Quantitative Research Study
Down the Road: Does Phonological Awareness Predict Reading Ability in the Early Grades?
Denise K. Robinson
George Mason University
Author Note
Denise K. Robinson, College of Education and Human Development, George Mason University. This paper was prepared in partial completion for EDRS 811 Spring, 2012
Correspondence concerning this article should be addressed to Denise K. Robinson, College of Education and Human Development, George Mason University, Fairfax, VA
E-mail: [email protected]
Abstract
This quantitative study utilized previously collected longitudinal data to investigate the influence of kindergarten phonological awareness on reading levels for 104 students in kindergarten, first, and second grade. Additionally, gender and ethnicity were used as independent variables to determine their possible significance on phonological awareness and reading achievement. Findings indicated that kindergarten phonological awareness scores did significantly predict first and second grade reading achievement. Gender was found to have a significant effect only for second grade reading levels; the significance of ethnicity on reading was found to diminish from kindergarten to second grade. Down the Road: Does Phonological Awareness Predict Reading Ability in the Early Grades?
Phonological awareness is the knowledge that oral language is made up of a number of structures that are separate from meaning, such as syllables (units of spoken words), onset (consonants that begin syllables) and rime (the remainder of the syllable), and phonemes (the smallest meaningful unit of sound that make up syllables and words) (Burns, Griffin, & Snow, 1999; Goswami, 2000). In other words, phonological awareness is the understanding that speakers can manipulate the letters and sounds that make up the words we say. Well-developed phonological awareness has been found to be a strong predictor of a young student’s ability to read and spell (Goswami, 2000). Therefore, many school systems have begun to assess kindergarten students’ level of phonological awareness, so that interventions can be put in place for struggling young readers (Linklater, O’Connor, & Palardy, 2009; McCoach, O’Connell, Reis, & Levitt, 2006). Student reading achievement levels are also typically gathered at the end of each grade level. This study will seek to determine whether kindergarten students’ level of phonological awareness can predict future reading success.
Phonological Awareness and Early Literacy
Children bring a variety of early literacy experiences with them when they enter kindergarten. Some will have spent time in preschool classes or child care settings. Others will have experienced the world with their families and friends. Most will have developed strong oral language skills that will serve as the basis for learning to read and write (Burns, Griffith, & Snow, 1999). The ability to manipulate units of oral language—rhyming, alliteration, breaking words apart and putting them back together—underlies the child’s phonological awareness. Students whose background experiences and use of language do not prepare them to build strong phonological awareness are at risk for later reading difficulties (Linklater, et al., 2009).
Researchers have investigated the relationships between specific phonological skills and reading achievement and have found strong, positive correlations between skills such as identifying letters and their sounds (the alphabetic principle), dividing words into onset and rimes (the first sound and the remainder of the word or syllable), or phoneme (sound) deletion and blending and later reading success (Goswami, 2000). Building on these relationships, school systems have begun to assess young student’s phonological awareness in an attempt to focus classroom instruction and identify students in need of specific, targeted intervention (Linklater, 2009; McCoach, et al., 2006). This is important because research has shown that readers who struggle in the early grades are often identified as struggling readers in later grades (Goswami, 2000).
In a recent study, Nithart, Demont, Metz-Lutz, Majerus, and Poncelet (2011) investigated the simultaneous development of phonological and reading skills. Phonological discrimination, phonological awareness, and phonological memory were assessed at the end of the students’ kindergarten year and again at the end of first grade. Phonological discrimination and awareness were found to correlate strongly with reading skills in kindergarten. The researchers found that phonological awareness continued to grow from kindergarten to first grade, along with an increase in reading skills. Phonological memory also developed over this time span—first as a more separate construct than phonological awareness, and then converging with phonological awareness skills as reading skills developed.
In a study that utilized data from the Early Childhood Longitudinal Study (ECLS), Chatterji (2006) looked at the effects of different ethnic, gender, and socio-economic groups on first grade reading achievement. Using data gathered in both kindergarten and first grade, Chatterji found that gaps in literacy achievement based on gender (males lower than females) increased from kindergarten to first grade but that gaps based on ethnicity (African American and Hispanic students lower that Caucasian) were somewhat mitigated by instruction between kindergarten entry and first grade.
A strong body of research has been developed to document the importance of phonological awareness to reading achievement and to investigate many of the factors that influence the development of phonological awareness. At the same time, research has shown that readers who struggle in the early grades often remain struggling readers as time progresses. However, a gap in the research exists regarding the longitudinal effect of phonological awareness measured in kindergarten for the reading success of a sample of students in first and second grade. This study addresses that gap by using existing longitudinal data to determine whether phonological awareness can predict reading achievement over time.
Rationale and Purpose
In many school districts across the United States, children in kindergarten are assessed to determine their mastery of letters and their sounds, segmenting words into syllables, identifying rhyming words, and demonstrating an understanding of beginning reading skills. Later, reading fluency and comprehension are measured over time, as students move from grade to grade.
However, it is unclear whether data from a particular test of kindergarten phonological awareness can predict reading ability in first and second grade readers. It is the intent of this study to determine whether, in one school, scores on the DRA Word Analysis (WA; Pearson, 2005) can predict scores on the Developmental Reading Analysis (DRA; Pearson, 2005) for first and second graders at that school. Using longitudinal data from a suburban, Title 1 elementary school, scores from two consecutive school years were collected and analyzed to answer the following questions:
1. Are there differences on WA scores in kindergarten by gender and ethnicity?
Research findings regarding gender, ethnicity and phonological awareness to date have been mixed. Some suggest that early literacy experiences are more important to phonological awareness than gender. At the same time, students with limited English proficiency who have enough English to take the WA may struggle more with phonological awareness in a second language, even with good teaching. Based on these findings, we can hypothesize that there are no differences between boys and girls on WA scores and that Hispanic students will score lower on the WA than non-Hispanic students.
2. Are there associations between gender and ethnicity on kindergarten reading levels, on DRA scores in first grade, and on DRA scores in second grade? Like the research on phonological awareness, the research on early literacy skills and gender and ethnicity is mixed. Some studies suggest that gender and ethnicity have less of an impact on achievement in the early years and that this changes as students are faced with more difficult tasks. Therefore, we can hypothesize that there are no significant associations between gender and ethnicity and kindergarten reading levels but that there are significant associations between gender and ethnicity and DRA scores in first and second grade.
3. Is the first grade DRA score predicted by the WA score, kindergarten reading level, and gender? Similarly, is the second grade DRA score predicted by the same independent variables (WA, kindergarten reading level, and gender)? Based on research that suggests that phonological awareness and early reading skills are predictors of later reading success, we can hypothesize that the WA score and kindergarten reading level is a significant predictors of the first and second grade DRA scores. Further, we can hypothesize that due to changes over time gender is not a significant predictor of first grade DRA scores, but is a significant predictor in second grade.
4. To what extent does the interaction between gender and ethnicity account for the variance in kindergarten WA scores? I would expect that the interaction between gender and ethnicity will account for at least some of the variance in WA scores. I would hypothesize that Hispanic boys may have lower scores than other gender/ethnic group combinations.
