Poor oral discourse skills are the key cognitive-linguistic weakness of Chinese poor comprehenders: A three-year longitudinal study

Abstract

Prior research on reading difficulties has mainly focused on word decoding problems. However, there exists another group of children – poor comprehenders (PCs) – who have normal word decoding abilities but difficulties in reading comprehension. Less is known about PCs especially in non-alphabetic languages such as Chinese. This study identified three groups – poor decoders, PCs, and average readers – among 103 Chinese children at the end of first grade. Children’s performances in reading and cognitive-linguistic measures, both concurrently at first grade and retrospectively at kindergarten levels two and three, were then compared among the three groups. This study is the first to demonstrate the distinct cognitive profiles of poor decoders and PCs in Chinese. The key cognitive-linguistic weaknesses of Chinese PCs were found to be in oral discourse skills and working memory. The retrospective data further revealed their oral discourse weakness as early as in preschool years at age 5. Practically, the necessity of developing assessment and intervention tools that focus on oral discourse skills for Chinese PCs is highlighted.

Keywords

Chinese cognitive-linguistic skills poor comprehenders poor decoders reading comprehension

Introduction

Given that reading is an essential part of learning, children with reading difficulties are more prone to academic failure. Early identification and special intervention for children with reading difficulties are thus critical. Since reading difficulties might have more than one cause, researchers have proposed different subtypes of reading difficulties. In English, the most widely studied subtype of reading difficulty is dyslexia, which affects approximately 10% of school-age children (Snowling, 2000). Dyslexic children are characterized by difficulties in basic word decoding and thus are sometimes referred to as poor decoders (Aaron, Kuchta, & Grapenthin, 1988; Palladino & Ferrari, 2013). Such difficulties further undermine their ability to read and extract meaning from written texts, and thus lead to poor reading comprehension (Perfetti & Hogaboam, 1975; Perfetti, Landi, & Oakhill, 2005). Another subtype of reading difficulty is specific reading comprehension difficulties, which affects also approximately 5–10% of school-age children (Nation & Snowling, 1997; Yuill & Oakhill, 1991). These children are often referred to as poor comprehenders (PCs). By definition, PCs suffer from poor reading comprehension despite having average or even good word reading ability. Unlike children with dyslexia or poor decoding, PCs behave normally in basic word decoding and their problem is only revealed when they are asked to interpret the meanings of texts. This specific feature of PCs makes it relatively difficult for educators to identify children who have this ‘hidden’ type of reading difficulties (Nation, 2005; Nation, Clarke, Marshall, & Durand, 2004). Since PCs were found to encounter reading comprehension difficulties as much as dyslexic children (Catts, Adlof, & Weismer, 2006) and given their similarly high prevalence, equal attention should be given to the identification of and intervention for dyslexic children and PCs.

Researchers have acknowledged the reading problems of PCs and have successfully identified the key features of PCs in English. It was found that a selective combination of strengths and weaknesses of cognitive-linguistic skills contributes to the unique feature of PCs – they have normal or good phonological skills but have poor oral language skills and working memory (Cain, Oakhill, & Bryant, 2004; Cain, Oakhill, & Elbro, 2003; Nation & Snowling, 2000). It has been suggested that the good phonological skills of PCs account for their skilled word reading, while poor oral language skills account for their difficulties in comprehension. It has further been suggested that the comprehension difficulties of English-speaking PCs are caused by general oral language weaknesses at the word, sentence, and discourse levels (Catts et al., 2006). Such a conclusion was based on findings showing that these children exhibited a range of oral language problems, including in vocabulary, grammar and discourse skills such as story comprehension, inference-making, and comprehension monitoring (Catts et al., 2006; Nation et al., 2004; Oakhill, Hartt, & Samols, 2005). For example, Nation and colleagues (2004) found that 8-year-old PCs performed more poorly than typical readers in vocabulary and semantic similarity judgments. Moreover, PCs were reported to have grammatical deficits and poor understanding of explicit and inferential information in listening comprehension tests (Cain et al., 2001; Catts et al., 2006).

A number of longitudinal studies further pointed out that English-speaking PCs showed early signs of oral language weaknesses before school entry (Catts et al., 2006; Justice, Mashburn, & Petscher, 2013; Nation, Cocksey, Taylor, & Bishop, 2010). For example, the four-year longitudinal study by Nation and colleagues (2010) identified PCs and average readers among a group of 8-year-olds, and retrospectively examined their cognitive profiles when they were aged 5.5, 6, and 7. Their findings showed that prospective PCs were inferior to prospective average readers in vocabulary, grammatical, and story comprehension at all ages. Similar findings were reported by Justice and colleagues (2013), who demonstrated the early language difficulties of PCs during preschool years. These researchers concluded that early oral language weaknesses are the causal antecedents of specific reading comprehension difficulties, and that PCs can be identified and given intervention before they experience reading failure at school.

