Introduction

Considerable scholarly endeavors have been made to unravel the development of L2 academic writing, probably due to its gate-keeping role in successfully accessing membership of diverse international academic communities. Nevertheless, among these endeavors, form-focused linguistic development has been the research focus constantly, that is, to understand L2 academic writing development through examining CAF-based (Complexity, Accuracy, Fluency) elements (e.g. syntactic and lexical features, Maamuujav et al. 2021). Albeit such examination does facilitate our understanding of the linguistic development of L2 academic writing, the meaning-oriented development has still remained far more understudied. Furthermore, compared with a conventional task-based lens, a task-repetition (hereafter abbreviated as TR) counterpart to examine meaning-oriented language development seems to be necessary because a single task may fail to trigger a qualitative change in learners’ L2 system (Nitta and Baba 2014).

TR, a significant variable in task implementation, has drawn considerable attention from TBLT and SLA (Kim et al. 2020). Bygate and Samuda (2005) defined TR as either intact or partial repetition of identical or reformulated tasks. This echoes Ellis’ (2012) definition of TR in which learners are required to repeat “without any changes to the task or by modifying the design of the task or by manipulating one of the other implementation variables” (p. 202). With the definition aforementioned as premises, TR has been further categorized into exact TR (to repeat the identical task), procedural TR (to repeat different content with the same procedure), and content TR (to repeat the same content with a different procedure). The present study approaches TR specifically as Task-type Repetition (TtR), i.e. to repeat the same task type with different content.

With ample positive evidence pertinent to TR’s facilitative role in L2 oral acquisition (e.g. Bygate 1996; 2001) and certain commonalities shared by oral and written modality, it is theoretically reasonable to presume that the features of L2 academic writing discourse (i.e. high lexical density, authority, and preference to noun-dominant language with intra-clause reasoning rather than the inter-clause one, Halliday 1998; Hyland 2009; Schleppegrell and Colombi 2004) may afford a strong possibility of cross-modal transference of such facilitative role. It is also rational to hypothesize that the advantageous language learning potential of (L2) writing over the oral counterpart (i.e. permanence in keeping records and the resultant convenience to revise, as well as the recursive nature of writing activities) (Manchón 2014) may enable more affordances to augment the facilitative role. This is probably because written tasks, compared to their oral counterparts, feature distinct meaning-making spaces for teaching and learning (Bygate et al. 2014: ix, Cited in Manchón 2014, p.15).

From the perspective of language development in L2 writing with TR, two gaps deserve attention and require a response. First, the meaning-oriented studies juxtaposing TR with L2 writing have been scant, even the form-focused ones that gauged writing development by drawing upon diverse CAF measures were inadequate (Khezrlou 2020). Accordingly, a meaning-oriented approach (to gauge language development) should be embraced, because “a focus on meaning informs and guides the development of students’ ability to become competent multilingual creators of written texts” (Byrnes 2014, p. 95). The second is that a developmental perspective on TR’s role in L2 writing needs to be further underpinned. The perspective should have been valued because a single task, as aforementioned, may fail to trigger a qualitative change in learners’ L2 system (Nitta and Baba 2014). As such, it is the gaps above that motivate the current study. The study aims to facilitate our understanding of L2 writing development in a meaning-oriented way and deepen our understanding of the facilitative role TtR may play over the development.

Literature review

Task repetition

Although mixed results and claims have been yielded (e.g. the contradictory attitudes toward massed repetitions presented by Bygate 2001 and Suzuki and Hanzawa 2021), multitudes of studies have unveiled how TR can facilitate form-focused oral development of L1/2 (e.g. Bygate 1996; Bygate and Samuda 2005). Referring to Skehan’s (1998) presumption about limited attentional resources, the primary presumption shared by these studies was that the increasing familiarity with tasks enabled by TR may save attention resources available to L2 learners, thereby their task performance can be polished (Fukuta 2016).

Studies pertinent to TR in L2 writing share commonalities with these concentrating on oral modality. Theoretically, the cognitive-linguistic view aforementioned also considers writing an optimal venue to accommodate TR’s facilitative role in language development. To be specific, such facilitation seems to be intimately correlated to unique features of writing -- the convenience of revision and the less time pressure for output. These features may further promote more learners’ attention to form and meaning during the writing process (Wong 2001). Empirically, within L2 writing, the benefits from TR primarily clustered around the form-focused language development with CAF measures in grammatical accuracy (Weissberg 1998), fluency (Manchón and Roca de Larios 2011), content, structuring, lexical resources, as well as general quality (Liao and Wong 2010). Recently, TR as facilitation to L2 writing has been gaining finer-grained insights: By criticizing some previous studies for merely giving cross-sectional scenarios with a design of pre-/post-test rather than a longitudinal trajectory of L2 writing development, Nitta and Baba (2014) brought Dynamic Systems Theory as a lens to check how TtR longitudinally enabled L2 writing performance. They concluded that exact TR may not promote task performance, while TtR would enable grammatical and lexical development in the long run. Though Nitta and Baba’s findings above are convincing due to adequately long period (thirty weeks) and dense datasets, some inadequately specified variables (e.g. classroom instructions) in their study may attenuate the reliability of the facilitative effects exclusively ascribed to TtR on students’ writing development. This is because in some cases it is difficult to ascertain whether TR itself or the pedagogical intervention is the real promoter to written language development (Ellis 2009). More recently, studies pertinent to TR in L2 writing have seemingly prioritized the check on resonance between certain variables and TR, further underpinning TR’s facilitative role in L2 writing: Mehrang (2016) uncovered that the repetition of a structured task can elicit further development in complexity and accuracy than an unstructured one. Khezrlou (2020) checked how TR may resonate with direct written corrective feedback and reported different CAF developmental trajectories in three study conditions: The development of accuracy under the condition of TR with error correction; the immediate development in complexity (subordination) in the case of TR with reformulation, and the postponed development in written fluency under the condition of TR and TR with reformulation respectively.