Method
This quantitative study utilizes longitudinal data regarding student reading performance in kindergarten through second grade. Phonological awareness and reading achievement data is recorded for each K-3 student in a large, mid-Atlantic public school system. Schools throughout the district use the same assessment protocols and teachers administering the assessments must be trained in their use and scoring, allowing for some measure of consistency from school to school. Data used in this study was obtained from one elementary school in the school system.
Sample
The sample population for this study consisted of students who attended one suburban elementary school in a major mid-Atlantic metropolitan area. The school is a Title 1 school; more than 50% of the 520 students qualify for free and reduced lunch subsidies. It offers a full-day kindergarten program, as well as Head Start and preschool special education classes. Mobility for the school is approximately 30%. Data was collected for a total of 104 students (male, n = 60; female, n = 44) who attended the school from kindergarten through second grade, beginning in two consecutive school years (2007-08 and 2008-09). Demographic data for the sample is provided in Table 1.
= = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = =
Insert Table 1 about here
= = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = =
Data
The measures used in this study were part of a reading assessment program, the Developmental Reading Assessment program (DRA; Pearson, 2005), designed to determine a student’s level of phonological awareness and reading ability, including both fluency and comprehension. The assessment program begins in kindergarten and assesses learners as they progress through elementary school. Both the WA and DRA protocols are administered individually to students by the classroom teacher or qualified reading specialist in the fall and spring, with an optional assessment window available mid-year.
Measure of phonological awareness. The DRA Word Analysis (WA; Pearson, 2005) is a criterion referenced diagnostic assessment that provides the classroom teachers with information on how struggling and emergent readers understand words in context and isolation and to aid in planning for instruction. The measure is divided into five strands, sequenced in order of difficulty: phonological awareness; metalanguage used to talk about print; recognition of letters and high frequency words; phonics; and structural analysis and syllabication. The WA is made up of 40 tasks, 11 of which are administered to kindergartners. The score range for these 11 tasks is 0 – 193. The benchmark score is 61 for the fall administration and 167 for the spring; the same form of the instrument is used for each administration. The tasks reflect skills measured by the DRA as students gain in reading fluency and comprehension and are presented in order of difficulty. Typically, kindergarten students who do not meet the benchmark score are eligible for specifically targeted intervention services. Struggling readers in higher grades may be assessed using the remaining tasks, at the discretion of the school. The WA is administered by classroom teachers, who are trained in using and scoring the protocol. Information regarding reliability and validity of the WA is available from the publisher.
Measure of reading achievement. Reading achievement was measured at the end of kindergarten, first, and second grades using the DRA (Pearson, 2005). Like the WA, the DRA is a criterion referenced protocol, administered individually by trained classroom teachers or reading specialists. Following a strong introduction and picture preview, the student reads a leveled text while the teacher records miscues. After reading the text, the student retells the story from memory and answers specific comprehension questions. Scores are obtained in three areas: reading engagement, oral reading fluency, and comprehension; the student’s proficiency is rated as independent, instructional, or frustration. Testing is repeated using texts at the next level until an instructional level is reached. Benchmarks for each grade level have been established: kindergarten—levels 3-4 (emergent); first grade—level 16 (developing, fiction and non-fiction); second grade—level 28 (expanding, fiction and non-fiction). Levels below benchmark are considered below grade level, at benchmark are considered on grade level, and above benchmark are considered above grade level. The DRA is typically administered in the fall and spring, with an optional winter window; alternate assessments are available for each reading level. Information on the reliability and validity of the DRA is available from the publisher.
Data collection
Data for this study was collected from existing student records. Longitudinal reading data for all students in the school district are recorded on the K-3 Student Achievement Record. Year end literacy scores are entered on a sheet that is kept with each student’s school records. Scores from the K-3 Student Achievement Record used in this study were the kindergarten Word Analysis score (WA), the final kindergarten reading level (klevel), and the first and second grade DRA scores (dra1 and dra2). The kindergarten reading level is scored as 3 (below grade level), 4 (on grade level), or 5 (above grade level); no specific DRA levels are provided. The dra1 and dra2 scores provide the specific reading levels attained by the student.
Access to the student records was granted by the building principal. All records were kept strictly confidential. Student gender and ethnicity were noted along with reading scores; however neither student names nor identification numbers were recorded.
Statistical data analysis
Are there differences on WA scores in kindergarten by gender and ethnicity? To answer this question two independent samples t- test analyses were performed. For the first, the kindergarten WA scores were entered as the dependent variable; gender (male/female) was entered as the grouping (independent) variable. For the second t-test, ethnicity (Hispanic/non-Hispanic) was entered as the grouping (independent) variable; kindergarten WA scores remained as the dependent variable. A 95 percent confidence level was chosen.
Are there associations between gender and ethnicity on kindergarten reading levels, on DRA scores in first grade, and on DRA scores in second grade? A series of chi-square tests for association were undertaken in order to answer this question. To test the hypothesis that there are no significant associations between gender and ethnicity and kindergarten reading levels two chi-square tests were conducted: the first with gender (male/female) as the independent variable and three levels of kindergarten reading (below grade, on grade, and above grade) as the dependent variable; the second with ethnicity (Hispanic/non-Hispanic) as the independent variable and kindergarten reading levels as the dependent variable. To test the hypothesis that there are significant associations between gender and ethnicity and DRA scores in first and second grade, similar chi-square tests for association were made. Two chi-square tests each were made for the dra1and dra2 scores, with gender or ethnicity as independent variables and below, on, and above grade level scores as the dependent variables.
Is the first grade DRA score predicted by the WA score, kindergarten reading level, and gender? Similarly, is the second grade DRA score predicted by the same independent variables (WA, kindergarten reading level, and gender)? Research has shown that strong levels of phonological awareness are a predictor of higher reading ability in later grades. Similarly, students who struggle with phonological awareness often struggle with reading, even as early as kindergarten. Readers who struggle in the early grades often continue to exhibit reading difficulties in subsequent grades, sometimes in spite of targeted interventions. In order to understand if these scenarios are at play in this elementary school, multiple linear regressions were used in order to determine whether the DRA scores (dependent variable) attained by first or second grade students at the end of the school year can be predicted by their WA scores, kindergarten reading levels, and gender (independent variables).
To what extent does the interaction between gender and ethnicity account for the variance in kindergarten WA scores? Research has shown that well developed levels of phonological awareness are important prerequisites for higher reading achievement in higher grades. Therefore, in order to determine the extent to which interactions between gender and ethnicity account for the variance in WA scores, a two-way ANOVA analysis was run. Kindergarten WA scores were entered as the dependent variable, with gender and ethnicity entered as fixed factors.
Results
Are there differences on WA scores in kindergarten by gender and ethnicity?
Gender. The results of Levene’s test for equality of variances shows that equal variance can be assumed, F = 1.056 (p = .307). Further, the results from the independent t-test indicate that there is no significant difference between males and females on kindergarten WA scores, t (102) = -.87 (p = .384). Therefore, the null hypothesis, that there are no significant differences between boys and girls on WA scores in kindergarten is retained.
