February 2025 – Volume 28, Number 4
https://doi.org/10.55593/ej.28112a2
Amir Mahshanian
University of Isfahan
<mshn_amir
yahoo.com>
Mohammadtaghi Shahnazari
University of Isfahan
<m.shahnazarifgn.ui.ac.ir>
Ahmad Moinzadeh
University of Isfahan
<moinfgn.ui.ac.ir>
Abstract
This study investigates the relationships among working memory (WM), syntactic parsing ability (SP), and L2 reading performance across varying proficiency levels. A cohort of 120 L1-Persian EFL learners was categorized into beginner, intermediate, and advanced proficiency groups based on their IELTS scores. Participants completed a reading span test, a sentence segmentation task, and a reading test to measure WM, SP, and L2 reading ability, respectively. Statistical analysis revealed significant differences in syntactic parsing and reading abilities across proficiency levels, with advanced learners outperforming their intermediate and beginner counterparts. Moreover, significant relationships were identified between WM, SP, and L2 reading performance among participants in the beginner group, while such relationships were not observed in the intermediate and advanced groups. Additionally, the results identified SP as a robust predictor of L2 reading performance, specifically within the beginner proficiency level, suggesting that syntactic parsing abilities substantially contribute to variations in L2 reading outcomes at this proficiency level. Focusing on the intricate interplay between cognitive resources and reading processes, these findings underscore the essential role of syntactic parsing in enhancing L2 reading comprehension among beginner learners, highlighting the necessity for targeted instructional strategies to support the development of these skills.
Keywords: working memory, lower-level processing, syntactic parsing, L2 reading
Reading comprehension is a key factor in accounting for academic and occupational success across both first and second languages (Goldman & Pellegrino, 2015). It involves constructing mental representations of written language by extracting meaning at various levels, including words, sentences, and text (Brimo et al., 2017). The process of comprehension engages a complex interplay of skills that unfold interactively during reading (Lopez, 2008). Grabe and Stoller (2011) assert that reading comprehension encompasses the simultaneous processing of written information through two distinct levels of cognitive processes: lower-level and higher-level processing. Lower-level processing skills include word recognition (the ability to decode the individual words for meaning), semantic encoding (the ability to integrate word meanings with their grammatical structures to understand the meaning of a clause), and syntactic parsing (the grammatical ability to recognize word ordering structures in a sentence). On the other hand, higher-level processing refers to conscious strategies such as inference-making and comprehension monitoring, which readers employ to derive deeper meaning from text (Habók & Magyar, 2019).
Past research has demonstrated that syntactic parsing, as a key component of lower-level processing skills, significantly contributes to fluent reading (Chen et al., 2018). According to Grabe and Stoller (2011), syntactic parsing involves breaking down larger structural units into smaller ones to extract grammatical information from text and construct meaning at the clause level. This process allows for the identification of the grammatical relationships between components of a text’s sentences which helps readers gain an understanding of the meaning of the whole sentence (Tasci & Turan, 2020). While fluent readers, such as those reading in their first language (L1) or those with higher levels of proficiency in L2 reading, may perform syntactic parsing effortlessly and automatically, it is widely recognized that processing syntactic relationships poses cognitive challenges for less fluent readers, such as those at lower L2 proficiency levels (Lopez, 2008).
Given the critical role of lower-level processing skills in text comprehension, researchers argue that successful reading in a second language heavily relies on automatic processing at a lower level, such as syntactic parsing (Hamatani, 2020). That is, with lower- and higher-level processing both consuming cognitive resources, automatic lower-level processing of syntactic structures conserves cognitive resources, notably working memory (WM), which can then be allocated to more demanding higher-level processes (Reynolds, 2000). More precisely, individual differences in working memory capacity can significantly impact the efficiency of higher-level reading processes like inference-making (Linderholm et al., 2008). Hence, it could be posited that reading comprehension in a second language is subject to the constraints of cognitive resources such as WM (Liu et al., 2019). With slow and non-automatic syntactic processing, readers fall short of very limited (e.g., Torrens & Yagüe, 2018) WM resources, that are needed for processing at a higher level (e.g., gaining an inferential understanding of the text), which accounts for the poor reading performance of less proficient L2 readers.
Considering the importance of syntactic parsing within the constraints of WM, in accounting for individual differences in reading comprehension, the scarcity of L2 research on the interaction between WM, syntactic parsing, and L2 reading is notably surprising. While some studies in the L2 literature have explored relationships between WM and syntactic parsing (e.g., Hopp, 2014) or between syntactic parsing and L2 reading (e.g., Lopez, 2008; Tasci & Turan, 2020), the distinct contributions of WM and syntactic parsing to reading performance in a second language remain largely unexamined. Despite consensus regarding the relationship between WM and L2 reading comprehension (Linck et al., 2014; Shin, 2020), it is still unclear how underlying reading processes like syntactic parsing might influence the interplay between WM and L2 reading comprehension or whether the association among WM, syntactic parsing, and L2 reading varies by language proficiency level.
Exploring the interplay between working memory (WM) and syntactic parsing, and their collective impact on overall reading performance, provides a clearer understanding of the underlying mechanisms involved in second language (L2) reading comprehension. Therefore, this study aimed to investigate the relationship among WM, syntactic parsing, and reading performance, focusing particularly on whether and how L2 proficiency moderates this relationship. By examining the distinct contributions of WM and syntactic parsing ability to reading outcomes across different proficiency levels, this research seeks to inform material developers and educational policymakers on how best to cater to the diverse needs of L2 learners based on their WM and syntactic abilities.
Theoretical Framework and Literature Review
The complexity and multidimensionality of reading comprehension have led to the development of a multitude of models and theories aimed at accounting for its various components and processes (Haft et al., 2019). Among these, two prominent theories emphasize the role of working memory in text comprehension, particularly in relation to language proficiency: the compensatory encoding model and the capacity constraint reader model (CC reader). The compensatory encoding model, an adaptation of Perfetti’s verbal efficiency model (Walczyk, 1995, 2000), expands on how less-skilled readers, who may possess underdeveloped lower-level processing skills such as word recognition, syntactic parsing, and semantic encoding, can still comprehend texts (Naumann & Goldhammer, 2016).
The foundational premise of Walczyk’s (1995, 2000) model posits that automatic lower-level processing facilitates effective text comprehension, with successful reading relying on continuous adoption of compensatory strategies to address inefficiencies at both lower and higher levels of processing (Grabe, 2009). Highlighting the pivotal roles of WM and language proficiency, Walczyk (2000) maintained that proficient readers benefit from automatic lower-level processing, which allows them to save up space in their WM for higher-level cognitive tasks such as inference-making and comprehension monitoring. With reduced cognitive load on lower-level processes like word recognition or syntactic parsing, proficient readers experience less difficulty in higher-level processing, leading to enhanced reading performance. Conversely, less proficient readers face significant challenges as lower-level processes do not occur automatically, necessitating greater allocation of WM resources for processing at the word or sentence level. This inefficiency in lower-level processing limits the availability of cognitive resources for higher-level processing, contributing to poorer reading performance among less skilled readers (Walczyk, 2000).
Another influential theory of reading that explicitly addresses the role of working memory in text processing is Just and Carpenter’s (1992) CC reader model. Grounded in connectionist theory, this model conceptualizes reading as an activation-based system where both lower- and higher-level processes operate concurrently (Grabe, 2009). According to this framework, as information from the text is accessed during reading, associated concepts are also activated in long-term memory and brought into working memory for further processing. The capacity constraint, then, refers to the maximum number of activations available for both storage and processing within this system (Miyake et al., 1994).
Given that working memory capacity varies across individuals, the CC reader model posits that differences in reading performance stem from readers’ capacity to maintain and manipulate information in WM (Long & Freed, 2021). In second language (L2) reading, where lower-level text processing (such as word decoding, syntactic parsing, and semantic encoding) consumes a substantial portion of available WM resources, higher-level processing is hindered due to resource limitations, thereby impairing comprehension. Consequently, as resource availability restricts the amount of cognitive processing, skilled readers are distinguished from less-skilled readers by their cognitive ability to efficiently access and manage available resources (Reichle et al., 2000). These models underscore the critical role of cognitive resources in reading comprehension and have spurred extensive research aimed at investigating the relationship between WM and reading comprehension.
Working Memory and L2 Reading Comprehension
Working memory (WM) is a cognitive system characterized by a finite capacity to manipulate and temporarily store information crucial for a variety of complex cognitive tasks (Baddeley, 2003). The most influential WM model in psycholinguistics, according to Delage and Frauenfelder (2019), is the modular tripartite WM model developed by Baddeley and Hitch (1974). This model comprises three components: the “central executive,” responsible for attentional control; the “phonological loop,” which stores auditory and verbal information; and the “visuospatial sketchpad,” which stores visuospatial information. Later, the “episodic buffer” was added as a fourth component serving as an interface between the phonological loop and the visuospatial sketchpad, integrating and retrieving diverse information from long-term memory (Baddeley, 2019). It is well established in past research that working memory capacity (WMC) is closely linked to language processing skills, such as writing, listening, and reading comprehension (Alptekin & Ercetin, 2010).
Given the importance of WM in language processing, a body of research has also investigated its role in L2 reading (Demir & Erçetin, 2020). The significant contribution of WM to L2 reading has been supported by a considerable number of past studies (e.g., Alptekin & Ercetin, 2011; Chang et al., 2019; Harrington & Sawyer, 1992). For example, Harrington and Sawyer (1992) found a strong correlation between WMC and L2 reading comprehension, indicating that L2 learners with higher WMC perform significantly better on reading tests than those with lower WMC. Lesser (2007) carried out a similar study on the influence of WMC, in addition to topic familiarity, on L2 reading and found that WMC makes significant contributions to L2 reading and grammatical knowledge.
