INTRODUCTION
The psychological concept of working memory (WM) dates back to the 1960s.1,2 Over fifty years of scientific studies from psychology, biology, and/or neuroscience have been undertaken around this construct, even though psychology, biology, and neuroscience have still not arrived at a uniform categorization of its processes and functions.3,4 Despite this, WM is implicated in all mental and cognitive work, such as reading, writing, computation, comprehension, etc. For example, reading is theorized to be the product of two sub-processes: word identification and linguistic (or auditory) comprehension.5 Ample evidence exists to support this view from a diversity of alphabetic writings on the link between word identification and children’s linguistic comprehension.5,6
Human WM has a critical role in facilitating the understanding of words we read, while impaired WM can make it more difficult to segment new text to understand given information on a page. What this means is that a reader struggling with impaired WM has challenges and difficulties with comprehension. A typical example is the Auditory WM which plays a critical function in reading. Auditory WM facilitates the recall of sounds and the.0ir associations with letters. When reading long, unfamiliar words from text, such as “cytoplasmic,” the words have to be segmented into syllabi or sound chunks, such as cy-to-plasmic. As the little reader decodes the ‘cy’ and the ‘to’ simultaneously, the child holds the sounds in her or his mind until she or he works through ‘plas’ and ‘mic’ and connects the sounds together. What this means in practice is that the more proficient the Auditory WM of a child, the easier the process becomes for such a child. Thus, word recognition at the level of sensory memory (external receptors) plays a critical role in the early development of reading compared to linguistic comprehension. However, with time, the more fluent word reading becomes, the better language comprehension also takes shape as a source of reading skill.7 Reading skills are critical to learning generally as children progress from primary to junior high.8 Given that many rural children in Ghana perform abysmally in language at the Basic Examination Certificate,9 there is increasing psycho-pedagogical interest to learn more about the component skills in memory and comprehension monitoring that contribute substantially to language comprehension.
Statement of the Problem
Ample research has been conducted for the last 30-years to monitor the exact nature of children’s reading process and also how children on their own are able to evaluate their comprehension when engaged in reading.10,11,12,13,14 Specifically, comprehension monitoring skills in reading are crucial in both contemporary literate society and school. A child who has the capability to self-assess metacognitively her understanding of a text, becoming conscious of unclear passages, is more likely to be a proficient reader and successful in school. The ability to think about how one is thinking as one reads a text, such as the ability to detect internal inconsistencies, has been observed as one of the distinguishing marks between proficient readers and less proficient readers.15,16 This also explains the variance observed in reading comprehension development between the ages of 7 to 11-years. Research on deficient readers struggling with word identification and with poor reading comprehension confirms the notion that monitoring comprehension is essentially critical to reading.15,16 The implication is that impoverished readers, compared to proficient readers, are generally deficient in monitoring their comprehension.16
Even though comprehension monitoring is critical to effective reading from primary four through five, few studies in Ghana have investigated the relationship between WM failures and comprehension monitoring impairments at this level. Additionally, comprehension monitoring as a metacognitive skill (which is not that one either has it or does not have it but develops as one goes through academic progression) is not too clear in the literature on how it develops through academic progression. The factors determining this are less well-known.
Few studies have examined the nexus between WM failures and comprehension impairments and their roles as potential foundational skills for predicting future poor reading capability in upper primary readers, besides word recognition in the study area. This study aims to fill this gap.
Research Questions
1. In what ways are primary readers able to notice inconsistencies in the first and second sentences of a paragraph?
2. What effect does stronger academic language (vocabulary, story recall, and academic knowledge) have on re-reading times for mismatched words in sentences?