Ethnicity. Again, the results of Levene’s test for equality of variance shows that equal variance can be assumed, F = 1.22 (p = .272). However, the results from the independent t-test suggest that there is a statistically significant difference between Hispanic and non-Hispanic students on kindergarten WA scores, t (102) = -2.36 (p = .02). The 95% CI for the difference indicates that non-Hispanic students exceed Hispanic students by an amount between 1.4 and 16.4 points. Therefore, the null hypothesis must be rejected and the alternative hypothesis, that there is a significant difference in WA scores by ethnicity is retained.
Are there associations between gender and ethnicity on kindergarten reading levels, on DRA scores in first grade, and on DRA scores in second grade?
Kindergarten reading level. The results from the chi-square test for associations between gender and the klevel indicate that there is no significant association between the two variables. That is, boys and girls had no significant difference in their reading levels at the end of kindergarten. Therefore, we fail to reject the null hypothesis, that there is no association between gender and kindergarten reading level. However, the results of the chi-square test on ethnicity indicate that there is a significant association between ethnicity and reading level in kindergarten for the students in this study, χ2(2) = 12.28, p = .002. Further, the standardized residuals indicate that a major contributor to the association is the under-representation of Hispanic students in the above grade group (SR = -2.2). Thus, we must reject the null hypothesis, that there is no association between ethnicity and kindergarten reading scores.
First grade reading level. As with the results in kindergarten, the results of the chi-square test did not suggest a significant association between gender and reading level in first grade. Again, the null hypothesis is retained. However, while the chi-square test for ethnicity and reading level in first grade did suggest a significant association between the two for the students in this study, χ2 (2) = 7.32, p = .026, there does not appear to be a major contributor to this association, as indicated by the standardized residuals. Thus, while Hispanic students were somewhat under-represented in the above grade level group and over-represented in the below grade group, and non-Hispanic students were somewhat over-represented the above grade group, none of these associations were found to be significant. Nevertheless, the null hypothesis must be rejected.
Second grade reading level. Unlike the previous results for the kindergarten and first grade reading levels, the chi-square test for association between gender and second grade reading levels indicates a significant association between the two, χ2 (2) = 13.14, p = .001. Specifically, the standardized residuals indicate that major contributors to this association are the significant over-representation of females in the above grade level group and the under-representation of males in this group. This result confirms previous findings that gender impacts reading scores as students move to higher grades, but in relation to this study the null hypothesis (that there is no significant association between gender and second grade reading levels) must be rejected. Similar to results for first grade reading levels, the results of the chi-square test for association between ethnicity and reading in second grade show a significant association between the two, χ2 (2) = 8.09, p = .017. Again, no main contributors were indicated by the standardized residuals. The results do indicate that Hispanic students are somewhat over-represented in the below grade level group and over-represented in the above grade group, opposite to the findings for non-Hispanic students.
Is the first grade DRA score predicted by the WA score, kindergarten reading level, and gender? Similarly, is the second grade DRA score predicted by the same independent variables (WA, kindergarten reading level, and gender)?
First grade DRA score. The results from the omnibus F-test indicate that the prediction of DRA scores in first grade from kindergarten WA scores, reading level in kindergarten, and gender is statistically significant, F (3, 100) = 11.55, p = .000. Further, R2 = .257 indicates that 25.7% of the variance in first grade DRA scores (dra1) is explained by the variance in kindergarten WA (WA), readinglevels in kindergarten (klevel), and gender (gen). The regression equation in this case is = .085 ( The regression coefficient for WA is statistically significant (p = .000) but the regression coefficients for kindergarten reading level (p = .057) and gender (p = .264) are not. In other words, the unique contribution of the WA score to the prediction of first grade reading levels is statistically significant, but the contributions of kindergarten reading levels and gender are not. Specifically, the squared part correlation for WA (.316)2 =.10 shows that 10% of the variance in first grade DRA scores is explained by kindergarten WA scores.
Second grade DRA score. The results from the omnibus F-test indicate that the prediction of DRA scores in second grade from kindergarten WA scores, reading level in kindergarten, and gender is statistically significant, F (3, 100) = 11.382, p = .000. Further, R2 = .255 indicates that 25.5% of the variance in second grade DRA scores (dra2) is explained by the variance in kindergarten WA (WA), readinglevels in kindergarten (klevel), and gender (gen). The regression equation in this case is = .094 ( The regression coefficient for WA (p = .001) and kindergarten reading levels (p = .046) are statistically significant but the regression coefficient for gender (p = .096) is not. In other words the unique contributions of kindergarten WA and reading level scores significantly predict end of the year DRA scores in second grade, but the contribution of gender does not. Specifically, the squared part correlation for WA (.291)2 =.084 shows that 8.4% of the variance in second grade DRA scores is explained by kindergarten WA scores and the squared part correlation for kindergarten reading level (.174)2 = .03 shows that 3% of the variance in second grade scores is explained by kindergarten reading levels.
To what extent does the interaction between gender and ethnicity account for the variance in kindergarten WA scores?
In order to answer this question, a two-way ANOVA analysis was run, with kindergarten WA scores entered as the dependent variable and gender (male, female) and ethnicity (Hispanic, Caucasian, African American, Asian) entered as fixed factors. The means and standard deviations are reported in Table 2.
= = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = =
Insert Table 2 about here.
= = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = =
The Levene’s test for equal variances indicates that the assumption of homogeneity of variances is not met, F (6, 97) = 2.428, p = .031. The results of the omnibus F-test indicate that there is a statistically significant main effect for ethnicity, F (3, 97) = 2.97, p = .036, pη2 = .084. There is no significant main effect for gender (p = .569) or for gender and ethnicity together (p = .403). Further, 8.4% of the variance in WA scores in kindergarten is accounted for by the unique contribution of differences in ethnicity, controlling for all other effects. The results of the analysis of variance for WA scores are provided in Table 3.
= = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = =
Insert Table 3 about here.
= = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = =
Additionally, the results of the post-hoc Tukey tests show that there are no significant differences in kindergarten WA scores between any ethnic groups. In other words, although there is a significant main effect for ethnicity, it is combination of interactions, rather than unique contributions by interactions between any two ethnic groups that account for that effect. The results of the multiple comparisons between ethnic groups for WA scores are provided in Table 4.
= = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = =
Insert Table 4 about here.
= = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = =
In summary, the results of the two-way ANOVA analysis indicate that the null hypothesis, that there are no differences on kindergarten WA scores based on gender and ethnicity, is rejected.
Discussion
This study looked at longitudinal data obtained from one school in a large school district to determine whether early phonological assessment measures could predict later reading achievement scores. Based on previous research, it seemed likely that a positive relationship between the two measures would be found. However, the influences of student gender and ethnicity were variables identified in the literature as having a possible effect on both phonological awareness and early literacy. Therefore, the associations between gender and ethnicity and the kindergarten WA score and the DRA scores in first and second grade were assessed.
Statistical analysis showed that there were no statistical difference between males and females on the kindergarten WA scores. Since the research in this area has been mixed, it may be that data taken from another school year may produce different results. However, in this instance, phonological awareness does not reflect significant differences by gender. However, Hispanic students were found to perform at a statistically significant lower level on the WA than non-Hispanic students in kindergarten. Based on the literature, this may be due to lower English proficiency or to the types of oral language experiences they have had prior to beginning school (Chatterji, 2006; Linklater, et al., 2009).
