Development of word order and morphosyntactic skills in reading comprehension among Chinese elementary school children

Learning and Individual Differences 47 (2016) 61–69

Contents lists available at ScienceDirect

Learning and Individual Differences journal homepage: www.elsevier.com/locate/lindif

Development of word order and morphosyntactic skills in reading comprehension among Chinese elementary school children Carrey Tik-Sze Siu a, Connie Suk-han Ho a,⁎, David Wai-ock Chan b, Kevin Kien-hoa Chung c a b c

Department of Psychology, The University of Hong Kong, Pokfulam Road, Hong Kong Faculty of Education, The Chinese University of Hong Kong, Hong Kong Department of Special Education and Counselling, The Hong Kong Institute of Education, Hong Kong

a r t i c l e

i n f o

Article history: Received 3 April 2015 Received in revised form 17 November 2015 Accepted 23 December 2015 Available online xxxx Keywords: Syntactic skill Reading comprehension Chinese Word order Morphosyntax

a b s t r a c t The present paper reported two studies which compared the roles of word order and morphosyntactic skills in reading comprehension among Chinese elementary school children. In Study 1, we found that over and above the effects of age, nonverbal intelligence and word reading, word order skill was a stable predictor of reading performance throughout grades 1 to 6, whereas morphosyntactic skill was associated with reading comprehension at grades 3 and 4. Study 2 was a three-year longitudinal study which showed that morphosyntactic but not word order skill at grade 2 longitudinally predicted sentence comprehension at grade 3 beyond the control variables and the auto-regressor; word order rather than morphosyntactic skill at grade 2 contributed significant variances to passage comprehension at grades 3 and 4. The findings suggested a differential dependence of reading on word order and morphosyntactic skills at different ages and in reading comprehension at different levels. © 2015 Elsevier Inc. All rights reserved.

1. Introduction Syntactic skills constitute a conscious understanding of the rules of grammar and an ability to manipulate the grammatical structure of sentences in a language (Gombert, 1992). Note that syntactic skills concern not only a reflection upon grammaticality of sentences, but also the use of linguistic strategies or operations to fix grammatical anomalies. In the reading literature, evidence has pointed to considerable syntactic influences on both word recognition (e.g. Bowey, 1986a,b; Cain, 2007; Muter & Snowling, 1998; Plaza & Cohen, 2003; Rego & Bryant, 1993; So & Siegel, 1997) and reading comprehension (e.g. Chik et al., 2012a, Demont & Gombert, 1996, Muter, Hulme, Snowling, & Stevenson, 2004, Siegel & Ryan, 1988, Tong, Tong, Shu, Chan, & McBride-Chang, 2014, Willows & Ryan, 1986, Yeung et al., 2011). The prior work has typically studied the global relationship between syntax and literacy development, albeit using an array of syntactic measures. Nevertheless, syntax in many languages comprises at least two important dimensions, namely the linear arrangement of words (i.e., word order) and morphological marking of constituents (i.e., morphosyntax), and their respective roles in reading development have yet to be contrasted. In the present studies, we examined children's developing syntactic skills in relation to their reading performance across ages. We were particularly interested in how word order and morphosyntactic skills might ⁎ Corresponding author. E-mail addresses: [email protected] (C.T.-S. Siu), [email protected] (C.S. Ho), [email protected] (D.W. Chan), [email protected] (K.K. Chung).

http://dx.doi.org/10.1016/j.lindif.2015.12.023 1041-6080/© 2015 Elsevier Inc. All rights reserved.

differentially contribute to reading comprehension at varied levels in Chinese children at different grades. 1.1. The importance of syntactic skills in reading To conceptualise the cognitive processes underlying reading, Bishop and Snowling (2004) have extended the triangle model and highlighted the role of syntactic skills in decoding words and comprehending connected texts. First, it has been argued that knowledge of how constituents are canonically ordered in a sentence allows children to capitalise on word order constraints to decipher unfamiliar words (Rego & Bryant, 1993; Tunmer, 1989; Tunmer & Chapman, 1998). In particular, word order knowledge enables inferences about the word class of novel words, which facilitates interpretation of unfamiliar words by limiting their probable meaning in the sentential context (Bishop & Snowling, 2004). This conjecture has been supported by the findings that children's syntactic skills significantly contributed to their single word recognition (e.g. Bowey, 1986a,b, Cain, 2007, Muter & Snowling, 1998, Rego & Bryant, 1993, Tunmer, 1989, Tunmer & Hoover, 1992, Willows & Ryan, 1986). For instance, Cain (2007) reported that the word order skill of English-speaking children was uniquely associated with their word reading even when the effects of vocabulary, grammatical knowledge, and memory were controlled. More importantly, Bishop and Snowling (2004) claimed that reading comprehension particularly recruited knowledge and skill in syntax. Previous research has indicated that syntactic skills were involved in reading comprehension (Demont & Gombert, 1996; Gaux & Gombert,

62

C.T.-S. Siu et al. / Learning and Individual Differences 47 (2016) 61–69

1999; Goff, Pratt, & Ong, 2005; Plaza & Cohen, 2003, 2004; Tunmer, 1989; Tunmer, Nesdale, & Wright, 1987; Willows & Ryan, 1986). For instance, in a longitudinal study, Muter et al. (2004) found that morphosyntactic (grammatical) skill at age 4 was a significant predictor of reading comprehension one year later. To explain why reading comprehension demands syntactic skills, Gough and Tunmer (1986) argued that a child who has syntactic deficits is likely to have a poor understanding of connected texts because they cannot work out the complex clausal structures. It has also been suggested that a weak mastery of grammatical morphemes such as tense or aspect markers is likely to result in an imprecise understanding of the time reference of events in texts. Furthermore, Tunmer and Bowey (1984) held that word order knowledge allows children to use the current input to anticipate syntactic categories, thus expediting integration and monitoring of ideas in comprehension. For example, upon hearing the subject “he”, a child who understands the canonical word order of subject-verb-object in English would expect a verb that can assign a thematic role to it. Such ongoing prediction of upcoming words is especially important in reading long texts because additional semantic information can be solicited from the context to smooth the text interpretation. Overall, syntactic skills serve to build up a context in which unfamiliar words and ideas could be interpreted and integrated so as to ease the comprehension processes (Adams, 1990; Bishop & Snowling, 2004).

with morphosyntactic performances. The results indicated that these two popular syntactic measures differed in language and memory requirements, thus suggesting that they were not equivalent in testing syntactic skills. In view of this, we think the findings produced from studies using disparate syntactic tasks are not readily comparable to establish the role of syntactic skills in proficient reading. Following up on Cain's (2007) findings, we called for a refined conceptualisation of syntactic skills as constituting two separate facets, namely word order and morphosyntactic skills. Apparently, these two abilities have to do with two distinct types of grammaticality. In terms of assessed knowledge, the morphosyntactic task primarily tests one's evaluation of the morphemic agreement between constituents, which is essentially different from the understanding of legitimate word order tested in the word order task. In addition, the word order task requires a rearrangement of words into a correct sequence, a process that necessitates reflection on the entire sentential context to position all the constituents. Contrastively, the manipulation in the morphosyntactic task is more localised to isolated words (Pratt, Tunmer, & Bowey, 1984). Therefore, we support Cain's (2007) notion that the syntactic knowledge and processing demands in these two tasks are by nature different. In this research, we attempted to consider anew the relationship between syntax and literacy by delineating the roles of different facets of syntactic skills in reading comprehension among Chinese school-aged children.

