Local Grammar Approach to Investigating Advanced Chinese EFL Learners’ Development of Communicative Competence in Academic Writing: The Case of ‘Exemplification’

4.1 Research Design

The present study adopts a programmatic longitudinal design, i.e., tracing EFL learners’ language development across different levels of study as defined by academic programmes (e.g., BA, MA, PhD), which is motivated by Ortega and Iberri-Shea’s (2005, p. 30) call that “an increase of longitudinal investigations of this kind [programmatic longitudinal study] would be a positive development in future instructed SLA research”. Specifically, it explores advanced Chinese EFL learners’ development of communicative competence by examining their use of exemplification in MA and PhD degree dissertations.

It also merits noting that, similar to Candarli (2021), this study did not use L1 or expert writers’ use of language (e.g., published research articles) as reference for comparison, as it follows the multilingual turn and acknowledges the value of researching EFL learners’ use of language in its own right. In fact, Gablasova, Brezina, and McEnery (2017) have discussed the disadvantages of comparing learner language with L1/expert language use and raised issues concerning data representativeness, comparability, and interspeaker variation. For instance, they argue that “not all differences between corpora can be automatically attributed to the difference between L1 and L2 use, because L1 corpora are very likely to differ from each other as well” (Gablasova, Brezina, and McEnery 2017, p. 18). Moreover, comparing degree dissertations with expert writing, published research articles for instance, may raise issue of comparability, as the two types of writing are stylistically different, written for different purposes and have different readership, all of which might also result in differences in language use.

In a nutshell, the study adopts a programmatic longitudinal design and aligns with learner corpus research in that it focuses on product, i.e., learners’ performance of language use, and analyses relatively a large number of naturally occurring language data. Hence, the study employs a whole-corpus, rather than text-based, approach to data analysis. While this approach would not allow reporting each individual learner’s use of exemplification from MA to PhD study, it helps to reveal the general trend of language development of this specific group of EFL learners. Additionally, acknowledging the usefulness of tracing individual learners’ language development, it is likely that each individual may show a distinct pattern of development, which consequently raises issues concerning the generalisability of the results (see also Gass, Behney, and Plonsky 2013; Granger, Gilquin, and Meunier 2015).

4.2 Corpora and Data Retrieval

Following the programmatic longitudinal design, we used China National Knowledge Infrastructure (CNKI; https://www.cnki.net/) to collect MA and PhD degree dissertations and thereby compiled two corpora for the current investigation, which were termed CoMA and CoPhD respectively. The dissertations were produced by a group of the same 55 Chinese EFL learners at the level of MA and PhD study in universities in mainland China. At this point, it is worth noting that student writers who have the same L1 background (i.e., Chinese) were selected because, as reported in Lu and Ai (2015), learners with different L1 backgrounds, even at the same level of proficiency, may not develop the same patterns of L2 syntactic complexity. “As a result, a study on learner academic writing by students from the same L1 background could potentially yield more reliable results” (Ansarifar, Shahriari, and Pishghadam 2018, p. 61).

All the students were majored in Applied Linguistics, because it is one of the very few disciplines that requires students to write their degree dissertations in English (see also Wu and Paltridge 2021). We focused on these advanced EFL learners because discourse features are more crucial for language learning at more advanced levels (see also Neff-van Aertselaer 2015). MA and PhD students, by the time they complete their degree dissertations, can be reasonably regarded as advanced EFL learners who, arguably, have little difficulty producing grammatically well-formed sentences and a reasonably high level of pragmatic competence. This indicates that MA and PhD degree dissertations could provide a rich source that is especially appropriate for investigating advanced EFL learners’ development of communicative competence in academic context.

The corpora have the unique features such as the MA and PhD dissertations were written by the same group of writers across a relatively long time, but neither they started nor ended the study at the same time. Moreover, these students were enrolled in a wide range of universities in China rather than were selected from one university, thus partly responding to Siyanova-Chanturia’s (2015) suggestion that longitudinal research will need to examine texts produced by a larger variety of L2 participants. These could ensure that the data could be largely representative of advanced Chinese EFL learners’ academic writing. Finally, the two corpora would also be comparable because they were compiled of texts not only authored by the same group of writers, but also stylistically the same, i.e., both compiled of degree dissertations.

