Abstract
Purpose
This study aims to clarify the construct of literature review (LR) writing skills and examine its impact on research proposal (RP) writing performance among postgraduate students in Hong Kong SAR, China.
Design/Approach/Methods
A total of 286 postgraduate students completed an LR writing task and an RP writing task within a Chinese academic writing course. Twelve skill variables were analyzed using exploratory factor analysis (EFA) to define the LR writing skill construct. Latent profile analysis (LPA) was employed to categorize students into skill-based profiles and evaluate their influence on RP writing performance.
Findings
The EFA identified four key factors underlying LR writing skills: Citation and Synthesis, Research Question Elicitation, Language Expression and Organization, and Contextual Awareness. The LPA classified students into two groups: students with low LR writing skills and students with moderate-to-high LR writing skills. Students in the moderate-to-high skill group performed significantly better in RP writing than their low-skill counterparts.
Originality/Value
This study provides a comprehensive framework for understanding the construct of LR writing skills and highlights its practical implications for academic writing instruction. By identifying skill-based student profiles and their relationship with RP writing performance, the findings offer insights for designing targeted pedagogical interventions to enhance LR and overall academic writing skills.
Introduction
The literature review (LR) plays a critical role in demonstrating the author's understanding of current research (Nelson & King, 2023; Paltridge, 2004). Within higher education, LR writing holds importance in the instruction of academic language and research methods. Beyond serving as a requisite component in academic writing, the LR contributes to nurturing students’ comprehension and interest in actively participating in academic research (Saeed et al., 2021).
Despite the wealth of knowledge derived from previous studies on LR writing, particularly concerning text quality and challenges encountered by students during the writing process (e.g., Boote & Beile, 2005; Chen et al., 2016; Cooper, 1985), there remains a need for deeper exploration into the underlying skills shaping LR writing performance. Existing research has predominantly focused on discrete skills such as opinion generation and language use (e.g., Badenhorst, 2019; Hu & Gao, 2018), leaving the overall construct of LR writing skills inadequately understood. Moreover, the impact of LR writing skills on broader academic writing, such as research proposals and dissertations, has been acknowledged (Chatterjee-Padmanabhan & Nielsen, 2018; Saeed et al., 2021), yet a comprehensive investigation into this area is lacking.
Furthermore, the majority of existing research in this domain has been conducted in English learning contexts, predominantly utilizing qualitative methodologies. To address these gaps, this study endeavors to elucidate the construct of LR writing skills within the context of Chinese language learning through exploratory factor analysis (EFA). Additionally, it seeks to differentiate between LR student writers employing a person-centered approach, namely latent profile analysis (LPA). By exploring potential variations in academic writing tasks, such as research proposals, among students with differing levels of LR writing skills, this research aims to establish a connection between LR writing skills and overall performance in academic writing across genres, including research proposals. In essence, the outcomes of this research endeavor are anticipated to enrich the current understanding of LR writing by encompassing participants from diverse linguistic contexts. Moreover, the findings are poised to inform instructional strategies for LR writing in both first and second language English contexts.
Literature Review
Previous Research on LR Writing
Previous LR writing studies have focused mostly on two aspects of students’ writing performance: text quality and challenges that students encounter during the writing process. Regarding text quality of LR writing, existing literature has documented the shift in focus from the normative nature of the writing procedure (including planning, structuring and writing, and dissemination) to the discursive formation of writing content (e.g., Boote & Beile, 2005; Cooper, 1985), the disciplinary characteristics of LR writing (e.g., Bolderston, 2008; Paul & Criado, 2020), and different types of review articles including systematic reviews, meta-analyses, and integrative reviews (e.g., Paul et al., 2021; Snyder, 2019).
Regarding the writing challenges, research has indicated that students were most likely to be trapped in linguistic challenges (e.g., Bitchener & Basturkmen, 2006; Qian & Krugly-Smolska, 2008; Sarid et al., 2021), cognitive challenges (e.g., Badenhorst, 2018; Chen et al., 2016; Sun et al., 2022), and ontological challenges (e.g., Kamler & Thomson, 2008; Sun et al., 2022; Warburton & Macauley, 2014). These challenges may hinder students from applying language conventions, conducting complex cognitive processing, and being proactive in writing a complete LR. Despite their insightful findings on understanding the progress and needs of students learning LR writing, these studies have barely touched on the construct of LR writing skills.
LR Writing Skills
To comprehensively examine the requisite writing skills for LR writing, our literature search was expanded to encompass various educational fields—including language, business, medical, and justice education—as these areas provide diverse perspectives on LR writing practices.
