability deficits will interfere with text generation by diverting working memory resources away from composing the ideas and towards the basic symbolic encoding of the words. Spelling and decoding deficits also make it more difficult to revise text due to higher frequency of spelling errors and inefficient reading abilities. Inadequate reading abilities will also make it more difficult to access external texts necessary for building up topic knowledge for the writing task (e.g. from web material, textbooks, reference books).
Careful consideration of these specific literacy profiles leads to the possibility that students with language impairments face unique challenges during writing tasks. The purpose of this study was to extend the current writing motivation literature by investigating students with LI. Because very few studies of writing motivation with at-risk writers have intentionally sampled students with LI, this study has the potential to help form a more refined understanding of writing motivation in general, as well as begin to uncover the role language ability might play in maintaining writing motivation. With this knowledge, SLPs and related educators will be better equipped to design interventions that increase their students’ language and literacy abilities.
1 Motivational perspective: Self-determination theory
Self-determination theory (SDT; Ryan and Deci, 2000) proposes that individuals may be intrinsically or extrinsically oriented towards a task. ‘Intrinsic motivation’ refers to performing a behavior because the activity is inherently interesting or enjoyable. Individuals who are intrinsically interested in an activity are more likely to demonstrate increased persistence and experience more positive emotional well-being (Burton et al., 2006; Grolnick and Ryan, 1987; Reis et al., 2000). ‘Extrinsic motivation’ occurs when an individual engages in an activity for external reasons (e.g. rewards, ego), or for reasons other than inherent interest. Extrinsic motivation is not a unitary construct, however, as there is a continuum of the quality of extrinsic motivation. If an individual feels motivated by external controls (e.g. punishment, threats, preserving ego) and less by personal choice, this type of motivation is less self-determined and is likely to result in a lower quality of motivation. At the other end of the continuum, individuals may feel more autonomous and experience more of a sense of personal agency and enjoyment during the activity. As Ryan and Deci (2000) point out, many educational tasks are not inherently interesting, so it is important that educators are able to facilitate more self-determined forms of extrinsic motivation.
SDT proposes four differentiated types of external motivation: external regulation, introjection, identification and integration. As can be seen in Figure 1, these subcategories of extrinsic motivation can be differentiated based on the external or internal orientation of an individual’s behavior. At the far left of Ryan and Deci’s continuum is ‘amotivation’. Students who are ‘amotivated’ lack the intention to act. This may be due to a perceived lack of competence, control, or task value. To the right of amotivation are four levels of extrinsically motivated orientations. The first level of extrinsic motivation, external regulation, is generally a matter of compliance focused on an external demand or consequence. External regulation also is accompanied by feelings of external locus of control. During introjection, individuals complete activities with a focus on avoiding negative emotions (e.g. guilt, anxiety) or to enhance one’s ego. These activities are focused on maintaining contingent self-esteem and self-worth. Identification reflects a more internally regulated style of extrinsic motivation. Here the person has recognized the activity’s importance and is internally regulating the necessary behaviors to complete the task. The most autonomous form of extrinsic motivation, integration, occurs when the reasons for an activity have been integrated with one’s sense of self, values and needs.
A taxonomy of human motivation
As SDT has developed, it is generally accepted that external regulation and introjection represent a controlled motivation style (Figure 1). In contrast, ‘autonomous motivation’ consists of the identified, integrated and intrinsic forms of motivation (Ryan and Deci, 2000). Autonomous motivation is conceptualized as engaging in an activity for internal, as opposed to external, purposes. Numerous studies have linked autonomous motivation with improved academic outcomes and emotional well-being (for a review, see Guay et al., 2008). For this reason, it is important for educators to understand and utilize methods that will facilitate autonomous motivational orientations towards academic tasks. The distinction between controlled and autonomous motivation also is important to the design of this study, as autonomous motivation was a primary dependent variable.
SDT proposes three essential psychological needs that are necessary for the development of more autonomous forms of extrinsic motivation: relatedness, competence and autonomy. A key reason people will complete activities for an external reason is to feel related to significant others. If the activity is valued by significant other(s), students are more likely to value it as well (Furrer and Skinner, 2003). In addition to feeling a sense of relatedness, a sense of autonomy is necessary towards the development of intrinsic motivation and internally regulated extrinsic motivation (Katz and Assor, 2007; Vansteenkiste et al., 2004). Autonomy refers to ‘the need to feel a sense of control, agency, or autonomy in interactions in the environment’ (Schunk et al., 2008: 248). Finally, it is unlikely that students will internalize an activity if they do not perceive that they are competent and capable of being successful. This sense of perceived competence is central to this study, as it is reasonable to expect that difficulties with language could negatively influence students’ perceived writing competence.