Nonetheless, the studies aforementioned, intentionally or unintentionally, have highlighted form-focused language development with CAF measures, leaving the meaning-oriented one veiled. Namely, neither the textual representation of the internal dimensions of tasks (learners’ interpretation of tasks and strategies to perform tasks, Manchón 2014) nor learners’ agency in meaning-making activities (Byrnes 2014), decision-making activities to align meaning with form for instance, has been adequately probed.

Language development from a meaning-oriented perspective

To value meaning-making in writing requires a meaning-oriented theory, and the optimum turns out to be Systemic Functional Linguistics (SFL) which considers language primarily meaning-making. In SFL, meaning is realized across strata of content (further layered into lexicogrammar and semantics) and expression (Martin and Rose, 2008). The theory, as Byrnes (2014) claimed, is a meaning- and choice-oriented one integrating meaning and form, contributing to inspiring and detailing what constitutes language development.

The present study highlights nominalization in GM (Grammatical Metaphor) in SFL. Put simply, GM refers to the incongruent realization of meaning by repacking grammar (e.g. Halliday 1993, 1998). In everyday use of language, the meaning is realized congruently across the strata of lexicogrammar and semantics: Nominal group as PARTICIPANT of human experience, verbal groups as PROCESS of enacting the experience, prepositional/ adverbial groups as CIRCUMSTANCE to accommodate the experience, and conjunction as RELATOR to interconnect the experience (Liardét 2013, p. 164). By contrast, because academic language typically requires the condensation of abstract and specialized knowledge as well as lexical resources into compact texts (Hyland 2009), the meaning is frequently realized incongruently across the strata aforementioned (Halliday 2002). See the two sentences for an instance:

  1. a.

    The students have effectively learned about ancient China by reading classic literature because the literature describe a lot about people’s lives and the cultural traditions in old times. (congruent)

  2. b.

    Reading literature has enabled the students’ effective learning about ancient China due to the abundant description about people’s life and the cultural traditions involved. (incongruent)

In sentence a, the meaning is realized congruently: The nominal groups (students, classic literature) function as PARTICIPANT/ENTITY of human experience, the verb as PROCESS (learned), the adjective as EPITHET/QUALITY(classic), the prepositional phrase as CIRCUMSTANCE (in old times), and the conjunction as RELATOR (because). By contrast, in sentence b, the meaning is realized incongruently, such as the PROCESS “read”, “ learn”, and “describe” are incongruently reconstrued as PARTICIPANT/ENTITY “reading”, “ learning” and “description” respectively; the CIRMUSTANCE “effectively” as the EPITHET “effective”, and the RELATOR “because” as the CIRCUMSTANCE “due to”, etc. In this way, the incongruence between the strata of lexicogrammar and semantics enables a shift from daily-used language to an academic counterpart featuring dense information and abstractness. Furthermore, it is obvious from the instance above that nominalization (i.e. to realize PROCESS, EPITHET, CIRCUMSTANCE, etc., into nominal groups as PARTICIPANT/ENTITY) is more common than other incongruent realizations of meaning.

As such, GM and nominalization are important linguistic resources to achieve valued academic written discourse and also central to academic registers (Halliday and Matthiessen 1999). GM has been valued considerably in academic writing as meaning-making activities because “it is a measure of its significance for SFL that GM participates in phylogenetic, ontogenetic, and logogenetic (wording-generating) dynamics of meaning-making” (Byrnes 2009, p. 52), and it may afford an extra platform to make meaning (ibid). As the central construct of GM, nominalization is intimately correlated to unique properties of written language as Halliday (2002) stated: “writing as dense, structured, crystalline, oriented towards things (entities, objectified processes), product-like, tight, with meanings related as components” (p. 350). Namely, the denser information chunk of writing modality (than the oral one) and the inclination to reconstrue dynamic processes into static entities in academic written texts can be optimally manifested by nominalization. As such, nominalization is an effective lens to check a meaning-oriented developmental trajectory of L2 writing.

The framework

The present study drew upon Liardét’s framework of nominalization (2013, 2016), which consists of metaphorical control, metaphorical enrichment, and logogenetic impact (i.e. textual impact), to unravel meaning-making activities in L2 academic writing development. The framework was valued primarily due to the Chinese-EFL- learner-related datasets underpinning the framework in a series of studies by Liardét. The datasets to a large extent share commonalities with the current study particularly in the aspect of the context of data collection. Specifically, metaphorical control refers to the extent to which a congruent realization of meaning can be completely reconstrued into an incongruent GM case, primarily into nominalization (Liardét 2016); Metaphorical enrichment refers to the degree of enrichment of metaphorical meaning when the semantically related paradigmatic alternatives are employed respectively; logogenetic impact refers to the textual effects enabled by nominalization including four key patterns resonating with cohesion construction: anaphoric reconstrual (the metaphorical reconstrual of the previous congruent meaning realization as Theme to elicit new meaning); elaborated nominal groups (the condense reconfiguration of concurrent meanings into a lexically dense nominal group); cause and effect metaphorical net (the concurrent reconfiguration of experiential and logical meanings, conventionally from inter-clause to intra-clause, to form metaphorical causal relations), and meaning accumulation (the accumulation by restating or repeating certain meanings) (Liardét 2013; 2016).