Despite the above findings in English, the characteristics of PCs in non-alphabetic languages, such as Chinese, have yet to be explored. Given the unique characteristics of the Chinese language, it is believed that the contributing factors of Chinese reading difficulties could be different from those in English. In fact, previous studies have demonstrated that the cognitive bases of dyslexia, or poor word decoding, are distinctly different from those in English (Ho, Chan, Lee, Tsang, & Luan, 2004; Shu, McBride-Chang, Wu, & Liu, 2006; Wu, Packard, & Shu, 2009). The basic orthographic unit in Chinese is the character, which is visually made up of complex stroke-patterns with no elements that are associated with phonemes. Characterized by their heavy semantic component, most Chinese words are compound words that are made to represent units of meaning (Zhang et al., 2012). For example, the Chinese word for elevator is 升降機, which is constructed by combining three individual morphemic meanings – 升 (up) + 降 (down) + 機 (machine). Meanwhile, Chinese is known to use relatively different linguistic systems of grammar and discourse to convey meaning at the sentence and text levels (Chik et al., 2012; Yeung, Ho, Chan, Chung, & Wong, 2013). Despite the unique features of Chinese, a tailor-made approach for the identification of and intervention for Chinese PCs is currently lacking. Thus, there is a need to investigate the underlying factors of Chinese PCs. In addition, it would be highly valuable to make a direct comparison between the factors contributing to PCs and the factors contributing to poor decoders among Chinese children.

One special feature of the educational system in Chinese-speaking regions, such as Hong Kong, is that children learn to read very early, beginning at age 3 or 4. As soon as they enter primary school, children are already required to have sufficient ability to comprehend reading passages. In such an educational context, it is sensible to assume that specific reading comprehension difficulties may occur earlier among school-age children in societies such as Hong Kong, even though such difficulties are known as a kind of ‘late-emerging’ reading disability in Western societies (Lipka, Lesaux, & Siegel, 2006). In line with previous research that highlighted the importance of early identification and intervention for PCs (Adlof, Catts, & Lee, 2010; Justice et al., 2013; Nation et al., 2010), we conducted a three-year longitudinal study, in which a sample of Hong Kong children were followed from preschool age (4–6 years old) until they completed first grade (7 years old). This study design enabled us to retrospectively investigate the cognitive and linguistic weaknesses contributing to the PCs identified at the end of first grade. Such results provide valuable information for understanding whether first-grade PCs show early signs of weaknesses before entering elementary school.

In view of the lack of information regarding Chinese PCs, the current study took the initiative to examine the underlying cognitive-linguistic factors that contribute to PCs in Chinese. Children were to be tested on a range of cognitive-linguistic measures that are known to be important for learning to read in Chinese: phonological skills, orthographic skills, oral language skills (i.e., vocabulary, grammatical, and discourse skills), and working memory (Chik et al., 2012; Ho & Bryant, 1997; Ho, Ng, & Ng, 2003; McBride-Chang & Ho, 2000; So & Siegel, 1997; Yeung et al., 2011). Using a three-year longitudinal design, the current study identified three groups of children at the end of first grade: poor decoders, PCs, and average readers. The cognitive-linguistic skills of these three groups of children were compared with each other, concurrently at first grade as well as retrospectively at the last two levels of preschool years (i.e., K2 and K3). The findings were expected to provide novel data to address two specific questions on Chinese PCs: to what extent do the cognitive-linguistic weaknesses of Chinese PCs differ from those of Chinese poor decoders? Do the cognitive-linguistic weaknesses of Chinese PCs begin to show early in preschool years at K2 and K3?

Method

Identification of PCs, poor decoders, and average readers among the participants in grade 1

Participants

A total of 103 Cantonese-speaking Chinese children in Hong Kong (56 boys and 47 girls) participated in the present study. All children were tested at three points: kindergarten level two (K2), kindergarten level three (K3), and elementary school grade one (G1), at which their mean ages were 5.1 years, 6.0 years, and 7.2 years respectively. The sample was initially recruited from five kindergartens that were all mainstream schools using Cantonese as the medium of instruction. Per parents’ and teachers’ reports, all children were typically developing children without any visual, hearing, or language impairments. Prior to the first testing at K2, parents of each participating child received an informed consent form that explained the purpose of the current study and asked for their approval for the participation of their children. The test started after all signed informed consents were received. The current study was approved by the Human Research Ethics Committee for Non-Clinical Faculties, the University of Hong Kong, and was conducted in accordance with the Committee’s code of ethical conduct in research with human participants.

Measures

Table 1 shows an overview of all measures used at each time point. Three measures were employed for the classification of the 103 participants into different groups. All participants completed a nonverbal intelligence (IQ) test at K2, and they also completed a word reading test and a passage reading comprehension test at G1. Detailed descriptions of each measure are as follows:

Table 1.

An overview of all measures at each time point.

Domain	Measure	K2	K3	G1
Nonverbal IQ	Nonverbal IQ	✓
Reading	Word reading	✓	✓	✓
Reading	Reading comprehension			✓
Reading-related cognitive skills	Phonological skill	✓	✓
Reading-related cognitive skills	Orthographic skill	✓	✓	✓
Oral language	Vocabulary skill	✓	✓	✓
	Grammatical skill	✓	✓	✓
	Discourse skill (Story comprehension)Discourse skill (Comprehension monitoring)	✓	✓	✓
				✓
Working memory	Working memory			✓

Nonverbal IQ

Sets A and B of the Raven’s Colored Progressive Matrices (Raven, Court, & Raven, 1995) were conducted at K2 as a proxy of their nonverbal reasoning ability. The Cronbach’s alpha reliability was .70.

Word reading

The Chinese Word Reading subtest of the Hong Kong Test of Specific Learning Difficulties in Reading and Writing (HKT-P[II]; Ho et al., 2007) was administered at G1. HKT-P(II) is a standardized assessment tool for diagnosing dyslexic children in Hong Kong. The test consists of three literacy subtests: Chinese Word Reading, One-Minute Reading, and Chinese Word Dictation. The Chinese Word Reading subtest was used to assess children’s word reading ability in this study. In this test, children were required to read aloud 150 Chinese two-character words. The test was discontinued when the child scored zero in 15 consecutive words. In this task, the maximum raw score and scaled score a child could get was 150 and 19 respectively (the mean scaled score of the test was 10 and the standard deviation was 3). The Cronbach’s alpha reliability was .98.