2 SDT and academic motivation
Numerous studies have validated these tenets of SDT, as well as the strong relationship between autonomous motivation (intrinsic, identified, integrated) and positive academic outcomes (Burton et al., 2006; Grolnick and Ryan, 1987; Lepper et al., 2005; Reeve, 2002; Reis et al., 2000; Ryan and Connell, 1989; Vansteenkiste et al., 2004). In addition, there is strong evidence for this perspective on human motivation across ethnic, gender and age groups (Guay et al., 2008; Ryan and Deci, 2000). In an exploratory study, Ryan and Connell (1989) sampled from urban (n = 112), suburban (n = 156) and rural (n = 450) elementary schools. Within this sample, there was an ordered correlation, providing strong validation for their model of differentiated motivational orientations. More recent studies have linked different types of motivation to academic and affective outcomes. For example, Lepper et al. (2005) sampled 797 elementary and middle school children and found that intrinsic motivation was positively related to both grades and standardized test scores, while extrinsic motivation was negatively correlated with these outcomes. This research by Lepper and colleagues had a diverse sample and found very few differences across gender or ethnicity groups.
Relatedness, autonomy and perceived competence also are highly relevant considerations related to the motivation of students with specific learning difficulties. For example, Deci et al. (1992) investigated the relationship between students with specific learning difficulties’ perceptions of autonomy and competence with their academic achievement and social adjustment, and, consistent with SDT, perceptions of competence and autonomy were primary predictors of adjustment and achievement. Similarly, Grolnick and Ryan (1990) compared the perceived competence and autonomous motivation of students with specific learning difficulties to three other groups: matched for IQ, low-achievers and randomly selected typically developing (TD) peers. In this research, the students with specific learning difficulties and lower achievers reported lower levels of academic perceived competence and were less autonomously motivated. In addition, teachers rated the students with LD as significantly less competent and less motivated.
3 Writing competence and motivation in students with LI
As previously discussed, students with LI face unique difficulties in the writing process. Writing is a complex cognitive task that requires the individual to balance multiple processes in working memory (Hayes, 2000). However, students with LI often have impoverished working memory capacity (Montgomery et al., 2010). In addition, students with language impairments will be less proficient at retrieving and using linguistic knowledge during various writing processes. The writer must efficiently and effectively use semantic, morphosyntactic and discourse knowledge during text generation and revision processes. Because of the heavy language demands of these writing-related tasks, writing is a particularly demanding cognitive task for students with LI, and this proposition is supported by recent research comparing the written language of students with LI with their same-aged peers (Dockrell et al., 2007). For this reason, it seems realistic to hypothesize that these language-specific difficulties negatively influence the perceived writing competence of students with LI.
As discussed earlier, these language-related difficulties seem to present distinct challenges from those faced by students with dyslexia, who have a core deficit in using the alphabetic code of written language to recognize and spell words. While these students may not experience as much difficulty formulating language for writing, they must devote more cognitive resources towards spelling words, formulating letters or reading text for revision. For example, Berninger et al. (2008) found that the writing of students with dyslexia and intact oral language was characterized by increased difficulty with word spelling, which in turn had a direct influence on the participants’ quality of written composition. Some researchers have tested the hypothesis that inadequate writing ability relates to lower levels of writing motivation in students with specific learning difficulties, and the available evidence generally supports this proposition (for a review, see Graham, 2006). For example, Shell et al. (1995) did not specifically study students with specific learning difficulties, but found significant differences in three motivational variables (self-efficacy, attribution and outcome expectancies) when comparing high achieving writers to low achieving writers. Graham et al. (1993) found that students with specific learning difficulties had a less favorable view of writing than their TD peers, but no significant differences were found when comparing the motivational variable between these groups. While some intervention studies have discovered a relationship between improved competence and writing motivation (Berninger and Hidi, 2007; Graham and Harris, 1989), others have not (Garcia-Sánchez and Fidalgo-Redondo, 2006; Graham et al., 2005; Harris et al., 2006; Page-Voth and Graham, 1999). Because of these mixed results, it is necessary to continue investigation into the writing motivation of students with specific learning difficulties in order to form a more complete and better-validated model of their motivation.
While previous research related to writing motivation and specific learning difficulties provides some insight for understanding writing motivation in students with LI, caution must be exercised when making this generalization. Because many students have dyslexia and LI (American Speech-Language-Hearing Association, 2001), it is likely that many of the participants in these studies had significant difficulties with the language demands of the writing process. However, most of the studies reviewed for this study have utilized a broad definition of specific learning difficulties as a significant discrepancy between ability and achievement, and did not specifically measure oral language functioning. One study (Harris et al., 2006) did reference language impairments in how its sample was defined, but did not examine the unique motivational profile of this subgroup. A study focusing on writing motivation in students with LI, a previously understudied population, would extend the current writing motivation literature.