After testing for associations between gender and reading levels in kindergarten, first and second grade, no significant associations were found in kindergarten and first grade. However, in second grade, gender significant gender differences were noted, as girls were over-represented and boys were under-represented in the above grade DRA groups. Similar results have been noted by previous researchers, who note a widening achievement gap based on gender as students move into higher grades (Chatterji, 2006).
Associations between ethnicity and reading levels moved in the opposite direction. Overall, there were significant associations between ethnicity and reading achievement. In kindergarten, Hispanic students were found to be significantly under-represented among the above grade readers; in first and second grade Hispanic students were somewhat under-represented in the above grade level group and over-represented in the below grade group, and non-Hispanic students were somewhat over-represented the above grade group (and under-represented in the below grade group in second grade), none of these associations were found to be significant. The very small sample of African American students in this study may have impacted our results. However, the general finding that the influence of ethnicity on achievement is somewhat mitigated by instruction is supported here.
In looking at the kindergarten WA, kindergarten reading level, and gender as predicators for DRA reading scores in first and second grade, the WA was found to be a significant predictor in both grades; the kindergarten reading level was a significant predictor only for second grade. This is important for administrators and teachers in the school, because it underscores the importance of strong phonological awareness skills as a foundation for future reading success. The WA is designed not only to identify students in need of targeted intervention, but also as a tool to help teachers design appropriate lesson plans (Pearson, 2005). Building phonological awareness in kindergarten could be a first step in building stronger readers in the primary grades (Burns, Griffith, & Snow, 1999).
Congruent with previous findings, only ethnicity was found to have a significant main effect on WA scores, and that that variance could not be attributed to a specific ethnic group. Thus, while teachers may need to be aware that ethnicity may be a factor in phonological awareness skills and should be taken into account when planning instruction, it is not such a strong factor that students from various ethnic groups would not be able to succeed on measures such as the WA.
Limitations and Implications
This study was designed to provide statistical analysis of the influences on phonological awareness and reading achievement for one school in a large school district. The measures were in place throughout the district and were designed to be used in conjunction with one another. Other schools in other school districts using other measures may find that their results differ from those reported here. On the other hand, this study points to the value of conducting a statistical review of data readily available within a school. These results do not shed light on how best to teach Hispanic boys (for example) how to read, but they do show that with good instruction students from all ethnicities can become strong readers.
It is also important to recognize that this study did not take into account any classroom teaching or targeted interventions (such as kindergarten interventions for students who did not make the fall benchmark on the WA or Reading Recovery for struggling first grade readers). These interventions may have influenced both WA and DRA scores. However, the research here was designed to provide a more global picture of reading achievement at one school.
Future Research
This study utilized only kindergarten, first and second grade longitudinal data. Future research could extend this study by including year-end data for third through sixth graders. Alternately, information on the students’ age, home languages or level of English proficiency, and preschool experience could be added to the existing data to explore other associations and relationships.
Conclusion
This quantitative study utilized previously collected longitudinal data to investigate the influence of kindergarten phonological awareness on reading levels for 104 students in kindergarten, first, and second grade. Additionally, gender and ethnicity were used as independent variables to determine their possible significance on phonological awareness and reading achievement. Findings indicated that kindergarten phonological awareness scores did significantly predict first and second grade reading achievement. Gender was found to have a significant effect only for second grade reading levels; the significance of ethnicity on reading was found to diminish from kindergarten to second grade.
References
Burns, S., Griffin, P., & Snow, C. (1999). Starting out right: A guide to promoting children’s reading success. Washington, DC: National Academies Press.
Chatterji, M. (2006). Reading achievement gaps, correlates, and moderators of early reading achievement: Evidence from the Early Childhood Longitudinal Study (ECLS) kindergarten to first grade sample. Journal of Educational Psychology, 98, 489-507. doi: 10.1037/0022-0663.98.3.489
Goswami, U. (2000). Phonological and lexical processes. In M. L. Kamil, P. B. Mosenthal, P. D. Pearson, & R. Barr, (Eds.), Handbook of reading research (Vol. III, pp. 251 - 267).
Mahwah, NJ: Lawrence Erlbaum.
Linklater, D. L., O’Connor, R. E., & Palardy, G. J., (2009). Kindergarten literacy assessment of English Only and English language learner students: An examination of the predictive validity of three phonemic awareness measures. Journal of School Psychology, 47, 369-394. doi: 10.1016/j.jsp.2009.08.001
McCoach, B. D., O’Connell, A. A., Reis, S. M., & Levitt, H. A., (2006). Growing readers: A hierarchical linear model of children’s reading growth during the first 2 years of school. Journal of Educational Psychology, 98(1), 14-28. doi: 10.1037/0022-0663.98.1.14
Nithart, C., Demont, E., Metz-Lutz, M-N., Majerus, S., & Poncelet, M. (2011). Early contribution of phonological awareness and later influence of phonological memory throughout reading acquisition. Journal of Research in Reading, 34, 346-363. doi: 10.1111/j.1467-9817.2009.01427.x
Pearson Education, (2005). Developmental Reading Analysis, 2nd Edition, Parsippany, NJ: Celebration Press.
Table 1
Frequency Distributions for Ethnicity
______________________________________________________________________________ Ethnicity n %
______________________________________________________________________________
Hispanic 41 39
Caucasian 28 27
African American 7 7
Asian 28 27
Table 2
Means and Standard Deviations for WA Scores by Gender and Ethnicity
______________________________________________________________________________
Male Female Total
_________________ __________________ _________________
Ethnicity N M SD N M SD N M SD
______________________________________________________________________________
Hispanic 26 174.62 17.72 15 170.27 24.67 41 173.02 20.34
Caucasian 16 181.44 14.80 12 189.67 5.07 28 184.96 12.22
African American 7 171.43 32.87 0 - - 7 171.43 32.87
Asian 11 179.73 18.45 17 182.65 16.79 28 181.50 17.18
Table 3
Analysis of Variance for Kindergarten WA Scores
____________________________________________________________________
Source df F pη2 p
____________________________________________________________________
Gender 1 0 .33 .003 .569
Ethnicity 3 2.97 .084 .036
Gender X Ethnicity 2 0 .92 .019 .403
S within group error 97 (353.99)
____________________________________________________________________
Note. The value enclosed in parentheses is the mean square error (MSw). S = subjects.
Table 4
Multiple Comparisons for Kindergarten WA Scores Among Ethnic Groups
______________________________________________________________________________
Ethnic Groups ∆M SE∆M 95% CI for ∆M
______________________________________________________________________________
Hispanic – Caucasian -11.94 4.61 -24.00 .12
Hispanic – African American 1.60 7.69 -18.52 21.71
Hispanic – Asian -8.48 4.61 -20.53 3.58
Caucasian – African American 13.54 7.95 -7.25 34.32
Caucasian – Asian 3.46 5.03 -9.68 16.61
African American – Asian -10.07 7.95 -30.86 10.71
______________________________________________________________________________
Note. ∆M = Mean difference. SE∆M = Standard error of ∆M.