Alptekin and Erçetin (2011) also reported that WM and topic familiarity affect L2 reading ability, particularly in inferential text comprehension. In a study involving Korean L2 learners, Joh (2018) observed significant differences in L2 reading performance as a result of different strategy use when readers with low WMC processed difficult texts, whereas no such differences were noted with easier texts or among readers with higher WMC. In another L2 study on Korean learners, Jung (2018) argued that WM plays a crucial role in predicting variance in the reading performance of Korean L2 learners. These findings align with those of Chang et al. (2019), who identified WMC as a strong predictor of reading comprehension, as well as L2 grammar and writing ability.
Notwithstanding the aforementioned findings, several studies have argued that there is either no relationship between WMC and L2 reading or that WMC’s effect is only marginally significant compared to other cognitive factors such as participants’ prior knowledge (e.g., Chun & Payne, 2004; Georgiou & Das, 2016; Joh & Plakans, 2017). For instance, Chun and Payne (2004) found no significant relationship between the reading ability of German L2 English learners and their WMC. Similarly, Georgiou and Das (2016) reported that WMC is not a predictor of second-language reading proficiency. Additionally, Joh and Plakans (2017) suggested that WM becomes a significant factor in predicting L2 reading ability only when readers possess some prior knowledge of the text.
In a recent study, Shahnazari (2023) suggested that while WMC is an important component in predicting the performance of less proficient readers, it is a non-significant factor in predicting the performance of proficient L2 readers. However, the link between WM and the underlying reading mechanisms (e.g., higher- and lower-level processing abilities) and how the interplay between WM and these processing skills might impact L2 reading performance is absent from these findings. These gaps, along with the contradictory results in L2 research on the link between WM and reading, leave the possibility open for the relationship between WM and reading to be influenced by other underlying reading processes such as word recognition, syntactic parsing, etc., and hence call for a closer examination of each underlying aspect of reading comprehension in minute detail (Huang et al., 2022).
Syntactic Parsing and L2 Reading Comprehension
Syntax, a crucial component of grammar, determines the structure of words and how they are combined to form phrases, clauses, and sentences (Radford, 2009). Syntactic knowledge allows language learners to analyze the grammatical structure of sentences. This process, known as syntactic parsing, involves the analysis, understanding, and production of various grammatical structures within the context of a sentence (Cutting & Scarborough, 2006). In the context of second language (L2) learning, however, no consensus has been reached about the unique contributions of syntactic parsing to reading comprehension. While some studies highlighted the importance of syntactic parsing for fluency and accuracy in L2 reading (e.g., Nergis, 2013; Shiotsu & Weir, 2007), others stressed that syntactic parsing is either non-significant or only partially significant in explaining variations in L2 reading comprehension (e.g., Lopez, 2008; Ulijn & Strother, 1990).
Initial investigations into the role of syntactic parsing in reading comprehension began by examining the effects of syntactic simplification on L2 reading. In a study on syntactic simplification, Yano et al. (1994) suggested that learners who read syntactically simplified texts performed significantly better on L2 reading comprehension tests compared to those who read original, unmodified materials. This indicates that L2 readers with an adequate grasp of syntactic structures are less likely to encounter difficulties in understanding textual structures. In comparing different linguistic factors that predict variance in L2 reading performance, Shiotsu and Weir (2007) identified syntactic parsing as the most significant predictor of L2 reading ability. The key role of syntactic parsing ability in L2 reading was further highlighted in studies comparing the relative contributions of syntactic and lexical knowledge to reading comprehension. These studies suggested that the ability to parse syntactic structures is the most important predictor of reading achievement in a second language (Atai & Nikuinezhad, 2012; Maftoon & Tasnimi, 2014).
More recently, Tasci and Turan (2020) investigated the relative contributions of lexical breadth, lexical depth, and syntactic parsing to L2 reading comprehension. Their findings, consistent with their subsequent study (Tasci & Turan, 2021), indicated that syntactic parsing most significantly related to L2 reading comprehension. Furthermore, Chen et al. (2018) explored the instructional effects of syntactic parsing on the reading abilities of Chinese EFL learners. They found a strong relationship between syntactic parsing and reading speed, suggesting that instruction in syntactic parsing could enhance the reading rate of L2 learners. However, it is important to note that Chen et al.’s (2018) study primarily focused on reading speed rather than overall reading ability, emphasizing how syntactic parsing skills influence the pace at which learners can read texts.
In contrast to studies emphasizing the important role of syntactic parsing in explaining individual differences in reading comprehension, several L2 investigations have contested the presence of a significant relationship between syntactic parsing and L2 reading ability (e.g., Ulijn & Strother, 1990; Yamashita & Ichikawa, 2010). For instance, Ulijn and Strother (1990) examined whether syntactic simplification enhances L2 reading comprehension, arguing that modifying texts for syntactic simplicity does not necessarily lead to improved or faster reading performance. Similarly, Liu and Bever (2002) and Yamashita and Ichikawa (2010) concluded that the syntactic structure of texts—whether conventional, syntactically parsed, randomly segmented, or word-by-word segmented—does not affect the reading speed of L2 learners, and variations in reading comprehension cannot be solely attributed to syntactic abilities. In another study investigating the impact of syntax on L2 reading achievement, Lopez (2008) also found no significant correlation between syntactic parsing and readers’ ability to comprehend a given text.
The conflicting findings from various studies, which either support a significant relationship between syntactic parsing and L2 reading ability (e.g., Shiotsu & Weir, 2007; Tasci & Turan, 2020; Chen et al., 2018) or refute the influence of syntactic parsing on reading comprehension (Lopez, 2008; Ulijn & Strother, 1990), have impeded our understanding of the role of syntax in L2 reading. This divergence suggests that besides syntactic parsing, other linguistic variables and cognitive factors, such as working memory capacity, may play pivotal roles in deriving meaning from textual data. Therefore, a thorough investigation into how these underlying linguistic and cognitive processes interact during L2 reading comprehension is essential for advancing current knowledge in L2 research.
Working Memory, Syntactic Parsing, and L2 Reading Comprehension
While theoretical frameworks such as those proposed by Perfetti (1985) and Stanovich (2000) have established the independent contributions of WM and syntactic parsing to reading comprehension, empirical evidence supporting the relationship between WM, syntactic processing, and L2 reading performance remains inconclusive (Chen et al., 2018). Previous studies have consistently linked WMC with syntactic processing (Miyake & Friedman, 1998; Sagarra, 2007; Sagarra & Herschensohn, 2010) and reading comprehension (Abu-Rabia, 2006).Top of Form However, these studies primarily focused on specific aspects of syntactic parsing ability at the sentence level. For instance, Sagarra (2007) investigated how WMC influences L2 learners’ sensitivity to gender agreement violations in English sentences, finding that higher WMC correlates with heightened sensitivity to such violations, whereas lower-WMC learners show reduced sensitivity. Similarly, Sagarra and Herschensohn (2010) explored the impact of proficiency and WM on processing gender and number agreement in English, revealing that sensitivity to gender violations is influenced by WMC among higher proficiency learners but not among lower proficiency learners. These findings also align with Skorobogatova et al. (2021) broader conclusions that significant relationships exist between L2 syntactic parsing ability and WMC.
In another study, Abu-Rabia (2006) explored the foundational role of working memory capacity and syntactic parsing ability in reading development, particularly within Arabic orthography. Abu-Rabia suggested that deficits in reading comprehension in Arabic could be attributed to inefficient WMC or syntactic processing skills. However, this study specifically focused on Arabic orthography and did not investigate whether WMC or syntactic processing skills influence reading performance in a second language (L2), nor did it explore how L2 proficiency might modulate such relationships.
Among the few studies examining the links between working memory, syntactic parsing, and language processing, Hopp (2014) conducted research comparing native and nonnative speakers of English to analyze their syntactic parsing preferences, focusing particularly on individual differences influenced by WMC. Hopp investigated whether attachment preferences in ambiguous relative clauses differ based on WMC in both first language (L1) and second language (L2). Specifically, he distinguished between low-attachment and high-attachment preferences. Low attachment occurs when a relative clause is interpreted as pertaining to the last noun phrase in a sentence (e.g., “I saw the son of the president who is tall“), where the adjective clause “who is tall” is preferred to relate to “the president” rather than “the son.” In contrast, high attachment happens when the adjective clause “who is tall” is preferred to relate to the first noun phrase “the son.”
Two significant findings emerged from Hopp’s (2014) study. First, L2 learners with lower working memory capacities demonstrated a tendency to chunk the sentence into smaller units (e.g., the entire noun phrase and the relative clause), leading them to attach the relative clause to the first noun. In contrast, learners with higher WMC showed less inclination toward chunking and preferred to interpret the relative clause as pertinent to the second noun in the sentence. Second, the study highlighted that syntactic processing strategies, influenced by WMC constraints, operate similarly in both L1 and L2 contexts. Thus, WMC’s impact on syntactic processing in L1 mirrors its effect on L2 syntactic parsing ability.
What has been notably absent in past research is an exploration of whether the interplay between cognitive resources such as WMC and linguistic factors like syntactic parsing governs general reading ability in a second language. Furthermore, a significant gap in the current literature pertains to whether the impacts of WMC and syntactic parsing abilities, if they exist, on reading performance can be moderated by proficiency levels in a second language. Therefore, further investigation is warranted both to elucidate the relationship between WMC and underlying reading processes and, more importantly, to empirically validate hypotheses posited in memory-based reading comprehension models (e.g., the CC reader and compensatory encoding).