4 The hypothesized model
Figure 2 represents the hypothesized model of autonomous writing motivation that was tested in this study. Beginning on the left side of Figure 2, LI status was hypothesized to influence the perceived writing competence of the participants. LI status refers to the students’ eligibility for special education services in the area of language, and students in this study are categorized as either LI or TD. It is hypothesized that students with LI will report lower levels of perceived writing competence due to the linguistic challenges of the writing process.
Proposed model of autonomous writing motivation
As previously described, perceived competence generally is considered to be a primary determinant of autonomous motivation (Deci and Ryan, 2000). For this reason, in the present study perceived writing competence is hypothesized to have a direct relationship with students’ autonomous writing motivation. It also is hypothesized that perceived writing competence will mediate the relationship between LI status and autonomous writing motivation.
5 Control variables
Researchers have identified a number of variables other than language that may be related to writing motivation. Spelling ability, gender and grade were control variables in the present study. The rationale for their inclusion as control variables will be discussed in the following sections.
a Spelling ability:
Researchers have proposed at least two possible reasons that spelling deficits are negatively related to a student’s perceptions of writing competence. First, spelling errors are likely to negatively influence an adult’s perception of a child’s writing (Marshall and Powers, 1969). That is, writing with spelling errors is more likely to receive critical feedback. Second, spelling difficulties may interfere with fluent writing composition (e.g. Graham et al., 2002). Students need to be able to transcribe their message efficiently so they can devote working memory resources towards other components of the writing process. Because spelling can influence students’ attitudes towards writing and many students with LI have difficulty with spelling, it is an important control variable in this study.
b Gender:
It also could be expected that gender influences writing motivation. Gender differences related to motivation have been observed across domains, cultures and age (Hyde and Durik, 2005). However, the nature of these gender differences appears to be, at least in part, domain-specific. For example, boys report higher levels of perceived competence than girls in science, math and athletics (Meece et al., 2006). On the other hand, girls report that they feel more competent than boys in language arts (Jacobs et al., 2002). In a review of writing-related literature, Pajares and Valiante (2006) found that, in general, girls feel more competent about their writing skills than boys.
c Grade:
Developmental differences in writing motivation might also be expected within this study’s participants. Research conducted from variety of motivational perspectives suggests that student motivation generally declines as a student progresses through school (Eccles et al., 1998; Harter, 1981; Schunk et al., 2008). This trend has been observed in a number of domains, including writing (Pajares, 2003; Pajares et al., 2007).
6 Summary, purpose and research questions
Research with more homogenous groups within the wide range of students with specific learning difficulties is needed to form models of writing motivation better matched to these groups. Students with language impairments are an example of such a homogeneous group, and the nature of this disability holds important implications and challenges related to the writing process. Because language skills are essential to the writing process, it seems likely that the language difficulties faced by students with LI could negatively influence the perceived writing competence and subsequent quality of writing motivation. Thus, it would be beneficial to investigate this relationship further. It is currently unclear how significant language impairments relate to students’ perceptions of competence and to their overall writing motivation. It also still remains to be determined whether language impairments have a unique relationship with writing motivation that is independent of spelling ability, gender and grade. Research focused on these possibilities has the potential to help inform the design of interventions aimed at improving writing motivation in students with language impairments, as well as educators’ ability to provide greater specificity in matching instruction with student exceptionalities. Three research questions were proposed:
Are there differences in perceived writing competence and autonomous writing motivation between students with LI and TD peers?
Does LI status predict perceived writing competence, after controlling for grade, gender and spelling ability?
Does perceived writing competence mediate the relationship between LI status and autonomous writing motivation, after controlling for spelling ability, gender and grade?
The hypotheses follow from these questions and the relationships shown in Figure 2. The first of these is that students with LI will report lower levels of perceived writing competence and autonomous writing motivation than their TD peers. It is also hypothesized that perceived writing competence mediates the relationship between LI status and autonomous writing motivation.
This hypothesized mediated relationship was tested by following the logic of Baron and Kenny (1986), who describe four steps in determining the presence of a mediating variable:
determining if there is a direct link between the initial variable (LI status) and the dependent variable (autonomous writing motivation);
determining if the initial variable (LI status) is related to the mediator (perceived writing competence;
determining if there is a link between the mediator (perceived writing competence) and the dependent variable (autonomous writing motivation); and
determining if the link between the initial variable (LI status) and dependent variable (autonomous writing motivation) becomes insignificant when the mediator (perceived writing competence) is present.
II Methods
1 Participants
Thirty-three students with specific language impairments (LI) and 239 typically developing (TD) peers, aged 8 to 10 years, and in 11 schools in a midwestern state of the USA participated in the study. Students in a total of 31 classrooms were surveyed. Demographic and geographic characteristics of the participating schools are presented in Table 1. In general, the students attended schools in rural towns (population <10,000) or micropolitan cities (population 10,000–50,000). The median age for each group is reported in Table 2.