1.2. Different facets of syntactic skills Although researchers have suggested a general notion that understanding of syntax is a critical component of literacy acquisition, they focused more on word order than morphosyntax in explaining the syntactic influence on reading. However, we argue that there is a need to reconsider the findings and carefully differentiate the parts of different syntactic skills in developing reading ability. Reviewing the above studies informs us that past research has been diverse in measuring syntactic skills. Earlier work often used an oral cloze design to tap syntactic skills (e.g. Chen, Lau, & Yung, 1993; Siegel & Ryan, 1988; So & Siegel, 1997; Willows & Ryan, 1986), in which the participant was asked to supply a correct word to complete a sentence grammatically (for example, “It_____very windy outside yesterday.”). More recent studies, however, have tended to use either or a combination of morphosyntactic (grammatical) and word order tasks to assess syntactic skills. The morphosyntactic judgment and/or correction task requires the participant to first decide whether the sentences are grammatically correct and then to locate and fix the errors to restore the grammaticality of the sentences; for example, eat should be replaced by ate in “Yesterday Mary eat an apple.” (e.g. Cain, 2007; Davidson, Raschke, & Pervez, 2010; Gottardo, Stanovich, & Siegel, 1996; Jongejan, Verhoeven, & Siegel, 2007; Reder, Marec-Breton, Gombert, & Demont, 2013; Tsang & Stokes, 2001; Yeung, Ho, Chan, Chung, & Wong, 2013). In other words, this task calls for both grammatical knowledge and a conscious ability to supply appropriate words to repair the sentences. The word order correction task is another popular syntactic measure (e.g. Cain, 2007, Chen & Wong, 1991, Chik et al., 2012a, Muter et al., 2004, Nation & Snowling, 2000, Tsang & Stokes, 2001, Yeung et al., 2011, 2013), where the scrambled segments of a sentence have to be rearranged into a correct order (for example, “driving/a/Sally/car/was” should be corrected as “Sally was driving a car.”). Note that the oral cloze task is a more general measure of syntactic skills because participants must call on both word order and morphosyntactic knowledge to come up with a word to complete the sentence. However, word order and morphosyntactic operations are assessed respectively in the word order and morphosyntactic correction tasks, and processing demands in these two measures are thought to be rather different. In a study that intended to compare the commonly used syntactic tasks, Cain (2007) found that after controlling for receptive vocabulary, memory explained unique significant variances in word order skill, whereas grammatical knowledge was uniquely correlated

1.3. Syntactic skills and reading in Chinese Recent years have witnessed a surging interest in the involvement of syntactic skills in reading in Chinese. Chen and colleagues reported initial findings that syntactic skills, in terms of rearranging words into a sentence (Chen & Wong, 1991), detecting grammatical errors and filling in missing words (Chen et al., 1993), were important correlates of reading proficiency among Chinese school-aged children. So and Siegel (1997) also showed that children's ability to insert missing words in an oral cloze task improved steadily from grades 1 to 4, and such syntactic skill was a reliable predictor of their word reading ability. A few more recent studies have also substantiated the significance of syntactic skills in reading comprehension in Chinese. At the sentence level, Chik et al. (2012a) demonstrated that syntactic skills (including word order, connectives and morphosyntactic knowledge) at grade 1 were longitudinally related to sentence comprehension at grade 2, after taking the children's age, nonverbal intelligence, phonological, orthographic, and morphological processing into consideration. Yeung et al. (2011) also found that syntactic skill assessed with an oral cloze task uniquely predicted sentence and passage comprehension among Chinese first graders above and beyond the effects of word reading ability and other reading-related skills. Some others (Chik et al., 2012b; Tong et al., 2014; Yeung et al., 2013) obtained complementary findings among grades 4 and 5 children that syntactic skills accounted for unique variances of sentence- and passage-level reading comprehension after controlling for the influence of word-level reading-related skills. 1.4. Development of individual syntactic skills in relation to reading in Chinese According to C. N. Li and Thompson (1981), Chinese is unique in its simple structure of words, in which a word is a single morpheme that cannot be decomposed into smaller units. Hence, Chinese is characterised as an isolating language that has an impoverished inflectional system (Matthews & Yip, 1994). For instance, nouns in Chinese are not marked for their grammatical function in a sentence; case is instead expressed by the use of prepositions and word order. Plurality is also indicated by the use of a separate word rather than transforming the noun itself. Furthermore, there is no subject-verb agreement in Chinese. Verbs in Chinese are also not inflected for tense, number, and

C.T.-S. Siu et al. / Learning and Individual Differences 47 (2016) 61–69

gender. These functions are generally taken over by other devices such as adverbs, function words or particles (Lin, 2006). Compared to morphemic transformation, word order is a more significant syntactic device to express meaning in Chinese. It has been suggested that Chinese speakers rely on word order to interpret Chinese sentences (Chang, 1992; Chao, 1968). Young children as early as age three are sensitive to the canonical word order and draw on this incipient word order knowledge to understand Chinese sentences (Lee, 1996). Tsang and Stokes (2001) examined the development of syntactic awareness in Cantonese-Chinese children; they found that the children performed better on both judgment and revision of word order violations than morphosyntactic correction. The findings also showed that the development of word order skill was largely complete by grade 1 and preceded that of morphosyntactic skill, which was still fledgling among the first graders. Some studies have started to contrast children's performance in word order and morphosyntax but did not ask whether these syntactic skills changed in their predictive relationship with reading ability across grades (Hakes, 1980; Tsang & Stokes, 2001). We find such an investigation important because it informs us of the extent to which individual syntactic skills are at work in acquiring literacy across ages. Considering that word order is fundamental in conveying meaning in Chinese, we expected that word order skill is more closely associated with reading comprehension than morphosyntactic skill is. 1.5. The present study To summarise, the objective of the present research was to examine the development of two dimensions of syntactic skills (i.e., word order and morphosyntactic skills) in relation to reading comprehension in Chinese school-aged children. The present research consisted of two studies. Study 1 was a cross-sectional study which compared the relative roles of word order and morphosyntactic skills in reading comprehension among Chinese children from grades 1 to 6. Given the significance of word order to communicate meaning in Chinese, we hypothesised that children's word order skill was more predictive of their reading performance than their morphosyntactic skill was. The parts of individual syntactic skills in reading comprehension at different levels were further explored in Study 2. Study 2 adopted a three-year longitudinal design that contrasted the relative significance of word order and morphosyntactic skills to reading comprehension at sentence- versus passage-levels. We administrated two syntactic tasks and separate measures of sentence and passage comprehension to a group of Chinese second graders across a span of three years. We predicted that the two syntactic skills might differentially contribute to sentence and passage reading comprehension. 2. Study 1 2.1. Method 2.1.1. Participants Participants were 387 Chinese school-aged children (181 boys and 206 girls) recruited from 18 typical local primary schools in Hong Kong, China. The participants were at six grade levels in primary school and were divided into three cohorts: (a) junior graders (grades 1 and 2 children), (b) intermediate graders (grades 3 and 4 children), and (c) senior graders (grades 5 and 6 children). There were 86 junior graders (37 boys and 49 girls; mean age = 7.05 years; SD = .63 years), 133 intermediate graders (64 boys and 69 girls; mean age = 9.47 years; SD = .67 years) and 168 senior graders (80 boys and 88 girls; mean age = 11.21 years; SD = .71 years) in our sample. All participants had normal intelligence with a mean IQ score of 102.04 (SD = 12.35). There were no reports of physical, emotional, behavioural or literacy-related abnormalities from the parents and teachers.