This Study
Since the inception of research on working memory in second language acquisition (SLA), scholars have focused on understanding how WM influences learners’ language skills (Wen, 2019). The examination of WM’s connections with second-language reading comprehension has been a pivotal aspect of this inquiry (e.g., Leeser & Herman, 2022; Shin, 2020). However, conflicting findings in previous studies regarding the specific relationship between WM and L2 reading have hindered a comprehensive understanding of how WM impacts reading performance in a second language. Absent from past research on WM and L2 reading is an investigation of how WM interacts with syntactic processing skills in the course of reading. Additionally, there remains considerable uncertainty regarding the role of language proficiency in the relationship between WM and L2 reading ability.
To address these gaps, the current study aimed to investigate the relationships between WM, syntactic parsing, and L2 reading performance. To determine whether these connections, if any, varied among proficiency levels, this study also examined the links between WM, syntactic parsing, and L2 reading across three proficiency groups. Considering these objectives, the following research questions guided this study:
- Is there a significant difference in working memory capacity, syntactic parsing skills, and L2 reading ability across different proficiency levels?
- Is there a significant relationship between working memory, syntactic parsing skills, and L2 reading ability at different proficiency levels?
- Can working memory and syntactic parsing skills predict L2 reading ability across different proficiency levels?
Method
Participants
A total of 120 Persian EFL learners were selected from a larger pool of 151 university students majoring in translation studies, literature, medicine, nursing, pharmacy, and dentistry to participate in this study. Participants were screened for their English language proficiency using an IELTS mock test. Based on their overall scores on the test, participants were categorized into three proficiency groups: beginner, intermediate, and advanced. Appendix I provides detailed information on the grouping criteria, participants’ majors, as well as demographic details including age and gender.
The grouping method was based on Cambridge Assessment Scales in accordance with the Common European Framework of Reference (CEFR). Following this reference, the IELST scores were converted to a CEFR scale which was based on proficiency levels. These levels included A1 and A2 (beginner users with IELTS scores ranging between 0 and 4), B1 and B2 (intermediate users with IELTS scores ranging between 4 and 6.5), and C1 and C2 (advanced users with IELTS scores ranging between 7 and 9). This categorization was necessary as regression analysis assumes a linear relationship among variables, whereas we suspect the possibility of a curved shape in the data. By establishing these proficiency groups, we can examine distinct performance patterns more effectively while remaining open to nonlinear relationships that may emerge.
Within this framework, beginner users are defined as those who exhibit insufficient competence in using English in daily contexts, intermediate users as those who can demonstrate effective interaction, communication, and comprehension skills in daily English, and advanced users as those who possess a fully-developed command over the English language, capable of comprehending a variety of complex and lengthy texts and interpreting implied meanings. It should also be noted that all participants were informed about the study’s purpose and procedures, and informed consent was obtained prior to their participation. Participants were also assured of the confidentiality of their data and that their participation was entirely voluntary. Table 1 summarizes descriptive statistics of participants’ scores within each proficiency level.
Table 1. Descriptive Statistics of Language Proficiency Scores Across Different Proficiency Levels
| CEFR Scale | IELTS Range | Proficiency Levels | N | M | SD | Min | Max |
| A1-A2 | 0-4 | Beginner | 40 | 3.6250 | 0.44936 | 3.00 | 4.00 |
| B1-B2 | 4.5-6.5 | Intermediate | 40 | 5.6875 | 0.61694 | 4.50 | 6.50 |
| C1-C2 | 7-9 | Advanced | 40 | 7.5125 | 0.44560 | 7.00 | 8.50 |
| A1-C2 | 0-9 | Total | 120 | 5.6083 | 1.67305 | 3.00 | 8.50 |
Note. N=Number; M=Mean; SD=Standard Deviation; Confidence interval: 95%.
Materials
In this study, participants engaged in a series of tasks across four sessions, spaced one week apart, to assess their language proficiency, working memory (WM), syntactic parsing skills, and L2 reading ability. The tasks included an IELTS mock test, a reading span test (RST), a sentence segmentation task, and a multiple-choice reading test, which act as representative measures of each.
It should be noted that while the same proficiency test (IELTS) and memory measure (RST) were administered to all participants, the reading measures varied by proficiency level. This differentiation was necessary because it was not possible to adequately discriminate between levels of reading ability using a single general test applicable to all proficiency groups. This decision was motivated by the observation that the IELTS reading section was particularly challenging for the beginner group due to their limited vocabulary and grammatical knowledge. By using a reading test that accommodates the varying abilities of participants, we aimed to ensure an accurate assessment of true reading ability, allowing us to evaluate how reading skills at each level relate to working memory and syntactic parsing without the confounding influence of a test that might not reflect the competencies of beginner learners.
Test of English Language Proficiency
Participants’ English language proficiency was assessed using the aforementioned IELTS mock test (Cambridge University Press and Cambridge Assessment, 2020, 2021, 2022), which included four sections: listening, reading, writing, and speaking. Scores from these sections were averaged and categorized into bands ranging from 1 to 9. The speaking and writing sections were evaluated by two examiners, each with ten years of experience in EFL and IELTS instruction. The inter-rater reliability for the speaking and writing sections were checked using Pearson’s correlation because there were only two raters. Both showed acceptable levels of reliability; speaking = 0.91, writing = 0.88. The listening and reading sections were scored based on the provided answer keys.
The Reading Span Test (RST)
In this study, working memory was measured using an L1 version of Daneman and Carpenter’s (1980) Reading Span Test (RST). The rationale for employing this test is threefold. First, research has demonstrated that RSTs have higher reliability estimates compared to simple span tests like digit span or word span tests (In’nami et al., 2021). Second, RSTs tap into both aspects of WM (i.e., storage and processing) making them the preferred and mostly widely used method for assessing WM (Shin, 2020). Finally, L1 RSTs are advantageous over L2 RSTs as they minimize the potential effects of L2 knowledge deficiencies. Since WM is considered language-independent (Miyake & Friedman, 1998; Osaka & Osaka, 1992), L1 RSTs provide more reliable results, focusing solely on WM capacity rather than on construct-irrelevant factors like language proficiency (Shahnazari, 2023). Therefore, an L1 version of the test, originally developed by Shahnazari (2023), was used in this study. The published reliability of the test, indicated by Cronbach’s alpha, was 0.834 for processing and 0.737 for storage (Shahnazari, 2023).
Similar to an L2 RST, participants in an L1 RST are required to silently read a set of sentences to determine the acceptability of each statement in terms of semantics or syntax, which assesses the processing aspect of WM. After reading the entire set, participants were then asked to report the last word of each sentence, which were all Persian verbs, to measure the storage aspect of WM. The task included 54 active sentences, each ranging from 8 to 13 words in length. The number of sentences in each set increased as the task progressed, with four sets containing 3, 4, 5, and 6 sentences respectively. Half of the 54 sentences were syntactically impossible based on Persian clause structure, while the other half were syntactically acceptable in Persian.
For the scoring of the processing aspect of WM, one point was assigned for each correct verification of sentence acceptability. Similarly, for the scoring of the storage aspect, one point was awarded for the correct and in-order recall of the final words in each sentence. With a total of 54 sentences, each participant received separate scores for processing and recall, each ranging from 0 to 54. As all participants demonstrated a processing capacity above 85%, in accordance with Conway et al.’s (2005) guidelines, the recall scores were used as indicators of participants’ working memory capacity.
The Sentence Segmentation Task
A paper-based sentence segmentation task was employed to measure participants’ syntactic parsing ability. Four major types of sentences were incorporated into this task. The items included simple (N = 20), compound (N = 20), complex (N = 20), and compound-complex sentences (N = 20). A simple sentence is an independent clause containing a subject, a verb, and a complete thought (e.g., “Mary reads novels”). A compound sentence refers to two independent clauses joined by a coordinating conjunction (e.g., “Mary reads novels, but Jack reads comics”). A complex sentence is defined as a dependent clause (headed by a subordinating conjunction or a relative pronoun) joint to an independent clause (e.g., “While Mary reads novels, Jack prefers comics”). A compound-complex sentence includes two independent clauses joined to one or more dependent clauses (e.g., “Mary reads novels, but Jack reads comics because they are interesting”). Each sentence type was designed to assess different aspects of syntactic parsing, ensuring a comprehensive evaluation of participants’ abilities to analyze and segment various grammatical structures.
Consistent with previous research (Chen et al., 2018; Yamashita & Ichikawa, 2010), the syntactic parsing measure in this study adhered to eight segmentation principles: 1) separation of subjects from predicates (e.g., Mary / took the bus; My friend and I / took the bus); 2) separation of coordinating conjunctions from the rest of the sentence (e.g., Mary / took the bus /but/ Alan/ took a taxi); 3) separation of subordinating conjunctions from the clause following it (e.g., While / Mary / took the bus/ Alan / took a taxi); 4) separation of prepositional phrases from the rest of the sentence (e.g., Mary / waited / at the bus station); 5) separation of adverbs from the rest of the sentence (Yesterday / Mary / waited / at the bus station); 6) separation of transition words functioning as an adverb from the rest of the sentence (e.g., Mary / took the bus. /Therefore / she / arrived / on time); 7) separation of the complementizer “that” from the rest of the sentence (e.g., I / knew /that/ Mary / will take the bus); 8) separation of combined noun and verb modifiers in case of class-shift words (e.g., parents’ ideas change / influence children’s behavior c.f., “parents’ ideas / change influence children’s behavior”, which results from employing wrong separation principles). These principles were meticulously designed to ensure precise and systematic evaluation of participants’ syntactic parsing abilities, providing a detailed assessment of their capacity to correctly segment and comprehend various grammatical structures.