The tripartite framework of nominalization values meaning-making in different aspects: Metaphorical control and metaphorical enrichment denote the capability to reconstrue meaning across content-expression strata with rich alternatives, while logogenetic impact focuses on cohesion-based meaning-making. This focus, to a large extent, echoes the three selected indices of Latent Semantic Analysis (LSA, a unique measurement on intra/inter-sentence and intra/inter-paragraph semantic overlap as well as the overlap between explicit words and implicitly similar or meaning-related words, MacNarmara et al. 2014, p. 60–77) of Coh-Metrix employed in the study. Explicit to see among four patterns’ definitions included in logogenetic impact above are their shared foci on cohesion and semantic overlap to different extents. Specifically, anaphoric reconstrual and elaborated nominal groups resonate with the index of LSAGN (a proxy for how much given vs. new information exists in each sentence in a text, compared with the content of prior text information) and LSASS1 (LSA similarity between adjacent sentences) in highlighting local cohesion and semantic overlap, while LSASSp (LSA similarity between all possible paired sentences within a paragraph) underscores the two at a broader level.

To sum up, the current study discerns meaning-making in L2 academic writing from the perspective of the tripartite framework aforementioned and aims to examine how the meaning-making would resonate with TtR. Rather than to attain highly generalizable findings, the study aims to uncover the fine-grained developmental trajectory of the resonance, i.e. the individual idiosyncrasy embedded in the meaning-making development.

Two research questions are highlighted. Within the domain of L2 academic writing and under the framework selected:

  1. 1.

    In what trajectory may cohesion-/nominalization-based meaning-making develop along with the implementation of TtR in the case?

  2. 2.

    Along the trajectory, what noteworthy changes regarding cohesion-/nominalization-based meaning-making may emerge throughout the implementation of TtR?

The study and the instruments

The study context

The course-based study is situated in a semester-long EFL writing course for MTI students (Master of Translation and Interpreting) at a university in the Chinese mainland. Two course objectives are institutionally valued: To familiarize students with degree-requirement-related knowledge of academic English writing (e.g. a translation report) and to polish their linguistic rhetorical repertoire. Students attended the course weekly, before which a written essay was required to submit to an online co-editable file for the convenience of teachers’ examination and subsequent oral pedagogical feedback in class. The feedback was both form- and meaning-oriented, including evaluation and discussion on how students deployed their rhetorical resources, how they presented cohesion, to what extent their task performances were qualified, etc. Strategies, for instance, on-the-spot reformulation of a student’s essay, were deployed to inform students of a think-aloud trajectory of decision-making activities over the meaning-making process of writing. However, direct teaching of nominalization and GM was excluded for the purpose of eliciting natural data to address the perplexity of whether the facilitation of language development stems from TtR or the pedagogical intervention (Ellis 2009).

The tasks

The tasks in the study required an untimed essay (from 200 to 500 words) in which any academic invocation would be encouraged. All ten writing tasks were designed by the researcher as home assignments for students, with full reference to the course objectives aforementioned and Ellis’ (2009) four-criterion definition about the task. Generally, each task required students to formulate a written response to a picture showing a public sign that contains inappropriate Chinese-to-English translation (See Appendix A for a sample task and Appendix B for ten essay topics excerpted from the pictures). Although the teacher did not explicitly require students to represent any specific genre in writing, the genres presumably involved were descriptive, expository, argumentative, persuasive, etc., due to the prompts and task requirements. In this sense, the tasks can immerse students in meaning-making in writing by allowing them to textually present their individual opinions or inferences in meaning, thereby deepening their understanding of content knowledge of academic English writing. The ten tasks implemented over a ten-week session were considered TtR tasks. Over the session, the same task type was repeated but the substance contained in the tasks for eliciting written responses was different. The one-task-per-week set echoed Bui et al.’s (2019) finding in that a one-week interval between tasks is presumably the most contributive to structural complexity, which is of implicit importance to meaning-making.

Specifically, the task prompts firstly required a description of the public sign, and then encouraged expository, evaluative, and analytic comments on the inappropriateness of the translation in the sign, ending with a conclusion containing an individual translation based upon the previous reasoning. Therefore, a genre staging pattern DESCRIPTION ^ EXPOSITION ^ CONCLUSION was formed. Though with divergent definitions, genre here follows the SFL strand and thus is defined as “a staged, goal-oriented social process” evolving over an array of stages (Martin and Rose 2003, p.8). In this sense, essential subgenres for formulating a translation report are juggled successfully by the task design: the narrative and descriptive potentially valued by DESCRIPTION, the evaluative, the expository, the analytic, the persuasive, etc., by EXPOSITION and CONCLUSION. Virtually, the pattern, together with the three-paragraph structure elicited, can be pervasively identified in all student essays in the case, which has been confirmed by the pilot check.

The participants

As for the check on the general-level developmental trajectory of L2 academic writing in the case, 25 MTI students were invited as study participants (N = 25; Male = 7; Female = 18; Mean of age = 24). The students were in their first semester of the first academic year and enrolled in the writing course as a compulsory module. The established experiences of academic English writing of all participants were confined to thesis writing for their English language-related Bachelor degrees, such as Business English, English Literature, English Translation, etc. None of them had experience publishing both in English and Chinese. The majority of students had participated in different national tests for English language proficiency in China: Around one-third of them were qualified for TEM-8 (Chinese National Test for English Majors --Band 8), with the rest mostly qualified for CET6 (Chinese National test for Non-English majors). All tests above involve argumentative writing module, which is similar to writing task 2 in IELTS (Academic). No student has participated in other kinds of English proficiency tests. As such, they were considered emergent writers of academic English writing with intermediate language proficiency on average. In regard to the check on ontogenetic developmental trajectory, three students with low (pseudonym: Lee), intermediate (Monica), and high level (Jack) English proficiency were invited as participants, after consulting their course teachers about students’ English proficiency and checking their TEM-8 scores. Namely, the proficiency level of the three was placed by the comments from the course teachers and TEM-8 results: Lee failed in TEM-8 (around B1 level in CEFR) while Monica passed TEM-8 (B2) and Jack possessed much higher scores (more than 80 out of 100, C1).