63

2.1.2. Materials We administered five measures to the participants in Study 1, including a standardised test of nonverbal intelligence, two Chinese literacy measures (word reading and passage reading comprehension), and two syntactic tasks in Chinese (word order and morphosyntactic corrections). Because the participants included children at grades 1 to 6, the difficulty level of the test items was adjusted appropriate to the respective grades. The test items were also validated in pilot studies to ensure adequate discriminative power of the measures. 2.1.2.1. Nonverbal intelligence. We used Raven's Standard Progressive Matrices (Raven, Court, & Raven, 1996) to assess children's nonverbal intelligence. This standardised measure consisted of 60 black-andwhite items organised in five sets of 12 items each. In each item, a target geometric design with one missing part was presented. The test involved the children choosing one out of six or eight option patterns that best completed the target design. Children over 8.5 years old completed the full test, whereas 8.5-year-olds or younger did the first three sets of the test. Nonverbal intelligence was derived with reference to the local norms established by the Education Department of The Hong Kong Government (Hong Kong Education Department, 1986). 2.1.2.2. Word reading. We tested children's word reading ability with the Chinese word reading subtest of the Hong Kong Test of Specific Learning Difficulties for Reading and Writing (HKT-SpLD; Ho, Chan, Tsang, & Lee, 2000). We used a two-character word reading test, instead of a singlecharacter reading test, because a majority of Chinese words are disyllabic words (Modern Chinese frequency dictionary; Wang & Chang, 1985) to effectively differentiate meaning between words. In this subtest, a total of 150 Chinese two-character words, arranged in order of increasing graded difficulty, were presented. The child was asked to read the words aloud one by one. Testing was discontinued when the child failed to read 15 consecutive words. One point was given when both characters in a two-character word were pronounced correctly. The maximum score of this task was 150 and the reported internal reliability of this HKT-SpLD subtest was .96. 2.1.2.3. Passage reading comprehension. We assessed children's understanding of Chinese written passages with a reading comprehension task. Chinese texts were written based on The Hong Kong Corpus of Primary School Chinese (Leung & Lee, 2002) for words that are familiar to local school-aged children. All written texts were narrative or expository in nature, which are familiar genres to the target children. The length of the passages ranged from 67 to 324 Chinese characters. The test engaged the child in reading three to five Chinese passages in silence, followed by answering comprehension questions relevant to each passage. The intermediate and senior graders were asked to read one and two more passages, respectively, to fully capture their individual differences in reading comprehension ability for subsequent analyses. The older children might reach ceiling performance for the easier items from the first three passages; thus we asked them to read more passages and attempted more items to avoid the potential ceiling effect. The questions were either in multiple-choice format with four option answers, or in open-ended format expecting short answers from the child. Across different grades, the questions also varied in the level of understanding of the texts, including items that (a) asked explicit information or ideas which could be directly retrieved from the passages, (b) required straight-forward inferencing of the passage contents, (c) demanded integration of information and ideas from different parts of the passages, and (d) involved a global evaluation and judgment of the passage contents. Two practice items were given prior to the test items. One point was awarded for each correct answer to the multiple-choice questions and two points to that in the open-ended questions. The overall reliability (Cronbach's alpha) of this task was good at .86.

64

C.T.-S. Siu et al. / Learning and Individual Differences 47 (2016) 61–69

2.1.2.4. Word order skill. A word order correction task was developed to test children's knowledge of basic word order patterns in Chinese, including canonical sentence types such as subject-verb-object or subject-verb-verb, and special sentence types using /baa2/ (把) or / bei2/(被). After reviewing The Hong Kong Corpus of Primary School Chinese (Leung & Lee, 2002) and the Chinese textbooks published for local school-aged children, Chinese sentences that were of appropriate difficulty to the targeted children were written. These sentences were then segmented into three to seven fragments and were reshuffled, so that the basic word order was disrupted. The child was required to rearrange the jumbled segments to form a syntactically correct sentence. There were 13 to 15 test items; the senior graders attempted more items to ensure adequate individual variations in their responses for subsequent analyses. One practice item with correct feedback preceded the test items. The reliability over all items was good at .80.

Table 2 Partial and zero-order correlations among all measures by groups in Study 1.

2.1.2.5. Morphosyntactic skill. A morphosyntactic correction task was constructed based on previous studies that tested children's ability to detect and repair morphosyntactic violation in Chinese sentences (Chik et al., 2012b; Yeung et al., 2013). Each item involved a sentence with an error in the part of speech (e.g., 我們要‘珍貴’ [an adjective] 食 物。‘珍惜’[A verb] should be used instead; We have to treasure our food.). The child was required to identify the error and provide a correct word to restore the grammaticality of the sentence. There were 12 to 15 test items at different grade levels; the intermediate and senior graders were asked to do two and three more items, respectively, so that we could capture their individual variations in morphosyntactic skill more adequately for subsequent analyses. Each item constituted three points. One point was awarded for locating the morphosyntactic error in the sentence. Another point was given if the child fixed the error but changed the meaning of the sentence. Full mark was awarded when the child corrected the error and retained the intended meaning of the sentence. Two practice items with corrective feedback preceded the test items. The internal reliability of this task was high at .93.

Note. Partial correlations with age and nonverbal intelligence controlled are shown below the diagonal while zero-order correlations are shown above the diagonal. * p b .05. ** p b .01. *** p b .001.

Variable Junior graders (N = 86) 1. Word order correction 2. Morphosyntactic correction 3. Word reading 4. Passage comprehension Intermediate graders (N = 133) 1. Word order correction 2. Morphosyntactic correction 3. Word reading 4. Passage comprehension Senior graders (N = 168) 1. Word order correction 2. Morphosyntactic correction 3. Word reading 4. Passage comprehension

1

2

3

4

– .50*** .52*** .55***

.56*** .41** .48***

.43*** .31** .46***

.60*** .55*** .37** –

– .62*** .57*** .62***

.64*** – .49*** .57***

.56*** .49*** – .61***

.65*** .59*** .60*** –

– .66*** .41*** .61***

.65*** – .41*** .50***

.40*** .41*** – .34***

.60*** .49*** .34*** –

cohorts separately. The internal reliability of all the measures was good with Cronbach's alpha coefficients ranging from .80 to .93. In addition, zero-order correlations and partial correlations of the children's performance on all syntactic and literacy measures, after controlling for the effects of age and nonverbal intelligence, are shown in Table 2. The correlational results in different cohorts were reported separately in order to examine any developmental changes in the interrelationships between syntactic and literacy variables. Results showed that across different age cohorts there were moderate and significant correlations between the two syntactic skills, rs = .50 to .66, ps b .001. The two literacy measures also moderately correlated with one another, rs = .34 to .61, ps b .001. More importantly, the results revealed a modest to strong association between syntactic skills and reading performance. For example, word order skill was significantly associated with word reading and reading comprehension, rs = .41 to .62, ps b .001. Morphosyntactic skill was also significantly correlated with word reading and reading comprehension, with correlation coefficients ranging from .41 to .57, ps b .01. In addition, we observed that such interrelationships among different measures were shared by all the three age groups.