Based on the outlined principles, participants were instructed to parse sentences in the task using slashes. This instruction was provided in one session during their course, 35 days before the actual experiment, to minimize instructional effects on the test results and to accurately reflect participants’ true syntactic parsing abilities. Additionally, eight sample sentences were practiced with the participants before the experiment. For scoring, correctly segmented sentences were assigned one point, while incorrectly segmented sentences received a score of 0. Incomplete or partially correct segmentations also received a score of 0. Examples of incorrect segmentations included placing slashes where they were not required or failing to separate phrases as dictated by the criteria. With a total of 80 sentences, participants’ scores ranged between 0 and 80.
The Reading Test (RT)
The reading test (RT) was administered in three distinct forms tailored to participants’ English language proficiency levels: Forms A, B, and C, corresponding to the beginner, intermediate, and advanced groups, respectively. Each form consisted of two reading passages followed by 20 multiple-choice questions. Passages, ranging in length from 550 to 833 words, were sourced from Active Skills for Reading 2-4 (Anderson, 2008, 2013). To tailor the forms, vocabulary complexity and sentence structures were adjusted to align with the respective proficiency levels. For instance, Form A incorporated simpler vocabulary and shorter sentences to accommodate beginner learners, while Form C included more complex structures suitable for advanced learners. The effectiveness of these adjustments was subsequently verified using the Flesch-Kincaid readability measure (Kincaid et al., 1975), which confirmed that Form A was the least difficult and Form C the most challenging in terms of text complexity. This approach aimed to minimize the influence of L2 knowledge on reading performance, particularly at lower proficiency levels.
Drawing on Day and Park’s (2005) taxonomies, the questions following each passage covered a range of items, including vocabulary, pronoun references, paraphrasing, positive and negative factual information, text purpose, and inferential understanding. Examples of these questions are provided in Appendix II. With respect to scoring, correct responses were awarded one mark, while incorrect and unanswered questions received no marks. To ensure the validity of the task, four English teachers with over ten years of EFL teaching experience verified the content and face validity using the content validity index. They assessed and provided feedback on the representativeness, simplicity, and clarity of the test items. Reliability coefficients, calculated using the Kuder–Richardson formula 20, ranged from 75% to 79% across the three forms, indicating acceptable levels of internal consistency. The final version of the task was piloted with 30 EFL learners who were similar to the actual participants in terms of proficiency level, educational background, and age, and the results indicated that the test completion time was approximately 30 minutes for most participants, regardless of their language proficiency.
Procedure
Data collection started with participants taking the IELTS mock test. All written sections (listening, reading, and writing) were administered on the same day, while the speaking section was conducted one week later. Participants then engaged in three subsequent testing sessions, each separated by one week. Initially, a reading span test was administered to measure participants’ working memory capacity. This test focused on assessing both the storage and processing aspects of WM. Subsequently, a paper-based sentence segmentation task was employed to measure participants’ syntactic parsing ability. In the final session, participants’ reading comprehension was assessed using a multiple-choice reading test (RT). The RT consisted of three forms—A, B, and C—tailored to the respective proficiency levels (beginner, intermediate, and advanced), with passages selected to ensure readability and comparability across proficiency groups. This structured approach to data collection aimed to systematically examine the relationships between working memory, syntactic parsing, and L2 reading performance across different proficiency levels among Persian EFL learners.
It is also important to note that this research is part of a broader investigation into the relationship between working memory, lower- and higher-level reading processes, and L2 reading performance. Both this study and a related investigation on semantic processing draw upon the same participant pool and dataset. This ensures consistency across related investigations while focusing specifically on the interplay between working memory, syntactic parsing, and reading comprehension in our current analysis.
Data Analysis
To address the research questions, the data underwent comprehensive analysis in four sections. First, descriptive statistics, including the mean of raw scores and standard deviations, were computed for all measures. Second, the normality and homogeneity assumptions of the variables were checked using Levene’s and Kolmogorov-Smirnov tests. The normality of the data was further verified using Skewedness and Kurtosis statistics (ranging between -2 and +2), ensuring robustness in the analysis. Upon confirming the normality of the data and considering the equal sample size, a one-way ANOVA was carried out to compare performance on the measures across the proficiency groups. Effect sizes were reported as η² and interpreted according to Plonsky and Oswald (2014). In the fourth section, correlations between working memory, syntactic parsing, and reading comprehension were explored. This analysis provided insights into the relationships among these key variables, shedding light on their interplay in the context of second language reading comprehension. Finally, multiple regression and dominance analyses (Mizumoto, 2023) were conducted to elucidate the extent to which reading performance could be predicted by WM and syntactic parsing ability across proficiency levels, and the relative importance of each within the model.
Prior to conducting the multiple regression analysis, all requisite assumptions were systematically examined to ensure the robustness of the statistical model. The assumption of linearity was confirmed through scatterplot inspection, which indicated a linear relationship between the independent and dependent variables. Additionally, the Durbin-Watson statistic was calculated, yielding values of 1.181 and 1.309, which fall within the acceptable range of 1.5 to 2.5, confirming no significant autocorrelation in the residuals. Homoscedasticity was assessed by plotting standardized residuals against predicted values, revealing a random scatter that met this condition. Moreover, to evaluate multicollinearity, Variance Inflation Factor (VIF) values were computed, which ranged from 1.108 to 1.380, well below the critical threshold of 5, indicating that the predictors contributed uniquely to the model. Lastly, the One-Sample Kolmogorov-Smirnov (K-S) test for normality of residuals returned a non-significant result (p = .200), suggesting an approximately normal distribution. Collectively, these diagnostic checks confirm that the assumptions underlying both multiple regression and dominance analysis were adequately met, thereby ensuring the validity and reliability of the findings.
Results
Descriptive Statistics
The descriptive statistics for memory capacity (RST), reading comprehension (RT), and overall syntactic parsing (SP) ability are presented in Table 2. The results suggest that while there appeared to be some differences in the RST amongst the groups, no statistically significant differences were found; F (2,117) = 1.553, p = 0.216, η² = 0.026. In contrast, significant differences with large effect sizes were observed in the groups’ performance on the IELTS test (F (2, 117) = 581.167, p < 0.001, η² = 0.909) and the reading test (RT) (F (2,117) = 29.236, p < 0.001, η² = 0.333), with the advanced group demonstrating superior performance. Similarly, significant differences with a large effect size were found for syntactic parsing; (F (2, 117) = 16.442, p < 0.001, η² = 0.219).
Table 2. Descriptive Statistics for Reading Span Test (RST), Syntactic Parsing (SP), and Reading Test (RT)
| RST | Overall SP | RT | |||||
| Groups | N | Mean | SD | Mean | SD | Mean | SD |
| Beginner | 40 | 16.3250 | 8.95970 | 47.200 | 11.440 | 12.4250 | 3.37325 |
| Intermediate | 40 | 14.9750 | 6.71198 | 54.600 | 11.000 | 15.2500 | 2.58943 |
| Advanced | 40 | 13.5000 | 5.38278 | 60.400 | 8.2206 | 16.8000 | 1.45355 |
| Total | 120 | 14.9333 | 7.20496 | 54.066 | 11.581 | 14.8250 | 3.15073 |
Confidence interval: 95%.
Based on the above results, more detailed comparisons were made for syntactic parsing. Detailed mean scores and standard deviations for syntactic parsing ability across sentence types (simple, compound, complex, and compound-complex) are provided in Appendix III and ANOVA results comparing the differences in specific types of syntactic parsing ability are presented in Table 3. These results suggest that while there were no significant differences in parsing simple and compound sentences, significant differences were found for the parsing of complex and compound-complex sentences, with the former exhibiting a large effect size and the latter showing medium effect size.
Table 3. ANOVA Results and Effect Sizes for Syntactic Parsing Ability (SP)
| Sum of Squares | df | Mean Square | F | Sig. | Effect Size () | ||
| SP | Between Groups | 3501.867 | 2 | 1750.933 | 16.442 | 0.000 | 0.219 |
| Within Groups | 12459.60 | 117 | 106.492 | ||||
| Total | 15961.46 | 119 | |||||
| Simple | Between Groups | 24.9500 | 2 | 12.475 | 2.965 | 0.055 | 0.048 |
| Within Groups | 492.250 | 117 | 4.207 | ||||
| Total | 517.200 | 119 | |||||
| Compound | Between Groups | 47.5170 | 2 | 23.758 | 1.427 | 0.244 | 0.024 |
| Within Groups | 1948.450 | 117 | 16.653 | ||||
| Total | 1995.967 | 119 | |||||
| Complex | Between Groups | 792.6170 | 2 | 396.308 | 24.595 | 0.000 | 0.296 |
| Within Groups | 1885.250 | 117 | 16.113 | ||||
| Total | 2677.867 | 119 | |||||
| Compound-Complex | Between Groups | 392.3170 | 2 | 196.158 | 9.331 | 0.000 | 0.138 |
| Within Groups | 2459.650 | 117 | 21.023 | ||||
| Total | 2851.967 | 119 | |||||
A post-hoc LSD test was conducted to perform multiple comparisons between groups regarding their overall syntactic parsing scores, as well as their ability to parse complex and compound-complex sentences, reported in Table 4. The results indicated significant differences in all levels.
