Fluent reading is an indication of proficient decoding and comprehension. Studies have demonstrated that oral reading fluency (ORF) correlates positively with reading comprehension. The ORF assessment comprises a set of individually administered 1-minute reading probes. These tests are designed to monitor learning progress toward predetermined instructional goals and identify children who may require additional instructional support. 

Although many studies have explored Mandarin ORF performance levels at each grade, the reading probes used are generally adopted from textbooks. However, results have indicated that the curriculum-based ORF is 50 characters per minute more than that measured by the standard probe assessment. This suggests that reading fluency levels are affected by the equivalence of the reading probes. This study is the first to use standard probes to investigate the developmental and expected growth of Mandarin ORF in the second and third grades of elementary schools. 

The number of correct words read per minute (the level) and the growth rate of student reading fluency are two key indicators of reading progress. However, few studies have investigated the growth rate of Mandarin ORF. Early studies tended to assert that within-year ORF growth is linear rather than quadratic, but recent research has indicated that growth rates differ within a school year. This study examined the growth model of reading fluency and estimated its growth rate. 

Students with a large vocabulary tend to have a higher ORF level. According to the Matthew effect, students with higher ORF levels have a greater growth rate. Previous Mandarin ORF research does not support this inference. Furthermore, the correlation between initial ORF levels and growth rates has not been reported. However, this issue is important; if students with different ORF levels have different growth rates, then appropriate learning growth goals should be set based on their initial abilities when planning interventions. This study tracked the development of ORF skills over the course of one academic year to investigate the following research questions: (1) What are the growth models of ORF for second- and third-grade students in elementary schools? (2) What is the growth rate of ORF over the semesters for second- and third-grade students? (3) Is the growth rate for reading fluency related to the ORF level at the beginning of the semester? 

The participants comprised 57 second-grade and 51 third-grade students and 53 girls and 55 boys. The students underwent an ORF test (Chang, 2018). For each grade, 19 reading probes, with a text length of 350–400 characters, were used. The alternate-form reliability was between .80 and .89. The generalizability coefficients for the second- and third-grade conditions were approximately .84 and .85, respectively, when one reading probe was used to estimate ORF performance. When three reading probes were used, the generalizability coefficients were .94 and .95, respectively. The correlation coefficients between ORF score and reading comprehension test score were .60 and .67, respectively.

The ORF tests were administered at 3-week intervals during each semester in the 2012 academic year. Each student took ORF tests at 13 time points, with three reading probes administered at the beginning of the school year (pretest), at the end of the first semester (midterm test), and at the end of the second semester (posttest). The average score across the three reading probes was calculated to obtain reliable growth rates. At each of the other time points, a single ORF score was collected. 

Eight examiners were trained in the administration and scoring of the ORF tests in a 4-hour workshop, during which the assessment and scoring procedures were introduced and the examiners practiced the procedures in pairs to ensure they could conduct the procedure reliably. Overall interrater reliability was .99 for this study. 

A hierarchical linear model was used to assess the goodness of fit between the data and five models: the null model (M0), random intercept model (M1), random coefficient model (M2), piecewise growth model (M3), and quadratic growth model (M4). The intraclass correlation (ICC) of the M0 model for the second grade was .73, meaning that 73% of the difference in reading fluency scores originated in differences between students (level 2) and 27% of the variance originated from differences within individuals (level 1). The ICC for the third grade was .81. The results suggested high heterogeneity between students and the necessity of a hierarchical model. 

Among five models, a piecewise growth model had the lowest deviance value and thus the best fit with the data. Thus, the children had different growth rates in the two semesters within the same academic year. On average, the number of characters the second and third graders could read correctly per minute was 109 and 137 at the beginning of the academic year, respectively. At the end of the third grade, the mean ORF was 165 characters. In the fall, the second-grade ORF growth averaged 13 characters per semester, whereas the average was approximately 18 characters per semester in the spring. The third graders’ mean growth rates were 20 and 6 characters in the first and second semesters, respectively. The second-grade growth rates in the two semesters were similar, but ORF gains were more modest during the second half of the academic year for the third grade. 

The correlation between ORF and the growth rate for the second-grade students was not significant. However, in the first semester of the third grade, the correlation between ORF and growth rate was significant and had a correlation coefficient of .32 . In the second semester, the correlation between ORF and growth rate was -.43. In different grades and semesters, the relationships between initial ORF and growth rate varied. This finding reminds teachers that they should consider the students’ grades and initial ORF when setting teaching goals. 

The results of this study established the expected ORF level and growth rate for second- and third-grade students, enabling teachers to identify high-risk students that are falling behind in reading to determine the effects of intervention by examining their change in growth rate and adjust their instruction accordingly. Knowing student’s ORF scores may help teachers plan reading instruction that is more responsive to the individual. By collecting multiple assessment scores, educators and researchers no longer must assume linear developmental trends because nonlinear functions can be fit to the time-series data. The nonlinear growth pattern suggests that instructional goals derived from a linear change model may require revision and that student growth expectations may require modification at different points in the academic year. 

The results only partially support the Matthew effect. At the beginning of the semester, the growth rate of the second-year students with strong ORF was not significantly greater than that of those with weaker ORF. However, in the first semester of the third grade, the level of ORF and the growth rate was significantly and positively correlated. Participants with strong reading fluency grew faster in their reading ability during the fall semester than those with weaker ORF. Therefore, the results may enable educators to recognize the need for more intensive instructional support during periods when rapid growth is expected, which is the second semester of the second grade through the first semester of third grade.