Reading comprehension

Since an age-appropriate standardized comprehension test was unavailable for Hong Kong children at the first grade level, a non-standardized passage comprehension task was administered at G1. The task contained three narrative-story passages. The storyline of each passage was created by the first author in reference to the stories in two published Chinese literacy books for first graders (Lu, 2008; Pan Lloyds Publishers, 2006). In our task, children answered eight to nine literal and inference questions after reading each of the three short written narratives. The task contained a total of 25 questions. Each correct answer was rewarded 1 mark, except for the three questions for which a complete answer was needed to address two separate points. A score of 0, 1 or 2 was applied to these three questions. The maximum total score of this task was 28. The Cronbach’s alpha reliability was .79.

Identification of poor decoders

In the current study, the Chinese Word Reading subtest of the HKT-P[II] was employed to identify children with poor word decoding at G1. As mentioned above, HKT-P(II) is a norm-referenced standardized assessment tool of dyslexia validated for diagnosing Cantonese-speaking Chinese children in Hong Kong. Based on the local norm of the Chinese word reading subtest, those who obtained a scaled score of 7 or below were defined as poor decoders.

Identification of PCs and average readers

A regression technique proposed by previous researchers (Li & Kirby, 2014; Tong, Deacon, & Cain, 2014; Tong, Deacon, Kirby, Cain, & Parrila, 2011) was used to identify PCs and average readers within this sample. Our data met the assumptions of residual normality, linearity, homoskedasticity, multicollinearity, and univariate and multivariate outliers required for regression analysis. To identify PCs and average readers from our sample at G1, a multiple regression analysis was first conducted to predict the reading comprehension of children at first grade. The multiple predicting variables included their concurrent age, nonverbal IQ, and word reading ability. After that, participants’ actual reading comprehension scores were plotted against the standardized predicted scores obtained from this regression analysis. Children who fell below the lower 70% confidence interval of the regression line were identified as PCs, while those who scored within the 20% confidence intervals were defined as average readers (Figure 1). Children with predicted scores below 1 standard deviation (SD) of the mean were excluded. Additionally, children with extremely high or low scores in word reading (above or below 1.5 SD of the mean) were also excluded.

Figure 1.

The scatterplot for the distribution of poor decoders, poor comprehenders, and average readers, as well as those who were not selected for analyses, in the regression predicting reading comprehension from age, nonverbal IQ, and word reading ability.

Comparison of cognitive-linguistic skills among the three groups (K2 to G1)

Participants

The targeted participants in the three groups were 14 poor decoders (7 boys and 7 girls), 16 PCs (11 boys and 5 girls), and 16 average readers (9 boys and 7 girls). They were identified from the whole sample of 103 children using the screening criteria described in the section above (Figure 1).

Procedures and measures

Each child from the three groups was tested on a range of cognitive-linguistic and reading measures at K2, K3, and G1. Each child was tested individually for approximately one hour, either by the first author or by trained undergraduate students, in a quiet place at their schools or homes. Children were allowed to take a break whenever they needed it.

Details of the measures are described below:

Word reading

This part of the study included two Chinese word reading tasks. The first one was the Chinese Word Reading subtest of the HKT-P[II] (Ho et al., 2007) as mentioned above. The second was the word list of Ho and Bryant (1997), which has been widely adopted as a word reading assessment for Chinese preschool children (McBride-Chang & Ho, 2000; Tong, McBride-Chang, Shu, & Wong, 2009). This preschool-level task was additionally included in our study because the HKT-P(II) was designed for school-age children and thus might not be age appropriate for testing the ability of our sample of children when they were assessed in preschool. This task required children to read aloud 27 simple one-character words and 34 simple two-character words. A ceiling rule was applied when 10 consecutive words were incorrectly read. Taken together, children’s word reading abilities in each round of testing were indexed by the total number of words they could accurately read in the two word reading tasks combined (i.e., the maximum word reading score a child could get was 211). The Cronbach’s alpha reliabilities were within the range of .97–.98.

Cognitive-linguistic measures

Phonological skills. The phonological skills of the participants were assessed using a syllable deletion task (e.g., McBride-Chang & Ho, 2000). A total of 12 test items were administered in our task. In each item, a three-syllable word was presented aurally to children, who were then asked to orally delete one syllable from the word. Among the 12 items, the children were required to delete either the initial syllable (4 items), middle syllable (4 items), or final syllable (4 items). For the items within each of the three categories, one was a real word, e.g., 維他命 (vitamin) /wai4 taa1 ming6/ while the other three were non-words, e.g., 勤老線 /kan4 lou5 sin3/. The maximum score of the task was 12. The Cronbach’s alpha reliabilities were within the range of .79–.84.

Orthographic skills. The design and procedures of the Left-Right Reversal subtest of the HKT-P(II) was adapted (Ho et al., 2007) to measure children’s knowledge about the proper orientation of highly frequent orthographic units. In the task, the children were presented with 28 simple Chinese characters in print, half of which were left-right reversed. The children were required to verbally report to the examiner whether the orientation of each character was correct or not. One mark was given for each correct answer. The task was administered at all three testing points. The Cronbach’s alpha reliabilities were within the range of .70–.80.