In addition, it is perhaps worth mentioning that the method of data collection used in the current investigation may suggest a new way to compile longitudinal corpora. It is quite challenging to follow a large group of learners for a relatively long time, not only because of time and labour, but also because it is impossible “to tell at the outset which learners would keep learning the L2 for that long” (Myles 2015, p. 316). In consequence, “longitudinal L2 data is rather rare, and when it does exist, it tends to document learner development over a small number of months/years for a small number of learners” (Myles 2015, p. 316). While recently there has been few studies such as Yuksel et al. (2023) which collected data over quite a long time, relatively large longitudinal corpus is still scarce. Hence, the method of data collection used in the present study, which emphasises the criterion of collecting written texts produced by the same group of writers but does not require the texts be produced within the same time period, might offer some methodological insights into collecting longitudinal data for future studies.

All the sampled degree dissertations were subsequently edited and cleaned, excluding author names, affiliation, acknowledgements, header and footer, etc. The metainformation of the two corpora was shown in Table 4.

4.3 Exemplificatory Markers in MA and PhD Dissertations

Based on previous studies (e.g., Hyland 2007; Paquot 2008; Su and Zhang 2020; Triki 2021), we identified a set of exemplificatory markers (Table 5) and subsequently searched for them and their possible variants (e.g., an example of, a good/typical example of) to retrieve instances of exemplification in the two corpora. Admittedly, searching these markers may left undetected those instances which do not co-occur with these markers; nevertheless, doing so could allow us to retrieve a substantial set of instances of exemplification, as Triki (2021) reported that nearly 90 % of instances of exemplification co-occur with one of these markers. Another advantage of using computerised search to retrieve instances of exemplification, rather than by manual reading, is that it could guarantee the consistency in data retrieval across corpora, as manual reading may subject to intuition and result in inconsistency.

As shown in Table 5, overall, the student writers under examination used these markers in quantitatively similar ways from MA to PhD levels and no statistically significant difference was observed, which indicates that as advanced EFL learners, both MA and PhD students are indeed quite aware of the importance of exemplification in academic writing. As for individual markers, these student writers show statistically significant differences in the use of the three markers for example, for instance, and e.g. The student writers at MA level used significantly more frequently for example but less frequently for instance and e.g., whereas PhD student writers used more frequently for instance and e.g. This difference may be attributed to instructional effect. In EFL contexts, for example and for instance “are often taught as two synonyms and as a result, learners use them interchangeably. Learners presumably write for instance instead of for example when they have already used for example in their essay and want to change for the sake of variety” (Paquot 2008, p. 110). This instructional effect may also account for the higher frequency of e.g. in PhD student writers’ writing. That is, compared with PhD education, the appropriate use of e.g. probably has not been well taught at the MA level, which inevitably resulted in MA student writers’ less frequent use of e.g.

As discussed in Section 3, however, it may not be reliable to explore EFL learners’ performance of exemplification by simply examining their uses of exemplificatory markers, partly because not all instances containing these markers realise exemplification and partly because the quantification of their use of exemplification is oversimplified. More importantly, simply counting the frequency of each mark cannot reveal the complexity of exemplification patterns (see also Su, Zhang, and Chau 2022); for instance, instances of for example can realise a wide range of patterns at different degrees of complexity, as has been shown in Table 2. Therefore, below we present a more fine-grained and systematic analysis of their use of exemplification from the perspective of local grammar.

4.4 Local Grammar Analyses of Exemplification

Local grammar uses context-specific terminologies for analysing instances of the target discourse function or act. For the current investigation, we adopted the terminologies used in Su, Zhang and Lu (2021) and reproduced them in Table 6. The majority of these terminologies are self-evident, except Exemplification-SC and Exemplification-RS. This distinction was made on the basis that strategies of exemplification can be generally classified into two types (Su, Zhang, and Lu 2021). One is exemplifying by presenting subtypes or subcategories of an abstract, superordinate category, labelled Exemplification-SC; and the other is exemplifying by citing relevant studies, labelled Exemplification-RS. The distinction has been shown to be useful for exploring the performance of exemplification by different groups of writers (Su, Zhang, and Chau 2022), as will be further discussed in the present study.

As previous studies have discussed comprehensively the local grammar analyses of exemplification (e.g., Su, Zhang, and Lu 2021; Su, Zhang, and Chau 2022), we will not dwell on the analyses here but only present the local grammar patterns of exemplification, with an example given of each, in Table 7.

4.5 Results and Discussion

Based on the analyses, we examined the distribution of the local grammar patterns of exemplification across the two corpora, as shown in Table 8.