The instruments

In the study, Coh-Metrix was firstly employed to map out the general development of cohesion and semantic overlap with three selected LSA indices aforementioned. Rather than to enable quantitative analysis, the check by Coh-Metrix aimed to reveal and compare longitudinal developmental trajectories of the general-level performance (the whole class on average) and the three students’ respectively. With the trajectories as premise, the key timings over the development (the timing at which obvious changes of Coh-Metrix indices can be detected) were roughly located by descriptive comparison.

Then the frequency of nominalization at three genre stagings (DESCRIPTION^EXPOSITION^CONCLUSION) was counted respectively to enable comparisons for different purposes. The typical cases of nominalization identified in the tasks produced at each key timing were listed, compared, and analyzed under the framework. In addition, the semi-structured follow-up interviews were employed to delve into the quantitative findings by detailing the human agents’ interpretation of their writing activities and the tasks.

The study was approved by the School of Foreign Languages of the university concerned. Consent forms have been sent to all participants and successfully collected to gain eligibility of data use and publication.

Coh-Metrix

Coh-Metrix is distinctive in checking text meaning and underlying discourse (MacNamara et al. 2014). Admittedly, the quantitatively validated indices generated by automated measurements on writing may run a risk of misunderstanding written texts as decomposable physical entities rather than an intact ensemble interconnected by logical flow of meaning. Nevertheless, many indices yielded by Coh-Metrix show an intimate correlation to textual meaning-making, particularly cohesion. To be specific, amongst 106 validated indices generated by Coh-Metrix 3.0 (the public version), three indices of LSA were selected and compared: LSASS1, LSASSp, and LSAGN. The index of LSAPP1 (semantic overlap between adjacent paragraphs) was not included because the three genre stagings elicited by task prompts in the study have induced the three-paragraph essay structure, as aforementioned.

The interviews

As aforementioned, a plethora of studies about nominalization or GM has been valuing a reader’s /researcher’s perspective to interpret writers’ GM-related textual presentation. This then necessitates the interviews in the current study as a think-aloud channel to verbalize how writers themselves interpret their agentive decisions in meaning-making activities. To be specific, the semi-structured follow-up interviews would be conducted if any preliminary findings had emerged. This aims to unveil cognitive reasons behind behaviors, for instance, to reveal student authors’ original intention behind using a certain nominalization case.

The data and analysis

Eventually, the study collected 248 pieces of essays with 97,924 words (Mean = 394.9, SD = 1.87, Min = 207.9, Max = 581.9) from 25 students throughout a ten-week TtR session in total. In addition, two teachers successfully finished the identification of the nominalization cases contained in 30 essays produced by three participants. Throughout the whole TtR session, 11 follow-up Chinese interviews, each lasting around fifteen minutes, were successfully conducted, then transcribed (in an indirect way to enable better informational logic and readability), and translated by the researcher.

The researcher employed Coh-Metrix firstly to check the ten corpora (each including all essay data of a given week) respectively to reveal the general-level task performance of the intact class. Then the essays produced by three participants were checked respectively. The former check attempted to map out how L2 academic writing may develop synoptically at the group level and set a benchmark for comparison, while the latter intended to compare different ontogenetic developmental trajectories. The essay data were firstly checked for rectifying punctuation (adding the full stop unconsciously omitted by the student, for instance) and spelling errors by peer students (postgraduates) with the assistance of the rectification function embedded in Office Word. The researcher rechecked the data at the end. Next, the LSA index scores of three students and the general-level’s were compared diachronically to identify any possible developmental stage with any key timing (the timing when obvious changes of Coh-Metrix indices can be detected by comparison) over the developmental trajectory.

As to analyzing the data of nominalization, firstly the researcher and a colleague (an experienced EFL writing teacher) have manually identified all cases of nominalization in three students’ essays. The two-round manual identification of nominalization (as for two coders’ coding, Cronbach’s alpha = 0.817) conformed to derivation (primarily by checking suffix) and agnation (by unpacking the incongruent into the congruent) proposed by Ravelli (1988, 1999). The first-round check occurred concurrently with the proceeding of the TtR session, i.e. the identification was immediately conducted as soon as the weekly tasks were submitted. The second round occurred at the end of the session after all essay tasks had been successfully collected. The identification was conducted according to three different genre stagings: DESCRIPTION^EXPOSITION^CONCLUSION. At each staging, the frequency of nominalization was counted respectively for the purpose of genre-based comparison. Due to the meaning-oriented aim of the current study, the count of frequency only included the nominalization of realizing different ideational experiences, that is, the repetitively employed case at each staging was counted only once. If any controversy over identification, it would not be settled until the agreement was reached by the two teachers. To count and then compare different frequencies of nominalization cases is primarily for delineating the longitudinal development of students’ deployment in nominalization.

Additionally, the genre-staging-based qualitative analysis of selected nominal cases was employed to detail the different meaning-making intentions of the three. Specifically, at each staging different typical nominalization cases were selected from the texts produced by three students for analysis. Among all nominalization cases, prioritized were the ones (1) collected from the tasks at key timings identified by Coh-Metrix, and (2) the ones with a higher frequency of repetition throughout all ten tasks (i.e. the cases repetitively identified throughout a certain TtR session, continuously or discontinuously).

Findings and discussion

Every section under each sub-heading starts with showcasing the findings, and then extends to relevant discussion for the purpose of inference or triangulation, if necessary.