Table 4. Post-hoc LSD Test Results for Syntactic Parsing Ability Across Sentence Types
| Dependent Variable | Overall SP | CX Sentences | CCX Sentences | ||||
| (I) Level | (J) Level | M-Diff (I-J) | M-Diff (Sig). | M-Diff (I-J) | M-Diff (Sig). | M-Diff (I-J) | M-Diff (Sig). |
| Beginner | Intermediate | -7.40** | 0.002 | -3.925*** | 0.000 | -2.05* | 0.048 |
| Advanced | -13.20*** | 0.000 | -6.225*** | 0.000 | -4.425*** | 0.000 | |
| Intermediate | Beginner | 7.40** | 0.002 | 3.925*** | 0.000 | 2.05* | 0.048 |
| Advanced | -5.80* | 0.013 | -2.30* | 0.012 | -2.375* | 0.022 | |
| Advanced | Beginner | 13.20*** | 0.000 | 6.225*** | 0.000 | 4.425*** | 0.000 |
| Intermediate | 5.80* | 0.013 | 2.3* | 0.012 | 2.375* | 0.022 | |
Note. SP = Syntactic Parsing; M-Diff= Mean Difference; CX = Complex Sentences; CCX = Compound-complex Sentences; Confidence interval: 95%.
Correlations
Table 5 shows the results of the Pearson correlation coefficients across the three proficiency levels. As shown in the table, in the beginner group, working memory (WM), as measured by the RST, exhibited significant correlations with L2 reading comprehension and all syntactic parsing (SP) abilities, including overall syntactic parsing and each sub-category of sentence types (i.e., simple, compound, complex, and compound-complex sentences). Specifically, WM was most strongly correlated with the ability to parse compound-complex sentences. Additionally, overall syntactic parsing ability was significantly correlated with L2 reading comprehension in this group, with notable correlations found across all sentence types.
Conversely, no significant correlations were observed between WM, syntactic parsing, and L2 reading comprehension in the intermediate and advanced groups. These results suggest that the relationship between WM, syntactic parsing, and reading comprehension is more pronounced in beginner learners, with diminishing influence at higher proficiency levels. Table 5 provides a detailed summary of the correlational analyses across proficiency groups.
Table 5. Correlations Between RST, SP, and RT Across Proficiency Groups
| RST | RT | SP | SM | CP | CX | CCX | ||
| RST | Beginner | — | .334* | .569*** | .334* | .482** | .332* | .635*** |
| Intermediate | — | -.226 | -.084 | .064 | -.071 | -.075 | -.051 | |
| Advanced | — | -.024 | .087 | .030 | -.076 | .046 | .213 | |
| RT | Beginner | — | .537*** | .366* | .357* | 0.403* | .351* | |
| Intermediate | — | .056 | -.217 | .219 | -.102 | .061 | ||
| Advanced | — | .043 | .004 | -.023 | .084 | .133 |
Note. RST = Reading Span Test; RT = Reading Test; SP = Overall Syntactic Parsing; SM = Simple Sentences; CP = Compound Sentences; CX = Complex Sentences; CCX = Compound-complex Sentences. Significance: * 05;** .01; *** .001
Regression Analysis
To address the last research question regarding the predictors of L2 reading comprehension, we first conducted a multiple linear regression analysis with proficiency, working memory (WM), and syntactic parsing (SP) as predictors of L2 reading performance (Table 6). The overall regression model was significant and accounted for 38.3% of the variance.
Table 6. The Significance of the Multiple Linear Regression Model for Variables Predicting L2 Reading.
| Dependent variable | R² | Adj. R² | df 1 | df 2 | F | p | ηp² [95% CI] |
| RT | .383 | .367 | 3 | 116 | 23.986 | .000 | .383 |
The multiple linear regression and dominance analyses (Mizumoto, 2023) revealed that proficiency was the most significant predictor of L2 reading comprehension, contributing 63.85% to the dominance weight of the model (Table 7). This suggests that proficiency has a substantial and dominant role in predicting L2 reading comprehension. Syntactic parsing (SP) was another significant predictor of L2 reading, contributing 35.38% to the dominance weight. Although SP significantly predicts L2 reading, its influence is less pronounced compared to proficiency. In contrast, working memory emerged as the least significant predictor of L2 reading, accounting for only 1.03% of the dominance weight. This indicates that WM’s role in predicting L2 reading comprehension is relatively minor compared to proficiency and syntactic parsing.
Table 7. Summary of Multiple Linear Regression Analysis for Variables Predicting L2 Reading
| Independent variables |
b | SE | 95% CI (B) | β [95%CI] |
t | p | Dominance weight [95%CI] (%) |
||
| Upper | Lower | Raw weight | Rescaled weight | ||||||
| Intercept | 7.236 | 1.150 | 4.959 | 9.514 | 6.293 | .000 | |||
| Proficiency | 1.766 | .329 | 1.113 | 2.418 | .459 | 5.361 | .000 | 0.249 | 63.85% |
| SP | .069 | .023 | .023 | .115 | .253 | 2.950 | .004 | 0.138 | 35.38% |
| WM | .023 | .034 | -.044 | .089 | .052 | .675 | .501 | 0.004 | 1.03% |
Note. b = unstandardized regression coefficient; SE = standard error; β = standardized regression coefficient; CI = confidence interval. Dominance was assessed using the methodology outlined by Mizumoto (2023).
Since correlations were significant between WM, SP, and L2 reading only in the beginner group (Table 5), we also conducted a multiple linear regression analysis with syntactic parsing (SP) and working memory (WM) as predictors of L2 reading performance in this group (Table 8). The regression model was significant (F(2,37) = 7.787, p < .001), accounting for approximately 29.6% of the variance in reading comprehension, with a substantial effect size (ηp² = .296).
Table 8. The Significance of the Multiple Linear Regression Model for Variables Predicting L2 Reading in the Beginner Group
| Dependent variable | R² | Adj. R² | df 1 | df 2 | F | p | ηp² [95% CI] |
| RT | .296 | .258 | 2 | 37 | 7.787 | .296 | .296 |
The multiple linear regression and dominance analyses (Table 9) revealed that syntactic parsing (SP) was a significant predictor of L2 reading comprehension in the beginner group (β = .463, t(37) = 2.608, p = .013), contributing 70.61% to the dominance weight. This indicates a substantial role of SP in predicting L2 reading comprehension within this group. Conversely, WM’s contribution to the dominance weight was only 29.72%. This suggests that while SP is a robust predictor of L2 reading comprehension for beginners, WM’s role is comparatively minor.
Table 9. Summary of Multiple Linear Regression Analysis for Variables Predicting L2 reading in the Beginner Group
| Independent variables |
b | SE | 95% CI (B) | β [95%CI] |
t | p | Dominance weight [95%CI] (%) |
||
| Upper | Lower | Raw | Rescaled | ||||||
| Intercept | 5.263 | 2.027 | 1.156 | 9.371 | 2.596 | .013 | |||
| SP | .136 | .052 | .030 | .242 | .463 | 2.608 | .013 | 0.209 | 70.61% |
| WM | .044 | .067 | -.091 | .180 | .117 | .661 | .513 | 0.088 | 29.72% |
Note. b = unstandardized regression coefficient; SE = standard error; β = standardized regression coefficient; CI = confidence interval. Dominance was assessed using the methodology outlined by Mizumoto (2023).
Overall, these findings emphasize the predominant role of syntactic parsing in predicting L2 reading comprehension among beginner learners, whereas working memory demonstrates limited predictive utility in this context.
Discussion
The present study aimed to investigate the relationships among working memory (WM), syntactic parsing ability (SP), and L2 reading performance across different proficiency levels. Our findings indicated significant differences in syntactic parsing and reading abilities across these proficiency groups, with advanced learners demonstrating superior performance on both the syntactic parsing measure and the reading test, while beginner learners exhibited the lowest scores in the same tasks. Correlational analyses revealed a noteworthy association between working memory capacity (WMC) and syntactic parsing ability (SP), specifically among less proficient learners, underscoring the importance of cognitive resources such as WM in syntactic processing. Additionally, a significant relationship emerged between syntactic parsing and L2 reading comprehension among less proficient readers.
These findings are pivotal as they provide empirical evidence supporting interactive models of reading comprehension, which posit that reading involves simultaneous higher- and lower-level processing. According to these models, failure at one processing level, such as inaccurate or inadequate syntactic parsing, can be compensated for by employing diverse strategies to construct a mental model and comprehend the text (Perfetti, 1985). However, the adoption of these reading strategies is contingent upon the capacity of readers’ WM. Syntactic parsing, particularly for less-proficient readers, demands substantial WM resources to analyze the syntactic structures of the text. Consequently, as highlighted by Nahatame (2020), less proficient readers allocate a significant amount of attentional resources to syntactic parsing, thereby leaving limited WM capacity for higher-level processing tasks, such as adopting reading strategies to inferentially comprehend the text.
In exploring linguistic and cognitive predictors of second language reading, this study also found that 29.72% of the variability in reading performance among less proficient readers can be attributed to their working memory capacity. Additionally, through multiple regression and dominance analyses, it was determined that syntactic parsing ability explains 70.61% of the variance observed in their reading performance. While WMC and syntactic parsing ability account for a substantial portion of the variability in L2 reading performance among lower-proficiency readers, other factors are implicated in predicting reading ability at higher proficiency levels (Burton & Daneman, 2007). These factors may include lower-level processing skills such as the ability to decode the individual words for their meanings and integrate word meanings with grammatical structures (Grabe, 2009), as well as higher-level processing skills of inference-making, comprehension monitoring, and goal-setting (Habók & Magyar 2019).