The investigation reveals both similarities and statistically significant differences between Chinese English-major MA and PhD student writers’ use of exemplification in degree dissertations. With regard to similarities, the investigation shows that the overall use of exemplification by the two groups of student writers was quite the same and no significant difference was found, suggesting that their use of exemplification remained steady across the two levels of study. This may be due to that exemplification is one most basic discourse act in academic writing and both MA and PhD students, as advanced EFL learners, are well aware of its importance in academic writing. This consequently raises another question, that is, would EFL learners’ use of other more complex discourse acts such as defining terms and expressing cause-effect relationship show a developmental pattern that is similar to or different from the one reported here? This is one direction worthy of future investigation.

Another similarity observed is that PA 1 (Exemplified – Indicator – Exemplification-SC) is the pattern used most frequently by both groups of writers and there is no statistically significant difference. This pattern allows the writer to provide specific subtypes or subcategories to illustrate abstract concepts or categories (e.g., Verdictives: giving a judgement or verdict; for example, acquit, judge, etc.), making the abstract more concrete or specific and thereby contributing to the accessibility of academic writing. The high frequency of this pattern may be partly attributed to these student writers’ limited repertoire of exemplifying phraseologies (Paquot 2008) and partly to L1 transfer, i.e., there is a direct counterpart of this pattern in Chinese (Su, Zhang, and Chau 2022). This in turn points to the necessity of teaching various ways for providing examples and, by extension, for performing other discourse acts in pedagogical practice, which could arguably be helpful to enrich EFL learners’ phraseological repertoire.

Apart from the similarities, there are also notable differences worth discussing. The first is concerned with their use of the two strategies of exemplification. As noted earlier, exemplification may be used either to specify abstract superordinate categories, or to support an argument/observation. Accordingly, exemplification may be realised by presenting subcategories or by citing relevant studies, which were labelled Exemplification-SC and Exemplification-RS respectively. We thus examined the overall frequency and the distribution of patterns with the two labels across the two corpora, as shown in Table 9.

As can be seen in Table 9, the analyses show statistically significant differences in the use of the two exemplifying strategies by the two groups of student writers. Specifically, MA student writers used significantly more frequently the strategy of exemplifying by presenting subcategories or subtypes, whereas PhD student writers used significantly more frequently the other strategy, i.e., exemplifying by citing relevant studies. This may suggest that, while the two groups of writers are aware of the significance of exemplification in academic writing, their preferred strategy is different: MA student writers tend to exemplify abstract concepts or superordinate categories, whereas PhD student writers are more likely to illustrate or exemplify arguments or observations, which are discussed in turn. The reason for why MA student writers prefer the strategy of exemplifying by presenting subcategories may be that these student writers had not considered many relevant studies in their field; or alternatively, this may be due to that this strategy has not been well taught at the level of MA study in Chinese universities and consequently is something relatively unfamiliar to them, as instruction may affect language learning and/or development (Candarli 2021). This in turn points to the significance of teaching discourse acts and their associated lexicogrammatical realisations in EFL classrooms, which will be discussed in greater detail below.

PhD student writers used the other strategy, i.e., exemplifying by citing relevant studies, much more frequently than their MA counterparts. What is also noteworthy is that expert writers in the discipline of Applied Linguistics have been reported to use even more frequently (487.00 times per million words) this strategy (Su, Zhang, and Chau 2022). This appears to suggest that, if expert writers’ performance of exemplification can be regarded as the norm, the advanced Chinese EFL learners’ performance of exemplification (at least in the use of this particular strategy) shows a linear pattern of development – MA student writers used the least, expert writers the most, and PhD student writers in-between, indicating development of communicative competence over time.

In short, given that the overall use of exemplification by the two groups of Chinese EFL learners remained steady, PhD student writers’ more frequent use of the strategy of exemplifying by citing relevant studies indicates a change from exemplifying primarily by presenting sub-categories to a more balanced performance of exemplification by using both strategies. As the more balanced and appropriate use of exemplification contributes to the overall quality of academic writing, the differences discussed above would suggest that advanced Chinese EFL learners’ communicative competence improves with proficiency advancing.

Furthermore, it is observed that the two groups of writers’ use of patterns 6 and 8 also show statistically significant differences. Specifically, MA student writers used the two patterns more frequently than their PhD counterparts. Instances realising these two patterns are highly formulaic and relatively colloquial (e.g., Take the voices of animals for example), which means that MA student writers relied upon more formulaic but less formal sequences. This is consistent with Duan and Shi’s (2021, p. 24) observation that EFL learners “at a lower level, rely more heavily on memorization-based formulaic patterns in their production and move slowly toward self-constructed, more native-like language as proficiency increases, a process accompanied by a decrease in formulaic language”. In contrast, PhD student writers’ lower use of such patterns may suggest that they are more aware of genre appropriateness, i.e., formal, rather than colloquial, language is normally preferred in academic writing, which offers additional evidence for their development of communicative competence in academic writing.