Coh-Metrix indices’

The comparison of LSA indices found that over the course of TtR, the written task performance in cohesion and semantic overlap dynamically resonated with different-level language proficiency, and a staged developmental trajectory was identified. After comparing the trends of three LSA indices, four key timings (at around T4 and T8, with T1 as the start point and T10 as the end being naturally qualified) were respectively identified. At these key timings, important changes were obviously detectable. For instance, Lee’s three LSA scores show a constant decline to be lower/ the lowest at around T4 (See Figs. 1–3; the higher the index scores are, the better the performance in the corresponding domain is).

Fig. 1: Comparison of LSASS1.
figure 1

Figure 1, 2, 3 compare three key index scores of LSA from Coh-Metrix (LSASS1, LSASSp, LSAGN) and find a commonality pertinent to the less proficient participant (Lee): Over the beginning session of TtR (T1- around 4), Lee’s index scores of LSASS1, LSASSp, and LSAGN to different extents have remained above the general-level counterpart representing the average task performance by the whole class and the other two’s, while the other two have remained below the general level roughly (LSASSp and LSAGN) and partially (LSASS1).

Full size image
Fig. 2
figure 2

Comparison of LSASSp.

Full size image
Fig. 3
figure 3

Comparison of LSAGN.

Full size image

Firstly, based on the definitions of three LSA indices, it is safe to presume that the less proficient student (Lee) outperformed the other two in both local and global cohesion over the period from T1 to around T4. Figures 1–3 compare three key index scores of LSA from Coh-Metrix (LSASS1, LSASSp, LSAGN). Specifically, the comparison of the scores reveals: Over the beginning session of TtR (T1- around T4), Lee’s index scores of LSASS1, LSASSp, and LSAGN, to different extents, were constantly higher than the general-level counterparts (representing the average task performance by the whole class) and the other two’s. By contrast, the other two have remained below the general level roughly (LSASSp and LSAGN) and partially (LSASS1).

Namely, Lee’s (low-level) performance in inter-sentential cohesion seemed to be polished better than others’ during the period of the first four/five repetitions, instantiated by apparent increases in the scores of LSASS1 (to around 0.26, Fig. 1) and LSASSp (to around 0.31, Fig. 3) respectively. By contrast, a roughly better performance in the same aspect by Monica (intermediate level) than Jack (high level) was also detectable: Jack’s index scores have roughly remained the lowest (LSASSp and LSAGN) and not outnumbered the general-level ones until certain timing after T3 (LSASS1).

Another finding is that the middle session of TtR (T4–7/8) witnessed differences: The performance in inter-sentential cohesion and semantic overlap of the intermediate-level student (Monica) was generally better than others’. For instance, the scores of Monica in LSASS1 stood the highest at around 0.22 at T5 and 0.24 at T7 respectively, while Lee’s (low-level) and Jack’s (high-level) have remained fluctuating (LSASS1) around or right below the general-level scores (LSASSp, LSAGN) throughout the session.

Nominalization’s

The findings at the step are showcased to align with the three genre stagings: DESCRIPTION, EXPOSITION, and CONCLUSION. At each staging, firstly the frequency of nominalization cases produced by three students was compared. Then the selected nominalization cases (the ones collected from different tasks produced at key timings or the ones with a higher frequency of repetition) were analyzed.

At DESCRIPTION

At the genre staging of DESCRIPTION (the opening of the essay), nominalization has been absent throughout the beginning session of TtR, while made its debut till T4 (Jack) and T5 (Lee and Monica) (Fig. 4).

Fig. 4
figure 4

Frequency of nominalization in DESCRIPTION (the opening).

Full size image

The absence of nominalization may indicate these emergent student writers were not adequately informed of genres valued by academic English writing, and the debut of nominalization may imply a change in their genre knowledge throughout TtR.

Table 1 compares the two descriptive openings before and after the debut of nominalization in Lee’s case. In the excerpt, Lee firstly realized his descriptive meaning congruently by deploying clause-based patterns in T4 “they are translated into …”, while then the similar meaning was incongruently reconstrued as a nominal group --- “…with its English translation as…” in the next task (T5).

Table 1 Comparison of pre- and post-employment of nominalization at DESCRIPTION by Lee.
Full size table

The change above can be triangulated by Lee’s response in the interview:

To write like this [in T5] reads more objective because firstly people as subjects should be omitted, as a teacher for my B.A. thesis suggested… Secondly the latter [writing in T5] seems to be more academic, just like what I have read in a journal recommended by the teacher [tutor].

(Strongly agreed by other interlocutors)

Lee’s words above may imply that his employment of nominalization possibly resulted from his changing individual perception of valued genre features of academic English writing, which is rooted in his prior experience. This echoes that for the purpose of attenuating the unfamiliarity brought by new genres, writers primarily resort to their established writing experiences including “antecedent genres” as Devitt (2004) defined, or to the familiar genres previously experienced (Tardy 2012). What should not be ignored here is that it was TtR that accommodated such change.

Nevertheless, an interesting finding here is that nominalization in the case cannot be consistently considered meaning-making attempts of students. The case of “translation” aforementioned by Lee was the only one and dictated his employment of nominalization at DESCRIPTION. Such monotonous and repetitive employment is problematic in that it may reveal a trace of linguistic fossilization. A more typical case is Monica’s, where her employment of a nominalization case (…, the intention of which is to …) at this genre staging was presented once a task continuously from T5 to T9. This may indicate such stable and repetitive employment can be ascribed to a hypothesis that the nominal phrase has been fossilized into her inter-language system of meaning-making. Table 2 showcases the concordance of this fossilized nominal phrase.