Building upon prior research emphasizing WM as a significant predictor of L2 reading ability (Demir & Erçetin, 2020; Jung, 2018), this study extends our understanding by revealing that WMC may also correlate with underlying processes of reading comprehension, particularly syntactic parsing ability. Furthermore, in line with previous investigations (Dussias & Pinar, 2010; Joh & Plakans, 2017), this study underscores the role of L2 proficiency as a crucial predictor of L2 reading. The findings also provide empirical support for two prominent theories of reading comprehension, the capacity-constrained reader model (Just & Carpenter, 1992) and the compensatory-encoding model (Walczyk, 1995), both highlighting the critical role of WM in effective text comprehension.
According to the capacity-constrained reader model (CC Reader), text comprehension results from simultaneous lower- and higher-level processing, which are significantly influenced by limitations in WMC (Liu et al., 2019). The model posits that there should be a balance between available cognitive resources and the demands for their activation. When the demand for cognitive processing exceeds available resources, such as in situations requiring syntactic parsing, WM resources are allocated to lower-level processing, making higher-level processing, like inferential understanding, challenging or even unachievable. This imbalance can slow down information processing during reading and lead to the forgetting of previously activated concepts (Miyake et al. 1994). The findings of this study support these theoretical insights, indicating that less proficient L2 learners expend a substantial portion of their cognitive resources on syntactic processing due to their limited automatization of L2 knowledge. Consequently, this leaves minimal WM capacity available for higher-level processing.
The findings of this study also lend empirical support to Walczyk’s (1995) compensatory-encoding model, which posits that efficient comprehension stems from automatic lower-level processing, with working memory (WM) playing a crucial role in reading comprehension (Grabe, 2009). Consistent with the findings of this study, the model presumes that proficient readers automatically process texts at a lower-level (e.g., through syntactic parsing), thereby conserving WM resources for higher-level tasks such as inference-making, strategic reading, and comprehension monitoring, resulting in enhanced reading performance. In contrast, less proficient readers lack automatization of lower-level processing skills such as syntactic parsing, necessitating greater reliance on WM resources for comprehension. This limitation leaves minimal WM capacity available for higher-level text processing, likely contributing to poorer reading outcomes.
By and large, the comprehension of written material in a second language (L2) involves intricate cognitive processes that vary significantly between low- and high-proficient learners. Overall, the findings of this study underscore the role of working memory and syntactic parsing in text comprehension, aligning closely with the multiple-processing model of L2 reading (Mahshanian, 2023). This model posits that while both low- and high-proficiency L2 learners can ultimately achieve comprehension, they employ different cognitive pathways (see Figure 1).
Fig. 1. The Multiple-processing Model of L2 Reading (Mahshanian, 2023)
Within this multiple-processing framework, high-proficient L2 learners primarily rely on higher-level processing skills, particularly comprehension monitoring (i.e., the conscious and strategic monitoring of one’s understanding while reading; Zargar et al., 2020), to achieve text comprehension. In contrast, low-proficient L2 learners predominantly depend on WM to process written information at a lower level, particularly through syntactic parsing. For less proficient learners, syntactic parsing may not be automatized, requiring significant cognitive resources to decode and interpret sentence structures. This reliance on WM for lower-level processing leaves fewer resources available for higher-level comprehension tasks, such as making inferences or integrating information across paragraphs (Walczyk, 1995; Liu et al., 2019). This latter premise of the multiple-processing model of L2 reading is supported by the findings of this study, indicating that working memory and syntactic parsing play substantial roles in fostering reading comprehension performance among low-proficiency L2 learners.
Conclusion, Implications, and Limitations
The findings of this study highlighted the significant relationship between working memory, syntactic parsing ability, and reading performance among less proficient readers. Specifically, the results indicated that L2 reading at lower proficiency levels is highly contingent on syntactic parsing skills. Inefficient reading comprehension in a second language may be partly due to L2 learners’ insufficient parsing abilities. Additionally, working memory demonstrated a strong correlation with syntactic parsing abilities among beginner learners. Unlike skilled readers, less proficient readers appear to rely more heavily on their WM to analyze sentence structures and comprehend texts.
A key contribution of this study lies in its methodological approach of conducting separate regression analyses across different proficiency levels, rather than assuming uniformity in the predictive value of these abilities across all learner groups. This nuanced approach acknowledges that the relative importance of cognitive and linguistic predictors, such as working memory (WM) and syntactic parsing, may vary substantially depending on learners’ proficiency levels.
This methodological differentiation provides a potential explanation for some of the inconsistencies observed in prior studies, which often employed singular models without accounting for proficiency-level variation. Specifically, discrepancies in previous findings may stem from differences in participant composition; studies with predominantly low-proficiency learners might emphasize the role of syntactic parsing and WM, whereas studies with higher-proficiency learners could underscore the influence of higher-level processes such as inference-making and comprehension monitoring. By contrast, this study highlights how the unique interplay between cognitive resources and linguistic skills shifts across proficiency levels, offering a more refined understanding of the factors contributing to L2 reading comprehension. Importantly, these insights can guide foreign language instructors, materials developers, and language testers in designing pedagogical approaches, instructional materials, and assessment tools that address the critical roles of WM and syntactic parsing in reading performance across different proficiency levels.
An important implication of this study pertains to the enhancement of WM and its potential impact on reading comprehension. A central question in WM research within language learning contexts is whether targeted WM training can enhance the reading comprehension abilities of less proficient L2 learners. Previous studies (Carretti et al., 2007; Ericsson & Kintsch, 1995; McNamara & Scott, 2001) have investigated this question, suggesting that improving working memory capacity among less proficient readers may significantly enhance certain aspects of reading comprehension (Carretti et al., 2009). Building on these findings, the present study establishes a robust relationship between WM and L2 reading comprehension at lower proficiency levels. It posits that enhancing WM capacity in poor comprehenders could potentially improve their ability to integrate disparate textual elements and retain relevant information, thereby facilitating a more comprehensive understanding of the text.
Furthermore, developers of educational materials and task designers should focus on providing less proficient L2 learners with reading materials and tasks designed to promote automatic syntactic parsing skills. By automating syntactic parsing, a foundational lower-level processing skill, less proficient learners can allocate their cognitive resources, such as working memory, to higher-level processing tasks, thereby enhancing their comprehension of written texts. Teaching strategies that emphasize the identification of cohesive textual units, such as phrases and sentences, through chunk learning techniques (Abney, 1991), can accelerate syntactic processing among lower-proficiency L2 learners and improve overall reading performance.
However, despite these educational implications, the findings of this study were constrained by several methodological limitations. First, syntactic parsing was operationally defined in this study as the ability to segment sentences based on predetermined criteria (see Section 2.2.3). Yet, parsing encompasses broader syntactic structures (e.g., null constituents, head movements, pied-piping, unaccusative predicates, feature deletion, A-movements, etc.) that were not included in our segmentation task. Therefore, while this study provides valuable insights into the interaction between working memory WM, syntactic parsing, and reading comprehension in second language (L2), further research is needed to explore how various types of syntactic structures relate to WM and reading comprehension.
Second, vocabulary knowledge, a well-established predictor of L2 reading comprehension (Bernhardt, 2005; Cho, 2015; Lockiewicz & Jaskulska, 2015), was not controlled for in our study. Research on glossing, which provides information on unfamiliar lexical items, has shown that vocabulary plays a crucial role in the relationship between WM and reading performance. For instance, Jung (2021) demonstrated that WM capacity moderates the effects of glossing on L2 reading among Korean learners. Future studies on WM and reading should consider controlling for participants’ vocabulary knowledge to enhance the generalizability of the results.
Third, our examination of lower-level reading processes was limited primarily to syntactic parsing. However, prior research emphasizes that other lower-level skills, such as word recognition, significantly influence reading comprehension (Grabe, 2009). Future investigations should therefore explore how WM interacts with these additional processing skills, such as word recognition and semantic encoding, to better understand their combined impact on L2 reading comprehension.
Finally, while the study included a sample of 120 Persian EFL learners, this sample size may limit the generalizability of the findings to other populations of EFL learners. A larger sample size across diverse language backgrounds would increase the robustness of the findings and their applicability to a broader range of L2 learners. Future research should aim to include a more diverse and larger sample to strengthen the external validity of the results.
About the Authors
Amir Mahshanian, Ph. D., is a lecturer in Applied Linguistics at the University of Isfahan, Iran. His research interests include second language acquisition, psycholinguistics, and language assessment. His professional work focuses on the cognitive and metacognitive aspects of second language acquisition. His special interest in eye-tracking reading research led him to propose a comprehensive working memory and reading model in his PhD dissertation in 2022. ORCID ID: 0009-0009-3678-3741
Mohammadtaghi Shahnazari, Ph.D. is an Associate Professor in Applied Linguistics at the University of Isfahan, Iran. His areas of expertise include, but are not restricted to, language acquisition and psycholinguistics. His main research interest is the links between working memory and learning a second language. He received his PhD degree in Applied Linguistics from Auckland University, New Zealand in 2012. ORCID ID: 0000-0002-6839-5307
Ahmad Moinzadeh, Ph.D. is an Associate Professor in Applied Linguistics at the University of Isfahan, Iran. His areas of expertise, among other things, are first and second-language syntax, first-language acquisition, and second-language learning. He received his PhD degree in Applied Linguistics from the University of Ottawa, Canada in 2000. ORCID ID: 0000-0002-5588-8751
To Cite this Article
Mahshanian, A., Shahnazari, M., & Moinzadeh, A. (2025). Does reading performance vary according to available working memory capacity and syntactic parsing skills? Teaching English as a Second Language Electronic Journal (TESL-EJ), 28(4). https://doi.org/10.55593/ej.28112a2
References
Abney, S. (1991). Principle-based parsing: Computation and psycholinguistics. Kluwer Academic Publishers.