By studying the complex writing process from input to output, researchers have identified various writing skills that support LR composition. The skill that has received the most attention is discourse synthesis, which describes the process wherein students undergo “selection, organization, and connection” when processing input literature sources (Golparvar & Khafi, 2021; Nelson & King, 2023; Segev-Miller, 2007). This skill involves selecting appropriate literature on a specific topic, as well as summarizing and synthesizing ideas within and across multiple selected literature sources (Badenhorst, 2018, 2019; Jaidka et al., 2013; Nelson & King, 2023; Numrich & Kennedy, 2017). Indeed, LR writing is a kind of integrated writing (IW), which requires the skill of discourse synthesis (Spivey, 1997; Vandermeulen et al., 2023). An empirical study using hierarchical regression analysis revealed that discourse synthesis skill accounted for 63.6% of the overall variance in L1 Chinese IW performance (Zhu et al., 2021). Moreover, referencing management (e.g., applying consistent reference formats and styles in the text) is also crucial for literature source use (Badenhorst, 2019; Bell, 2010; Numrich & Kennedy, 2017), and reflects student writers’ awareness of keeping academic conventions and avoiding the plagiarism problem.
After processing the input literature sources, students engage in the output production phase of LR writing through opinion generation and written text development. The skill of opinion generation involves commenting on and identifying research gaps in previous research, as well as raising students’ own research questions (Chen & Li, 2019; Hunt et al., 2018; Kwan, 2006; Paul & Criado, 2020). The key objectives are to critically review current research progress, point out deficiencies, and underline relevance to one's own research (Kwan, 2006; Xie, 2017). In other words, students should maintain specificity and alignment between the current research gap and their own research. As the other aspect of output production, written text development comprises the linguistic features of LR writing, including the language use and text organization (Chen et al., 2016; Rabie & Boraie, 2021; Randolph, 2009). It is also important to employ an appropriate title as well as appropriate headings and subheadings throughout the text organization to suggest a clear research agenda (Denney & Tewksbury, 2013; Paul & Criado, 2020).
Besides, academic writing highlights the research context, both theoretical and practical, to situate the research on a solid basis (Boote & Beile, 2005). Recent research has underscored the significance of establishing research context as an important aspect of writing skills in LR writing (Boote & Beile, 2005; Hart, 2018; Kwan, 2006; Snyder, 2019). Specifically, students should be able to establish a literature-based theoretical background for their own research and connect the research to practical issues (Hart, 2018; Kwan, 2006; Snyder, 2019).
While the previously discussed writing skills have undergone individual analyses across different studies, they have been treated as discrete entities. Consequently, there exists a notable gap in empirical research that comprehensively examines and integrates these skills into a cohesive skill construct. The research on integrated writing (IW) has laid the groundwork for exploring such a skill construct, providing valuable insights (e.g., Cheong et al., 2019; Zhu et al., 2021; Zhu et al., 2024; Zhu et al., 2016). In a pivotal study, Zhu et al. (2016) identified 10 writing skill variables, such as identification, synthesis, argument, and organization, drawing from prior IW-related research to analyze the construct of IW writing skills. The outcomes of this analysis revealed that these variables could be consolidated into four underlying skills within the construct. This operationalization serves as a guiding framework for the present study.
Based on our review of LR writing research, we extracted various skills supporting the entire writing process from input to output. Specifically, prior research has identified skills associated with LR composition, with the input phase involving selecting appropriate literature, summarizing, synthesizing ideas, and embedding referencing. In the output phase, two sub-processes are discernible: opinion generation—which further includes commenting on current research, identifying gaps, and raising research questions—and written text development, which involves titling, language expression, and text organization. Additionally, previous studies have emphasized the importance of both theoretical and practical contexts in establishing a robust foundation, further justifying the inclusion of these skill sets.
In total, 12 LR writing skill variables were extracted, as outlined in Table 1. Subsequently, we embarked on the exploration of how these skill variables could be amalgamated into specific skills, thereby forming the skill construct for LR writing.
12 LR Writing Skill Variables.
LR Writing Skills and Academic Writing Performance
The relationship between literature review (LR) writing skills and overall academic writing performance is of paramount significance, given the integral role that LR writing plays in academic endeavors, including research proposals and dissertations (Chatterjee-Padmanabhan & Nielsen, 2018; Saeed et al., 2021; Sun et al., 2022). Notably, Chatterjee-Padmanabhan and Nielsen (2018) conducted a qualitative case study examining the challenges faced by doctoral students in crafting a research proposal. The findings suggested that a poorly executed LR could signal potential deficiencies in other sections, such as the methodology, within the research proposal. This insight implies that students struggling in LR writing may encounter difficulties in establishing their scholarly voice in academic writing. One possible explanation is that there are several shared characteristics between LR writing and academic writing, such as research proposal writing. Scholars have underscored various writing skills essential for effective research proposal composition, encompassing the extraction and synthesis of ideas from source materials, articulating critical arguments, utilizing academic language, and structuring the writing cohesively (Hyland, 2017; Knoch & Sitajalabhorn, 2013). All these skills are extremely akin to the above-identified LR writing skill variables. Therefore, it is anticipated that a positive association exists between LR writing skills and the performance of academic writing, such as research proposals.