No values were significant at p < .05.
Denise K. Robinson
George Mason University
Author Note
Denise K. Robinson, College of Education and Human Development, George Mason University. This paper was prepared in partial completion for EDRS 811 Spring, 2012
Correspondence concerning this article should be addressed to Denise K. Robinson, College of Education and Human Development, George Mason University, Fairfax, VA
E-mail: [email protected]
Abstract
This quantitative study utilized previously collected longitudinal data to investigate the influence of kindergarten phonological awareness on reading levels for 104 students in kindergarten, first, and second grade. Additionally, gender and ethnicity were used as independent variables to determine their possible significance on phonological awareness and reading achievement. Findings indicated that kindergarten phonological awareness scores did significantly predict first and second grade reading achievement. Gender was found to have a significant effect only for second grade reading levels; the significance of ethnicity on reading was found to diminish from kindergarten to second grade. Down the Road: Does Phonological Awareness Predict Reading Ability in the Early Grades?
Phonological awareness is the knowledge that oral language is made up of a number of structures that are separate from meaning, such as syllables (units of spoken words), onset (consonants that begin syllables) and rime (the remainder of the syllable), and phonemes (the smallest meaningful unit of sound that make up syllables and words) (Burns, Griffin, & Snow, 1999; Goswami, 2000). In other words, phonological awareness is the understanding that speakers can manipulate the letters and sounds that make up the words we say. Well-developed phonological awareness has been found to be a strong predictor of a young student’s ability to read and spell (Goswami, 2000). Therefore, many school systems have begun to assess kindergarten students’ level of phonological awareness, so that interventions can be put in place for struggling young readers (Linklater, O’Connor, & Palardy, 2009; McCoach, O’Connell, Reis, & Levitt, 2006). Student reading achievement levels are also typically gathered at the end of each grade level. This study will seek to determine whether kindergarten students’ level of phonological awareness can predict future reading success.
Phonological Awareness and Early Literacy
Children bring a variety of early literacy experiences with them when they enter kindergarten. Some will have spent time in preschool classes or child care settings. Others will have experienced the world with their families and friends. Most will have developed strong oral language skills that will serve as the basis for learning to read and write (Burns, Griffith, & Snow, 1999). The ability to manipulate units of oral language—rhyming, alliteration, breaking words apart and putting them back together—underlies the child’s phonological awareness. Students whose background experiences and use of language do not prepare them to build strong phonological awareness are at risk for later reading difficulties (Linklater, et al., 2009).
Researchers have investigated the relationships between specific phonological skills and reading achievement and have found strong, positive correlations between skills such as identifying letters and their sounds (the alphabetic principle), dividing words into onset and rimes (the first sound and the remainder of the word or syllable), or phoneme (sound) deletion and blending and later reading success (Goswami, 2000). Building on these relationships, school systems have begun to assess young student’s phonological awareness in an attempt to focus classroom instruction and identify students in need of specific, targeted intervention (Linklater, 2009; McCoach, et al., 2006). This is important because research has shown that readers who struggle in the early grades are often identified as struggling readers in later grades (Goswami, 2000).
In a recent study, Nithart, Demont, Metz-Lutz, Majerus, and Poncelet (2011) investigated the simultaneous development of phonological and reading skills. Phonological discrimination, phonological awareness, and phonological memory were assessed at the end of the students’ kindergarten year and again at the end of first grade. Phonological discrimination and awareness were found to correlate strongly with reading skills in kindergarten. The researchers found that phonological awareness continued to grow from kindergarten to first grade, along with an increase in reading skills. Phonological memory also developed over this time span—first as a more separate construct than phonological awareness, and then converging with phonological awareness skills as reading skills developed.
In a study that utilized data from the Early Childhood Longitudinal Study (ECLS), Chatterji (2006) looked at the effects of different ethnic, gender, and socio-economic groups on first grade reading achievement. Using data gathered in both kindergarten and first grade, Chatterji found that gaps in literacy achievement based on gender (males lower than females) increased from kindergarten to first grade but that gaps based on ethnicity (African American and Hispanic students lower that Caucasian) were somewhat mitigated by instruction between kindergarten entry and first grade.
A strong body of research has been developed to document the importance of phonological awareness to reading achievement and to investigate many of the factors that influence the development of phonological awareness. At the same time, research has shown that readers who struggle in the early grades often remain struggling readers as time progresses. However, a gap in the research exists regarding the longitudinal effect of phonological awareness measured in kindergarten for the reading success of a sample of students in first and second grade. This study addresses that gap by using existing longitudinal data to determine whether phonological awareness can predict reading achievement over time.
Rationale and Purpose
In many school districts across the United States, children in kindergarten are assessed to determine their mastery of letters and their sounds, segmenting words into syllables, identifying rhyming words, and demonstrating an understanding of beginning reading skills. Later, reading fluency and comprehension are measured over time, as students move from grade to grade.
However, it is unclear whether data from a particular test of kindergarten phonological awareness can predict reading ability in first and second grade readers. It is the intent of this study to determine whether, in one school, scores on the DRA Word Analysis (WA; Pearson, 2005) can predict scores on the Developmental Reading Analysis (DRA; Pearson, 2005) for first and second graders at that school. Using longitudinal data from a suburban, Title 1 elementary school, scores from two consecutive school years were collected and analyzed to answer the following questions:
1. Are there differences on WA scores in kindergarten by gender and ethnicity?
Research findings regarding gender, ethnicity and phonological awareness to date have been mixed. Some suggest that early literacy experiences are more important to phonological awareness than gender. At the same time, students with limited English proficiency who have enough English to take the WA may struggle more with phonological awareness in a second language, even with good teaching. Based on these findings, we can hypothesize that there are no differences between boys and girls on WA scores and that Hispanic students will score lower on the WA than non-Hispanic students.
2. Are there associations between gender and ethnicity on kindergarten reading levels, on DRA scores in first grade, and on DRA scores in second grade? Like the research on phonological awareness, the research on early literacy skills and gender and ethnicity is mixed. Some studies suggest that gender and ethnicity have less of an impact on achievement in the early years and that this changes as students are faced with more difficult tasks. Therefore, we can hypothesize that there are no significant associations between gender and ethnicity and kindergarten reading levels but that there are significant associations between gender and ethnicity and DRA scores in first and second grade.
3. Is the first grade DRA score predicted by the WA score, kindergarten reading level, and gender? Similarly, is the second grade DRA score predicted by the same independent variables (WA, kindergarten reading level, and gender)? Based on research that suggests that phonological awareness and early reading skills are predictors of later reading success, we can hypothesize that the WA score and kindergarten reading level is a significant predictors of the first and second grade DRA scores. Further, we can hypothesize that due to changes over time gender is not a significant predictor of first grade DRA scores, but is a significant predictor in second grade.
4. To what extent does the interaction between gender and ethnicity account for the variance in kindergarten WA scores? I would expect that the interaction between gender and ethnicity will account for at least some of the variance in WA scores. I would hypothesize that Hispanic boys may have lower scores than other gender/ethnic group combinations.