2.1.3. Procedures Parental consent for the children's participation was sought prior to testing. A team of trained experimenters administered all the assessments in quiet rooms in the respective schools. Each child received a single testing session which lasted approximately 75 min on a school day. Brief breaks were given whenever necessary. The presentation order of the measures was randomised.

2.2.2. Hierarchical regression analyses Separate hierarchical regression analyses were conducted on the three cohorts to compare the relative contributions of different dimensions of syntactic skills to reading comprehension across ages. As depicted in Table 3, age and nonverbal intelligence were entered in the first and second block, respectively, in all regressions as control variables. Word reading was entered next because it was shown to be a strong correlate of Chinese reading comprehension (e.g., T. Li,

2.2. Results 2.2.1. Descriptive statistics and correlations among variables Table 1 presents the means and standard deviations (in percentage scores) of the children's performance on all measures for the three

Table 1 Reliability coefficients and mean performances on all measures by groups in Study 1. Variable

Junior graders (J) (N = 86) α

Age (in months) Nonverbal intelligence Syntactic measures Word order correction Morphosyntactic correction Literacy measures Word reading Passage comprehension

Intermediate graders (I) (N = 133)

Senior graders (S) (N = 168)

M

SD

M

SD

M

SD

84.54 104.57

7.58 11.37

113.69 102.63

8.08 11.08

134.47 109.34

8.54 14.02

.80 .93

.53 .38

.20 .17

.75 .50

.23 .23

.97 .54

.32 .27

s# .86

.49 .38

.26 .14

.72 .49

.18 .22

.86 .67

.12 .35

– s

Note. Performance on all measures except nonverbal intelligence is reported in percentage scores. # HKT-SpLD word reading is a standardised measure with a reported split-half reliability of .96. * p b .05. ** p b .01. *** p b .001.

C.T.-S. Siu et al. / Learning and Individual Differences 47 (2016) 61–69

65

Table 3 Hierarchical regression analyses predicting passage comprehension by groups in Study 1. Junior graders (N = 86)

Intermediate graders (N = 133)

Senior graders (N = 168)

Step and variable

Final β

R2

R2 change

F change

Final β

R2

R2 change

F change

Final β

R2

R2 change

F change

Predicting passage comprehension Step 1: Age Step 2: Nonverbal intelligence Step 3: Word reading Step 4: Word order correction Step 5: Morphosyntactic correction Step 4: Morphosyntactic correction Step 5: Word order correction

.20 −.02 .19 .33 .23 .23 .33

.13 .13 .32 .43 .46 .40 .46

.13 .00 .19 .11 .03 .09 .06

9.54** .01 17.56*** 12.49** 3.70 8.98** 6.90**

.08 .07 .33 .30 .22 .22 .30

.00 .06 .42 .52 .55 .50 .55

.00 .06 .35 .10 .03 .09 .04

.80 14.70*** 139.37*** 49.37*** 13.43*** 39.91*** 21.93***

.06 −.07 .09 .48 .15 .18 .44

.00 .00 .12 .38 .39 .27 .39

.00 .00 .11 .26 .01 .15 .12

.00 .02 21.45*** 69.87*** 3.27 34.50*** 32.93***

Note. * p b .05. ** p b .01. *** p b .001.

McBride-Chang, Wong, & Shu, 2012; Yeung et al., 2013). Two regressions were run in each group of children. In the first regression, word order and morphosyntactic skill were entered in the fourth and fifth block, respectively. Their order of entry was reversed in the second regression. These two aspects of syntactic skills were entered in separate blocks so as to evaluate their unique contribution to predicting reading comprehension when the effect of another was statistically controlled. Results of these regressions organised by different cohorts were summarised in Table 3. Overall, the results indicated that age, nonverbal intelligence, word reading, word order and morphosyntactic skills together significantly predicted performance on passage comprehension across the three groups of children, explaining about 39% to 55% of its total variances. Over and above the effects of age, nonverbal intelligence and word reading, the two syntactic skills jointly explained an additional 14%, 13%, and 27% of variances in passage comprehension among the junior, intermediate, and senior graders, respectively. Nevertheless, as illustrated in Table 3, we observed a change in their relative contributions to reading comprehension across grades. Among the junior graders, word order skill was a significant predictor of passage comprehension surviving controls for age, nonverbal intelligence, word reading and morphosyntactic skill. However, a different predictive pattern was shown among the intermediate graders. In intermediate grades, morphosyntactic skill emerged as another significant predictor of passage comprehension when all other predictors in the model were controlled for. Together with word order skill which remained to have significant contribution to passage comprehension, syntactic skills explained an additional 13% of the variances in passage comprehension. Finally, among the senior graders, morphosyntactic skill did not explain additional significant variances when word order skill was first entered into the regression equation, whereas word order skill continued to serve as a significant predictor of passage comprehension after taking the effect of morphosyntactic skill into consideration. 2.3. Discussion Study 1 represents an initial endeavour to contrast the involvements of different dimensions of syntactic skills in reading comprehension among grades 1 to 6 Chinese children. Such an attempt was motivated by the earlier finding that word order and morphosyntactic corrections required different syntactic knowledge and processing demands (Cain, 2007). Our results suggested that individual syntactic skills were differentially associated with reading performance at different grades. Throughout grades 1 to 6, word order skill consistently explained significant variances of reading comprehension beyond and over the effects of age, nonverbal intelligence, word reading and morphosyntactic skill. Nevertheless, morphosyntactic skill did not make independent contribution to reading comprehension at grades 1 and 2, but became a significant correlate of reading comprehension among grades 3 and 4 children. These disparate predictive relationships are consistent with Tsang and Stokes (2001) that Chinese young children's word order skill