To get a closer look at the relationship between students’ varied LR writing skills and academic writing performance, a person-centered approach (e.g., latent profile analysis, LPA) may be considered. Unlike variable-centered analysis (e.g., correlations and regression), LPA is suitable for “questions that concern group or individual differences in patterns of development” (Laursen & Hoff, 2006, p. 379). This approach has been employed in existing research on writing education to explore the nuances of students’ writing competence (e.g., Yao et al., 2021; Yung & Cai, 2020). For instance, Yung and Cai (2020) identified four profiles of academic writing performers (i.e., complex, high, medium, and low) among a sample of 917 university students in Hong Kong SAR, China. The results showed that students with different profiles varied in different aspects (i.e., citation, argument, language, and structure) of academic writing performance. This study suggested that researchers could use LPA to classify students into distinct profiles and then investigate their diverse performance in academic writing.
Overview of the Literature and the Present Study
Thus far, we have drawn three major findings from the extant literature on LR writing. First, many studies have explored the text quality of LR writing and the challenges encountered when writing an LR (e.g., Boote & Beile, 2005; Chen et al., 2016). However, most of these studies focused on undergraduate or doctoral students’ LR writing in English as a first or second language with qualitative research approaches (e.g., case studies and interviews). It is worth noting that research on LR writing in the Chinese language remains significantly underrepresented within this field of study. In comparison, empirical research on LR writing in a Chinese context is notably less extensive than that conducted in English contexts. Existing studies within the Chinese context have primarily concentrated on aspects such as the extent to which students are proficient in LR writing (e.g., Chu et al., 2021). However, there remains a notable research gap in the assessment of LR writing skills within the context of L1 Chinese writing.
Second, existing research has documented several disparate LR writing skills (e.g., Badenhorst, 2019), without a comprehensive analysis of these skills. It is warranted to examine the LR writing skill construct based on the identified variables to provide a more comprehensive view of LR writing skills. Third, previous research has reported an association between students’ LR writing skills and their overall academic writing performance, but these studies mainly used qualitative research approaches with relatively small samples (e.g., Chatterjee-Padmanabhan & Nielsen, 2018). Quantitative research with a relatively large sample is therefore warranted to provide more compelling evidence for their association. In addition, LPA may serve as an effective method for analyzing students’ nuances in LR writing skills, as well as their impacts on academic writing performance.
Given this, the present study, contextualized within Chinese-language education at a university in Hong Kong SAR, China, aims to: (a) explore the construct of LR writing skills, (b) delve into the profiles of students’ LR writing skills, and (c) examine the differences in academic writing performance as a function of the different profiles.
Building upon an extensive body of research primarily conducted in the context of English writing, it is imperative to recognize that academic writing is a universal evaluative practice—a means to demonstrate disciplinary knowledge and proficiency in critical thinking, interpretation, and presentation (Irvin, 2010), regardless of linguistic background. This universality suggests that insights garnered from English writing studies can be effectively applied to the Chinese academic context despite language differences. Moreover, because academic writing frameworks typically do not distinguish between native and non-native contexts, findings derived from Chinese L1 writing can, in turn, offer valuable perspectives for refining LR writing strategies in both L1 and L2 English settings.
Overall, the following research questions guided the study:
What constitutes the construct of literature review (LR) writing skills? What distinct profiles emerge concerning LR writing skills? To what extent do students with different profiles differ in academic writing performance?
Methods
Participants
The participants were 286 postgraduate students (male = 89, female = 197; mean age = 24.63, SD = 1.04) in language communication sciences from a research-intensive university in Hong Kong SAR, China. They were L1 Chinese speakers enrolled in a Chinese language academic writing course. The course aimed to provide students with systematic training on research methodology and academic writing in Chinese language teaching and language studies over 39 hr in one semester. The students’ participation in this study was voluntary and uncompensated, and written informed consent was obtained from all participants. The research project was approved by the institutional research ethics board of The Hong Kong Polytechnic University (PolyU), Hong Kong SAR, China.
Instruments
LR Writing Task and Scoring Rubrics
The purpose of the LR writing task was to assess the students’ LR writing performance. The participants were allowed to choose any topic in the field of Chinese language teaching and language studies and select 6–10 relevant academic papers in consideration of their own research interests and academic backgrounds. The entire LR was less than 5,000 Chinese characters in length, including references but excluding appendices. The participants were required to submit their completed LRs to an online platform in the seventh week of the semester.
The five-level scoring rubrics for the LR writing task were developed based on the above-formed 12 skill variables (see Appendix A). Detailed descriptions of each level were provided, with each level ranging across two marks. This design was adapted from Cheong et al. (2019) and had the advantage of displaying varying performance levels in detail through intensive mark distribution. There were five levels for each variable, from the highest level (9–10 marks) to the lowest level (1–2 marks). Thus, the highest possible total score of the 12 variables was 120 (12*10). A pilot study was conducted with 10 participants. The scoring rubric was amended based on the pilot outcomes and experts’ suggestions. For example, several descriptors (e.g., synthesis and titling) were rephrased for clarity.
