The aim of this study was not to build the best model of cognitive component and reading ability but to examine the hypothesized relationships between cognitive component and reading ability. Because the model fit is satisfactory, its regression coefficients were examined to validate hypotheses 1–6.
Table 9
Standardized path coefficients of the relationships.
Estimate
SE CR P
Standardized Unstandardized
COG T2 <--- COG T1 0.332 0.221 0.083 2.672 **
COG T2 <--- Read T1 –0.274 –0.032 0.014 –2.218 *
Read T2 <--- Read T1 0.201 0.208 0.062 3.385 ***
Read T2 <--- COG T1 –0.768 –4.581 0.371 –12.363 ***
COG T1 <--> Read T1 –0.883 –0.002 0.000 –10.353 ***
COG T2 <--> Read T2 –0.157 0.000 0.000 –2.226 * Note 1. REC = reading comprehension, WOR = word recognition, COG = cognitive component, READ = reading ability, T1 = data from time 1 and T2 = data from time 2.
Note 2. *** = p< .001, ** = p< .01, * = p< .05
60
Table 9 reports the standardised path coefficients of the relationships amongst the major variables without covariates in three types of the relationship including;
(1) Auto-regressive effects of the two variables.
The auto-regressive effect (C1C2) of the cognitive component at time-1 (COG T1) on the cognitive component at time-2 (COG T2) was significantly correlation, with the standard coefficient 0.332 (p< .01). And the auto-regressive effect (R1R2) of the reading ability at time-1 (REC T1) on the reading ability at time-2 (REC T2) was significantly correlation, with the standard coefficient 0.202 (p< .001).
(2) Synchronous correlations between the two variables.
The unconditional zero order correlation (C1R1) between the cognitive component (CCG T1) and the reading ability (REC T1), both at time-1 was significantly negative correlation, with the standard coefficient -0.883 (p< .001). And the zero order correlation (C2R2) between the cognitive component (COG T2) and the reading ability (REC T2), both at time-2 was significantly negative correlation, with the standard coefficient -0.157 (p< .05).
These negative correlations indicate an inverse relationship between the cognitive component time-1 and reading ability time-1, and also inverse relationship between the cognitive time-2 and reading ability time-2.
(3) Reciprocal effects between the two variables over time.
The Reciprocal effects (C1R2) of the cognitive component at time-1 (COG T1) on reading ability at time-2 (REC T2) was significantly negative relationship, with the standard coefficient -0.768 (p< .001). And the effect (R1C2) of reading ability at time-1 (REC T1) on cognitive component at time-2 (COG T2) was significantly negative relationship, with the standard coefficient -0.274 (p< .05). These negative effected indicate an inverse relationship between the cognitive component time-1 and reading ability time-2, and also inverse
relationship between the reading ability time-1.and cognitive time-2
61
Figure 6 illustrates the standardised path coefficients of the relationship between
cognitive component and reading ability for hypothesis testing by cross-lagged SEM analysis.
Figure 6. Standardised path coefficients of the relationships determined for hypothesis testing by cross-lagged SEM.
The negative correlations of C1R1, C2R2, R1C2, and C1R2 may involve with the negative correlation between rapid naming tests (rapid colour naming and rapid number naming) and the other tests. In order to assess the causal relationship from three cognitive components (morphological awareness, decoding-related and rapid naming), a cross-lagged SEM was analyses again by separated the cognitive variables as a latent factors, and assess the relationship between three cognitive variables and reading ability.
COG T1
Read T2 COG T2
C1C2 =0.332**
C1R2 = –0.768***
C2R2 = –0.157*
C1R1 = –0.883***
R1R2 = 0.201***
R1C2 = –0.274*
Read T1
62
Figure 7. The model of the relationships between morphological awareness, decoding-related and rapid naming, and reading ability.
Table 10
Standardized path coefficients of the relationships.
Estimate
63
Table 10 (continued). Standardized path coefficients of the relationships.