Abu‐Rabia, S. (1995). Learning to read in Arabic: Reading, syntactic, orthographic and working memory skills in normally achieving and poor Arabic readers. Reading Psychology, 16(4), 351-394. https://doi.org/10.1080/0270271950160401
Anderson, N. J. (2013). Active skills for reading 3. Cengage Learning.
Anderson, N. (2008). Active skills for reading: Books 1-4. Heinle.
Alptekin, C., & Erçetin, G. (2010). The role of L1 and L2 working memory in literal and inferential comprehension in L2 reading. Journal of Research in Reading, 33(2), 206-219. https://doi.org/10.1111/j.1467-9817.2009.01412.x
Alptekin, C., & Erçetin, G. (2011). Effects of working memory capacity and content familiarity on literal and inferential comprehension in L2 reading. TESOL Quarterly, 45(2), 235-266. https://doi.org/10.5054/tq.2011.247705
Atai, M. R., & Nikuinezhad, F. (2012). Vocabulary breadth, depth, and syntactic knowledge: Which one is a stronger predictor of foreign language reading performance? Iranian Journal of Applied Linguistics (IJAL), 15(1), 1-18.
Baddeley, A. D. (2003). Working memory and language: an overview. Journal of Communication Disorders, 36 (3), 189–208. https://doi.org/10.1016/S0021-9924(03)00019-4
Baddeley, A. D., & Hitch, G. (1974). Working memory. Psychology of Learning and Motivation, 8, 47–89.
Brimo, D., Apel, K., & Fountain, T. (2017). Examining the contributions of syntactic awareness and syntactic knowledge to reading comprehension. Journal of Research in Reading, 40(1), 57-74. https://doi.org/10.1111/1467-9817.12050
Burton, C., & Daneman, M. (2007). Compensating for a limited working memory capacity during reading: Evidence from eye movements. Reading Psychology, 28(2), 163-186. https://doi.org/10.1080/02702710601186407
Carretti, B., Borella, E., Cornoldi, C., & De Beni, R. (2009). Role of working memory in explaining the performance of individuals with specific reading comprehension difficulties: A meta-analysis. Learning and Individual Differences, 19(2), 246-251. https://doi.org/10.1016/j.lindif.2008.10.002
Carretti, B., Borella, E., & De Beni, R. (2007). Does strategic memory training improve the working memory performance of younger and older adults? Experimental Psychology, 54(4), 311-320. https://doi.org/10.1027/1618-3169.54.4.311
Chang, X., Wang, P., Cai, M. M., & Wang, M. (2019). The predictive power of working memory on Chinese middle school students’ English reading comprehension. Reading & Writing Quarterly, 35(5), 458-472. https://doi.org/10.1080/10573569.2019.1579126
Chen, X., Li, H., & Gui, M. (2018). Instructional effects of syntactic parsing on Chinese college sudents’ EFL reading Rates. Journal of Education and Training Studies, 6(11), 176-185. https://doi.org/10.11114/jets.v6i11.3470
Choi, Y. H. (2015). Roles of L1 and L2 derivational morphological awareness in L2 reading through the mediation of L2 vocabulary knowledge. Journal of Asia TEFL, 12(3).
Chun, D. M., & Payne, J. S. (2004). What makes students click: Working memory and look-up behavior. System, 32(4), 481-503. https://doi.org/10.1016/j.system.2004.09.008
Conway, A. R., Kane, M. J., Bunting, M. F., Hambrick, D. Z., Wilhelm, O., & Engle, R. W. (2005). Working memory span tasks: A methodological review and user’s guide. Psychonomic Bulletin & Review, 12, 769-786. https://doi.org/10.3758/BF03196772
Cutting, L. E., & Scarborough, H. S. (2006). Prediction of reading comprehension: Relative contributions of word recognition, language proficiency, and other cognitive skills can depend on how comprehension is measured. Scientific Studies of Reading, 10(3), 277-299. https://doi.org/10.1207/s1532799xssr1003_5
Day, R. R., & Park, J. S. (2005). Developing reading comprehension questions. Reading in a Foreign Language, 17(1), 60–73.
Delage, H., & Frauenfelder, U. H. (2020). Relationship between working memory and complex syntax in children with Developmental Language Disorder. Journal of Child Language, 47(3), 600-632. https://doi.org/10.1017/S0305000919000722
Demir, B., & Ercetin, G. (2020). Exploring the role of proficiency as a mediator of the relationship between working memory and text comprehension in the L2. Turkish Online Journal of English Language Teaching, 5(3), 156–178.
Dussias, P. E., & Piñar, P. (2010). Effects of reading span and plausibility in the reanalysis of wh-gaps by Chinese-English second language speakers. Second Language Research, 26(4), 443-472. https://doi.org/10.1177/0267658310373326
Ericsson, K. A., & Kintsch, W. (1995). Long-term working memory. Psychological Review, 102(2), 211. https://doi.org/10.1037/0033-295X.102.2.211
Goldman, S. R., & Pellegrino, J. W. (2015). Research on learning and instruction: Implications for curriculum, instruction, and assessment. Policy Insights from the Behavioral and Brain Sciences, 2(1), 33-41. https://doi.org/10.1177/2372732215601866
Georgiou, G. K., & Das, J. P. (2016). What component of executive functions contributes to normal and impaired reading comprehension in young adults? Research in Developmental Disabilities, 49, 118-128. https://doi.org/10.1016/j.ridd.2015.12.001
Grabe, W. (2009). Reading in a second language: Moving from theory to practice. Cambridge University Press.
Grabe, W., & Stoller, F. L. (2011). Teaching and researching reading. Pearson Education.
Habók, A., & Magyar, A. (2019). The effects of EFL reading comprehension and certain learning-related factors on EFL learners’ reading strategy use. Cogent Education, 6(1), 1616522. https://doi.org/10.1080/2331186X.2019.1616522
Haft, S. L., Caballero, J. N., Tanaka, H., Zekelman, L., Cutting, L. E., Uchikoshi, Y., & Hoeft, F. (2019). Direct and indirect contributions of executive function to word decoding and reading comprehension in kindergarten. Learning and Individual Differences, 76, 101783. https://doi.org/10.1016/j.lindif.2019.101783
Hamatani, S. (2020). Construction of a Syntactic Processing Automaticity Measurement for Japanese EFL Learners. ARELE: Annual Review of English Language Education in Japan, 31, 145-160. https://doi.org/10.20581/arele.31.0_145
Harrington, M., & Sawyer, M. (1992). L2 working memory capacity and L2 reading skill. Studies in Second Language Acquisition, 14(1), 25-38. https://doi.org/10.1017/S0272263100010457
Hopewell, S. (2011). Leveraging bilingualism to accelerate English reading comprehension. International Journal of Bilingual Education and Bilingualism, 14(5), 603-620. https://doi.org/10.1080/13670050.2011.564274
Hopp, H. (2014). Working memory effects in the L2 processing of ambiguous relative clauses. Language Acquisition, 21(3), 250-278. https://doi.org/10.1080/10489223.2014.892943
Huang, L., Ouyang, J., & Jiang, J. (2022). The relationship of word processing with L2 reading comprehension and working memory: Insights from eye-tracking. Learning and Individual Differences, 95, 102143. https://doi.org/10.1016/j.lindif.2022.102143
Joh, J. (2018). Interplay of working memory, strategy use, and task difficulty in L2 reading comprehension. Journal of Asia TEFL, 15(4), 900. https://doi.org/10.18823/asiatefl.2018.15.4.1.900
Joh, J., & Plakans, L. (2017). Working memory in L2 reading comprehension: The influence of prior knowledge. System, 70, 107-120. https://doi.org/10.1016/j.system.2017.07.007
Johnson, R. E. (1970). Recall of prose as a function of the structural importance of the linguistic units. Journal of Verbal Learning and Verbal Behavior, 9(1), 12-20. https://doi.org/10.1016/S0022-5371(70)80003-2
Jung, J. (2018). Effects of task complexity and working memory capacity on L2 reading comprehension. System, 74, 21-37. https://doi.org/10.1016/j.system.2018.02.005
Jung, J. (2021). The role of glossing and working memory capacity in second language reading comprehension. Journal of Asia TEFL, 18(2), 438. http://dx.doi.org/10.18823/asiatefl.2021.18.2.4.438
Just, M. A., & Carpenter, P. A. (1992). A capacity theory of comprehension: individual differences in working memory. Psychological Review, 99(1), 122.
Kincaid, J. P., Fishburne, R. P., Rogers, R. L., & Chissom, B. S. (1975). Derivation of new readability formulas (automated readability index, fog count, and flesch reading ease formula) for Navy enlisted personnel. Research Branch Report 8–75. Chief of Naval Technical Training: Naval Air Station Memphis.
Koda, K. (2005). Insights into second language reading: A cross-linguistic approach. Cambridge University Press.
Le Fevre, D. M., Moore, D. W., & Wilkinson, I. A. (2003). Tape‐assisted reciprocal teaching: Cognitive bootstrapping for poor decoders. British Journal of Educational Psychology, 73(1), 37-58. https://doi.org/10.1348/000709903762869905
Lesser, M. J. (2007). Learner‐based factors in L2 reading comprehension and processing grammatical form: Topic familiarity and working memory. Language Learning, 57(2), 229-270. https://doi.org/10.1111/j.1467-9922.2007.00408.x
Linck, J. A., Osthus, P., Koeth, J. T., & Bunting, M. F. (2014). Working memory and second language comprehension and production: A meta-analysis. Psychonomic Bulletin and Review, 21(4), 861–883. https://doi.org/10.3758/s13423-013-0565-2
Linderholm, T., Cong, X., & Zhao, Q. (2008). Differences in low and high working-memory capacity readers’ cognitive and metacognitive processing patterns as a function of reading for different purposes. Reading Psychology, 29(1), 61-85. https://doi.org/10.1080/02702710701568587
Liu, X., Wang, W., & Wang, H. (2019). Age differences in the effect of animacy on Mandarin sentence processing. PeerJ, 7(e6437), 1-24. https://doi.org/10.7717/peerj.6437
Long, D. L., & Freed, E. M. (2021). An individual differences examination of the relation between reading processes and comprehension. Scientific Studies of Reading, 25(2), 104-122. https://doi.org/10.1080/10888438.2020.1748633
Lopez, A. L. (2008). To what extent does grammar knowledge account for competence in FL reading comprehension in university students? Revista española de lingüística aplicada, (21), 181-200.