Demographic and geographic characteristics of participating schools
Participant characteristics by grade, gender and LI status
a Language impairment group:
Students with LI (n = 33) had been evaluated or re-evaluated within three years of this study, and met the same state eligibility requirements for speech language impaired in the area of language as determined by their local IEP (individualized education program) team. All students in the LI group had an assessed IQ above 85 and scored more than 20 standard score points lower than their assessed ability level on at least one comprehensive measure of language. In addition, the students’ local districts had determined that their language abilities were having a significant impact on their educational development. All students who met the state eligibility requirements for speech language impaired were evaluated by a certified speech language pathologist who held a master’s degree and the American Speech-Language Hearing Association’s certificate of clinical competence. The primary investigator gathered eligibility information in collaboration with the speech language pathologists at the participating schools. Tests used included the Clinical Evaluation of Language Fundamentals, 4th edition (CELF-4; Semel et al., 2003), Comprehensive Assessment of Spoken Language (CASL; Carrow-Woolfolk, 1999), Test of Language Development: Primary, 4th edition (TOLD-P:4; Hammill and Newcomer, 2006), Wechsler Intelligence Scale for Children, 4th edition (WISC-IV; Wechsler, 2003) and Comprehensive Test of Nonverbal Intelligence, 2nd edition (CTONI-2; Hammill et al., 2009). The mean reported IQ score for students with LI was 95.57 (SD = 6.38), while the average reported comprehensive language score (e.g. CELF-4 Core Language, CASL Core Composite) was 74.82 (SD = 6.89).
Because language impairments often co-exist with other disabilities, several exclusion/inclusion decisions were made. Students with intellectual disabilities (IQ standard score < 85), hearing loss, syndromic conditions and autism were excluded because these disabilities present pervasive challenges that are not language specific. Students with co-existing non-pervasive disabilities, such as speech production disabilities and reading disabilities, were included.
b Typically developing (TD) peer group:
TD peers in the same classroom as the students with language impairments also were surveyed. These students were not eligible for special education services and did not receive special education services of any kind. In the participating schools, parent consent letters were sent home with all students in classrooms that had at least one student with LI. The overall response rate for all schools was 53%.
Participant characteristics are presented in Table 2. There were more boys with LI (n = 20, 61%) than girls (n = 13, 39%). However, there were an almost equal number of boys and girls in the TD group (male: n = 121; female: n = 118). In the overall sample, there were fewer 5th graders (n = 79) than 3rd graders (n = 101) and 4th graders (n = 92).
2 Measures
a Spelling ability:
The measure of spelling included in the present study was designed to provide an indicator of the students’ ability to spell single words of varying difficulty. The primary researcher asked students to spell 20 grade-appropriate spelling words from the Test of Written Spelling, 4th edition (TWS-4; Larsen et al., 1999). The authors of the TWS-4 suggested that these 20 words were appropriate for group administration. Students were given a sentence to accompany each spelling word, and each item was scored as either correct or incorrect. The students’ responses in each grade were equally distributed and possessed adequate reliability, as measured by Cronbach’s alpha (third grade = .78, fourth grade = .83, fifth grade = .77). Students’ raw scores were converted to within-grade z-scores.
b Student perceived writing competence:
In this study, LI status and spelling ability were hypothesized to affect perceived writing competence. It also was predicted that perceived writing competence is a determinant of autonomous writing motivation. For these reasons, a measure of perceived writing competence was included in this study. Students completed an adapted version of Bouffard et al.’s scale (2003), which originally was designed to measure perceived competence within an academic domain. While the original scale measured perceptions of competence in the domains of reading and mathematics, it was adapted in the present study to reflect perceptions of writing competence. For each of the nine items, two groups are depicted and the students must indicate which group more closely describes their beliefs about writing. In addition, the student indicates if that group’s beliefs are ‘Really true for me’ or ‘sort of true for me’. Example statements include ‘Some kids think that they are among the best in writing in their class’ and ‘Some kids find writing very difficult to them.’ An example item is below:
Factor analysis confirmed that all items loaded on one factor. Internal consistency for this measure as determined by Cronbach’s alpha was acceptable for both groups (LI group = .80, TD group = .78).
c Autonomous writing motivation:
Six items from the Academic Self-Regulation Questionnaire: Adapted (SRQ-A; Deci et al., 1992), were adapted to assess students’ style of writing motivation, and were administered by the primary investigator. As mentioned earlier, autonomous motivation is conceptualized as engaging in an activity for internal, as opposed to external, purposes. The students were presented with reasons for engaging in writing activities that are representative of an autonomous writing orientation. The students then responded on a 4-point Likert-type scale, reporting how often (always, most of the time, sometimes, never) they demonstrate that behavior. An example item is ‘I try to do well on my writing assignments because I like doing a good job on my school work.’
The two autonomous subscales (identified regulation, integrated regulation) were combined to create one autonomous motivation scale (Cronbach’s alpha = .78). Factor analysis was used to determine that the six items used for the autonomous motivation scale loaded onto one factor.