was developed in place earlier than their morphosyntactic skill was. It has been assumed that word order is a syntactic device of paramount importance in Chinese; children as young as three years old are aware of the canonical word order in Chinese and used it to interpret Chinese sentences (Chang, 1992; Lee, 1996). The early development of word order skill and its central role in Chinese language thus explain its significance to reading comprehension in Chinese throughout the elementary school years. In contrast, Chinese is an isolating language in which the use of inflections is relatively limited (C. N. Li & Thompson, 1981; Matthews & Yip, 1994). Tsang and Stokes (2001) posited that the flawed morphosyntactic structures in Chinese did not substantially transform the overall meaning of sentences. This is quite a contrast to the case in English in which inflections carry a heavy functional load in expressing meaning. Hence, it is until a later age when Chinese texts take a more sophisticated use of morphosyntax to qualify words that the morphosyntactic skill begins to play a more prominent role in understanding Chinese texts. Our results seemingly run counter to the findings of Chik et al. (2012a) that word order skill was less important to reading comprehension than morphosyntactic knowledge and connective usage were among Chinese junior graders. Nevertheless, note that the comprehension task used in Chik et al. (2012a) examined the understanding of Chinese sentences. Because rearranging sentence segments in the word order task calls for building coherence between constituents, we argue that it relates more closely with passage comprehension, which likewise requires considering the overall coherence of connected texts in attempting the questions. Manipulations in the morphosyntactic task, however, are more localised to specific words (Pratt et al., 1984). Even if the entire sentential context is not scrutinised, it is possible to come up with the correct answer by referring to one or a few isolated constituents, a situation shared by the sentence comprehension task. The proposed close relationship between word order skill and passage comprehension might also explain our results in senior graders in Study 1. We think that the contribution of morphosyntactic skill to understanding passages might be overtaken by word order skill because increasingly long texts were introduced in senior grades. In Study 2, we tried to reconcile the apparently inconsistent findings by relating different aspects of syntactic skills to reading comprehension at two different levels – sentence and passage reading in Chinese children. In addition, the cross-sectional design of Study 1 posed a challenge to drawing inferences about developmental changes of the parts of different syntactic skills in literacy development. We were not in a position to rule out the possibility of some unspecified cohort differences between the children at different grades. In Study 2, we sought to circumvent this limitation by adopting a longitudinal design which followed the children from grade 2 to grade 4. This allowed us to make a stronger claim to cast syntactic skills as an essential component of proficient reading. In sum, Study 2 aimed to examine how word order and morphosyntactic skills might be differentially associated with sentence- and passage-level reading comprehension among Chinese elementary school children.

66

C.T.-S. Siu et al. / Learning and Individual Differences 47 (2016) 61–69

3. Study 2

required the child to select the option that best completed the sentence among four option answers. The target answer and the distractors in each item shared the same word class but differed in meaning and usage – the child could not use the syntactic constraints of the sentence, but had to grasp the meaning of the option answers and sentence to choose the correct answer. There were two practice items before the test items. One point was awarded for each correct answer and the maximum score was 16. The reliability (Cronbach's alpha) of this task was satisfactory at .80 and .81 in grades 2 and 3, respectively.

3.1. Method 3.1.1. Participants In this three-year longitudinal study, 575 Chinese children (280 boys and 295 girls) were initially recruited from grade 2 classes in four local primary schools in Hong Kong. Our original sample of 575 second graders (mean age = 8.03 years, SD = .38) were followed from the end of grade 2 and were tested again at the end of grades 3 and 4 (i.e. three time points). Because some children dropped out of the study, the final sample included 528 fourth graders (271 boys and 251 girls) with a mean age of 9.98 years, SD = .42 years. Therefore, the data analyses were based on the 528 children who remained in the study till grade 4. All the children had normal intelligence with a mean IQ score of 110.32 (SD = 14.32). No physical, emotional, behavioural or literacy related abnormalities were reported for any of the children. Informed written consent for children's participation was obtained from parents or guardians before testing started.

3.1.2.2. Passage reading comprehension. We developed a passage comprehension task to assess children's understanding of short passages in grades 2 to 4 (Times 1 to 3). The design and procedure of this task were similar to those in Study 1. The children were asked to read three to five Chinese narrative and expository passages in silence, and then answered comprehension questions in multiple-choice or openended format. The levels of understanding required in the questions were manipulated in the same way as that in Study 1. One point and at maximum two points were given for each correct answer to the multiple-choice and open-ended questions, respectively. The reliability coefficients (Cronbach's alpha) of this task were .30, .79, and .83 in grades 2, 3, and 4, respectively.

3.1.2. Materials and procedures Study 2 included a standardised test of nonverbal intelligence, three Chinese literacy measures (word reading, sentence comprehension, and passage comprehension), and two Chinese syntactic tasks (word order correction and morphosyntactic correction). The children were assessed on syntactic skills and word reading at the end of grade 2 (Time 1), and nonverbal intelligence at the end of grade 4 (Time 3). They received testing on reading comprehension throughout the end of grades 2 to 4 (Time 1 to Time 3). Because the literacy measures were repeatedly given to the children across times, we considered the grade-level difficulties when constructing measures for children at different ages to ensure adequate discriminative power of the tasks. The assessments were conducted in quiet rooms in the respective schools by a team of trained experimenters. At each time point, the children finished all measures in a single testing session on a school day. The presentation order of the tasks was randomised across classes. Short breaks were given whenever it was necessary. The nonverbal intelligence test and the Chinese word reading test were the same as those in Study 1. Other measures in this study were described below.

3.1.2.3. Word order skill. We assessed children's knowledge of basic word order structures in Chinese at grade 2 (Time 1). This word order correction task comprised 20 items with canonical sentence types (e.g., SVO, SVOO and SVOVO) and special sentence types using /baa2/ (把) or / bei2/(被). In each item, the sentence was broken down into three to six fragments that were shuffled. The children who could successfully position the jumbled fragments into a correct order were scored one point for each item. One practice item was given before the test items. The reliability (Cronbach's alpha) over all items was high at .86. 3.1.2.4. Morphosyntactic skill. We tested children's ability to locate and fix morphosyntactic errors at grade 2 (Time 1) with a morphosyntactic correction task. The task design was similar to that in Study 1. There were 13 items, each presented a Chinese sentence with a morphosyntactic error. The child was asked to identify the error in each item and supply a syntactically appropriate word that best completed the sentence. The full mark for each item was three points and the scoring procedure followed that in Study 1. Two practice items with corrective feedback preceded the 13 test items. The internal reliability of this task was good at .81.

3.1.2.1. Sentence reading comprehension. We administered a sentence comprehension task in grades 2 and 3 (Times 1 and 2) to assess children's understanding of Chinese sentences. The Chinese sentences were constructed based on Chinese texts from the local curriculum according to The Hong Kong Corpus of Primary School Chinese (Leung & Lee, 2002). The sentence items were validated in pilot studies, which showed that they were neither too easy nor too difficult to children at the target age range. This task comprised 16 written cloze sentences with a noun, a verb, an adjective, or an adverb missing. Each item

3.2. Results 3.2.1. Descriptive statistics and correlations among all variables The children's mean performance on all measures administered at Times 1 to 3 is shown in Table 4. No floor or ceiling effects were

Table 4 Descriptive statistics of age and mean performances on all measures from Time 1 to Time 3 in Study 2. Variable

Age (in months) Nonverbal intelligence Syntactic measures Word order correction Morphosyntactic correction Literacy measures Word reading Sentence comprehension Passage comprehension

Time 1 (Grade 2)

Time 2 (Grade 3)

Time 3 (Grade 4)

α

M

SD

α

M

SD

α

M

SD

– -–

96.38 –

4.54 –

– –

110.58 –

4.79 –

– s

119.76 110.32

5.01 14.32

.86 .81

.79 .80

.21 .20

– –

– –

– –

– –

– –

– –

s# .80 .30

.64 .72 .52

.16 .21 .27

– .81 .79

– .76 .58

– .20 .19

– – .83

– – .53

– – .21

Note. Performance on all measures except nonverbal intelligence is reported in percentage scores. # HKT-SpLD word reading is a standardised measure with a reported split-half reliability of .96. * p b .05. ** p b .01. *** p b .001.