Maftoon, P., & Tasnimi, M. (2014). Syntactic knowledge, vocabulary breadth, and metacognitive awareness of reading strategies in reading comprehension in self-regulated vs. non-self-regulated readers. English Language Teaching, 1(4), 37-61.
Mahshanian, A. (2023). The relationship between working memory, lower- and higher-level reading processes, and L2 reading ability. PhD Dissertation, University of Isfahan.
McNamara, D. S., & Scott, J. L. (2001). Working memory capacity and strategy use. Memory & Cognition, 29, 10-17. https://doi.org/10.3758/BF03195736
Miyake, A., & Friedman, N. P. (2013). Individual differences in second language proficiency: Working memory as language aptitude. In Foreign Language Learning (pp. 339-364). Psychology Press.
Miyake, A., Carpenter, P. A., & Just, M. A. (1994). A capacity approach to syntactic comprehension disorders: Making normal adults perform like aphasic patients. Cognitive Neuropsychology, 11(6), 671-717. https://doi.org/10.1080/02643299408251989
Mizumoto, A. (2023). Calculating the relative importance of multiple regression predictor variables using dominance analysis and random forests. Language Learning, 73(1), 161-196. https://doi.org/10.1111/lang.12518
Nahatame, S. (2020). Revisiting second language readers’ memory for narrative texts: The role of causal and semantic text relations. Reading Psychology, 41(8), 753-777. https://doi.org/10.1080/02702711.2020.1768986
Nergis, A. (2013). Exploring the factors that affect reading comprehension of EAP learners. Journal of English for Academic Purposes, 12(1), 1-9. https://doi.org/10.1016/j.jeap.2012.09.001
Perfetti, C. (1985). Reading Ability. Oxford University Press.
Plonsky, L., & Oswald, F. L. (2014). How big is “big”? Interpreting effect sizes in L2 research. Language Learning, 64(4), 878-912. https://doi.org/10.1111/lang.12079
Radford, A. (2009). An introduction to English sentence structure. Cambridge University Press.
Reichle, E. D., Carpenter, P. A., & Just, M. A. (2000). The neural bases of strategy and skill in sentence–picture verification. Cognitive Psychology, 40(4), 261-295. https://doi.org/10.1006/cogp.2000.0733
Reynolds, R. E. (2000). Attentional resource emancipation: Toward understanding the interaction of word identification and comprehension processes in reading. Scientific Studies of Reading, 4(3), 169-195. https://doi.org/10.1207/S1532799XSSR0403_1
Sagarra, N. (2007). Online processing of gender agreement in low proficient English–Spanish late bilinguals. In M. J. Cabrera, J. Camacho, V. Deprez, N. Flores, & L. Sanchez (Eds.), Current issues in linguistic theory (pp. 240–253). John Benjamins.
Sagarra, N., & Herschensohn, J. (2010). The role of proficiency and working memory in gender and number agreement processing in L1 and L2 Spanish. Lingua, 120(8), 2022-2039. https://doi.org/10.1016/j.lingua.2010.02.004
Shahnazari, M. (2023). The role of WM in second language reading comprehension: Does L2 proficiency level matter? Learning and Motivation, 82, 101875. https://doi.org/10.1016/j.lmot.2023.101875
Shin, J. (2020). A meta-analysis of the relationship between working memory and second language reading comprehension: Does task type matter? Applied Psycholinguistics, 41(4), 873-900. https://doi.org/10.1017/S0142716420000272
Shiotsu, T., & Weir, C. J. (2007). The relative significance of syntactic knowledge and vocabulary breadth in the prediction of reading comprehension test performance. Language Testing, 24(1), 99-128. https://doi.org/10.1177/026553220707151
Shohamy, E. (1984). Does the testing method make a difference? The case of reading comprehension. Language Testing, 1(2), 147-170. https://doi.org/10.1177/02655322840010020
Skorobogatova, A. S., Henriques, A. S., Ruseishvili, S., Sekerina, I., & Madureira, S. (2021). Verbal working memory assessment in Russian-Brazilian Portuguese bilinguals. Cadernos de Linguística, 2(4), 1-24. https://doi.org/10.25189/2675-4916.2021.V2.N4.ID572
Stanovich, K. (2000). Progress in understanding reading: Scientific foundations and new frontiers. Guilford Press.
Taşçı, S., & Turan, Ü. D. (2020). The contribution of lexical breadth, lexical depth, and syntactic knowledge to L2 reading comprehension across different L2 reading proficiency groups. English Teaching & Learning, 45(2), 145-165. https://doi.org/10.1007/s42321-020-00065-z
Taşçı, S., & Turan, Ü. D. (2021). The contribution of syntactic and lexical knowledge to foreign language reading comprehension. Afyon Kocatepe Üniversitesi Sosyal Bilimler Dergisi, 23(1), 97-112. https://doi.org/10.32709/akusosbil.850830
Torrens, V., & Yagüe, E. (2018). The role of phonological working memory in children with SLI. Language Acquisition, 25(1), 102-117. https://doi.org/10.1080/10489223.2016.1187617
Ulijn, J. M., & Strother, J. B. (1990). The effect of syntactic simplification on reading EST texts as L1 and L2. Journal of Research in Reading, 13(1), 38-54. https://doi.org/10.1111/j.1467-9817.1990.tb00321.x
Walczyk, J. J. (1995). Testing a compensatory-encoding model. Reading Research Quarterly, 30(4), 396-408. https://doi.org/10.2307/747623
Yamashita, J., & Ichikawa, S. (2010). Examining reading fluency in a foreign language: effects of text segmentation on L2 readers. Reading in a Foreign Language, 22, 263-283.
Yano, Y., Long, M. H., & Ross, S. (1994). The effects of simplified and elaborated texts on foreign language reading. Language Learning, 44(2), 189–219. https://doi.org/10.1111/j.1467-1770.1994.tb01100.x
Zargar, E., Adams, A. M., & Connor, C. M. (2020). The relations between children’s comprehension monitoring and their reading comprehension and vocabulary knowledge: an eye-movement study. Reading and Writing, 33(3), 511-545. https://doi.org/10.1007/s11145-019-09966-7
Appendices
Appendix I. Distribution of Participants in Groups Based on University Major, Gender, and Age
| Gender and Age | Majors | |||||||||
| Groups | N | Male | Female | Age | Translation Studies |
English Literature |
Medicine | Nursing | Pharmacy | Dentistry |
| Beginner | 40 | 25 | 15 | 19-21 | 6 | 5 | 7 | 8 | 7 | 7 |
| Intermediate | 40 | 23 | 17 | 19-21 | 8 | 7 | 6 | 6 | 6 | 7 |
| Advanced | 40 | 18 | 22 | 19-22 | 8 | 9 | 7 | 5 | 5 | 6 |
| Total | 120 | 66 | 54 | 19-22 | 22 | 21 | 20 | 19 | 18 | 20 |
[back]
Appendix II. Examples of Questions Following Passages in the Reading Test (RT)
| Question Type | Examples |
| Vocabulary | The word plight in the 2nd paragraph is closest in meaning to which of the following? |
| Pronoun references | The pronoun those in the last paragraph refers to which of the following items? |
| Paraphrasing | What do the Arabian oryx, gray wolf, and whales have in common? |
| Positive Factual Information | Which animal species benefitted from a captive breeding program? |
| Negative factual information | Which is NOT mentioned as a threat to the animal populations in the passage? |
| Text purpose | What was the author’s purpose in writing this article? |
| Inferential Understanding | Which position regarding the ban on hunting would the author probably support? |
[back]
Appendix III. Mean and Standard Deviations of Syntactic Parsing Ability based on Sentence Type
| Overall SP | Simple | Compound | Complex | Comp-Complex | |||||||
| N | Mean | SD | Mean | SD | Mean | SD | Mean | SD | Mean | SD | |
| Beginner | 40 | 47.200 | 11.440 | 16.075 | 2.4639 | 12.075 | 4.3582 | 8.7500 | 4.6118 | 10.525 | 4.5063 |
| Intermediate | 40 | 54.600 | 11.000 | 16.875 | 2.0654 | 12.425 | 4.1441 | 12.675 | 4.0723 | 12.575 | 4.6623 |
| Advanced | 40 | 60.400 | 8.2206 | 17.150 | 1.5114 | 13.550 | 3.7138 | 14.975 | 3.2382 | 14.950 | 4.5850 |
| Total | 120 | 54.066 | 11.581 | 16.700 | 2.0847 | 12.683 | 4.0954 | 12.133 | 4.7437 | 12.683 | 4.8955 |
Note. SP = Syntactic Parsing; SD = Standard Deviation; Comp-Complex: Compound-complex; Confidence interval: 95%. [back]
| Copyright of articles rests with the authors. Please cite TESL-EJ appropriately. Editor’s Note: The HTML version contains no page numbers. Please use the PDF version of this article for citations. |