Method
This quantitative study utilizes longitudinal data regarding student reading performance in kindergarten through second grade. Phonological awareness and reading achievement data is recorded for each K-3 student in a large, mid-Atlantic public school system. Schools throughout the district use the same assessment protocols and teachers administering the assessments must be trained in their use and scoring, allowing for some measure of consistency from school to school. Data used in this study was obtained from one elementary school in the school system.
Sample
The sample population for this study consisted of students who attended one suburban elementary school in a major mid-Atlantic metropolitan area. The school is a Title 1 school; more than 50% of the 520 students qualify for free and reduced lunch subsidies. It offers a full-day kindergarten program, as well as Head Start and preschool special education classes. Mobility for the school is approximately 30%. Data was collected for a total of 104 students (male, n = 60; female, n = 44) who attended the school from kindergarten through second grade, beginning in two consecutive school years (2007-08 and 2008-09). Demographic data for the sample is provided in Table 1.
= = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = =
Insert Table 1 about here
= = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = =
Data
The measures used in this study were part of a reading assessment program, the Developmental Reading Assessment program (DRA; Pearson, 2005), designed to determine a student’s level of phonological awareness and reading ability, including both fluency and comprehension. The assessment program begins in kindergarten and assesses learners as they progress through elementary school. Both the WA and DRA protocols are administered individually to students by the classroom teacher or qualified reading specialist in the fall and spring, with an optional assessment window available mid-year.
Measure of phonological awareness. The DRA Word Analysis (WA; Pearson, 2005) is a criterion referenced diagnostic assessment that provides the classroom teachers with information on how struggling and emergent readers understand words in context and isolation and to aid in planning for instruction. The measure is divided into five strands, sequenced in order of difficulty: phonological awareness; metalanguage used to talk about print; recognition of letters and high frequency words; phonics; and structural analysis and syllabication. The WA is made up of 40 tasks, 11 of which are administered to kindergartners. The score range for these 11 tasks is 0 – 193. The benchmark score is 61 for the fall administration and 167 for the spring; the same form of the instrument is used for each administration. The tasks reflect skills measured by the DRA as students gain in reading fluency and comprehension and are presented in order of difficulty. Typically, kindergarten students who do not meet the benchmark score are eligible for specifically targeted intervention services. Struggling readers in higher grades may be assessed using the remaining tasks, at the discretion of the school. The WA is administered by classroom teachers, who are trained in using and scoring the protocol. Information regarding reliability and validity of the WA is available from the publisher.
Measure of reading achievement. Reading achievement was measured at the end of kindergarten, first, and second grades using the DRA (Pearson, 2005). Like the WA, the DRA is a criterion referenced protocol, administered individually by trained classroom teachers or reading specialists. Following a strong introduction and picture preview, the student reads a leveled text while the teacher records miscues. After reading the text, the student retells the story from memory and answers specific comprehension questions. Scores are obtained in three areas: reading engagement, oral reading fluency, and comprehension; the student’s proficiency is rated as independent, instructional, or frustration. Testing is repeated using texts at the next level until an instructional level is reached. Benchmarks for each grade level have been established: kindergarten—levels 3-4 (emergent); first grade—level 16 (developing, fiction and non-fiction); second grade—level 28 (expanding, fiction and non-fiction). Levels below benchmark are considered below grade level, at benchmark are considered on grade level, and above benchmark are considered above grade level. The DRA is typically administered in the fall and spring, with an optional winter window; alternate assessments are available for each reading level. Information on the reliability and validity of the DRA is available from the publisher.
Data collection
Data for this study was collected from existing student records. Longitudinal reading data for all students in the school district are recorded on the K-3 Student Achievement Record. Year end literacy scores are entered on a sheet that is kept with each student’s school records. Scores from the K-3 Student Achievement Record used in this study were the kindergarten Word Analysis score (WA), the final kindergarten reading level (klevel), and the first and second grade DRA scores (dra1 and dra2). The kindergarten reading level is scored as 3 (below grade level), 4 (on grade level), or 5 (above grade level); no specific DRA levels are provided. The dra1 and dra2 scores provide the specific reading levels attained by the student.
Access to the student records was granted by the building principal. All records were kept strictly confidential. Student gender and ethnicity were noted along with reading scores; however neither student names nor identification numbers were recorded.
Statistical data analysis
Are there differences on WA scores in kindergarten by gender and ethnicity? To answer this question two independent samples t- test analyses were performed. For the first, the kindergarten WA scores were entered as the dependent variable; gender (male/female) was entered as the grouping (independent) variable. For the second t-test, ethnicity (Hispanic/non-Hispanic) was entered as the grouping (independent) variable; kindergarten WA scores remained as the dependent variable. A 95 percent confidence level was chosen.
Are there associations between gender and ethnicity on kindergarten reading levels, on DRA scores in first grade, and on DRA scores in second grade? A series of chi-square tests for association were undertaken in order to answer this question. To test the hypothesis that there are no significant associations between gender and ethnicity and kindergarten reading levels two chi-square tests were conducted: the first with gender (male/female) as the independent variable and three levels of kindergarten reading (below grade, on grade, and above grade) as the dependent variable; the second with ethnicity (Hispanic/non-Hispanic) as the independent variable and kindergarten reading levels as the dependent variable. To test the hypothesis that there are significant associations between gender and ethnicity and DRA scores in first and second grade, similar chi-square tests for association were made. Two chi-square tests each were made for the dra1and dra2 scores, with gender or ethnicity as independent variables and below, on, and above grade level scores as the dependent variables.
Is the first grade DRA score predicted by the WA score, kindergarten reading level, and gender? Similarly, is the second grade DRA score predicted by the same independent variables (WA, kindergarten reading level, and gender)? Research has shown that strong levels of phonological awareness are a predictor of higher reading ability in later grades. Similarly, students who struggle with phonological awareness often struggle with reading, even as early as kindergarten. Readers who struggle in the early grades often continue to exhibit reading difficulties in subsequent grades, sometimes in spite of targeted interventions. In order to understand if these scenarios are at play in this elementary school, multiple linear regressions were used in order to determine whether the DRA scores (dependent variable) attained by first or second grade students at the end of the school year can be predicted by their WA scores, kindergarten reading levels, and gender (independent variables).
To what extent does the interaction between gender and ethnicity account for the variance in kindergarten WA scores? Research has shown that well developed levels of phonological awareness are important prerequisites for higher reading achievement in higher grades. Therefore, in order to determine the extent to which interactions between gender and ethnicity account for the variance in WA scores, a two-way ANOVA analysis was run. Kindergarten WA scores were entered as the dependent variable, with gender and ethnicity entered as fixed factors.
Results
Are there differences on WA scores in kindergarten by gender and ethnicity?
Gender. The results of Levene’s test for equality of variances shows that equal variance can be assumed, F = 1.056 (p = .307). Further, the results from the independent t-test indicate that there is no significant difference between males and females on kindergarten WA scores, t (102) = -.87 (p = .384). Therefore, the null hypothesis, that there are no significant differences between boys and girls on WA scores in kindergarten is retained.