Table 2 The concordance of fossilized nominal phrase at DESCRIPTION by Monica.
Full size table

As shown by Table 2, Monica incongruently reconstrued PROCESS “…intends to do…” into PARTICIPANT/ENTITY “…the intention of which is to …” as Head of the THEME in the subsequent clause to elicit different new meanings, (to realize different purposes in the cases above). Nevertheless, noteworthy here are the student’s attempts to strengthen cohesion by deploying nominalization as THEME of a relative sub-clause rather than a new clause, as Monica expounded in the interview:

I would have written “its intention is to …”, but the word “its”, in my opinion, is not an explicit anaphora because the previous sentence contains several singular nouns presumably referred by this word [“its”]….the relative word “which”, according to my grammatical knowledge, can refer to a previously-stated singular noun, a nominal group, and even the intact main clause. In this sense, I think the meaning would be more explicit and cohesively expressed. And to write like this, frankly, can also eliminate my grammatical errors and upgrade my syntax.

The interview data may underpin a finding that nuanced from anaphoric reconstrual (to grammar-metaphorically reconstrue previous congruent meaning into nominalization as THEME to elicit NEW meaning, Liardét 2016), Monica de facto employed nominalization as NEW meaning itself of THEME rather than the anaphoric reconstrual of the GIVEN information (“intention” had never been explicitly or implicitly mentioned in the previous sentence). However, the anaphoric purpose was achieved by complicating the syntactic relationship between clauses.

Furthermore, Monica’s deployment is also discrepant from elaborated nominal groups (Liardét 2016), because in the interview she confessed the absence of her knowledge pertinent to elaborated nominal groups. When the question “why to use a relative clause (“which”) rather than a nominal group containing previously mentioned information” was asked, she further explained that her possible attempts to use nominal groups (if informed of such knowledge) could have been hindered by her apprehension of “making grammatical errors” and thus “having no confidence to make clear the intended meaning”.

Above all, the fossilization may also shed new light on understanding metaphorical control in Liardét’s framework, because it seems to be problematic to identify “…the intention of which is…” as a full-control case: Though PROCESS “intend” was successfully reconstrued into ENTITY “intention”, the relative word “which” was not a GM case. Namely, anaphoric reconstrual may be individually mediated and realized (by grammatically complicating the syntax in the case).

Another possible reason for the fossilization above is that the attention resource of writers saved by the increasing familiarity with tasks enabled by TtR would be reallocated to linguistic forms rather than meaning-making. This echoes that syntactic focus would be more highlighted than meaning-making by students when re-encountering a task (Fukuta 2016). In this regard, the textual realization of meanings may be impeded by such familiarity, because writers’ attempts to formulate new meanings in new tasks may surrender from the easy retrieval of fossilized linguistic formulaic expression stored in working memory, as Monica mentioned in the interview:

I learned this expression […, the intention of which is to] from a journal article I read at week 2 or 3, and I think it is more academic than “in order to do something”. After using it for the first time, I found it can be used for all following tasks because the task prompts and the main content we should write in the opening [paragraph] are the same. In this sense why not to use such a better expression when the meaning is also understandable?

By contrast, the meaning-making activities of Jack are more intricate. In spite of the complete absence of nominalization over the beginning session (T1-3) and its intermittent presence throughout the rest TtR session, Jack seems to be more capable of making varied meanings-- the frequency of nominalization was constantly higher than the other two participants’, for instance, 3 cases in T8, the highest among all.

As Table 3 showcases, it seems to be rational to conclude that Jack, along with the implementation of TtR, outperformed the other two in the aspect of metaphorical enrichment by drawing upon different nominalizations to condense or conceptualize meanings at the genre staging of DESCRIPTION. For instance, diversified meanings were grammar-metaphorically reconstrued, such as from PROCESS “abbreviate” to ENTITY “abbreviation” (T4) and “express” to “expression” (T8), from EPITHET “significant” to ENTITY “significance” (T4), and “inappropriate” to “inappropriateness” (T10), etc. Nevertheless, the conclusion above is possibly perfunctory in that several cases concerned do not dovetail with the genre of the DESCRIPTION required by tasks, i.e. several cases such as “significance”, “information-provision”, and “inappropriateness” de facto serve the purpose of exposition and evaluation rather than description. This implies authors’ individual knowledge of the genre (Tardy 2009) may deeply resonate with their incongruent meaning-making activities, as Jack responded in the interview:

At the beginning, I did not think too much [about task prompt as “describing the picture”] at that time [of writing] and also thought that it should be acceptable to add these points [nominalization aforementioned] to express my opinions. After writing several essays, I felt that just to describe the picture, as I did in the first three tasks, made me a little confused. Is it too superficial to write like this, I mean, for academic writing?

Table 3 Comparison of nominalization in DESCRIPTION by three participants at key timings throughout TtR.
Full size table

As Johnstone et al. (2002) argued that students’ genre knowledge may grow tacit by repetitively re-encountering and engaging in similar genres, Jack’s case above may shed new light on this: TtR can arouse students’ reflection on such tacit genre knowledge when realizing the intended ideational meaning, instantiated by “Is it too superficial…”.

At EXPOSITION

At the genre staging of EXPOSITION (the body paragraph of the essay), the nominalization deployment reveals a more dynamic developmental trajectory (see Fig. 5). As such it is more significant to compare how the staged trends of the three may change rather than a detailed description of each. Overall, the middle session (from around T4 to T8) witnessed a relatively higher frequency of nominalization of all three than the counterpart over the rest of the TtR session.

Fig. 5
figure 5

Frequency of nominalization in EXPOSITION (the body).

Full size image

Specifically over this session, the frequency of employing nominalization by Jack has roughly remained above 10 from T4 to T8 (13 cases, the highest), while the other two’s have fluctuated around 6 and 5 cases respectively with both the highest emerging at T4 (9 and 7 cases respectively). As aforementioned, the repetitive employment of a nominal case in a task would be counted only once. In this regard, it is safe to presume that the higher frequency of nominalization may imply students’ increasing attempts to make new meanings over this TtR session.