To sum up, the investigation shows that the Chinese English-major MA and PhD student writers’ overall use of exemplification is quantitatively similar across the two levels of study, which suggest that they, as advanced EFL learners, are well aware of the important roles exemplification plays in academic writing. The investigation, however, also shows that PhD student writers are more familiar with the strategy of exemplifying by citing relevant studies and the lexicogrammatical realisations they used to realise exemplification are more formal and generically appropriate, which indicates that advanced Chinese EFL learners’ communicative competence does develop as proficiency increases. This yields important insights into researching, assessing and teaching academic writing in EFL context, as discussed below.

The present study examined advanced Chinese EFL learners’ use of exemplification in academic writing and found that, although their overall use of exemplification remained steady from MA to PhD levels of study, there are also notable differences in their lexicogrammatical choices they used to perform exemplification. It is, however, unclear whether their performance of other discourse acts would show a similar or different developmental pattern. For academic writing research, this points to the need to systematically explore EFL learners’ development of communicative competence in academic context by examining their performance of other discourse acts such as defining terms, making comparison and contrast, expressing cause-effect relationship, and so on. Such investigation, arguably, could deepen our understanding of EFL learners’ development of communicative competence in academic context.

Turning to academic writing assessment. The present study found that PhD student writers used significantly more frequently the strategy of exemplifying by citing relevant studies than MA student writers did. As appropriately using this exemplifying strategy contributes substantially to the persuasiveness, accessibility and the overall quality of academic writing, this indicates that EFL learners’ performance of discourse acts might also be a useful indicator of their writing proficiency and should thus be considered in academic writing assessment. In fact, early studies have similarly argued that the discoursal or communicative aspects are crucial for writing assessment. For example, Chiang (1999, p. 229) argues that “the discoursal aspects of written essays take precedence over grammatical accuracy in native speakers’ judgement of writing quality”. Plakans (2014, p. 1399) stresses that academic writing assessment “should consider aspects of text that capture the communicative nature of language”. The discoursal or communicative aspects, however, have so far been rarely considered in the assessment of academic writing (see Barkaoui and Hadidi [2020] for a comprehensive review). It is thus suggested that future research in academic writing assessment consider EFL learners’ pragmatic performance of discourse acts as an additional indicator of L2 writing proficiency or quality.

The final implication relates to the teaching of academic writing. First of all, given the crucial roles discourse acts play in academic writing and the appropriate performance of discourse acts may indicate L2 writing proficiency, it is necessary to raise academic writing teachers’ awareness of discourse acts when teaching academic writing. This requires efforts to be made by both researchers and teaching practitioners of academic writing. As Durrant and Mathews-Aydınlı (2011) argue,

For language learners, the key information about formulas is rarely which word sequences are the most frequent per se. Rather, learners need to know what functions they are likely to need to express, what forms most appropriately fulfil those functions, and what variation those forms permit when they are fitted into specific contexts.

(Durrant and Mathews-Aydınlı 2011, p. 61)

Indeed, previous studies have found that EFL learners usually have difficulty performing specific communicative functions in academic writing (Hinkel 2004; Paquot 2008), which may be attributed to that the lexicogrammatical realisations available to perform such functions have not been adequately documented, described, and taught (Su and Zhang 2020). For academic writing researchers, this points to the necessity to adequately document and describe the lexicogrammatical realisations of each discourse act and to present the findings in a way that is accessible to teaching practitioners. To that end, the proposed local grammar approach, representing a more fine-gained type of functional analyses, would be useful, because local grammar is in essence a focus-on-function approach to linguistic description and “[o]nly a function-driven annotation makes possible the identification and documentation of a near-complete inventory of lexico-syntactic means used to express various communicative functions in discourse” (Callies 2015, p. 54). Local grammars have been shown to be particularly useful for developing such inventories (e.g., Hunston and Su 2019; Su and Wei 2018; Su and Zhang 2020).

Subsequently, with the inventories of lexicogrammatical realisations available to perform a given discourse act offered by local grammars, academic writing teachers should attempt to incorporate such inventories into the practice of teaching discourse acts. One possible way to do so is to integrate the inventories of each discourse act and its associated local grammar patterns into pedagogical materials and to explicitly teach the target discourse act, its rhetorical functions and lexicogrammatical realisations (Zhang and Su 2022). Doing so would give EFL learners ample opportunities for exposure and learning of discourse acts, which could in turn contribute to EFL learners’ development of communicative competence and, ultimately, academic literacy.