Ethnicity. Again, the results of Levene’s test for equality of variance shows that equal variance can be assumed, F = 1.22 (p = .272). However, the results from the independent t-test suggest that there is a statistically significant difference between Hispanic and non-Hispanic students on kindergarten WA scores, t (102) = -2.36 (p = .02). The 95% CI for the difference indicates that non-Hispanic students exceed Hispanic students by an amount between 1.4 and 16.4 points. Therefore, the null hypothesis must be rejected and the alternative hypothesis, that there is a significant difference in WA scores by ethnicity is retained.
Are there associations between gender and ethnicity on kindergarten reading levels, on DRA scores in first grade, and on DRA scores in second grade?
Kindergarten reading level. The results from the chi-square test for associations between gender and the klevel indicate that there is no significant association between the two variables. That is, boys and girls had no significant difference in their reading levels at the end of kindergarten. Therefore, we fail to reject the null hypothesis, that there is no association between gender and kindergarten reading level. However, the results of the chi-square test on ethnicity indicate that there is a significant association between ethnicity and reading level in kindergarten for the students in this study, χ2(2) = 12.28, p = .002. Further, the standardized residuals indicate that a major contributor to the association is the under-representation of Hispanic students in the above grade group (SR = -2.2). Thus, we must reject the null hypothesis, that there is no association between ethnicity and kindergarten reading scores.
First grade reading level. As with the results in kindergarten, the results of the chi-square test did not suggest a significant association between gender and reading level in first grade. Again, the null hypothesis is retained. However, while the chi-square test for ethnicity and reading level in first grade did suggest a significant association between the two for the students in this study, χ2 (2) = 7.32, p = .026, there does not appear to be a major contributor to this association, as indicated by the standardized residuals. Thus, while Hispanic students were somewhat under-represented in the above grade level group and over-represented in the below grade group, and non-Hispanic students were somewhat over-represented the above grade group, none of these associations were found to be significant. Nevertheless, the null hypothesis must be rejected.
Second grade reading level. Unlike the previous results for the kindergarten and first grade reading levels, the chi-square test for association between gender and second grade reading levels indicates a significant association between the two, χ2 (2) = 13.14, p = .001. Specifically, the standardized residuals indicate that major contributors to this association are the significant over-representation of females in the above grade level group and the under-representation of males in this group. This result confirms previous findings that gender impacts reading scores as students move to higher grades, but in relation to this study the null hypothesis (that there is no significant association between gender and second grade reading levels) must be rejected. Similar to results for first grade reading levels, the results of the chi-square test for association between ethnicity and reading in second grade show a significant association between the two, χ2 (2) = 8.09, p = .017. Again, no main contributors were indicated by the standardized residuals. The results do indicate that Hispanic students are somewhat over-represented in the below grade level group and over-represented in the above grade group, opposite to the findings for non-Hispanic students.
Is the first grade DRA score predicted by the WA score, kindergarten reading level, and gender? Similarly, is the second grade DRA score predicted by the same independent variables (WA, kindergarten reading level, and gender)?
First grade DRA score. The results from the omnibus F-test indicate that the prediction of DRA scores in first grade from kindergarten WA scores, reading level in kindergarten, and gender is statistically significant, F (3, 100) = 11.55, p = .000. Further, R2 = .257 indicates that 25.7% of the variance in first grade DRA scores (dra1) is explained by the variance in kindergarten WA (WA), readinglevels in kindergarten (klevel), and gender (gen). The regression equation in this case is = .085 ( The regression coefficient for WA is statistically significant (p = .000) but the regression coefficients for kindergarten reading level (p = .057) and gender (p = .264) are not. In other words, the unique contribution of the WA score to the prediction of first grade reading levels is statistically significant, but the contributions of kindergarten reading levels and gender are not. Specifically, the squared part correlation for WA (.316)2 =.10 shows that 10% of the variance in first grade DRA scores is explained by kindergarten WA scores.
Second grade DRA score. The results from the omnibus F-test indicate that the prediction of DRA scores in second grade from kindergarten WA scores, reading level in kindergarten, and gender is statistically significant, F (3, 100) = 11.382, p = .000. Further, R2 = .255 indicates that 25.5% of the variance in second grade DRA scores (dra2) is explained by the variance in kindergarten WA (WA), readinglevels in kindergarten (klevel), and gender (gen). The regression equation in this case is = .094 ( The regression coefficient for WA (p = .001) and kindergarten reading levels (p = .046) are statistically significant but the regression coefficient for gender (p = .096) is not. In other words the unique contributions of kindergarten WA and reading level scores significantly predict end of the year DRA scores in second grade, but the contribution of gender does not. Specifically, the squared part correlation for WA (.291)2 =.084 shows that 8.4% of the variance in second grade DRA scores is explained by kindergarten WA scores and the squared part correlation for kindergarten reading level (.174)2 = .03 shows that 3% of the variance in second grade scores is explained by kindergarten reading levels.
To what extent does the interaction between gender and ethnicity account for the variance in kindergarten WA scores?
In order to answer this question, a two-way ANOVA analysis was run, with kindergarten WA scores entered as the dependent variable and gender (male, female) and ethnicity (Hispanic, Caucasian, African American, Asian) entered as fixed factors. The means and standard deviations are reported in Table 2.
= = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = =
Insert Table 2 about here.
= = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = =
The Levene’s test for equal variances indicates that the assumption of homogeneity of variances is not met, F (6, 97) = 2.428, p = .031. The results of the omnibus F-test indicate that there is a statistically significant main effect for ethnicity, F (3, 97) = 2.97, p = .036, pη2 = .084. There is no significant main effect for gender (p = .569) or for gender and ethnicity together (p = .403). Further, 8.4% of the variance in WA scores in kindergarten is accounted for by the unique contribution of differences in ethnicity, controlling for all other effects. The results of the analysis of variance for WA scores are provided in Table 3.
= = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = =
Insert Table 3 about here.
= = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = =
Additionally, the results of the post-hoc Tukey tests show that there are no significant differences in kindergarten WA scores between any ethnic groups. In other words, although there is a significant main effect for ethnicity, it is combination of interactions, rather than unique contributions by interactions between any two ethnic groups that account for that effect. The results of the multiple comparisons between ethnic groups for WA scores are provided in Table 4.
= = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = =
Insert Table 4 about here.
= = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = =
In summary, the results of the two-way ANOVA analysis indicate that the null hypothesis, that there are no differences on kindergarten WA scores based on gender and ethnicity, is rejected.
Discussion
This study looked at longitudinal data obtained from one school in a large school district to determine whether early phonological assessment measures could predict later reading achievement scores. Based on previous research, it seemed likely that a positive relationship between the two measures would be found. However, the influences of student gender and ethnicity were variables identified in the literature as having a possible effect on both phonological awareness and early literacy. Therefore, the associations between gender and ethnicity and the kindergarten WA score and the DRA scores in first and second grade were assessed.
Statistical analysis showed that there were no statistical difference between males and females on the kindergarten WA scores. Since the research in this area has been mixed, it may be that data taken from another school year may produce different results. However, in this instance, phonological awareness does not reflect significant differences by gender. However, Hispanic students were found to perform at a statistically significant lower level on the WA than non-Hispanic students in kindergarten. Based on the literature, this may be due to lower English proficiency or to the types of oral language experiences they have had prior to beginning school (Chatterji, 2006; Linklater, et al., 2009).