C.T.-S. Siu et al. / Learning and Individual Differences 47 (2016) 61–69

67

Table 5 Partial and zero-order correlations among all measures across times in Study 2. Variable

1

2

3

4

5

6

7

8

1. Word order correction (T1) 2. Morphosyntactic correction (T1) 3. Word reading (T1) 4. Sentence comprehension (T1) 5. Sentence comprehension (T2) 6. Passage comprehension (T1) 7. Passage comprehension (T2) 8. Passage comprehension (T3)

– .45*** .55*** .59*** .53*** .20** .38*** .46***

.52*** – .35*** .45*** .42*** .20** .26*** .31***

.60*** .42*** – .74*** .73*** .19** .53*** .61***

.67*** .52*** .76*** – .74*** .22*** .56*** .59***

.60*** .49*** .75*** .78*** – .22*** .52*** .53***

.30*** .28*** .26*** .32*** .31*** – .20** .24***

.47*** .34*** .58*** .62*** .58*** .29*** – .54***

.56*** .40*** .65*** .66*** .60*** .33*** .60*** –

Note. Partial correlations with age and nonverbal intelligence controlled are shown below the diagonal while zero-order correlations are shown above the diagonal. * p b .05. ** p b .01. *** p b .001.

observed across the time points. The internal reliability of most of the measures was satisfactory with Cronbach's alpha coefficients ranging from .79 to .86 (except for passage comprehension at grade 2). Table 5 shows the zero-order correlations (above diagonal) and partial correlations (below diagonal) among the children's performance on syntactic and literacy measures across times, after partialling out the effects of age and nonverbal intelligence. Correlational analyses results revealed a moderate correlation between the two syntactic skills administered at Time 1, r = .45, p b .001. As shown in Table 5, all literacy measures, including word reading, sentence-level and passage-level reading comprehension, were significantly associated with one another across time points (rs = .19 to .74, ps b .01). Of primary interest was that word order skill at Time 1 was significantly correlated with word reading, sentence comprehension and passage comprehension at Times 1 to 3, rs = .20 to .59, ps b .01, if applicable. Similarly, morphosyntactic skill was also related to literacy performance at different levels, with correlation coefficients ranging from .20 to .45, ps b .01. 3.2.2. Hierarchical regression analyses We were interested in how different dimensions of syntactic skills were related to reading comprehension at different levels; thus we performed separate analyses regressing sentence and passage comprehension on syntactic skills. In all regressions, age and nonverbal intelligence were entered first as control variables. Word reading at Time 1 and the auto-regressor (i.e., reading comprehension at the preceding time point) were entered in the second and the third block respectively. We ran three sets of regressions, each with sentence comprehension at Time 2, passage comprehension at Time 2 or that at Time 3 as the dependent variable (see Table 6).

Within each set, word order skill and morphosyntactic skill were entered in the fourth and fifth block, respectively, in the first regression, whereas their order of entry was reversed in the next regression. These two dimensions of syntactic skills were entered in separate blocks to examine their relative contribution to sentence and passage comprehension when the effect of another was statistically controlled. Table 6 summarises the results of the regression analyses. For sentence-level reading comprehension, age, nonverbal intelligence, word reading and the auto-regressor together explained 70% of the total variance in sentence comprehension at Time 2. The two syntactic skills explained an additional 1% of its total variance. While morphosyntactic skill at Time 1 significantly predicted sentence comprehension at Time 2, word order skill at Time 1 did not. Regarding passage-level reading comprehension, the full model with all predictors explained 46% and 53% of total variances in passage comprehension at Time 2 and Time 3, respectively. After controlling for the effects of age, nonverbal intelligence, word reading and the auto-regressor, word order skill at Time 1 was significantly associated with passage comprehension at Time 2, whereas morphosyntactic skill at Time 1 did not explain significant additional amount of its variance. Word order skill at Time 1 was still a longitudinal correlate of passage comprehension at Time 3, whereas morphosyntactic skill at Time 1 remained non-predictive of later passage comprehension. Overall, the two aspects of syntactic skills were differentially related to reading comprehension at varied levels: grade 2 morphosyntactic skill significantly predicted grade 3 sentence comprehension, while grade 2 word order skill was a longitudinal predictor of grades 3 and 4 passage comprehension.

Table 6 Hierarchical regression analyses with Time 1 syntactic skills predicting Times 2 and 3 reading comprehension in Study 2. Time 2 (Grade 3) Step and variable Predicting sentence comprehension Step 1: Age Nonverbal intelligence Step 2: Word reading Step 3: T1 Sentence comprehension Step 4: T1 Word-order correction Step 5: T1 Morphosyntactic correction Step 4: T1 Morphosyntactic correction Step 5: T1 Word-order correction Predicting passage comprehension Step 1: Age Nonverbal intelligence Step 2: Word reading Step 3: T1 Passage comprehension Step 4: T1 Word-order correction Step 5: T1 Morphosyntactic correction Step 4: T1 Morphosyntactic correction Step 5: T1 Word-order correction Note. * p b .05. ** p b .01. *** p b .001.

Time 3 (Grade 4)

Final β

R2

R2 change

.04 .07 .43 .37 .03 .08 .08 .03

.15

.15

25.29***

.63 .70 .71 .71 .71 .71

.48 .07 .00 .01 .01 .00

364.88*** 64.53*** 1.50 4.08* 5.15* .46

.11 .15 .46 .31 .11 .03 .03 .11

.12

.12

19.82***

.38 .44 .46 .46 .45 .46

.26 .06 .01 .00 .00 .01

114.42*** 26.64*** 4.97* .32 1.98 3.17

F change

Final β

R2

R2 change

F change

– – – – – – – –

– – – – – – – –

– – – – – – – –

– – – – – – – –

−.02 .15 .28 .32 .16 .04 .04 .16

.18

.18

29.41***

.44 .51 .53 .53 .52 .53

.26 .08 .02 .00 .01 .01

125.33*** 42.59*** 8.95** .54 2.64 6.78**

68

C.T.-S. Siu et al. / Learning and Individual Differences 47 (2016) 61–69

3.3. Discussion Study 2 was a three-year longitudinal study exploring the interrelationships between different facets of syntactic skills and varied levels of reading comprehension in Chinese children at grades 2 to 4. We have found in Study 1 that word order and morphosyntactic skills changed in their contributions to passage-level reading comprehension across ages. Study 2 extended Study 1's findings that the two syntactic skills in question were differentially involved in comprehension at different levels. When we controlled for the effects of age, nonverbal intelligence, word reading and the auto-regressive effect of reading comprehension, morphosyntactic skill at grade 2 rather than word order skill was a longitudinal correlate of sentence comprehension at grade 3. In contrast, the prediction of passage comprehension manifested a different pattern – word order, but not morphosyntactic skill, significantly contributed to passage comprehension at grades 3 and 4. 4. General discussion In the present research, we set out to evaluate the roles of different dimensions of syntactic skills in reading comprehension in Chinese using a cross-sectional design in Study 1 and a longitudinal design in Study 2. Overall, the findings from both studies were in line with prior work which has revealed an involvement of syntactic skills in reading comprehension in alphabetic (e.g. Muter et al., 2004; Plaza & Cohen, 2003) and Chinese language (e.g.Chik et al., 2012a, Tong et al., 2014), in support of the triangle model extended (Bishop & Snowling, 2004). Our results also suggested an overall trend of development of syntactic skills from grades 1 to 6, albeit the age difference in morphosyntactic skill was less visible in senior graders. These findings added on Tsang and Stokes (2001) that syntactic skills demonstrated consistent growth throughout elementary school years.