Vocabulary skills. A word definition task, based on that of Zhang and colleagues (2012), was used to assess children’s vocabulary skills at all testing points. The task required children to explicate the meaning of 15 words including nouns, verbs, and adjectives. The correct definition of each word was determined based on the suggestions of Chinese dictionaries. For example, the definition of the word 頑皮 /wann4-pei4/ (naughty) was ‘to describe someone who does not do what they are told to’. A correct definition such as this was rewarded 2 marks in each item. One mark was given to a partially correct answer. For example, ‘it is a bad behavior’ is a partially correct definition for the word naughty because it helps to define naughty as some kind of misbehavior, but it fails to precisely explicate its meaning. A score of zero was given to a definition that did not help explain the term at all, e.g., ‘it means a naughty child’. The maximum score of the task was 30. The Cronbach’s alpha reliabilities were within the range of .71–.79.

Grammatical skills. An oral cloze task was used to tap children’s understanding of Chinese grammar (So & Siegel, 1997). Children were orally presented with 22 sentences (via MP3 recordings), each of which had a word left out and replaced by a bell-like sound. Children were instructed to figure out the missing words, which could be a noun, a verb, an adjective, a classifier (e.g., 塊 piece of), or a grammatical morpheme (e.g., 是 is/am/are). One sample sentence item was ‘媽媽嘅蛋糕好香’ (Mother a cake that smells very good). There could be more than one possible answer for each item, e.g., 整 (made), 焗 (baked), 買 (bought), as long as they were grammatically correct. One point was given for each correct answer. The maximum score of this task was 22. The Cronbach’s alpha reliabilities were within the range of .69–.71.

Discourse skills. Two tasks were used to assess children’s comprehension at the discourse level: story comprehension and comprehension monitoring.

Story comprehension. Children listened to a short story, which was about a boy who loved to draw and took part in a drawing competition. After that, children were asked to answer a total of 10 comprehension questions, of which five were literal questions, e.g., ‘What did Lok Lok (the main character) draw for the competition?’, and the other five were inference-type questions, e.g., ‘Do you think Lok Lok was good at drawing? Why?’ Among the 10 questions, five were yes/no or multiple choice questions, three required a one-word answer, and the other two were open-ended questions. One mark was given for each correct answer. Both of the two open-ended questions required an answer that addressed two key points, and one score was given for each correctly answered key point. The maximum score of the task was 12. The Cronbach’s alpha reliabilities were within the range of .71–.76.

Comprehension monitoring. An inconsistency detection task was administered at G1 to assess the children’s ability to monitor the consistency of ideas in a story discourse (Kim & Phillips, 2014). In this task, children listened to 10 short stories through MP3 recordings. Each story contained three to five sentences, in which conflicting information was presented across sentences. For instance, one of the stories talked about a girl who ran out of sweets and could therefore not give any sweets to a boy, but the story concluded by saying that the boy wrote a thank-you card to the girl for the sweets she gave him. The children’s task was to think about which part of the story did not make sense. For each item, 2 points were given if the children were able to accomplish two things: first, to identify the part that went wrong (e.g., ‘it does not make sense that the boy wrote a thank-you card to the girl’) and second, to give reasons for their judgment (e.g., ‘it does not make sense, because at the beginning, the story tells us that the girl did not give any sweets to the boy’). If the children failed to give a full-mark answer, the examiner gave verbal prompts, e.g., ‘why do you think this is wrong?’ Only one point was given if the children were able to point out the problematic statement, but failed to give reasonable explanation to back up their judgment. Zero marks were given if the children said there was nothing wrong with the story or if they incorrectly identified some other parts of the story as problematic. The maximum score of the task was 20. The interrater reliability of Cohen’s kappa was .77. The Cronbach’s alpha reliability was .82.

Working memory. A backward digit span task was employed to measure children’s nonverbal working memory capacity at G1. For each item, the examiner read a string of digits aloud to children, who were then asked to reproduce the digits they heard in reverse order. The 16 items contained increasingly longer series of digits (span of 2 to 7). To receive 1 mark, the children needed to recall all the digits in an item in the correct order. The testing was discontinued if the child failed in two consecutive trials of the same span length. The total score a child could get was the number of items he/she answered correctly before the stopping rule was applied (if there was any). The maximum score of the task was 16. The Cronbach’s alpha reliability was .82.

Results

Identification of poor decoders, PCs, and average readers among the participants at grade 1

Among the current sample of 103 children, three groups of children with varied levels of word decoding abilities and/or reading comprehension were identified. The means and standard deviations of age, nonverbal IQ, word reading, and reading comprehension for the three groups are reported in Table 2. These variables were compared among the three groups using one-way analysis of variance (ANOVA) with Bonferroni post hoc tests. As shown in Table 2, the three groups did not differ significantly in age and nonverbal IQ, but their reading profiles were distinctive from each other, as was expected: when compared to average readers, poor decoders performed significantly worse in both word reading and reading comprehension (p < .001), while PCs performed significantly worse in reading comprehension only (p < .001). PCs showed similar word reading abilities to average readers, but their reading comprehension abilities were as poor as those of the poor decoders (n.s.).

Table 2.

Means, standard deviations, and univariate ANOVA results for poor decoders, poor comprehenders, and average readers on age, nonverbal intelligence, word reading, and reading comprehension.

		Poor decoders(n = 14)		Poor comprehenders (n = 16)		Average readers (n = 16)		F (2, 43)	Post hoc^{a b}
		Mean	SD	Mean	SD	Mean	SD
Age	K2	5.01	.37	5.09	.29	5.05	.32	2.75	n.s.
	K3	6.00	.37	6.06	.31	6.01	.31	.272	n.s.
	G1	7.13	.35	7.15	.30	7.11	.29	.060	n.s.
Nonverbal IQ	K2	10.00	2.75	10.31	2.85	11.09	3.82	.475	n.s.
Word reading^c	G1	6.46	1.20	13.17	2.82	12.38	2.40	34.51***	PD<PC=AR
Reading comprehension	G1	13.00	2.25	11.94	2.54	17.81	2.01	29.87***	PD=PC<AR

***

p < .001.