After testing for associations between gender and reading levels in kindergarten, first and second grade, no significant associations were found in kindergarten and first grade. However, in second grade, gender significant gender differences were noted, as girls were over-represented and boys were under-represented in the above grade DRA groups. Similar results have been noted by previous researchers, who note a widening achievement gap based on gender as students move into higher grades (Chatterji, 2006).
Associations between ethnicity and reading levels moved in the opposite direction. Overall, there were significant associations between ethnicity and reading achievement. In kindergarten, Hispanic students were found to be significantly under-represented among the above grade readers; in first and second grade Hispanic students were somewhat under-represented in the above grade level group and over-represented in the below grade group, and non-Hispanic students were somewhat over-represented the above grade group (and under-represented in the below grade group in second grade), none of these associations were found to be significant. The very small sample of African American students in this study may have impacted our results. However, the general finding that the influence of ethnicity on achievement is somewhat mitigated by instruction is supported here.
In looking at the kindergarten WA, kindergarten reading level, and gender as predicators for DRA reading scores in first and second grade, the WA was found to be a significant predictor in both grades; the kindergarten reading level was a significant predictor only for second grade. This is important for administrators and teachers in the school, because it underscores the importance of strong phonological awareness skills as a foundation for future reading success. The WA is designed not only to identify students in need of targeted intervention, but also as a tool to help teachers design appropriate lesson plans (Pearson, 2005). Building phonological awareness in kindergarten could be a first step in building stronger readers in the primary grades (Burns, Griffith, & Snow, 1999).
Congruent with previous findings, only ethnicity was found to have a significant main effect on WA scores, and that that variance could not be attributed to a specific ethnic group. Thus, while teachers may need to be aware that ethnicity may be a factor in phonological awareness skills and should be taken into account when planning instruction, it is not such a strong factor that students from various ethnic groups would not be able to succeed on measures such as the WA.
Limitations and Implications
This study was designed to provide statistical analysis of the influences on phonological awareness and reading achievement for one school in a large school district. The measures were in place throughout the district and were designed to be used in conjunction with one another. Other schools in other school districts using other measures may find that their results differ from those reported here. On the other hand, this study points to the value of conducting a statistical review of data readily available within a school. These results do not shed light on how best to teach Hispanic boys (for example) how to read, but they do show that with good instruction students from all ethnicities can become strong readers.
It is also important to recognize that this study did not take into account any classroom teaching or targeted interventions (such as kindergarten interventions for students who did not make the fall benchmark on the WA or Reading Recovery for struggling first grade readers). These interventions may have influenced both WA and DRA scores. However, the research here was designed to provide a more global picture of reading achievement at one school.
Future Research
This study utilized only kindergarten, first and second grade longitudinal data. Future research could extend this study by including year-end data for third through sixth graders. Alternately, information on the students’ age, home languages or level of English proficiency, and preschool experience could be added to the existing data to explore other associations and relationships.
Conclusion
This quantitative study utilized previously collected longitudinal data to investigate the influence of kindergarten phonological awareness on reading levels for 104 students in kindergarten, first, and second grade. Additionally, gender and ethnicity were used as independent variables to determine their possible significance on phonological awareness and reading achievement. Findings indicated that kindergarten phonological awareness scores did significantly predict first and second grade reading achievement. Gender was found to have a significant effect only for second grade reading levels; the significance of ethnicity on reading was found to diminish from kindergarten to second grade.
References
Burns, S., Griffin, P., & Snow, C. (1999). Starting out right: A guide to promoting children’s reading success. Washington, DC: National Academies Press.
Chatterji, M. (2006). Reading achievement gaps, correlates, and moderators of early reading achievement: Evidence from the Early Childhood Longitudinal Study (ECLS) kindergarten to first grade sample. Journal of Educational Psychology, 98, 489-507. doi: 10.1037/0022-0663.98.3.489
Goswami, U. (2000). Phonological and lexical processes. In M. L. Kamil, P. B. Mosenthal, P. D. Pearson, & R. Barr, (Eds.), Handbook of reading research (Vol. III, pp. 251 - 267).
Mahwah, NJ: Lawrence Erlbaum.
Linklater, D. L., O’Connor, R. E., & Palardy, G. J., (2009). Kindergarten literacy assessment of English Only and English language learner students: An examination of the predictive validity of three phonemic awareness measures. Journal of School Psychology, 47, 369-394. doi: 10.1016/j.jsp.2009.08.001
McCoach, B. D., O’Connell, A. A., Reis, S. M., & Levitt, H. A., (2006). Growing readers: A hierarchical linear model of children’s reading growth during the first 2 years of school. Journal of Educational Psychology, 98(1), 14-28. doi: 10.1037/0022-0663.98.1.14
Nithart, C., Demont, E., Metz-Lutz, M-N., Majerus, S., & Poncelet, M. (2011). Early contribution of phonological awareness and later influence of phonological memory throughout reading acquisition. Journal of Research in Reading, 34, 346-363. doi: 10.1111/j.1467-9817.2009.01427.x
Pearson Education, (2005). Developmental Reading Analysis, 2nd Edition, Parsippany, NJ: Celebration Press.
Table 1
Frequency Distributions for Ethnicity
______________________________________________________________________________ Ethnicity n %
______________________________________________________________________________
Hispanic 41 39
Caucasian 28 27
African American 7 7
Asian 28 27
Table 2
Means and Standard Deviations for WA Scores by Gender and Ethnicity
______________________________________________________________________________
Male Female Total
_________________ __________________ _________________
Ethnicity N M SD N M SD N M SD
______________________________________________________________________________
Hispanic 26 174.62 17.72 15 170.27 24.67 41 173.02 20.34
Caucasian 16 181.44 14.80 12 189.67 5.07 28 184.96 12.22
African American 7 171.43 32.87 0 - - 7 171.43 32.87
Asian 11 179.73 18.45 17 182.65 16.79 28 181.50 17.18
Table 3
Analysis of Variance for Kindergarten WA Scores
____________________________________________________________________
Source df F pη2 p
____________________________________________________________________
Gender 1 0 .33 .003 .569
Ethnicity 3 2.97 .084 .036
Gender X Ethnicity 2 0 .92 .019 .403
S within group error 97 (353.99)
____________________________________________________________________
Note. The value enclosed in parentheses is the mean square error (MSw). S = subjects.
Table 4
Multiple Comparisons for Kindergarten WA Scores Among Ethnic Groups
______________________________________________________________________________
Ethnic Groups ∆M SE∆M 95% CI for ∆M
______________________________________________________________________________
Hispanic – Caucasian -11.94 4.61 -24.00 .12
Hispanic – African American 1.60 7.69 -18.52 21.71
Hispanic – Asian -8.48 4.61 -20.53 3.58
Caucasian – African American 13.54 7.95 -7.25 34.32
Caucasian – Asian 3.46 5.03 -9.68 16.61
African American – Asian -10.07 7.95 -30.86 10.71
______________________________________________________________________________
Note. ∆M = Mean difference. SE∆M = Standard error of ∆M.
No values were significant at p < .05.