they differed in the choices of syntactic as well as comprehension tasks. Chik et al. (2012a) compared the relative contributions of three syntactic skills and concluded that morphosyntactic skill and connectives usage were stronger predictors of sentence comprehension than word order skill was. It accords with our finding that understanding sentences is dependent more on morphosyntactic than word order skill. In addition, Tong et al. (2014) reported that beyond the effects of other linguistic-related variables, only performance in conjunction cloze task uniquely predicted passage comprehension, whereas the ability to judge and correct grammatical errors did not. This finding is also consistent with ours that repairing isolated morphosyntactic errors is apparently not involved in interpreting passages. Our results suggest that understanding sentences and passages differentially rely on word order and morphosyntactic skills, thus explaining the seemingly inconsistent findings in the literature. When we consider also the results of Study 1 that word order and morphosyntactic skills differ in their predictive relations with reading comprehension at different ages, our findings collectively reinforce the notion that these two syntactic operations were not comparable (Cain, 2007). We think that these two syntactic skills capture different dimensions of grammaticality, thus both skills should be jointly measured to have a comprehensive assessment of syntactic skills. To the best of our knowledge, the measurement of syntactic skill in the field has been rather diverse; the construct has been indexed by a mixture of syntactic tasks across studies. We attribute this phenomenon to the fact that the concept of syntax is so broad that essentially covers a wide spectrum of grammatical structures in different languages. Despite this, we maintain that syntactic skills should at least be represented by two components – word order and morphosyntax, because word order pattern and morphemic system are two basic syntactic structures available in many languages. To address the issue on the role of syntactic skills in developing literacy, we urge for more studies that consistently include both facets in assessing syntactic skills.

4.1. Contributions of word order and morphosyntactic skills to sentence and passage comprehension

4.2. Limitations and suggestions for future research

More importantly, our results in Study 2 demonstrated a distinction between word order and morphosyntactic skill in terms of their contributions to reading comprehension at different levels. As discussed earlier, after reading a sentence, repairing the morphosyntactic violation is usually restricted to specific words or phrases. For example, in the item ‘Yesterday I eat an apple’, the word yesterday sufficiently hints the child that the verb eat was incorrect in tense agreement. This approach applies to the sentence comprehension task in which one nearby constituent might provide adequate ideas to pick up the plausible option answer. Therefore, the manipulations in these two tasks are thought to be operating on the word level. Comparatively speaking, both word order correction and passage comprehension necessitate taking a more global perspective in approaching the items. Reordering shuffled segments into a grammatical sentence requires the child to examine every constituent to build up logical coherence within the sentence. In the item “the supermarket/mother/every day/my/goes to”, no matter what the child has to digest all presented segments before he or she can successfully rearrange them into a correct sentence. Similarly, the questions in the passage comprehension task demand the child to integrate ideas from different parts of the text in formulating inferences and evaluations. As such, the ability to consider the overall context and to establish coherence is clearly implicated in the process. Hence, morphosyntactic skill is at work in understanding sentences, in contrast to word order skill as being more involved in interpreting passages. The present results are important because they help explain and integrate the current scattered findings on the relation between syntactic skills and reading performance in Chinese. Both Chik et al. (2012a) and Tong et al. (2014) demonstrated a significant role of syntactic skills in Chinese reading comprehension in a longitudinal context. However,

The major limitation of the present research is a lack of vigorous experimental manipulations in the study design. The findings should be taken with the caution that any strong causal interpretations are dubious. In future research, we suggest collecting longitudinal data of syntactic skills, so that hierarchical linear modelling analyses can be performed to estimate children's individual growth in syntactic skills. We may also assess whether this individual syntactic development predicts later reading performances. By doing this, more compelling evidence can be gathered to support a contributing role of syntactic skills in developing reading ability. Training studies (e.g., Layton, Robinson, & Lawson, 1998) are also called for to help disentangle causal relationships between syntactic skills and reading development. Another caveat is that because we did not include measures of related skills such as phonological awareness, morphological awareness or vocabulary, our conclusion might be tempered by their shared variances with reading comprehension. Nevertheless, we have controlled for in our regressions the effect of word reading ability, which is assumed to be an overarching decoding ability within the word level. Given that we consistently partialled out the impact of this important word-level literacy skill, we are still positive in claiming a significant part of syntactic skills in reading comprehension. Nevertheless, we see the value of including these variables in our future research to improve interpretability of the findings. Finally, the low reliability of the passage comprehension task at Time 1 also requires some caution in interpretation. To conclude, our study provided further empirical evidence to understand the syntactic influence on reading performance. We found a differential reliance of reading comprehension on word order and morphosyntactic skill across grades (Study 1) and in comprehension at different levels (Study 2). The findings suggest that the two syntactic