3 Data collection
The primary investigator then administered the measures to students with parental consent for participating in the research. The students’ classroom teacher determined whether a small group or classroom setting would be optimal for the students and the class schedule. For this reason, the measures were administered both in small and larger groups. Following the procedures of Graham et al. (2007), the primary investigator introduced the measures by stating:
I am interested in how you and other kids feel about writing. So, I am going to ask you some questions. This is not a test or anything that you need to worry about. It won’t affect your grades. I will be the only person who sees your answers. I will not be sharing them with your teacher. Just try to answer my questions as honestly as you can.
The primary investigator also gave instructions for completing the scales, which were read aloud. All data were collected between November and February of the same school year. It took students approximately 30 minutes to complete all measures.
III Results
This study examined the relationship between LI status (LI or TD) and writing motivation by investigating if students with language impairments reported lower levels of writing motivation. The current study also investigated if perceived writing competence mediates the relationship between LI status and autonomous writing motivation. Descriptive data are presented first, followed by t-tests comparing the perceived writing competence and autonomous motivation of students with LI and their TD peers. Next, the results of two stepwise regression analyses are presented to determine if:
LI status is a significant predictor of perceived writing competence after controlling for spelling ability, gender and grade; and
if perceived writing competence mediates the relationship between LI status and autonomous writing motivation after controlling for spelling ability, gender and grade.
1 Descriptive data
a Spelling:
The TD (M = 14.7, SD 3.6) group performed significantly better on the spelling task, t(36) = 7.0, p < .01, than the students with LI (M = 8.4, SD 5.2), and there was more variability of scores within the LI group (Table 3). In addition, girls (M = 14.5, SD = 3.8) scored significantly higher than boys (M = 13.5, SD = 4.8) in spelling in the overall sample, t(270) = 2.4, p = .01. There were no gender differences in the LI group, t(31) = .83, p = .41.
Comparison of spelling raw scores by grade, gender and LI status
Note: spelling range = 0–20
b Perceived writing competence:
Perceived writing competence means are presented in Table 4. There was a significant difference between males (M = 2.8, SD = .64) and females (M = 3.0, SD = .53), t(250) = 2.6, p = .01. There was also a significant difference between the LI group (M = 2.4, SD = .68) and the TD group (M = 3.0, SD = .55), t(250) = 5.3, p = .01. There were no gender differences within the LI or TD groups.
Comparison of perceived writing competence mean by grade, gender and LI status
Note: range = 1–4
c Autonomous writing motivation:
In the TD group, there was a significant difference between males (M = 2.78, SD = .75) and females, (M = 3.10, SD = .58), t(233) = 3.70, p < .01. However, there were no gender differences in the LI group, t(31) = 1.44, p = .161, and also no significant differences between grades, F(2, 265) = 2.06, p = .129 (for descriptive data, see Table 4).
As seen in Table 5, the strongest relationships observed were between spelling and LI status (r = .50), perceived writing competence and autonomous motivation (r = .45), perceived competence and spelling (r = .38), perceived competence and LI status (r = .32), and autonomous motivation and gender (r = .24). Significant correlations also existed between perceived competence and gender (r = .16), spelling and gender (r = .15), autonomous motivation and LI status (r = .13), and spelling and autonomous motivation (r = .12).
Zero-order correlations between study variables
Notes: * = p < .05, ** = p < .01, NA = correlations between categorical variables were not reported
d Motivational differences between schools:
A one-way analysis of variance (ANOVA) was conducted to determine if between-school differences existed among the 11 participating schools for the two motivational variables, perceived writing and autonomous writing motivation. No between-school differences existed for perceived writing competence, F(10, 240) = 1.42, p = .17, or for autonomous writing motivation, F(10, 256) = 1.319, p = .22.
2 Analysis of the research questions
An independent-samples t-test was conducted to evaluate the hypothesis that students with LI would report lower levels of perceived writing competence and autonomous writing motivation than their TD peers (Table 6). Because there were no significant differences between schools on these measures, all students were included. Differences in perceived writing competence were significant, t(250) = 5.293, p = < .01. Students in the LI group (M = 2.42, SD = .68) reported significantly lower levels of perceived writing competence than students in the TD group (M = 2.98, SD = .55). Cohen’s d was .91, indicating a large effect size (Cohen, 1988). Differences on the autonomous writing motivation measure also were significant, t(266) = 2.05, p = .04. Students in the LI group (M = 2.68, SD = .72) reported significantly lower levels of autonomous writing motivation than students in the TD group (M = 2.94, SD = .68). Cohen’s d was.37, indicating a small to medium effect size.