3. How does self-reporting in early readers influence comprehension monitoring?
THEORETICAL FOUNDATIONS/LITERATURE REVIEW
Development of Comprehension Monitoring and Primary Readers
Monitoring comprehension is a complex cognitive skill. Nevertheless, it has been empirically verified that as early as the preschool period, one could observe prototypical monitoring behaviors with children under the age of 3-years, when the sequence of events in well-known stories is changed.17 Then, between ages 8 and 11, comprehension monitoring becomes more improved.18 Empirical studies on children’s errors show some high improvements among 5-year-olds in comprehension monitoring, with improvement observed at the age of 11.19 At the empirical level, three models have examined how comprehension monitoring contributes to reading. The first group of studies examined concurrent relations and children’s comprehension monitoring when other variables such as vocabulary and WM were controlled.20,21,22 Other groups of studies have also investigated comprehension, with children having difficulties understanding what they read due to the challenges of identifying internal inconsistencies. This inability is an indication of poor text integration and insufficient mental construction of what they read.4,23 Still, other studies investigated WM as the storage and processing of information concurrently from a text. These combined processes are critical for relating new incoming information to already-existing models in the mind. This model gives a theoretical underpinning for the connection between WM and reading comprehension generally, as well as comprehension monitoring in particular.21 It is particularly interesting to underscore the fact that in the literature, skills in monitoring comprehension explain the specific difference in reading measures in the ages of 8, 9 and 11-years and above of WM to reading. Findings such as this show that WM by itself is insufficient for skilled reading. A higher-level metacognitive process plays a critical role in predicting comprehension, both in reading and listening.20,21,22
Another important variable has to do with vocabulary knowledge as a critical predicting factor for both reading comprehension and skills in monitoring comprehension.24 Nevertheless, vocabulary is never a stand-alone variable, and it is intimately connected to WM.23,25 Recent studies on inference-making suggest the mediating roles of WM capacity and the size of vocabulary. Therefore, any attempt to understand the nexus between monitoring comprehension and, by implication, any comprehension impairment and reading comprehension correlates, such as WM and vocabulary, needs to be controlled for associated variance.
Comprehension Monitoring and Text Comprehension
Text comprehension is contingent upon the reader’s ability to mentally construct textual information. To do this, the reader has to go beyond literal words and sentences and integrate the meanings of words and sentences coherently.26,27 What is more important is that as the reading of the text unfolds, there are continuous refinements, and readers and listeners successively connect concepts and ideas into the mental representation.28 Therefore, seeing the text as both integrative and dynamic is essential to comprehension. What this means is that theoretically, comprehension monitoring or metacognition is critically constitutive in constructing a mental representation of what one is reading. Readers who consistently monitor what they are reading in the context of coherent mental representation will be better able to identify at what point in the reading they fail to know the meaning of some words, at what point in the reading the text fails to synergize with their previous knowledge, as well as when some information in the text can hardly come together to make meaning.
In situations such as these, readers with better monitoring skills are not only conscious of failure in comprehension, but more critically, they are able to take effective action to remedy gaps, such as re-reading the text, posing questions, making inferences, or cross-checking the meanings of unfamiliar words.29 Thus, what is theoretically evident in the literature on reading psychology is this: evaluating one’s comprehension during reading is significantly connected with integrative processes critical to constructing a coherent mental representation.30 On the other hand, readers who process text without synchronizing propositions and ideas into a coherent whole but serially, such as word-by-word or sentence-by-sentence, fail to detect conflicts. This impairs comprehension monitoring. Comprehension monitoring is then theorized as a higher-level linguistic skill that is distinct from such lower linguistic skills as vocabulary and grammar.
Comprehension Monitoring and Reading through Eye-tracking with Early Readers Learning English as L2
Literacy skills, such as reading a complex matter involving a lot of cognitive processes such as letter identification, combining letters to form words, and sounding them in the form of phenomena, as well as other linguistic skills31; for example, (1) lexical and syntactic competencies,32 (2) integrating information from prior knowledge and from text,33 (3) inference making,14 (4) disregarding non-relevant information,30 (5) readers’ motivation34 and (6) metacognitive abilities.35 Metacognition is crucially important for this current study since text comprehension monitoring is a critical component of metacognition. In particular, procedural metacognition36 is underscored by two mental processes: monitoring and control. Understandably, then, learning to read can be challenging for many early readers learning English as L2, using the maternal language in the home.37 Using English as an L2 proficiency is significantly related to success in formal education.38,39,40 Therefore, it is critical to understand these challenges. Children learning English as L2 compared to native or monolinguals often lag behind in reading comprehension. This is often linked to deficits in vocabulary size and depth.36,38-46 However, when it comes to comprehension monitoring, there is some evidence that learners of English as L2 or English Additional (EAL) tend generally to be less adeptly reactive to L1. They lag behind in metacognition processes such as checking and regulating whether or not they understand figurative and multi-word phrases.47