PD = Poor decoders; PC = Poor comprehenders; AR = Average readers.

The symbol = indicates no significant group difference; < indicates p < .05 or less.

Word reading scores refer to the HKT-P(II) scaled scores.

Comparison of cognitive-linguistic skills among the three groups

The mean scores and standard deviation of the three groups in the cognitive-linguistic measures are reported in Table 3. Repeated measures analyses of variance (ANOVA) were conducted to compare the groups on the measures that were administered at multiple testing points: word reading, phonological skills, orthographic skills, vocabulary skills, grammatical skills, and discourse skills of story comprehension. For each of these longitudinal measures, a two-way repeated measures ANOVA was conducted with time (K2, K3, G1) as the within-subject variable and group (poor decoders, PCs, average readers) as the fixed between-subject variable. One exception was phonological skills for which the time factor had two levels only (K2, K3).

Table 3.

Means, standard deviations, and group comparison results for poor decoders, poor comprehenders, and average readers on word reading, phonological skills, orthographic skills, oral language skills (vocabulary, grammatical, discourse skills), and working memory across times.^a.

		Poor decoders (n = 14)		Poor comprehenders (n = 16)		Average readers (n = 16)		Group comparison
		Mean	SD	Mean	SD	Mean	SD	F (2,43)	Post hoc^{b c}
Word reading and related cognitive skills
Word reading	K2	22.07	15.98	43.69	20.02	45.44	29.06	16.75***	PD<AR; PD<PC
	K3	45.57	16.56	94.13	30.86	77.38	31.59
	G1	91.00	10.82	139.06	21.79	136.38	20.55
Phonological skills	K2	5.57	2.73	8.25	3.53	8.38	2.80	6.33**	PD<AR; PD<PC
Phonological skills	K3	7.93	2.34	10.50	2.76	10.56	2.25
Orthographic skills	K2	16.86	3.90	19.00	3.81	20.63	4.40	6.86**	PD<AR; PD<PC
	K3	20.93	3.22	24.25	2.21	23.75	2.21
	G1	21.71	2.43	24.00	2.39	24.31	2.68
Oral language skills
Vocabulary skills	K2	5.61	3.05	8.88	4.88	10.00	5.24	4.42*	PD<AR
	K3	11.43	4.40	15.06	4.88	13.44	4.29
	G1	15.00	4.49	16.97	4.93	19.81	3.66
Grammatical skills	K2	7.71	2.43	8.88	3.36	8.25	2.35	2.68^	n.s.
	K3	9.07	4.09	10.13	3.83	10.94	3.34
	G1	12.43	1.77	14.19	3.41	15.50	3.43
Discourse skills – Story comprehension	K2	5.86	2.03	4.50	1.32	6.06	1.34	10.22***	PC<AR
	K3	6.79	1.85	6.56	1.86	8.31	1.62
	G1	7.86	1.61	7.06	1.88	8.81	1.64
Discourse skills – Comprehension monitoring	G1	6.14	4.52	4.63	3.20	11.31	5.67	9.31***	PC<AR; PD<AR
Working memory
Nonverbal working memory	G1	6.43	2.21	4.00	1.37	6.25	3.09	5.23**	PC<AR; PC<PD

p = .08, *p < .05, **p < .01, ***p < .001.

Repeated measure ANOVA was used to conduct group comparison for all variables, except for comprehension monitoring and nonverbal working memory, for which univariate ANOVA was used.

PD = Poor decoders; PC = Poor comprehenders; AR = Average readers.

Only significant results (p < .05) are shown.

As reported in Table 3, the results of repeated measures showed significant main effects of group for all longitudinal measures (F (2,43) = 4.42–16.75, ps < .02) except for grammatical skills. The main effects of time were also found to be significant for all the longitudinal variables (F (2, 43) = 27.42–381.56, ps < .001), while the interaction effects (time*group) were non-significant. Such results indicated that the children improved significantly in all these measures, and the three groups made similar progress over time. The development of these skills in the three groups of children is depicted in the Supplemental material.

Post hoc analyses of the above repeated measures ANOVAs were further conducted using the Bonferroni procedure for multiple comparisons. As reported in Table 3, poor decoders and PCs scored significantly lower than average readers in different cognitive-linguistic tasks over the years. Poor decoders showed significantly weaker performance than average readers in word reading (p < .001), phonological skills (p = .011), orthographic skills (p = .004), and vocabulary skills (p = .02), while PCs showed significantly weaker performance than average readers in discourse skills of story comprehension (p < .001).

Univariate ANOVAs with Bonferroni post hoc tests were further conducted to determine whether the above cognitive-linguistic weaknesses of PCs and poor decoders showed early in K2 and K3 respectively. Significant effects of group were found for all the longitudinal measures at both K2 and K3 (F (2,43) = 3.24–11.77, ps < .05) (except for vocabulary skills at K3 and grammatical skills at K2 and K3). The results of post hoc tests revealed that poor decoders scored significantly lower than average readers in word reading (ps < .03), phonological skills (ps < .05), orthographic skills (ps < .05), and vocabulary skills (ps < .04) at both K2 and K3 (except for vocabulary skills at K3). PCs were found to score significantly lower than average readers in discourse skills of story comprehension (ps < .03) at both K2 and K3. The results of univariate ANOVA post hoc tests for each longitudinal measure are shown in the Supplemental material.