Raw score averages on perceived writing competence scale and autonomous writing motivation scale for LI and TD students
Notes: * = p ≤ .05, ** = p ≤ .001; standard deviations appear in parentheses below means; range = 1–4
A multiple regression analysis was conducted to evaluate the second hypothesis, that LI status would predict perceived writing competence over and above gender, grade and spelling ability (Table 7). An ordered series of predictors was entered stepwise to the regression model. In the first two steps, gender and grade were entered into the model to act as controls for subsequent variables. In the third step, spelling ability accounted for a significant proportion of the perceived writing competence after controlling for gender and grade, R2 change = .13, F(1, 246) = 37.70, p < .01. In the fourth step, LI status accounted for a significant proportion of the perceived writing competence after gender, grade and spelling ability were controlled, R2 change = .02, F(1, 245) = 6.256, p = .01. The final model’s effect size, Cohen’s f2 = .25, indicated a medium to large effect. The interaction between LI status and spelling was not a significant predictor in the final model. Multicollinearity diagnostics were considered, but there was no evidence of problematic multicollinearity among variables.
Summary of hierarchical regression analysis for variables predicting perceived writing competence
Notes: * = p < .05, ** = p < .01
A second stepwise regression analysis was conducted to test the current study’s third hypothesis that perceived competence mediates the relationship between LI status and intrinsic writing motivation. More specifically, this analysis evaluated:
if there is a direct link between LI status and autonomous writing motivation;
if there is a link between perceived writing competence (hypothesized mediator) and autonomous writing motivation; and
if the link between LI status and autonomous writing motivation becomes insignificant when perceived writing competence is entered into the regression equation (Table 8).
Summary of hierarchical regression analysis for variables predicting autonomous writing motivation
Notes: * = p < .05, ** = p < .01
The control variables, gender and grade, again were entered in the first two steps of the ordered sets of predictors. In the third step, spelling ability did not account for a significant proportion of the variance after controlling for grade and gender, R2 change = .01, F(1, 243) = 1.61, p = .21. LI status also did not account for a significant proportion of the variance when added to the model in the fourth step, R2 change = .004, F(1, 242) = 1.21, p = .27. However, in the fifth step, perceived competence was a significant predictor of autonomous writing motivation above and beyond gender, grade, spelling and LI status, R2 change = .16, F(1, 241) = 49.79, p < .01. Again, Cohen’s f2 indicated a moderate to large effect size for the final model (f2 = .33)
Interaction terms were computed to determine if spelling and perceived competence had different effects based on group membership (e.g. gender and LI status). Interaction terms were calculated by multiplying the continuous zxzvariables by the categorical variables (e.g. gender × spelling) and then entered into the regression model as independent variables. None of the interaction terms were significant and thus are not reported in Table 8.
IV Discussion
The present study resulted in three main findings:
Students with LI do in fact report lower levels of perceived writing competence and autonomous writing motivation.
LI status was a significant predictor of perceived writing competence even when spelling ability, gender and grade are controlled; and
Perceived writing competence was a significant predictor of autonomous writing motivation when spelling ability, grade and gender are controlled, but LI status was not.
Therefore, the hypothesized mediated relationship was partially but not fully supported. The general implications of these findings will be discussed below.
The first analysis, which compared scores of LI students with those of their TD peers, suggests that students with language impairments are more likely to report lower levels of perceived writing competence and autonomous writing motivation. The effect size for these differences in perceived writing competence between LI students and their TD peers was large, while the effect size for the difference between groups on autonomous motivation was small to medium. From these analyses, it seems reasonable to suggest that students with LI are, on average, considerably less likely to be autonomously motivated to write and more likely to have lower perceived writing competence. Even though these simple comparisons suggest that students with LI are less motivated to write, they do not consider other variables that may explain this relationship. The regression analyses and subsequent follow-up analyses, discussed below, provide additional information useful for understanding these relationships.
The second analysis, which utilized multiple regression to predict the students’ perceived writing competence, provided additional insight into the relationship among the variables contributing to perceived writing competence. Each of the independent variables of grade, gender, spelling and LI status was significantly related to perceived writing competence (R = .44), with approximately 20% of the variance in perceived writing competence explained by these four variables. Spelling ability also was a significant predictor of perceived writing competence, even after controlling for gender and grade. This finding can be seen as consistent with Hayes’s (2000) writing cognition model, as difficulty with spelling presumably can demand more working memory resources during the writing process. It also may be that spelling errors are salient visual signals of low writing competence, and that these errors often are the target of negative feedback (Graham and Harris, 2003). However, these possible explanations cannot be determined from the present findings, but could be important areas for future research. LI status also was a significant predictor of perceived writing competence, even after controlling for spelling, grade and gender. In the final model, spelling and LI status had the strongest influence on perceived writing competence. The interaction between LI status and spelling was not significant, suggesting that spelling ability similarly influenced the perceived writing competence both of students with LI and their TD peers.