Many empirical studies assume that the reader’s ability to notice and react to mismatches between the reader’s background knowledge and text information in reading48 or in the view of other authors, between the reader’s interpretation of previous knowledge and later text segmentation constitutes comprehension monitoring.49,50 A typical approach to measuring comprehension monitoring is that children are asked to identify a mismatch inserted into a text, such as a syntactic mismatch between a pronoun and a verb. It could also be a mismatch between text information and respondents’ ability to verbalize or underline the inconsistency.9,16,51,52,53 Such studies demonstrate a significant relationship between comprehension and monitoring ability; in older children, skilled readers tend to have better comprehension and monitoring skills than younger children or less skilled readers.15,16,19,54 L2 learners of English, relative to L1, have been identified as being less likely to notice inconsistencies in the interpretation of figurative phrases and the story context.47
Besides, children with reading disorders are seen in reading comprehension to have a better predisposition than normal children to monitor their understanding of written text.55 Empirical evidence shows that children deficient in comprehending text due to challenges of language, WM, vocabulary, reason, etc. have also shown that metacognition (comprehension monitoring) may also be implicated.56 Evidence-based examples have to do with poor and slower readers having poor post-judging confidence ratings of prior information. This is attributed to lower accessibility to the information stored, which also affects readers’ confidence in their prior performance.57,58,59
MATERIALS AND METHODS
Participants
Participants were 150 pupils in the upper primary: primary four through primary six. They were enrolled in an eight-weekend (Friday evening-Saturday evening) online and on-site study designed to investigate whether or not there is a relationship between WM failures and comprehension monitoring impairments. Up to about 50 pupils, ranging from primary 4-6 were selected from each of the three (3) study sites in Kumasi, Bechem, and Sunyani in the 2022- 2023 academic year. In the three study sites, pupils were enrolled for this study in their first, second, and third years of study from selected local schools. Language teachers in these primary grades were also invited to participate. Schools that consented to participate were given a recruitment package with a short questionnaire and parental consent form. Teachers also completed screening forms to exclude any pupils with severe disabilities such that they would be unable to communicate in English. Out of these 150 pupils, 60% (90) were males, and the remaining 40% (60) were females. Forty percent (40%) lived in homes where English was the language spoken, while a majority (60%) spoke both English and Akan in the homes. The overall socioeconomic background of children was working class (90%), while 10% were from middle-class educated families.
Procedures and Measures
The procedure involved the assessment of pupils’ skills during each week of the six weeks of interaction. Prior to training, trainers had to complete training modules that were specific to particular measures, which included video narrative presentations and finishing online quizzes. Pupils were evaluated individually. Measures with responses and scores deemed complex were taped in audio. The subsequent five constructs were: (1) decoding, (2) vocabulary, (3) WM, (4) comprehension monitoring, and (5) reading comprehension. All five measures were assigned to all three grades of pupils who participated in the study.
Decoding: Decoding, as used in this study, was a latent variable with four measures. Two were from the woodcock reading mastery test—revised/normative update (WRMT-R/NU).60 One test measured pupils’ ability to apply phonic and structural analysis skills to not-so-familiar words. This consisted of 45 items, Reliability was 0.93. The second test was Word Identification. This measured pupils’ ability to identify isolated written words. This contained 106 items. The first six consecutive correct answers were used as the basis for achievement, Reliability was 0.97. The third test, Word Reading Efficiency (TOWRE-2).61 This measured the number of English words ranging from high to low that pupils could pronounce without any errors in 45-minutes, Reliability was 0.93.
Vocabulary: Vocabulary was measured using three variables: 1) expressive vocabulary test, Second Edition (EVT-2) uses standardized and norm-referenced tests to test expressive vocabulary. Participants were shown pictures and asked to give a word to label them, and then asked to give a word as a synonym for a target word. When both labels and synonyms were correct, the items were scored as correct using this scoring, (0=incorrect, 1=correct). The fourth test used the peabody picture vocabulary test (PPVT-4).62 It tested receptive vocabulary. The assessor asked participants to point to a picture out of four after reading a word that corresponds to the target’s word meaning, Internal consistency was found to be 0.97.