Apart from the above longitudinal measures, this study also had two concurrent measures (i.e., working memory and discourse skills of comprehension monitoring) that were administered at G1 only. For these two measures, univariate ANOVAs with Bonferroni post hoc tests were conducted to examine group differences. The results revealed a significant group effect for the two measures (F (2,43) = 5.23–9.31, p < .001). The results of Bonferroni post hoc tests further showed that poor decoders and PCs scored significantly lower than average readers in discourse skills of comprehension monitoring (ps < .02), while PCs also scored significantly lower than average readers and poor decoders in working memory (ps < .03). Figure 2 summarizes the present findings by comparing the cognitive-linguistic weaknesses between poor decoders and PCs.

Figure 2.

A summary of the cognitive-linguistic weaknesses of poor decoders and poor comprehenders in Chinese.

Discussion

The aim of the study was to understand specific reading comprehension difficulties (i.e., PCs) among Chinese children in early grades. Using a regression analysis method (Li & Kirby, 2014; Tong et al., 2014; Tong et al., 2011), 16 out of 103 first graders were identified as PCs, whose performances in a passage reading comprehension task were as poor as those of the poor decoders despite their average-to-good word reading abilities. The key cognitive-linguistic weaknesses of the identified Chinese PCs were found to be oral discourse skills and working memory. In all other cognitive-linguistic tasks, Chinese PCs demonstrated performance comparable to that of average readers in phonological skills, orthographic skills, oral vocabulary, and grammatical skills. Using a three-year longitudinal study design, the present study demonstrated that the oral discourse weaknesses of Chinese PCs could be revealed as early as at K2. Such retrospective findings are strong evidence in support of the key role of oral discourse weaknesses as a cause of specific reading comprehension difficulties among young Chinese children.

The underlying cognitive-linguistic factors of Chinese PCs share certain similarities with those well-established factors in the English language. For instance, poor working memory was found to be a common underlying factor of PCs in both languages (Cain et al., 2004; Seigneuric, Ehrlich, Oakhill, & Yuill, 2000). It appears that irrespective of language type, children with poor working memory may lack the mental capacity for simultaneous storage and processing of information, which may be associated with reading comprehension difficulties. Furthermore, PCs in both languages suffer from oral language weakness, but the specific nature of their oral language problems appears to have some differences.

In English, PCs tend to suffer from general oral language deficits at the word, sentence, and discourse levels (Catts et al., 2006), while we found that in Chinese, only the discourse level was involved. The present finding is in line with existing evidence and views that discourse skills play a key role in Chinese reading comprehension (Chik et al., 2012; Yeung et al., 2013). Since Chinese is a topic-prominent language, Chinese readers particularly need to integrate information beyond the sentence level to fully understand a passage. In contrast to subject-prominent languages such as English, the subject is not marked by position in Chinese and the need to name a subject is not as important as that in English. In topic-prominent Chinese, it is quite common in sentences that the ‘subject position’ is filled by a verb phrase or an adjective that functions as a ‘topic’ (Fung, 2009). More importantly, the topic can be omitted in a sentence, assuming that the topic has been clearly established in preceding sentences. Here is an example:

她是一位勤力的員工。從早到晚都在辦公室工作，很是忙碌。 ‘She is a hard-working employee. (She) works in office all day long, (and she is) very busy.’

Since sentence structure in Chinese is relatively loose and is not as transparent as it is in English, reading Chinese text requires readers to integrate information across sentences by using discourse skills, e.g., making inferences, monitoring their comprehension, and using context at the discourse level. Taken together, the current results indicated that oral discourse is a key weakness of Chinese PCs.

The present findings also indicate that poor vocabulary skills are a weakness for poor decoders but not for PCs in Chinese. One possible reason could be attributed to the special characteristic of the Chinese language. Chinese is a semantic-based orthography that represents meanings. To acquire word reading skills, children need to have a good grasp of word meanings. In other words, there is a strong association between word reading and vocabulary skills in Chinese (Zhang et al., 2012). Children who have vocabulary problems tend more to suffer from word reading difficulties (Shu et al., 2006). This may partly explain why poor decoders (i.e., children who failed to read Chinese words) tend to have vocabulary problems, while average readers and PCs (i.e., those who have normal word reading abilities) do not have such problems. While this could be one explanation, the underlying reasons that Chinese PCs do not have vocabulary deficits requires further investigation. Taken together, we conclude that Chinese PCs are as capable as average readers in recognizing words and understanding word meanings. However, they still have difficulties in reading comprehension because their weakness in oral discourse impairs their ability to comprehend texts adequately.

The current study also has significant implications for the research field of Chinese reading difficulties. Over the past decades, the primary focus of studies on Chinese reading difficulties has been dyslexia. As a result, researchers have tended to assess and intervene for children with reading comprehension difficulties using cognitive-linguistic tasks that are closely related to word reading problems, namely phonological, orthographic, and oral vocabulary skills (Ho et al., 2004; Shu et al., 2006; Wu et al., 2009). The current study, however, reveals that in addition to dyslexic readers, PCs also exist among Chinese children. Unlike poor decoders, PCs were found to have no problem in any of the above word-level and grammatical skills, but they still performed as poorly as poor decoders in reading comprehension. One implication of this result is that, in the Chinese language, even for children with normal abilities in recognizing print words and processing oral vocabulary and grammar, inadequate oral discourse skills and working memory may possibly make these children struggle to understand written passages. It follows that special attention and tailor-made assessment and intervention tools for Chinese PCs are needed.