To a certain extent, the increase may be ascribed to students’ growing knowledge of genre features of academic writing. As Jack responded in the interview:

Frankly I did not clearly know why I wrote like this [with nominalization]. I guess this lies in that after several times of writing, I found that I had to include all necessary meanings into a single paragraph as the task required, you know, such as to invoke and introduce the translation theory, to evaluate and discuss, and to analyze the translation inappropriateness, etc. In this sense, to write like this [with nominalization] can make my essay compact because I have so many points to write.

In this sense, nominalization was employed to achieve the compactness of written language, which is intimately correlated to the higher informational density privileged by genre conventions of academic writing. This may be perceived as the evidence of resonance between Jack’s growing genre knowledge and his meaning-making activities enabled by TtR, because “to invoke, introduce, evaluate, discuss, and analyze” mentioned above not only reveals his different meaning-making intention but also his individual perception on genre conventions concerned. For instance “to invoke and introduce” may primarily involve the genre of exposition, while “to evaluate, discuss, and analyze” would be correlated more to the persuasive.

Table 4 compares all instances of nominalization employed by the three students at four key timings (T1, 4, 8, and 10) at EXPOSITION. Overall, though the larger content capacity of this genre staging is likely to afford more potential chances of deploying nominalization, it is presumably reasonable that the genre of EXPOSITION may have more affordances for incongruent meaning-making.

Table 4 Comparison of nominalization in EXPOSITION by three participants at key timings throughout TtR.
Full size table

The first important finding is the nuanced improvement in metaphorical enrichment and metaphorical control presented by the three students. For instance, Jack incongruently reconstrued the congruent meaning “[Something] is translated in a wrong way” into “mistranslation” for the evaluative purpose at T1, while after three times of repetition it was metaphorically enriched into “inappropriateness” at T4 (inappropriateness = improper translation but may not be wrong) and further “inaccuracy” (T4) explicitly specifying the facet of “wrong”. Furthermore, unlike Zhou and Liu’s findings (2017) that intermediate GM was more prevalent in texts written by Chinese EFL student academic writers than Anglophone counterparts, the study found that the six patterns of intermediate metaphorical control (Liardét 2013; 2016) were rare throughout all textual data (only one case actually). However, the improvement in both metaphorical control and enrichment in this rare case can still be identified: At T4 Lee employed gerundive nominalization, “the understanding of…” to realize the meaning of “what we understand about …”. Then after six times of TtR, the similar congruent meaning was fully reconstrued as “comprehension” (understanding at a deeper level).

However, the deployment of the enriched metaphor above seems to be form-elicited without obvious writers’ agentive choice in meaning-making, as Lee mentioned in the interview:

Actually I am not clearly aware of difference between “comprehension” at T10 and “understanding” at T4. To some extent, I just paid attention to diction. I mean “comprehension” seems to be more formal and academic while “understanding” is too common and informal, possibly.

The second finding is that a wider range of incongruent meaning-making attempts enabled by subgenres involved in EXPOSITION may mitigate the fossilization identified at DESCRIPTION (Monica’s case). Though three students employed nominalization to reconstrue meanings in more diversified ways at EXPOSITION as shown by Table 4, the trace of fossilization has not been detected at this stage. As the typical case of fossilization at DESCRIPTION, Monica responded to this in the interview:

As to the phrase [“the intention of which is”] I repetitively wrote in my openings, it is more difficult, I think, to use something like this [the phrase] in the body paragraph, because though the [task] prompts are the same, what you are going to write here is different, unlike the opening you just need to describe. Actually it would read weird, you know, the meanings are possibly not naturally related with each other.

In this sense, seemingly TtR may also function as an affordance to enable the student’s reflection upon the alignment between form and meaning, and again the genre knowledge triggered by task requirements (as Monica mentioned “unlike the opening you just need to describe”) also facilitates such reflection.

At CONCLUSION

Figure 6 demonstrates the trajectories of the frequency of nominalization by three participants at CONCLUSION. Overall, the high-proficiency student (Jack) employed nominalization relatively more frequently on average, with the highest frequency being 4 cases at T4 (but none at T2 and T9). By contrast, the lower-proficiency students employed less nominalization, with more tasks completely excluding it (seven tasks for Lee and three for Monica respectively).

Fig. 6
figure 6

Frequency of nominalization in CONCLUSION (the ending).

Full size image

The first noteworthy is that Jack (high-level) and Monica (intermediate-level) shared a commonality: The frequency of the two ascended after T2 to reach the peak at T4 (4 cases and 3 respectively), while followed by immediate declines sustaining through several subsequent tasks repetitions. This may imply a non-linear development here. The second is that the absence of nominalization in Lee’s case was obvious (see Fig. 6). This may be ascribed to Lee’s uncertainty whether or not his linguistic repertoire can afford explicit meaning-making. More importantly, a sign of reflective awareness of generic features at CONCLUSION can be identified, as Lee responded in the interview:

I don’ know [why nominalization has been completely absent for seven tasks out of ten]. I just felt it was difficult to summarize what I had written above in the essay… too much content to be concluded in a clear way for me. And I thought here a noun, or a noun phrase could not express explicitly what I intended to summarize.

Table 5 illustrates the cases of nominalization in CONCLUSION at key timings employed by three participants respectively. The following discussion excludes Lee’s case because of the absence of nominalization in his case.