WM: The Woodcock-Johnson III NU Test of Cognitive Abilities21 was used to test auditory WM. Participants were asked to store and manipulate information as they listened to stories with both digits and words. They were required to do some reordering this way: first naming the words and the digit that followed, such as 4 followed by cocoa followed by 1 lion, and at the same time holding the sequential order in which they were earlier presented (e.g., cocoa-lion 4-1). A recording device was used to monitor, ensuring that timing gaps were the same for all. The reliability coefficient for all age groups was 0.90-0.97 with internal consistency at 0.80.
Comprehension monitoring: This was tested by asking participants to identify inconsistencies. Eight test stories, either consistent or inconsistent, were used. Three (3) were consistent stories, while five (5) were inconsistent. An instance of an inconsistent story was: “Last night Kwaku walked home through the park. There was no moonlight, so Kwaku could hardly see his way. Kwaku often takes this route home. He walked along a narrow path. The moon was so bright that it lit the way. Kwaku lives on the other side of the park. There was no moonlight, so Kwaku could hardly see his way. Kwaku often takes this route home. He walked along a narrow path.”
Inconsistent stories were to find out: (a) whether the story made sense, and (b) what was not right with the story using the scoring grade (score=0) or correct (score=1) for this task, which ranged from 0 to 8 points.
Reading comprehension: This construct was measured with two instruments: The Woodcock Reading Mastery Test—Revised/Normative Updated (WRMT)63 was used to test participants’ reading comprehension in short passages; the Gates-MacGinitie Reading Test64 was used to examine how well-participants could read.
RESULTS
Before applying multivariate analysis of variance (MANOVA) to test the assumptions for normality, linearity, univariate and multivariate outliers, homogeneity of variance-covariance matrices, and multicollinearity, we conducted a preliminary test. The various checks on the assumption that comprehension monitoring was dependent on WM detected no violations. We also tested gender differences in WM and comprehension monitoring using a one-way multivariate analysis of variance. We observed a statistically significant difference in the dependent variable between males and females on the combined dependent variables: F (2, 257)=13.85, p=0.000; Wilks’ Lambda=0.90; partial eta squared=0.10. We considered the results for the dependent variable separately using the Bonferroni adjusted alpha level of 0.25 to arrive at statistical significance: the scores on WM (F(1, 258)=27.25, p=0.000, partial eta square=0.10). An inspection of the mean scores indicated that males reported a higher-level of attitude towards statistics (M=90.71, 12.48) than females (M=82.06, 13.85) (Tables 1, 2 and 3).
Table 1. Descriptive Statistics of Children Assessment |
Measures |
Range |
N |
Mean (SD)
|
Word attached raw (P/4)
|
6-39
|
100 |
112 (20.89) |
Word Attach Standard (P/4) |
97-139 |
100
|
117.36 (8.67)
|
Word Identification raw (P/4)
|
19-83 |
100 |
49.19 (12.71) |
Word Identification Standard (P/4) |
93-146 |
100 |
118.98 (11.59)
|
Sight Word Raw (P/4)
|
12-73 |
100 |
45.14(14.74) |
Sight Word Standard (P/4) |
71-142 |
100 |
108.346 (15.30)
|
Phenomics Decoding raw (P/4)
|
2-42 |
100 |
20.06 (10.69) |
Phenomics Decoding standard (P/4) |
69-145 |
100 |
103.67 (14.61)
|
Vocabulary |
EVT-2 Raw (P/4)
|
58-140 |
100 |
97.86 (13.78) |
EVT-2 Standard (P/4) |
76-147 |
100 |
108.54.(12.28)
|
PPVT-4 Raw (P/4)
|
87-195 |
100 |
130.59 (16.37) |
PVT-4 Standard |
82-160 |
100 |
111.876.8 (12.64)
|
CELF- 4 Word classes Receptive Raw (P/4)
|
10-21 |
100 |
19.03 (1.84) |
CELF-4 Word classes Expressive (P/4) |
7-20 |
100 |
115.14.(1.84.)