The current study also provides empirical evidence that early identification of PCs prior to school entry is possible. By conducting a three-year longitudinal study, we were able to identify PCs at G1 and retrospectively traced back their early signs of discourse weaknesses as early as at K2. It seems that the use of oral discourse measures may be of particular help in increasing the accuracy of the identification of preschool children at risk of having specific reading comprehension difficulties after school entry. Moreover, targeted training in oral discourse skills may potentially be an effective approach for early intervention of Chinese PCs. We suggest conducting intervention studies to provide concrete evidence on the effectiveness of using such an approach to intervene for early-identified PCs. Another proposed future study is to develop a standardized Chinese reading comprehension test for students in early grades. Such a test would be normed for a large population similar to those developed for English-speaking children. Since such a standardized reading comprehension test has yet to be made available for Chinese students in early grades, the present study relied on a self-developed reading comprehension test to identify PCs. Therefore, the extent to which the current results are comparable with future findings on Chinese PCs remains an open question. This is a limitation of the current study, and in future studies, we will confirm the cognitive profile of Chinese PCs using standardized reading comprehension measures when such tools are available.

Conclusion

The current study longitudinally followed a total of 103 Chinese children from K2 to G1, among which 14 were identified as poor decoders and 16 were identified as PCs at G1. The Chinese PCs were found to perform as poorly in reading comprehension as the poor decoders did. However, PCs did not show weaknesses in cognitive-linguistic skills including phonological, orthographic, and oral vocabulary skills, which are traditionally known as the core contributing factors of Chinese reading difficulties in dyslexic children. Instead, oral discourse skills and working memory were found to be the key cognitive-linguistic weaknesses of Chinese PCs. The longitudinal data further confirmed that the oral discourse weaknesses of PCs started to show as early as at K2. One important implication is that dyslexia and specific reading comprehension difficulties in the Chinese language should be conceptualized as two distinct types of reading problems, each of which has a different set of cognitive-linguistic weaknesses.

Supplemental Material

Supplemental_Material – Supplemental material for Poor oral discourse skills are the key cognitive-linguistic weakness of Chinese poor comprehenders: A three-year longitudinal study

Supplemental material, Supplemental_Material for Poor oral discourse skills are the key cognitive-linguistic weakness of Chinese poor comprehenders: A three-year longitudinal study by Cathy Yui-Chi Fong and Connie Suk Han Ho in First Language

Footnotes

Funding

The authors received no financial support for the research, authorship, and/or publication of this article.

Supplemental material

Supplemental material for this article is available online.

ORCID iD

Cathy Yui-Chi Fong

References

Aaron

P. G.

Kuchta

Grapenthin

C. T.

(1988). Is there a thing called dyslexia? Annals of Dyslexia, 38(1), 31–49.

Adlof

S. M.

Catts

H. W.

Lee

(2010). Kindergarten predictors of second versus eighth grade reading comprehension. Journal of Learning Disabilities, 43, 332–345.

Cain

Oakhill

Barnes

M. A.

Bryant

P. E.

(2001). Comprehension skill, inference making ability, and their relation to knowledge. Memory & Cognition, 29, 850–859.

Cain

Oakhill

Bryant

P. E.

(2004). Children’s reading comprehension ability: Concurrent prediction by working memory, verbal ability and component skills. Journal of Educational Psychology, 96, 31–42.

Cain

Oakhill

J. V.

Elbro

(2003). The ability to learn new word meanings from context by school-age children with and without language comprehension difficulties. Journal of Child Language, 30, 681–694.

Catts

H. W.

Adlof

S. M.

Weismer

S. E.

(2006). Language deficits in poor comprehenders: A case for the simple view of reading. Journal of Speech, Language, and Hearing Research, 49, 278–293.

Chik

P. P.-M.

C. S.-H.

Yeung

P.-S.

Wong

Y.-K.

Chan

D. W.-O.

Chung

K. K.-H.

L.-Y.

(2012). Contribution of discourse and morphosyntax skills to reading comprehension in Chinese dyslexic and typically developing children. Annals of Dyslexia, 62, 1–18.

Fung

R. S.-Y.

(2009). Characteristics of Chinese in relation to language disorders. In Law

S.-P.

Weekes

B. S.

Wong

A. M.-Y.

(Eds.), Language disorders in speakers of Chinese (pp. 1–18). Clevedon, UK: Multilingual Matters.

C. S.-H.

Bryant

(1997). Phonological skills are important in learning to read Chinese. Developmental Psychology, 33, 119–147.

10.

C. S.-H.

Chan

D. W.-O.

Chung

Tsang

S.-M.

Lee

S.-H.

Cheng

R. W.-Y.

(2007). The Hong Kong Test of Specific Learning Difficulties in Reading and Writing for Primary School Students-Second Edition [HKT-P(II)]. Hong Kong: Hong Kong Specific Learning Difficulties Research Team.

11.

C. S.-H.

Chan

D. W.-O.

Lee

S.-H.

Tsang

S.-M.

Luan

V. H.

(2004). Cognitive profiling and preliminary subtyping in Chinese development dyslexia. Cognition, 91, 43–75.

12.

C. S.-H.

T.-T.

W.-K.

(2003). A ‘radical’ approach to reading development in Chinese: The role of semantic radicals and phonetic radicals. Journal of Literacy Research, 35, 849–878.

13.

Justice

Mashburn

Petscher

(2013). Very early language skills of fifth-grade poor comprehenders. Journal of Research in Reading, 36, 172–185.