Since LI status was a significant predictor of perceived writing competence, it seemed reasonable to hypothesize that LI status also would be related to students’ autonomous writing motivation. However, the third analysis did not support this hypothesis. LI status was not a significant predictor of autonomous writing motivation after spelling, gender and grade were controlled. In contrast, however, perceived competence was a strong predictor above and beyond the influence of grade, gender, LI status and spelling. The final model consisting of grade, gender, spelling, LI status and perceived competence showed that these variables were significantly related to autonomous writing motivation (R = .50) and explained 26% of the variance. Perceived competence was by far the strongest predictor of autonomous writing motivation, however. This finding is consistent with one of the central tenets of SDT: that perceived competence is one of the primary determinants of autonomous motivation. Interestingly, gender was the next most influential variable, as girls were more autonomously motivated to complete writing tasks. By comparison, LI status and spelling showed little direct relationship to autonomous writing motivation. Because there were significant autonomous writing motivation differences between the LI students and their peers as revealed by an earlier analysis, however, mediating and moderating variables were next considered.
As shown in Figure 2, it was hypothesized that perceived competence would mediate the relationship between the, LI status and the dependent variable, autonomous motivation. Overall, the data did not support this hypothesis. First, there was no direct link between LI status and autonomous writing motivation after controlling for spelling ability, gender and grade. Although the initial t-test indicated that students with LI reported lower autonomous writing motivation, this relationship was not significant in the regression analyses when spelling ability, gender and grade were entered as control variables. Second, the standardized beta weight for LI status was not significantly reduced when perceived competence was entered into the regression model.
One possible explanation for this unexpected finding begins with the realization that perceived competence is only one of the three major variables – perceived competence, autonomy and relatedness – that are proposed by SDT to contribute to autonomous motivation. While LI status was related to perceived competence, it seems less likely that LI status would be related to a student’s sense of autonomy or relatedness. Autonomy and relatedness could be considered sociocultural variables that are external to the student, whereas perceived competence is primarily determined by intra-individual skills. For example, the classroom teacher may employ teacher-centered controlling practices, such as pressuring communications or giving few student choices, which could reduce a student’s sense of autonomy. A teacher’s use of autonomy-supportive or controlling practices seems largely unrelated to a student’s language abilities, however. It is more likely that a teacher’s writing instruction approach would be consistent across students in the classroom regardless of ability.
On the surface, it appears that the students’ sense of relatedness also would be determined by interpersonal interactions that are relatively independent of language ability. However, there does seem to be a relationship between negative peer interactions and academic engagement in early elementary general education students (e.g. Buhs et al., 2006). To date, though, it is still unknown to what degree a student’s significant difficulties with language impact their sense of classroom relatedness and subsequent writing motivation. Nonetheless, it is likely that inclusion of these sociocultural variables, autonomy and relatedness, into a future study could significantly enhance a model relating this study’s variables to autonomous writing motivation.
In addition, the nature of the dichotomous LI status variable might have hindered its contribution to the proposed regression model. That is, students who may have had a great disparity in language ability were represented in the analysis by either a ‘0’ (TD) or ‘1’ (LI). If language ability had been measured by a continuous variable, however, such a variable presumably would be more sensitive to the relationship between language and autonomous writing motivation.
Gender’s role in predicting autonomous motivation also is an interesting finding in the present study. Consistent with a number of investigations (for a review, see Pajares, 2003), girls in this study reported higher levels of perceived writing competence and autonomous motivation for writing and performed significantly better than boys on the spelling measure. Gender also was influential in the regression analysis, as spelling and LI status were not significant predictors of autonomous motivation when controlling for grade and gender. Because there were no significant interactions with gender, it does not seem that gender moderates the relationship between spelling, LI status, perceived competence, and autonomous writing motivation. In follow-up testing, there were no between-gender autonomous motivation differences within the LI group. In addition, spelling and LI status were not significant predictors of autonomous writing motivation for either gender. It appears that the girls were simply stronger in each of the major study variables (perceived writing competence, autonomous writing motivation and spelling skills), and that gender contributed more to the autonomous motivation regression model than LI status.
1 Implications for practice
Writing is an ill-defined, cognitively complex task. This study suggests that it is possible this complexity may influence the writing motivation of students with LI. For this reason, educators would be advised to draw from the wealth of writing strategy research that has been conducted with at-risk writers and which is aimed at reducing the complexity of the writing process (e.g. Graham and Harris, 2006). These writing strategies have been shown to help students utilize writing goals and procedures that break the complicated writing task into more manageable parts. Theoretically, this should free working memory resources for students with LI and increase their sense of writing competence. Although there has not been extensive research utilizing writing strategies with students with LI, improved writing strategy use has been linked to greater writing competence in students with other literacy-related disabilities (Graham and Harris, 2006).