RP Writing Task and Scoring Rubrics
It is worth noting that the RP writing task was utilized as an academic writing task, except for the LR writing task, in the study for two reasons. First, the RP is commonly employed as a regular academic writing task in postgraduate language education in Hong Kong SAR, China (e.g., Yang et al., 2020). Second, the RP, as a research preparation, assists students in obtaining permission to begin work on a thesis and determining the direction in which to proceed (Wang & Yang, 2012). The participants were allowed to choose any topic within the field of Chinese language teaching and language studies, but the topic could not be the same as the one for the LR writing task. The RPs were fewer than 6,000 Chinese characters, including references but excluding appendices, and were required to be submitted to an online platform in the last week of the semester.
The scoring rubric for the RP writing task (see Appendix B), adapted from Saeed et al. (2021) and Wang and Yang (2012), measured students’ RP writing performance in three dimensions: (a) knowledge of the subject (20 marks); (b) research design and methods (20 marks); and (c) language expression and structure (10 marks). Each dimension was further divided into five fine-grained levels; each level ranged across four marks for the first two dimensions and two marks for the last dimension. The highest possible total score was 50. A pilot study was conducted with 10 participants, distinct from those in the LR writing task pilot. Based on the outcomes and expert feedback, the scoring rubric descriptions (e.g., for “knowledge of the subject”) were amended to enhance clarity.
Data Collection
This study was conducted over a 13-week semester. In week 1, the course lecturer conducted a 120-min session to introduce the course to students before their involvement in the LR and RP tasks. The session: (a) outlined the course objectives and teaching plans; (b) introduced the LR and RP writing tasks, including the requirements, scoring rubrics, and submission deadlines (week 7 and week 12, respectively); and (c) organized a classroom discussion on the two tasks and provided clarifications in case of confusion.
Furthermore, specific instructions on how to write a decent LR were provided in week 2. The course lecturer introduced the components of a literature review and corresponding writing skills, covering the entire writing process from information input to output production. Following this, the lecturer organized a small-group activity (4–5 students per group) in which groups were asked to evaluate an exemplar literature review and exchange their interpretations of the scoring criteria and their understanding of literature review writing. Subsequently, the lecturer conducted a whole-class Q&A session, encouraging students to ask questions about any aspects of literature review writing or its scoring criteria that were unclear. After these activities, students were presented with an article about LR writing, titled “The framework of multiple functions for literature reviews in empirical papers: Taking Chinese language integrated assessment research as an example” (Zhu & Lu, 2020). Students were recommended to refer to this article while completing their LR tasks. Data collection for the study was confirmed and approved by the University Research Ethics Board (Institutional Review Board).
Scoring
Two raters were recruited to score the LR and RP writing tasks based on the scoring rubrics. Both raters were native Chinese speakers with over 10 years of experience in teaching academic writing. Before scoring, to ensure that both raters used the rubrics accurately and consistently, a meeting was organized in which a thorough discussion on the two tasks and rubrics, as well as their design rationales, was conducted. The analytic rubrics were used because doing so enabled students’ writing performance to be meaningfully broken down into distinct components. To give the raters scoring practice, a trial scoring of 15 randomly selected scripts (both LRs and RPs) was conducted and the results were reviewed. The raters’ rating discrepancies were discussed until a consensus on the scoring was reached. Based on the trial results, the content and wording of the rubrics were adjusted slightly.
In the actual scoring, the LR writing scripts were scored first, followed by the RP writing scripts two weeks later. All LR and RP writing scripts were scored independently by the two raters, and the scores using whole numbers assigned by both raters were summed and averaged. When the scores assigned by the two raters differed by two marks or above for the LR variables, and four marks or above for the first two dimensions and two marks or above for the last dimension of RP, a third rater was involved, whose score was summed and averaged with the closest of the two scores assigned by the original raters. The inter-rater reliability (the intraclass correlation coefficient) for the 12 LR writing skill variables ranged from 0.74 to 0.94 and was 0.93 for the total score of the LR writing performance. The inter-rater reliability for the three RP writing dimensions ranged from 0.76 to 0.95 and was 0.94 for the total score of the RP writing performance. Cronbach's alpha for the LR writing task (with 12 variables) was a high 0.90, and for the RP writing task (with three dimensions) was a moderate 0.72. Both values were within an acceptable range.
Statistical Analyses
In the first stage, SPSS version 25.0 was used to calculate descriptive statistics (i.e., mean, standard deviation, skewness, and kurtosis) for further analysis. Subsequently, exploratory factor analysis (EFA) was conducted to explore the latent factor structure of LR writing skills comprising the 12 variables.
Finally, Mplus version 8.3 was used to perform LPA to categorize students into different profiles according to their diverse mean scores on the LR writing skill factors identified by EFA. In order to determine the optimal number of latent classes, several fit indices were adopted, including log-likelihood (LL), Akaike information criterion (AIC), Bayesian information criterion (BIC), sample size adjusted BIC (ABIC), Lo-Mendell-Rubin likelihood ratio test (LMRT), bootstrap likelihood ratio test (BLRT), and entropy. Lower absolute values for LL, AIC, BIC, and ABIC indicate a better model data fit. More importantly, researchers should be mindful of the fact that the values of certain indices will generally decrease as the number of profiles increases; when the decrease turns marginal, it suggests that adding a new profile provides little information about profile membership (Yung & Cai, 2020). Significant p values of LMRT and BLRT (<.05) indicate that the new model (e.g., the 3-profile model) fits the data better than the old model (e.g., the 2-profile model) (Nylund et al., 2007). An entropy value larger than 0.80 suggests a good classification quality with more than 90% precision (Lubke & Muthén, 2007). Further, Wald chi-square tests were performed using the BCH command implemented in Mplus to examine the differences in students’ LR writing skills and RP task performance across the identified profiles.