14.

Kim

Y.-S.

Phillips

(2014). Cognitive correlates of listening comprehension. Reading Research Quarterly, 49, 269–281.

15.

Kirby

J. R.

(2014). Unexpected poor comprehenders among adolescent ESL students. Scientific Studies of Reading, 18, 75–93.

16.

Lipka

Lesaux

N. K.

Siegel

L. S.

(2006). Retrospective analyses of the reading development of Grade 4 students with reading disabilities: Risk status and profiles over 5 years. Journal of Learning Disabilities, 39, 364–378.

17.

(2008). Chinese reading comprehension. Hong Kong: New Century Press.

18.

McBride-Chang

S.-H.

(2000). Developmental issues in Chinese children’s character acquisition. Journal of Educational Psychology, 92, 50–55.

19.

Nation

(2005). Children’s reading comprehension difficulties. In Snowling

M. J.

Hulme

(Eds.), The science of reading: A handbook (pp. 248–271). Oxford, UK: Blackwell.

20.

Nation

Clarke

Marshall

C. M.

Durand

(2004). Hidden language impairments in children: Parallels between poor reading comprehension and specific language impairments? Journal of Speech, Language, and Hearing Research, 47, 199–211.

21.

Nation

Cocksey

Taylor

J. S. H.

Bishop

D. V. M.

(2010). A longitudinal investigation of early reading and language skills in children with poor reading comprehension. Journal of Child Psychology and Psychiatry, 51, 1031–1039.

22.

Nation

Snowling

(1997). Individual differences in contextual facilitation: Evidence from dyslexia and poor reading comprehension. Child Development, 69, 996–1011.

23.

Nation

Snowling

(2000). Factors influencing syntactic awareness skills in normal readers and poor comprehenders. Applied Psycholinguistics, 21, 229–241.

24.

Oakhill

Hartt

Samols

(2005). Levels of comprehension monitoring and working memory in good and poor comprehenders. Reading and Writing, 18, 657–686.

25.

Palladino

Ferrari

(2013). Interference control in working memory: Comparing groups of children with atypical development. Child Neuropsychology, 19(1), 37–54.

26.

Pan Lloyds Publishers. (2006). Chinese reading comprehension. Hong Kong: Author.

27.

Perfetti

C. A.

Hogaboam

(1975). Relationship between single word decoding and reading comprehension. Journal of Educational Psychology, 67, 461–469.

28.

Perfetti

C. A.

Landi

Oakhill

(2005). The acquisition of reading comprehension skill. In Snowling

M. J.

Hulme

(Eds.), The science of reading: A handbook (pp. 227–247). Malden, MA: Blackwell.

29.

Raven

J. C.

Court

J. H.

Raven

. (1995). Raven’s colored progressive matrices. Oxford, UK: Oxford Psychologists Press.

30.

Seigneuric

Ehrlich

M.-F.

Oakhill

J. V.

Yuill

N. M.

(2000). Working memory resources and children’s reading comprehension. Reading and Writing, 13, 81–103.

31.

Shu

McBride-Chang

Liu

(2006). Understanding Chinese developmental dyslexia: Morphological awareness as a core cognitive construct. Journal of Educational Psychology, 98, 122–133.

32.

Snowling

M. J.

(2000). Dyslexia (2nd ed.). Oxford, UK: Blackwell.

33.

Siegel

L. S.

(1997). Learning to read Chinese: Semantic, syntactic, phonological and working memory skills in normally achieving and poor Chinese readers. Reading and Writing, 9, 1–21.

34.

Tong

Deacon

S. H.

Cain

(2014). Morphological and syntactic awareness in poor comprehenders: another piece of the puzzle. Journal of Learning Disabilities, 47, 22–33.

35.

Tong

Deacon

S. H.

Kirby

J. R.

Cain

Parrila

(2011). Morphological awareness: A key to understanding poor reading comprehension in English. Journal of Educational Psychology, 103, 523–534.

36.

Tong

McBride-Chang

Shu

Wong

A. M.-Y.

(2009). Morphological awareness, orthographic knowledge, and spelling errors: Keys to understanding early Chinese literacy acquisition. Scientific Studies of Reading, 13, 426–452.

37.

Packard

J. L.

Shu

(2009). Morphological deficit and dyslexia subtypes in Chinese. In Law

S.-P.

Weekes

B. S.

Wong

A. M.-Y.

(Eds.), Language disorders in speakers of Chinese (pp. 112–137). Clevedon, UK: Multilingual Matters.

38.

Yeung

P.-S.

C. S.-H.

Chan

D. W.-O.

Chung

K. K.-H.

Wong

Y.-K.

(2013). A model of reading comprehension in Chinese elementary school children. Learning and Individual Differences, 25, 55–66.

39.

Yeung

P.-S.

C. S.-H.

Chik

P. P.-M.

L.-Y.

Luan

Chan

D. W.-O.

Chung

K. K.-H.

(2011). Reading and spelling Chinese among beginning readers: What skills make a difference? Scientific Studies of Reading, 15, 285–313.

40.

Yuill

Oakhill

(1991). Children’s problems in text comprehension: An experimental investigation. New York, NY: Cambridge University Press.

41.

Zhang

McBride-Chang

Tong

Wong

A. M.-Y.

Shu

Fong

C. Y.-C.

(2012). Reading with meaning: The contributions of meaning-related variables at the word and subword levels to early Chinese reading comprehension. Reading and Writing, 25, 2183–2203.

Supplementary Material

Please find the following supplemental material available below.

For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.

For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.

0.00 MB

0.40 MB