In addition, educators should be mindful of the role feedback plays in developing writing skills and motivational orientations. Feedback is a multi-dimensional construct, and can be considered in terms of its focus, complexity and context. It can be directed towards student performance, task-related processes, student self-regulation, or a student’s sense of self (Hattie and Timperley, 2006). Numerous studies imply a link between different types of effective feedback and student motivation (e.g. Schunk et al., 2008), and certain types of feedback would likely provide positive support to strategy instruction. In particular, ‘strategy feedback’, which helps students focus their attention on strategy use rather than performance outcomes, is one type of feedback that may be an effective complement to strategy instruction with students with LI. This type of feedback is designed to help students gain control over the self-regulatory processes and writing strategies, which subsequently may increase writing outcomes and perceived competence (Zimmerman and Martinez-Pons, 1992).
2 Limitations and conclusions
Studies of the writing motivation of students with specific learning difficulties have generally defined at-risk writers in terms of their reading or spelling abilities and have not gathered adequate information about students’ language ability. The current study, however, shows that significant language difficulties seem to hinder students’ perceived writing competence in the middle and late elementary grades.
Because writing increasingly is recognized as an effective context for inclusive language intervention (e.g. Nelson et al., 2004; Singer and Bashir, 2004), this study has the potential for informing the theory and design of interventions targeting writing competence and engagement. While teachers’ and clinicians’ observations of students with specific language impairments would seem to indicate that these students are less motivated to write than their TD peers, to date this had not been validated by research. If perceived writing competence is lower for students with language impairments, as the current study strongly suggests, this finding begins to establish the need to design instruction that addresses both writing skills and writing engagement in this population. In addition, this study supports the hypothesis that significant difficulties with language can contribute to lower levels of perceived competence in the early and middle elementary school years. Thus, this study is a first step in uncovering the unique motivational challenges that students with language impairments face as they attempt to develop as writers. As these challenges are further identified, educators will be more equipped to provide writing instruction that increases students’ both writing motivation and writing skills. In addition, the scales used in this study may also be useful in advancing the study of writing motivation and language impairments, as they are utilized or adapted for use in future studies.
One limitation of the current design was the reliance on local decision-making regarding eligibility for special education services. Even though the same state criteria were used and the SLPs were adequately credentialed, there is a possibility that some variation in eligibility decisions existed between schools. The author attempted to minimize this variation by obtaining test scores from highly-qualified speech language pathologists and making sure that all students met the inclusion criteria.
Future research utilizing including longitudinal and qualitative dimensions may be particularly useful in forming a more complete understanding of writing motivation in students with language impairments. These types of designs would allow researchers to investigate how the writing motivation of students with LI develops over time, as opposed to the current study’s design that only allows for a one-time measurement of student attitudes. It would also be beneficial to include more direct behavioral observations and a richer description of writing engagement in future studies to augment the self-report data gathered in the current study. Because the current study was conducted within a region that is generally more homogeneous than other areas of the United States, it will be valuable to include more diverse student populations in future studies. It also is important to stress that, while this study primarily focuses on student variables, factors that are both internal and external to students work reciprocally in the development of writing motivation and competence. As Berninger and Hidi (2007) have stated, ‘the quality and appropriateness of instruction is just as important as the psychological attributes and attributions of the learner in explaining motivated learning behavior in general and for writing in particular.’ It is clear that future studies must consider contextual factors and teacher beliefs that contribute to the writing motivation of students with language impairments. SDT posits that autonomy and relatedness, not measured in this study, also likely are variables worthy of future studies. In addition, while the focus of the current study was on perceived writing competence, gathering objective measures of writing competence in future studies would increase our understanding of the bi-directional relationship between the students’ perceptions of competence and their actual proficiency.
Ultimately, intervention studies are needed to determine conclusively if increased writing motivation can influence writing and language outcomes. It is not yet known if increased motivation during language intervention, in particular within writing contexts, can contribute to improved language competence in students with language impairments. By showing the differences that do exist in perceived competence and writing motivation, however, the current study provides a foundation for future studies investigating ties between increased writing motivation and subsequent language development in varying modes (e.g. speaking, listening and reading). For example, intervention studies could investigate the relationship between increased engagement produced by strategic written language instruction and overall language development.
Although the current study suggests that language and spelling deficits contribute to decreased perceived writing competence, the precise reason for these relationships is not entirely clear. As hypothesized earlier, it could be that these difficulties impose high levels of cognitive load that negatively influence students’ writing attitudes. This relationship could also be related to the feedback and social comparisons that result from inferior spelling and language skills. Future research is needed to determine key factors that may mediate the relationship between these skills and writing motivation.
Finally, this study’s findings provide a rationale for continued study of the relationship between LI and writing motivation and for the integration of motivational dimensions into writing instruction for students with LI. While it seems that students with LI may be more likely to report lower levels of perceived writing competence, this study also suggests that environmental variables also contribute to these students’ writing motivation. Research and research-based interventions targeted at students with LI has the potential to further our general understanding of writing motivation and effective instructional practices, which would be valuable in the pursuit of high quality language and literacy instruction for all students.