Results
Descriptive Statistics
The descriptive statistics are displayed in Table 2. The mean scores of the LR writing skill variables ranged from 4.37 (LR7 Research question) to 6.35 (LR1 Selection) out of 10. The mean task performance scores were 72.76 out of 120 for LR, and 32.83 out of 50 for RP. In addition, the absolute values of skewness and kurtosis were below 1, suggesting normal distributions of the variables (Kline, 2023).
Descriptive Statistics of Primary Variables.
EFA Results
EFA was conducted to identify the factor structure of LR writing skills. The adequacy of the sample and suitability of the data for EFA were then examined. The Kaiser–Meyer–Olkin (KMO) value was 0.90, exceeding the commonly suggested value of 0.6 (Hair et al., 2014), and Bartlett's test of sphericity was significant (χ2 = 1862.13, df = 66, p < .001). Generally, the data were deemed to be adequate for factor analysis. Therefore, principal components analysis with varimax rotation was conducted. The two initially extracted factors explained 60.88% of the total variance, but the two-factor model was difficult to interpret. A closer look at the initial eigenvalues shows that they were 6.118 and 1.187 for the first two factors, and 0.825 and 0.790 for the third and fourth factors, respectively. According to Jolliffe (1972, 1986), factors with an eigenvalue greater than 0.7 are worth retaining. Furthermore, the third and fourth factors added important elements of variance (i.e., 6.87% and 6.59%, respectively) to the total variance. For these reasons, we proceeded to explore a four-factor structure.
After stipulating the extracted number of factors to be four, four factors were ultimately extracted, and they explained 74.33% of the total variance. The four factors accounted for 50.98%, 9.90%, 6.87%, and 6.59% of the variance, respectively. The loadings of the variables on each factor are presented in Table 3. The results show that all variables had high loadings (i.e., absolute values were higher than 0.45) on the primary factors, ranging from 0.505 to 0.837 without any cross-loading variables (Hair et al., 2014). The variables, hence, fit their respective factors well. Cronbach's alpha coefficients were calculated to examine the internal consistency of each factor. The values for factors 1 to 4 were 0.85, 0.70, 0.82, and 0.65, respectively, larger than the acceptable threshold (0.60) in explorative research (Hair et al., 2014). Overall, these findings provide evidence of construct validity and reliability for the construct of LR writing skills.
Results of Factor Loadings of the LR Writing Task.
Notes. Extraction method: principal component analysis; rotation method: varimax with Kaiser Normalization. Bolded and underlined items indicate major loadings for each variable.
Regarding RQ1, the four factors shown in Table 3 can be interpreted as specific skills within the skill construct, as follows.
Factor 1 includes LR4 Reference, LR1 Selection, LR2 Summary, and LR3 Synthesis, which are all related to the processing of the input literature. This factor measures students’ ability to cite or summarize (LR2) information and main ideas from selected literature (LR1) using appropriate referencing styles (LR4) and to synthesize (LR3) the information and ideas in their own words. This factor is labeled Citation and Synthesis.
Factor 2 includes LR6 Research gap, LR5 Comment, and LR7 Research question, which are closely related to opinion output. This factor reflects the extent to which students can develop their own research questions (LR7) based on comments (LR5) on and research gaps (LR6) in previous research. This factor is named Research Question Elicitation.
Factor 3 includes LR10 Language expression, LR9 Organization, and LR8 Titling. This factor indicates the proper use of language by students, with an emphasis on whether ideas are presented clearly, cohesively, and coherently (LR10). It also concerns the appropriate use of structure (LR9), focusing on whether the text is well organized with a clear title as well as clear headings and subheadings (LR8). This factor is thus called Language Expression and Organization.
Factor 4 includes LR12 Practical context and LR11 Theoretical context. This factor reflects students’ ability to situate the research in a specific context. It assesses whether students are aware of the significance of placing research on a solid theoretical foundation (LR11) and relating research to real-world practical concerns (LR12). This factor is thus called Contextual Awareness.
LPA Results
The mean scores of the four factors identified by the EFA were used as indicators in the subsequent LPA to categorize students into different profiles. The one- to four-profile solutions were examined, and the model fits are shown in Table 4. The values of LL, AIC, BIC, and ABIC decreased as the number of profiles increased, and the decrease became marginal when the number of profiles reached 2. The 2-profile solution emerged as the optimal model because: (a) it had the highest entropy value (0.823), and (b) the LMRT test suggested that the 3-profile solution was not significantly better than the 2-profile solution (p = .057).
Comparisons of Models for Latent Profiles.