Table 5 Comparison of nominalization in CONCLUSION by three participants at key timings throughout TtR.
Full size table

As to the other two (Jack and Monica), first, a shared decrease in frequency after T4 may implicate the growth of students’ knowledge of genres. As shown by Table 5, both Jack and Monica incongruently reconstrued their intended meanings either for different analytic or reasoning purposes, which are usually excluded from valued genre conventions at CONCLUSION. Specifically, the nominalization cases “alternative”, “relation”, “contrast” by Jack and “intention”, “faithfulness”, and “easiness” by Monica at T4, explicitly or implicitly, serve for analysis-related purposes.

Second, certain traces of improvement in logogenetic impact and metaphorical enrichment can be identified. Take “…a polished alternative to the previous version can result from …” in T4 as an instance (see Table 5). As Jack interpreted in the interview, his exact intended meaning is “It is better for us to replace the previous version because …”. In this regard, the incongruent reconstrual of PROCESS “replace” into ENTITY “alternative” by Jack. This, on the one hand, partially manifests an elaborated nominal group (Liardét 2016) that the concurrent meanings (the EPITHET “better” and the PROCESS “replace”) can be reconfigured into an intact lexically dense nominal group “a polished alternative to translate”. On the other hand, higher-level of metaphorical enrichment can be identified (“alternative” = “a choice+ available to replace, and “polished” = “improved in nuanced ways ”, though the latter is not a GM case). However, the enriched meanings in the above cases may serve for the analytic purpose, manifested by the causal logical relator (“result from”). Likewise, “easiness” employed by Monica (T4) was obviously used as a warrant to support the claim. Both are not conventionally valued by the genres of writing a conclusion.

The inappropriate meaning-making at CONSLUSION aforementioned may result from student's inadequate genre knowledge required, specifically the rhetorical and formal knowledge of genres proposed by Tardy (2009). As Jack expounded in the in the interview:

I think it [to employ nominalization for analysis] is likely to be one of my writing habits because I get accustomed to offering explanation or analysis after a point has been made, because, you know, just a point is not reliable and persuasive enough…. I did not know that to add a new point in a conclusion is improper until the teacher told me at a certain class, and I tried to eliminate such inappropriateness from then on.

In this regard, it is notable that at T10, such inappropriate analysis-oriented meaning-making has been eliminated: both Jack and Monica employed nominalization to configure a conclusive statement without analytic reasoning.

Conclusion

Under Liardét’s (2013; 2016) framework of nominalization and with Coh-Metrix and interviews as instruments, the study aims to check the meaning-oriented developmental trajectory of students’ L2 academic writing and the noteworthy changes that emerged along with the implementation of TtR. The findings first suggest that the development of meaning-making in L2 academic writing in the case was presumably staged rather than constantly linear, during which language proficiency dynamically resonated with TtR. The development was presumably staged because: By examining each student case, the data of LSA indices (Figs. 1–3) and nominalization frequency (Figs. 4–6) seem to increase and decline periodically, that is, the linear development featuring proportional cumulation (e.g. along with the proceeding of TtR, the cumulative practice with more written tasks is supposed to elicit better performance) could not be constantly detected in three individual cases.

Specifically, the beginning session of TtR (from T1- around 4) witnessed that the low-proficiency student constantly outperformed the other two participants and the class average level in the aspect of inter-sentential cohesion and semantic overlap, while the intermediate-level student stood out over the middle session (T4- around 8). Second, at three genre stagings, students’ knowledge of genres deeply resonated with their grammar-metaphorically meaning-making activities throughout TtR, particularly in facilitating the individual reflection on their established writing knowledge. Also, the deficiency of genre knowledge may cause a wider range absence of nominalization at DESCRIPTION and CONCLUSION than EXPOSITION. The linguistic fossilization identified at DESCRIPTION implies that anaphoric reconstrual may be individually mediated (by grammatically complicating the syntax in the study). Finally, nuanced improvement in metaphorical enrichment and control was also identified.

As such, the study may add knowledge to our understanding of meaning-making in L2 academic writing development and shed light on the future pedagogical praxis: First, the staged developmental trajectory identified by the study reminds teachers of the importance of devising different and proper proficiency-based pedagogical intervention at key timings throughout TtR session, because in-time intervention at a certain key timing may elicit more desired progress. Second, teachers may draw on affordances provided by TtR as effective toolkits to enable students’ reflection on learning L2 academic writing for different purposes, for instance, to deepen students’ understanding of academically valued genre conventions by immersing students into the iterative learning process embedded in TtR.

Limitations and future research

There are several limitations of the study. First, in view of the elusiveness of the explicit definition of meaning, admittedly only a small portion of meaning has been unveiled in the current study, thus future study on other aspects of the concept in L2 academic writing should be encouraged. Second, though Coh-Metrix is meaning-making-related, a future probe is needed to specify the resonance between its various indices and the specific meaning-making activities in L2 academic writing, e.g. how an index score can better serve a pedagogically interpreting or diagnosing purpose. Third, the contextual limits (e.g. the institutionally set schedule of the course) constrained the frequency of TtR and the sample range in the study, which may not be adequate to reach the level of massed repetition advocated by Bygate (2001). Also admittedly, the aim of the study and the space limit impede the highly generalizable findings, thus calling for future research in this line with more participants. Finally, although the feature of TtR (to repeat the same task type with different content) may constrain students’ ideational attempts to some extent, the influence of nominalization on meaning-making can vary when more diverse genres or essay topics are involved.

I also reckon that the findings of the study may necessitate our reconsideration of the rapport between L2 learners’ attention and familiarity with writing tasks enabled by TtR, i.e. how the rapport may afford or constrain learners to acquire, apply, and reflect on the privileged L2 academic genre conventions, and how their genre knowledge may unfold concurrently with the proceeding of SLA when TtR is valued as a crucial variable. As such, a more inclusive theoretical perspective is needed, like Complex Dynamic Systems Theory or Sociocultural Theory.