C.T.-S. Siu et al. / Learning and Individual Differences 47 (2016) 61–69

measures are not comparable because they assess different aspects of grammaticality. The take-home message is that it is important to thoroughly capture syntactic skills by testing both dimensions of skills before drawing compelling conclusion. References Adams, M.J. (1990). Beginning to read: Thinking and learning about print. Cambridge, MA: MIT Press. Bishop, D.V. M., & Snowling, M. J. (2004). Developmental dyslexia and specific language impairment: Same or different? Psychological Bulletin, 130, 858–886. Bowey, J. A. (1986a). Syntactic awareness and verbal performance from preschool to fifth grade. Journal of Psycholinguistic Research, 15, 285–306. Bowey, J. A. (1986b). Syntactic awareness in relation to reading skill and ongoing comprehension monitoring. Journal of Experimental Child Psychology, 41, 282–299. Cain, K. (2007). Syntactic awareness and reading ability: Is there any evidence for a special relationship? Applied PsychoLinguistics, 28, 679–694. Chang, H. W. (1992). The acquisition of Chinese syntax. In H. C. Chen, & O. J. L. Tzeng (Eds.), Language processing in Chinese. Amsterdam: North Holland. Chao, Y. R. (1968). A grammar of spoken Chinese. Berkeley: University of California Press. Chen, M. J., & Wong, A. (1991). Component skills as predictors of Chinese reading proficiency in primary school children. International Journal of Psychology, 26, 53–70. Chen, M. J., Lau, L. L., & Yung, Y. F. (1993). Development of component skills in reading Chinese. International Journal of Psychology, 28, 481–507. Chik, P. P. -M., Ho, C. S. -H., Yeung, P. P. -S., Chan, D. W. -O., Chung, K. K. -H., Luan, H., ... Lau, W. S. -Y. (2012a). Syntactic skills in sentence reading comprehension among Chinese elementary school children. Reading and Writing, 25, 679–699. Chik, P. P. -M., Ho, C. S. -H., Yeung, P. P. -S., Wong, Y. -K., Chan, D. W. -O., Chung, K. K. -H., & Lo, L. -Y. (2012b). Contribution of discourse and morphosyntax skills to reading comprehension in Chinese dyslexic and typically developing children. Annals of Dyslexia, 62, 1–18. Davidson, D., Raschke, V. R., & Pervez, J. (2010). Syntactic awareness in young monolingual and bilingual (Urdu-English) children. Cognitive Development, 25, 166–182. Demont, E., & Gombert, J. E. (1996). Phonological awareness as a predictor of recoding skills and syntactic awareness as a predictor of comprehension skills. British Journal of Educational Psychology, 66, 315–332. Gaux, C., & Gombert, J. E. (1999). Implicit and explicit syntactic knowledge and reading in pre-adolescents. The British Journal of Developmental Psychology, 17, 169–188. Goff, D. A., Pratt, C., & Ong, B. (2005). The relations between children's reading comprehension, working memory, language skills, and components of reading decoding in a normal sample. Reading and Writing: An Interdisciplinary Journal, 18, 583–616. Gombert, J. E. (1992). Metalinguistic development. Chicago: University of Chicago Press. Gottardo, A., Stanovich, K., & Siegel, L. S. (1996). The relationships between phonological sensitivity, syntactic processing, and verbal working memory in the reading performance of third-grade children. Journal of Experimental Child Psychology, 63, 563–582. Gough, P. B., & Tunmer, W. E. (1986). Decoding, reading and reading disability. Remedial and Special Education, 7, 6–10. Hakes, D. T. (1980). The development of metalinguistic abilities in children. Berlin: SpringerVerlag. Ho, C. S. -H., Chan, D., Tsang, S. -M., & Lee, S. -H. (2000). The Hong Kong test of specific learning difficulties in reading and writing. Pokfulam Road, Hong Kong: University of Hong Kong, Hong Kong Specific Learning Difficulties Research Team. Hong Kong Education Department (1986). Hong Kong supplement to guide to the standard progressive matrices. Hong Kong, China: Hong Kong Government. Jongejan, W., Verhoeven, L., & Siegel, L. S. (2007). Predictors of reading and spelling abilities in first- and second-language learners. Journal of Educational Psychology, 99, 835–851. Layton, A., Robinson, J., & Lawson, M. (1998). The relationship between syntactic awareness and reading performance. Journal of Research in Reading, 21, 5–23. Lee, T. H. T. (1996). Theoretical issues in language development and Chinese child language. In C. T. J. Huang, & Y. H. A. Li (Eds.), New horizons in Chinese linguistics. Netherlands: Kluwer Academic Press. Leung, M. T., & Lee, A. W. Y. (2002, May). The Hong Kong corpus of primary school Chinese. Paper presented at the 9th meeting of the International Clinical Phonetics and Linguistics Association, Hong Kong. Li, C. N., & Thompson, S. A. (1981). Mandarin Chinese: A functional reference grammar. Berkeley: University of California Press.

69

Li, T., McBride-Chang, C., Wong, A., & Shu, H. (2012). Longitudinal predictors of spelling and reading comprehension in Chinese as an L1 and English as an L2 in Hong Kong Chinese children. Journal of Educational Psychology, 104, 286–301. Lin, Y. A. (2006). Against the deficit in computational grammatical complexity hypothesis: A corpus-based study. Concentric: Studies in Linguistics, 32, 59–70. Matthews, S., & Yip, V. (1994). Cantonese: A comprehensive grammar. London and New York: Routledge. Muter, V., & Snowling, M. (1998). Concurrent and longitudinal predictors of reading: The role of metalinguistic and short-term memory skills. Reading Research Quarterly, 33, 320–337. Muter, V., Hulme, C., Snowling, M. J., & Stevenson, J. (2004). Phonemes, rimes, vocabulary, and grammatical skills as foundations of early reading development: Evidence from a longitudinal study. Developmental Psychology, 40, 665–681. Nation, K., & Snowling, M. J. (2000). Factors influencing syntactic awareness skills in normal readers and poor comprehenders. Applied PsychoLinguistics, 21, 229–241. Plaza, M., & Cohen, H. (2003). The interaction between phonological processing, syntactic awareness, and naming speed in the reading and spelling performance of first-grade children. Brain and Cognition, 53, 287–292. Plaza, M., & Cohen, H. (2004). Predictive influence of phonological processing, morphological/syntactic skill, and naming speed on spelling performance. Brain and Cognition, 55, 368–373. Pratt, C., Tunmer, W. E., & Bowey, J. A. (1984). Children's capacity to correct grammatical violations in sentences. Journal of Child Language, 11, 129–141. Raven, J. C., Court, J. H., & Raven, J. (1996). Standard progressive matrices. Oxford, England: Oxford Psychologists Press. Reder, F., Marec-Breton, N., Gombert, J. E., & Demont, E. (2013). Second-language learners' advantage in metalinguistic awareness: A question of languages' characteristics. British Journal of Educational Psychology, 83, 686–702. Rego, L. L. B., & Bryant, P. E. (1993). The connection between phonological, syntactic and semantic skills and children's reading and spelling. European Journal of Psychology of Education, 13, 235–246. Siegel, L. S., & Ryan, E. B. (1988). Development of grammatical-sensitivity, phonological, and short-term memory skills in normally achieving and learning disabled children. Developmental Psychology, 24, 28–37. So, D., & 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. Tong, X., Tong, X., Shu, H., Chan, S., & McBride-Chang, C. (2014). Discourse-level reading comprehension in Chinese children: What is the role of syntactic awareness? Journal of Research in Reading, 37, S48–S70. Tsang, K. K. S., & Stokes, S. F. (2001). Syntactic awareness of Cantonese-speaking children. Journal of Child Language, 28, 703–739. Tunmer, W. E. (1989). The role of language-related factors in reading disability. In D. Shankweiler, & I. Y. Liberman (Eds.), Phonology and reading disability: Solving the reading puzzle (pp. 91–131). Ann Arbor, MI: University of Michigan Press. Tunmer, W. E., & Bowey, J. A. (1984). Metalinguistic awareness and reading acquisition. In W. E. Tunmer, C. Pratt, & M. L. Herriman (Eds.), Metalinguistic awareness in children. Berlin: Springer-Verlag. Tunmer, W. E., & Chapman, J. W. (1998). Language prediction skills, phonological recoding ability and beginning reading. In M. Joshi, & C. Hulme (Eds.), Reading and spelling: Development and disorders (pp. 33–67). Hillsdale, NJ: Erlbaum. Tunmer, W. E., & Hoover, W. (1992). Cognitive and linguistic factors in learning to read. In P. B. Gough, L. C. Ehri, & R. Treiman (Eds.), Reading acquisition (pp. 175–214). Hillsdale, NJ: Erlbaum. Tunmer, W. E., Nesdale, A. R., & Wright, A. D. (1987). Syntactic awareness and reading acquisition. British Journal of Developmental Psychology, 5, 25–34. Wang, H., & Chang, R. B. (1985). Modern Chinese frequency dictionary. Beijing Language University Press: Beijing. Willows, D. M., & Ryan, E. B. (1986). The development of grammatical sensitivity and its relationship to early reading achievement. Reading Research Quarterly, 21, 253–266. Yeung, P. -S., Ho, C. S. -H., Chan, D. W., 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. Yeung, P. -S., Ho, C. S. -H., Chik, P. P. -M., Lo, L. -Y., Luan, H., 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.