Notes. LL = log-likelihood; AIC = Akaike information criterion; BIC = Bayesian information criterion; ABIC = sample size adjusted BIC; LMRT = Lo-Mendell-Rubin likelihood ratio test; BLRT = bootstrap likelihood ratio test.
Regarding RQ2, the delineation of students’ LR writing skills is presented in Figure 1. Profile 1, encompassing 136 participants (constituting 47.6% of the total participants), is categorized as comprising students with low LR writing skills. This classification is based on mean scores hovering around 4.5 out of 10 on all four indicators. On the other hand, Profile 2, with 150 participants (making up 52.4% of the total participants), is identified as consisting of students with moderate-to-high LR writing skills, as their mean scores on the four indicators closely approach 6.0 out of 10. Remarkably, for both profiles, the lowest mean scores were observed on factor 2, Research Question Elicitation, with scores of 3.90 and 5.86 for students with low LR writing skills and students with moderate-to-high LR writing skills, respectively. This indicates that students in both profiles encountered challenges in eliciting research questions during LR writing. Further, the results of Wald chi-square tests, as shown in Table 5, demonstrate significant differences in all four indicators between the two profiles (all ps <.01), revealing that students classified as having moderate-to-high LR writing skills outperformed their counterparts with low LR writing skills across all four facets of LR writing.
The estimated means of the LR writing skills.
Comparisons of Means Between Groups.
Note. ** p < .01.
Regarding RQ3, Wald chi-square tests reported significant RP writing performance differences between the two profiles:
Discussion
This study extends the research on LR writing by revealing its underlying skill construct and explores an effective way of demonstrating the patterns of LR writing skill profiles among postgraduate students whose first language is Chinese.
The Construct of LR Writing Skills
In response to RQ1, the results of EFA identified four underlying skills within the LR writing skill construct: (a) Citation and Synthesis, (b) Research Question Elicitation, (c) Language Expression and Organization, and (d) Contextual Awareness. Unlike prior research that often focused on individual LR writing skills in isolation, this study offers a comprehensive view of LR writing skills by showcasing how various discrete skills collaborate to form a multidimensional construct. By adopting this holistic perspective, the study thus enhances the existing literature on LR writing education. Moreover, while extant research predominantly concentrates on LR writing in English as a first or second language (e.g., Bitchener & Basturkmen, 2006; Boote & Beile, 2005; Chen et al., 2016), our study, involving postgraduate students enrolled in a Chinese language academic writing course, provides empirical evidence from a different linguistic and educational setting. Future research could investigate the applicability of the identified LR writing skill construct in contexts such as L1 and L2 English LR writing. In the subsequent sections, we provide in-depth discussions on each of the four identified LR writing skill constructs, aiming to enrich the current understanding of LR writing skills and their implications for pedagogy.
Citation and Synthesis includes selection, summary, synthesis, and reference. This factor refers to the comprehension and use of input literature materials. It explains more than half of the variance in student LR writing skills, indicating that students’ capacity to process input literature materials is the most demanding component. It also emphasizes that students should actively seek necessary information from input materials, make sense of it, and translate it for their own use rather than passively consuming information. These interactions with input literature have been reported in previous research concerning discourse synthesis (Nelson & King, 2023; Segev-Miller, 2007; Spivey, 1997). As Segev-Miller (2007) suggested, two of the major processes of discourse synthesis in LR writing are conceptual and rhetorical restructuring. Students are required to continually distinguish key ideas from less important ones within and across multiple selected literature sources, summarize those ideas in their own words, and synthesize them into cohesive content. During this process, the use of academic conventions, such as normative citation formats, helps student writers build their academic identities (Nelson & Castelló, 2012).
Research Question Elicitation involves identifying research gaps, making comments, and raising research questions. This factor is relevant to opinion output, and includes higher-order thinking skills like evaluating and creating (Singh et al., 2018). As the term “literature review” suggests, students are required to undertake a comprehensive review of the literature. This is a critical dialogue with the literature (Badenhorst, 2019; Breuer, 2013) conducted with the aim of informing the students’ own research. During the process, students need to critically evaluate the existing literature, recognize current research advances, and diagnose weaknesses and discrepancies among literature sources (Chen & Li, 2019; Kwan, 2006; Paul & Criado, 2020). Students should make comments on and identify gaps in the existing literature, thus paving the way for the creation of their original research questions. In other words, critical thinking is heavily emphasized in this skill.
Language Expression and Organization consists of language expression, organization, and titling. In accordance with existing research on linguistic features of LR writing (Chen et al., 2016; Rabie & Boraie, 2021; Randolph, 2009), the results underscore the importance of skillful use of language. This dimension of LR writing skills concerns whether students are able to use language to make meaning (known as “languaging”) (Swain, 2010). Furthermore, the results also underline the significant role of text organization. In fact, good text organization is no less important than good language expression in academic writing, although it might be easily overlooked when students face linguistic challenges (Denney & Tewksbury, 2013; Paul & Criado, 2020). Moreover, showing a clear review agenda by using an appropriate title as well as appropriate headings of sections and subheadings within sections not only helps to develop students’ audience perspective in writing (Canagarajah, 2011) but also promotes the credibility and integrity of their written work (Denney & Tewksbury, 2013).