|
WJ- Auditory Memory raw (P/4)
|
0-86 |
100 |
4.56 (2.11) |
WJ- Auditory Memory Standard (P/4) |
81-152 |
100 |
113.22 (13.81)
|
Comprehension Monitoring |
Comprehension Monitoring (P/4)
|
0-8 |
100 |
4.56(2.11)
|
Comprehension Monitoring (P/5)
|
1-8 |
100 |
6.03 (1.65) |
Comprehension Monitoring (P/6) |
2-8 |
|
6.37 (1.43)
|
Reading Comprehension |
WRMT- Passage Comprehension raw (P/6)
|
23-53 |
|
37.95 (5.50) |
WRMT Passage Comprehension Standard (P/6) |
88-132 |
|
110.71 (9.33)
|
Gates Ma Ginite raw (P/6)
|
13-47 |
|
35.22 (8.42) |
Gates MacGinite Standard (P/6) |
2-9 |
|
59.3 (8.4)
|
Reading Comprehension |
Reading comprehension measure (P/6)
|
10-26 |
|
20.45 (4.04)
|
Note: WRMT=Woodcock Reading Mastery Test; EVT-2=Expressive Vocabulary Test; PPVT-4=Peabody Picture Vocabulary Test; G1=Grade 1; G2=Grade 2; Grade 3=Grade 3 |
Table 2. Ability to Notice Inconsistencies between Paragraphs |
Measures
|
Range |
N |
Mean (SD) |
Average Change between Grades |
M Plus Growth (between Grades 4-5, 5-6)
|
Children’s Grades |
Primary 4
|
73-93 |
100 |
4.55 (2.10) |
t (107) =6.91, p<0.0001, d=0.67 |
r=0.34
|
Primary 5
|
86-105 |
100 |
6.02 (1.66) |
t (107) =1.61, p<0.108, d=0.67 |
r=0.22 |
Primary 6 |
98-1118 |
100 |
6.35 (1.41) |
|
|
Across Board Performance of Groups between Skilled and Poor Readers |
Skilled readers
|
73-93 |
210 |
6.01(1.66) |
|
r=-0.51 |
Poor readers |
86-105 |
90 |
2.01 (3.66) |
|
|
Table 3. Respondents’ Scores on Working Memory and Comprehension Monitoring |
Dependent Variables
|
Males |
Females |
Between Subject-Effect |
Mean |
SD |
Mean |
SD |
Df |
M Sq |
F-value |
p-value
|
Working memory score
|
90.71 |
12.48 |
82.06 |
13.85 |
1 |
4622.82 |
27.25 |
0.000* |
Comprehension monitoring score |
39.87 |
8.29 |
37.90 |
9.34 |
1 |
240.49 |
3.10 |
0.075
|
Grade Related Ability to Notice Inconsistencies between Paragraphs
This paper investigated whether or not WM failures negatively affect comprehension monitoring. To address this, the topic was broken into three research questions: 1) In what ways are early readers able to notice inconsistencies in the first and second sentences of a paragraph? 2) What effect does stronger academic language (vocabulary, story recall, and academic knowledge) have on re-reading times for mismatched words in sentences? and 3) How does self-reporting in upper primary readers influence comprehension monitoring?
What is evident in reading psychology is the link between readers’ ability to integrate what they read into a coherent mental process.30 Hence, the first question assessed the extent of integration with comprehension monitoring. The second research question measured whether or not the size of vocabulary, general academic knowledge of readers, and ability to recall what readers have read were associated with the WM processes. For example, if readers have difficulty with eye tracking, letter identification, combining letters to form words, or forming words to read sentences, some of these are deemed to be automatic processes at the perceptual level. Finally, the third question examined children’s ability to self-report what they have read.
The overall purpose of this paper was to investigate whether or not WM failures negatively affect comprehension monitoring. To address this, this study broke the topic into three research questions: 1) In what ways are early readers able to notice inconsistencies in the first and second sentences of a paragraph? 2) What effect does stronger academic language (vocabulary, story recall, and academic knowledge) have on re-reading times for mismatched words in sentences? and 3) How does self-reporting in upper primary readers influence comprehension monitoring? What has become evident in reading psychology is that text comprehension is directly linked with readers’ ability to integrate what they read into a coherent mental process.30 Hence, the first question on readers’ ability to notice or fail to notice inconsistencies in related sentences was to assess the extent of integration with comprehension monitoring. Further, the second research question sought to measure whether or not the size of vocabulary, general academic knowledge of readers, and ability to recall what readers have read were associated with the WM processes. For example, if readers have difficulty with eye tracking, letter identification, combining letters to form words, or forming words to read sentences, some of which are deemed to be automatic processes at the perceptual level, then reading becomes a bit more difficult at a much higher-level. Finally, the third question examined children’s ability to self-report what they have read.