Estimate
SE CR P
Standardized Unstandardized
MA2 <--- Read1 0.093 0.003 0.083 0.035 0.075
Read2 <--- MA1 0.185 0.097 0.064 1.505 *
Read2 <--- DE1 0.154 0.606 0.619 0.980 **
Read2 <--- RN1 -0.267 -2.084 0.646 -3.228 **
DE2 <--- MA1 -0.023 -0.024 0.068 -0.358 0.721
RN2 <--- MA1 0.265 0.042 0.010 3.978 ***
MA2 <--- DE1 0.197 1.002 0.675 1.484 *
RN2 <--- DE1 -0.055 -0.086 0.099 -0.864 0.387
MA2 <--- RN1 -0.198 -0.699 0.692 -1.009 *
DE2 <--- RN1 -0.158 -1.131 0.671 -1.685 *
MA1 <--> DE1 0.455 0.001 0.000 7.602 ***
MA1 <--> RN1 -0.580 -0.001 0.000 -9.102 ***
DE1 <--> RN1 -0.524 0.000 0.000 -8.321 ***
DE1 <--> Read1 0.469 0.001 0.000 7.156 ***
RN1 <--> Read1 -0.769 -0.002 0.000 -9.456 ***
MA1 <--> Read1 0.490 0.007 0.001 7.458 ***
RN2 <--> Read2 -0.097 0.000 0.000 -1.567 0.117
DE2 <--> Read2 -0.021 0.000 0.001 -0.355 0.722
DE2 <--> Read1 0.132 0.000 0.001 0.539 *
DE2 <--> RN2 -0.587 -0.001 0.000 -7.372 ***
MA2 <--> RN2 -0.556 -0.001 0.000 -6.956 ***
MA2 <--> DE2 0.775 0.009 0.001 9.400 ***
64 Note. *** = p< .001, ** = p< .01, * = p< .05
The figure 7 shows that the model of the relationships between morphological awareness, decoding-related and rapid naming, and reading ability without covariates. The table 10 results show that the rapid naming was significantly negative relationship with other variable (e.g. rapid naming at time 1 has negative relationship with reading ability at time 2 with the standardised path coefficients is -0.267). These result indicated that if student have high ability in morphological awareness decoding-related variables and reading ability then they require less time to do rapid naming test.
However, when combine rapid naming, morphological awareness and decoding-related variables into the cognitive component, it can be assumed that cognitive component was significantly related with reading ability over the time and also reading ability was
significantly related with cognitive component over the time. In sum, the Table 11 reports the summary of finding of the cross-lagged SEM results with the hypothesis of this study.
Table 11
Summary of findings.
Hypotheses Findings
1. Cognitive component and reading ability are correlated with each other in time 1.
1. Supported: Cognitive component was significantly related to reading ability in time 1 (C1R1 =–0.883, p< .001).
2. Cognitive component and reading ability are correlated with each other in time 2.
2. Supported. Cognitive component was significantly related to reading ability ( C2R2 = –0.157, p< .05).
65 Table 11 (continued)
Summary of findings.
Hypotheses Findings
3. Cognitive component in time 1
predicted cognitive component in time 2.
3. Supported: Cognitive component at time 1 predicted cognitive component at time 2 (C1C2 = 0.332, p< .01).
4. Reading ability in time 1 positively predicts reading ability in time 2.
4. Supported: Reading ability at time 1 predicted reading ability at time 2 (R1R2 = 0.201, p< .001).
5. Reading ability in time 1 predicts cognitive component in time 2.
5. Supported: Reading ability in time 1 significantly predicted cognitive component in time 2 (C1R2 = –0.274, p< .05).
6. Cognitive component in time 1 predicts reading ability in time 2.
6. Supported: Cognitive component in time 1 significantly predicted reading ability in time 2 ( C2R2 = –768, p< .001).
66
CHAPTER FIVE
DISCUSSION
5.1 Summary
The relationship between cognitive component and reading ability has been studied from several perspectives but remains inadequately understood, especially in the context of the Thai language, which is characterised as a transparent orthography. Several models have been employed in explaining the reading process, in which the relationships amongst reading-related variables are analyzed in a complex manner, for example the simple view of reading (Gough & Tunmer, 1986), Bottom-up Approach Model (Carrell, 1988), Top-down Approach Model (Ausubel, 1956) and Interactive View of reading (Rumelhart (1990). This study sought to explore these relationships in the context of Thai—a direction that has not been previously pursued. This study looked into the relationship between cognitive component and reading ability in reading development to determine whether specific directionality patterns exist over a period of one calendar year in the learning process.
Two tests were developed to assess each of the three domains of cognitive component in the context of the Thai language. That are, the morphological production and
morphological structure tests were developed for the morphological awareness domain; the rapid colour naming and rapid number naming tests were created for the rapid automatized naming domain; and the phoneme isolation and rapid word segmentation tests were designed for the decoding-related domain. Two other tests were developed to assess reading ability: the reading comprehension and word recognition tests. The content validity of all the tests was evaluated by an expert panel. The reliability of all the tests was greater than 0.60 (ranged from 0.647 to 0.981).
67
The cognitive component and reading ability of the 357 students in the fourth grade were examined to determine how cognitive and reading ability relationships change over time when these students reached grade five. As previously indicated, first-time data collection was conducted in January–February 2014 (time 1), and second-time data collection was carried out in November 2014–January 2015 (time 2). Two-time cross-lagged panel models were assessed to ascertain the directionality of reading development.
After data collection, the frequency distribution of cognitive component and reading ability across all the eight tests for time-1 and time-2 data was evaluated in term of skewness and kurtosis statistics, which suggest that all the tests were characterised by normal
distribution. The intercorrelations between cognitive component and reading ability in times 1 and 2 two were significant (ranged from 0.286 to 0.964). In order to achieve the aim of the study, the cross-lagged SEM was used to determine the direction of the relationship between cognitive component and reading ability in time-1 and time-2 data. The following sections discuss the major findings and related considerations.