Contextual Awareness is related to both the theoretical context and the practical context. The results echo the argument that students are supposed to be conscious of their specific research contexts and situate their research based on contextual understanding (Boote & Beile, 2005; Kwan, 2006). Context can be characterized as either theoretical or practical. Theoretical context informs the “what” of research, while practical context highlights the “why” of research (Boote & Beile, 2005). The importance of this skill can also be explained by the communicative function of a literature review. Researchers have indicated that a literature review is a scholarly text where writers seek to demonstrate some authority over an issue or topic via communicating their ideas with readers (Badenhorst, 2018; Nelson & King, 2023). To convince their readers, writers need to provide a strong theoretical underpinning for their ideas and exhibit their sincere concern regarding real-world practical issues. In other words, it is vital to raise students’ theoretical and practical contextual awareness in order for them to create convincing texts.
Patterns of Student LR Writing Skills and Students’ Performance in Academic Writing
Regarding RQ2, the LPA results indicate that the students could be classified into two profiles: students with low LR writing skills (profile 1) and students with moderate-to-high LR writing skills (profile 2). Interestingly, the lowest mean scores of students in both profiles were on the factor of Research Question Elicitation (i.e., 3.90 and 5.86 out of 10, respectively). After checking the mean scores of the skill variables that constitute this factor, we discovered that the low mean scores on LR6 Research gap and LR7 Research question (i.e., 4.67 and 4.37 out of 10, respectively; see Table 1) may be the cause. This is consistent with the findings of Chen et al. (2016), who indicated that recognizing research gaps and raising research questions are the two most challenging tasks in LR writing. In fact, research gaps are generated from “disparate sets of information” (Jacobs, 2011, p. 128). Research gap identification, therefore, requires students to conduct frequent comparisons and deliberations among literature sources. This may be challenging for students who are not skilled in scrutinizing concepts, integrating similarities, and analyzing differences (Chen & Li, 2019; Müller-Bloch & Kranz, 2015). Moreover, research questions are commonly formulated based on identified research gaps. However, not all gaps can be converted into meaningful and feasible research questions. After identifying potential gaps, students further need to confirm their existence (Müller-Bloch & Kranz, 2015), evaluate their size, complexity, and significance (Alvesson & Sandberg, 2011), and analyze the feasibility of performing the study and the possibility of answering the research questions. In this complicated process, students may find it difficult to raise questions when they do not fully engage in these strategic actions.
Regarding the factor of Contextual Awareness, the relatively low mean scores (4.23 and 6.04 out of 10 for the two profiles) can be attributed to the low scores on its underlying components—LR 11 Theoretical Context and LR 12 Practical Context (mean scores of 5.36 and 5.03, respectively; see Table 1). As noted by Boote and Beile (2005), a key function of LR writing is to establish a robust research foundation by linking existing literature with one's own study. However, constructing such linkages is particularly challenging because it requires students to both deeply understand theoretical frameworks (as reflected in LR 11) and effectively apply these frameworks in practical research contexts (as reflected in LR 12). This dual demand—synthesizing theory with practice—necessitates advanced critical thinking, synthesis, and evaluative skills, which may account for the lower performance on this factor.
In contrast, students in both profiles performed relatively better on the factors of Citation and Synthesis (mean scores of 4.81 and 6.88, respectively) and Language Usage and Organization (mean scores of 5.23 and 6.70, respectively). This may be because these factors represent explicit skills in literature review writing—skills that align with the common perception of literature reviews as systematic, academically rigorous summaries of existing research. Consequently, these competencies are more familiar to students. Moreover, these skills are not only essential in literature review writing but are also routinely practiced in various academic writing tasks (e.g., research proposals, research reports, and conference papers), thereby providing students with more opportunities for practice and refinement, which likely contributes to their higher performance in these areas.
Concerning RQ3, the LPA results demonstrate that students from different LR writing skill profiles performed differently in RP writing. Particularly, students with moderate-to-high LR writing skill outperformed those with low LR writing skill in RP writing. This finding corroborates previous research (e.g., Saeed et al., 2021; Wang & Yang, 2012) suggesting that students’ LR writing skills may influence their academic writing performance. The results also suggest that LR and RP writing share certain required skills, and that both kinds of writing can be enhanced by developing students’ capacity in the skill dimensions of Citation and Synthesis, Research Question Elicitation, Language Expression and Organization, and Contextual Awareness.