Ability to Notice Inconsistencies between Paragraphs
The results showed that the ability to recognize inconsistencies in paragraphs by pupils increased from primary four (M=4.55, SD=2.10) to primary 5 (M=6.02, SD=1.66) and primary 6 (M= 6.35, SD=1.41). There was an average change between grades. Significant changes between the two early grades of primary 4 and primary 5 were observed as follows (t (107) =6.91, p<0.0001, d=0.67) (t (106) =1.61, p=0.108, d=0.17) respectively. What these scores suggested was that this skill had increased. To further understand the growth in comprehension monitoring based on this increase in the ability to notice inconsistencies, we needed to isolate and investigate the variability at primary 4. A latent growth model in Mplus65 was run. Across time, the pattern observed in the means of the ability to notice inconsistencies and hence the ability to monitor comprehension with time, the correlations between Primary 4 and Primary 5 (r=0.34) relative to Primary 5 and Primary 6 (r=0.22) offered more support for a trend that was non-linear.
In addition to the grade-level non-linear trend, across the board, participants were found to have performed far better in recognizing inconsistencies relative to those who performed poorly, the former (M=6.01, SD=1.66), and the latter (M=2.01, SD=3.66). The former were more than three times able to better monitor their comprehension. The correlation between the two groups was r=-0.51, indicating a significant inverse correlation between the inability to recognize inconsistencies and the ability to recognize consistencies.
Stronger Academic Language (vocabulary, story recall, academic knowledge) and Re-Reading Times for Mismatched Words in Sentences
The relationship between vocabulary and WM was measured across the three grade levels, focusing on the ability to recall stories and re-reading times for mismatched words in sentences. First, class-related changes in comprehension monitoring at all three levels were examined. Overall, there was an increase in monitoring comprehension. However, the growth in this construct was found to be faster between primary five and primary six than it was between primary four and primary five. So, there was reliable variance in the linear growth, apparently due to individual differences, such as the size of vocabulary and academic knowledge, which made higher graders perform better and quicker in re-reading times for mismatched words. Vocabulary and Comprehension monitoring had a stronger positive relationship (β=0.07, p<0.001) for primary five and primary six, while for primary four and five, it was found to be (β=0.04, p<0.001).
We tested whether this was similar to WM and the data indicated for primary five and six (β=0.06, p<0.001) and (β=0.05, p<0.001) for primary four and five, showing a close relationship between WM and Comprehension Monitoring. Again, this data showed the presence of reliable differences in linear growth parameters. In summary, these data suggest that comprehension monitoring increased significantly across grade levels from primary 4 through primary 6. However, with time, there was some deceleration based on vocabulary in the lower grades of primary four.
Self-Reporting in Primary Readers and Comprehension
This third question examined the reader’s ability to do self-reporting in reading, which could predict both future reading and comprehension monitoring. This was informed by the first experiment, in which the increase in reading ability was not initially found to be entirely linear but non-linear. Three predictors were measured in primary four using the following constructs: decoding, vocabulary, and WM, focusing on an increase in comprehension monitoring. In primary four, decoding was found to be significant and a good predictor of reading comprehension (β=0.77, p<0.0001). This explained about 58% of primary-six reading comprehension. Vocabulary and WM models were included in the measure. Decoding continued to be a leading predictor for reading (β=0.46, p<0.0001), vocabulary (β=0.37, p<0.0001) and WM (β=0.45, p<0.0001) Combining vocabulary and WM accounted for about 25% of the increased reading variance in primary six. When primary four pupils decoding, vocabulary, and WM were controlled, their reading comprehension at primary six was predicted to be (β=0.38, p<0.001). Additionally, comprehension monitoring was uniquely responsible for explaining an additional 8% of the variance in reading comprehension in primary six.