Implications, Limitations, and Future Research
The findings of the present study contribute to the literature in three ways. First, it identifies a construct of LR writing skills, comprising four underlying components within the Chinese language context. Given that academic writing is inherently a universal construct, the skills identified here are not limited by language boundaries. Consequently, our findings not only complement existing research on English LR writing but also offer practical insights for developing teaching strategies applicable to both L1 and L2 language English contexts. This bidirectional transfer of insights underscores the mutual benefits of integrating research from Chinese and English academic writing domains. Second, the confirmation of the impact of students’ LR writing skills on their research proposal writing performance adds to the growing body of evidence affirming the connection between LR writing skills and overall academic writing performance. This reinforces the understanding that proficiency in LR writing is a pivotal factor influencing success in broader academic writing tasks. Third, the utilization of LP as a person-centered approach for analyzing students’ diverse LR writing skills demonstrates its appropriateness. This methodological choice not only sheds light on the skill profiles among students but also suggests the potential efficacy of LPA as a tool for understanding and categorizing students based on their LR writing skills.
We also have some pedagogical suggestions. First, a compulsory course could be set to provide students with detailed LR writing instruction as a necessary starting point for academic writing for postgraduate students. Although the importance of LR writing is widely acknowledged, the majority of research textbooks understate the importance of LR writing by only giving it a tiny chapter or section and place a much greater emphasis on illustrating research design, data collection, and analysis (e.g., Boote & Beile, 2005). Thus, an exhaustive course on LR writing instruction is needed for students who may not fully learn LR writing on their own. Second, an explicit pedagogy should strive to cover all four dimensions of LR writing skills. To familiarize students with each skill dimension, teachers could provide samples of high-quality LRs and encourage students to engage in discussions about skill elements exhibited in these reviews. This would promote students’ involvement and engagement in LR writing learning (Ma & Bui, 2022; Ren & Zhu, 2023). Particularly, we found that Citation and Synthesis accounted for the largest variance in LR writing skills, revealing that it is the most necessary skill for LR writing. Therefore, in-class instruction and after-class training should strengthen students’ ability in literature source selection, source text summarization and paraphrasing, and multiple-source integration. For example, guided annotation exercises can help students critically evaluate and select relevant sources, while scaffolded summarization and paraphrasing tasks can improve their ability to distill key arguments without plagiarizing. Additionally, structured synthesis activities, such as comparative analysis of multiple texts or collaborative writing tasks where students integrate diverse sources into a coherent argument, can further reinforce their synthesis skills.
Our research also indicated that students generally scored the lowest on Research Question Elicitation. Teachers could introduce the concept-mapping method to help students identify research gaps. This approach allows for concept comparisons across literature sources (Müller-Bloch & Kranz, 2015). Teachers could then guide students to develop research questions by evaluating the size, complexity, and significance of the potential gaps that have been identified. Furthermore, problematization is a strategy that can be employed to raise novel research questions (Alvesson & Sandberg, 2011). Students can be taught to perform a dialectical interrogation of various positions and stances adopted in the literature. This process enables them to engage in a critical rethinking and reformulation of the basic elements of “received wisdom” (p. 39) used in previous literature (Alvesson & Sandberg, 2011). As a result, novel or even surprising research questions may be generated.
There are also some limitations that merit further research. First, this study only involved a small sample of postgraduate students in language communication sciences from one university. Caution should be taken when attempting to apply the results to other educational contexts and/or populations. Future research could collect data from larger and more diverse samples, including participants from various educational backgrounds (e.g., doctoral students). Second, this study primarily adopted a cross-sectional research design without considering dynamic changes in students’ writing ability. LR writing skills can be improved through appropriate classroom interventions (Chen & Li, 2019; Zhang, 2022). Future research could take a longitudinal perspective and examine the changes in students’ LR writing skills over one semester or one academic year. Such an approach would provide insights into the effectiveness of instructional strategies designed to improve LR writing, the retention and transferability of learned skills, and the developmental trajectories of students’ LR writing proficiency, thereby offering a more comprehensive understanding of how these skills evolve over time.
Supplemental Material
sj-docx-1-roe-10.1177_20965311251413599 - Supplemental material for Examining Chinese Writers’ Literature Review Writing Skills and Their Impact on Academic Writing Performance
Supplemental material, sj-docx-1-roe-10.1177_20965311251413599 for Examining Chinese Writers’ Literature Review Writing Skills and Their Impact on Academic Writing Performance by Xinhua Zhu (祝新华), Yuan Yao (姚远), Jun Lei (雷钧) and Qi Lu (陆琦) in ECNU Review of Education
Footnotes
Ethical Considerations
The studies involving human participants were reviewed and approved by the Department of Chinese and Bilingual Studies, The Hong Kong Polytechnic University (IRB approval number: HSEARS20201124003).
Author Contributions
Xinhua Zhu designed the methodologies of the research project. Qi Lu, Jun Lei, and Xinhua Zhu collected and analyzed the data collaboratively. Qi Lu drafted the initial manuscript (including the figures and tables). Yuan Yao and Xinhua Zhu later elaborated and revised the manuscript while maintaining close communication with Qi Lu. All authors contributed to the article and approved the submitted version.
Funding
The authors disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: This work was supported by Research Grants Council (RGC) Junior Research Fellow Scheme of Hong Kong SAR [grant number JRFS2526-4H03].
Declaration of Conflicting Interests
The authors declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.
Supplemental Material
Supplemental material for this article is available online.
References
Supplementary Material
Please find the following supplemental material available below.
For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.
For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.