DISCUSSION
The findings of this study are consistent with other empirical studies showing that WM is directly related to children’s reading comprehension as well as in adults.66,67 Given that reading is a combination of integrating knowledge and skills such as decoding, vocabulary size, and syntactic and semantic processing, the findings reported here, in which WM has been consistently found to be correlated to decoding and vocabulary (β=0.45, p<0.0001) point to the significance of this construct in reading comprehension. What this finding means in practice is that at the elementary perceptual level of reading, involving processes such as decoding (eye tracking), identifying letters, combining letters to form words, combining words to form sentences, etc., a reader has challenges; at a higher-level of reading one sentence after another to integrate meaning, one would have exhausted attentional resources to make meaning.27,28,29 This situation of memory failures becomes more critical at the higher-level of processing, such as monitoring one’s comprehension to find out if one has understood what is being read, which becomes even more challenging.26 The findings in this study were consistent in that, for each construct assessed to find a correlation between decoding and WM, the correlation was almost related. For example, when measuring the three constructs of decoding, vocabulary, and WM in the third research question, decoding continued to be a leading predictor for reading (β=0.46, p<0.0001), vocabulary (β=0.37, p<0.0001) and WM (β=0.45, p<0.0001), Combining vocabulary and WM, the latter was found to account for about 25% of the additional reading variance in primary six.
Similarly, in the second research question, vocabulary had a stronger positive relationship with comprehension monitoring (β=0.07, p<0.001) for primary five and primary six, while primary four and five were found to have a (β=0.04, p<0.001) an indication of the grade/age relationship, which was found to be linear. This particular finding was in direct contradiction to the finding of Yeoman68 in which this author found a non-linear correlation between first- and second-grade readers and second- and third-graders. Again, testing this in relation to WM, there was a close relationship for primary five and six (β=0.06 p<0.001) and (β=0.05, p<0.001) for primary four and five, showing some close relationship of WM to comprehension monitoring.
In general, the participants of this study, who were all L2 learners of English, compared to those who used the native language more in their homes than L2, lacked metacognitive processes such as self-reporting, which confirmed.69 Whereas those who used L2 frequently in the homes were about three times better at recognizing mismatches and better at interpreting previous and later text segmentation, which constituted comprehension monitoring49,50,70 it was less so across the three grades for those who used more of L1. This implies that WM failures precipitated by L1 also impaired comprehension monitoring in L2. This finding supports the theory of Nugba et al71 that since WM reflects ability in attention control, readers with high WM tend to have more resources for attentional control to focus on primary tasks. This finding also complements that of Yeomans-Maldonado72 of mind wandering, especially dealing with demanding reading.
Interventions for L2 Children with Reading Difficulty and Language Teacher Education
Challenges with literacy acquisition, especially reading, in many Ghanaian basic schools could be attributed to WM challenges. Based on the findings of this paper, the subsequent three pedagogical interventions are being suggested: First, teacher education needs to focus on increasing teacher trainees’ awareness of WM problems. Most teachers appear to be unaware of WM capacities and that children with impaired WM have memory failures that begin to manifest in poor reading acquisition. Second, teacher education needs to expose teacher trainees more than ever to adopt a form of teaching aimed at reducing memory loads, especially for already struggling students. Precisely because poor-reading pupils or students typically have problems related to lower perceptual processes in reading such as de-codification (letter and word identification as well as phenomes), it appears to be pedagogically more pragmatic to use audio to teach these lower processes first for pupils struggling with them. When these lower processes have become automatic, then teachers expose them to print text. This approach would be a strategy to adapt the child’s environment to minimize memory load and enhance classroom learning. This is especially true when we have instances of some disadvantaged children from rural home backgrounds who are more used to oral communication than printed communication, and they carry this initial disadvantage to learning L2. The third approach would be for reading teachers to teach children to use memory strategies to improve the efficiency of their working memories.
Additionally, an integral part of supporting children who have deficient WM in the classroom would be for teachers to monitor students, especially struggling ones, on how they cope with mentally challenging activities demanding more mental load in such areas as mathematics and science. The fourth approach would be for teachers to assess students’ learning activities to identify possible areas that are likely to be problematic for children with small WM capacities, such as a) activities imposing heavy mental demands on WM; b) those that are too long or have long sentences and paragraphs; and c) those having a large chunk of unfamiliar, complex, and difficult-to-understand material that imposes significant mental processing, especially in explaining complex concepts.
The findings in this study have underscored the correlation between WM and comprehension monitoring among children who use English as L2, measured along different constructs such as decoding, vocabulary, WM, comprehension monitoring, and reading comprehension. The overriding finding is that WM plays a critical role in digesting words, especially in monitoring comprehension. Deficient WM makes decoding new text cumbersome since WM reflects attention-control abilities. This means that individuals with more attentional control resources are more capable of focusing on primary tasks while simultaneously engaging in metacognitive processes than those with less attentional control resources.