November 2021 – Volume 25, Number 3
Shaun Justin Manning
Hankuk University of Foreign Studies, Seoul, South Korea
<sjmanninghufs.ac.kr>
Abstract
This study investigates the influence of a grammatical reconstruction task on talk and subsequent individual writing. In a South Korean university, two classes alternated learning conditions, between ‘read–summarize–write’ and ‘read–reconstruct–write’. In the summarize condition, small groups collaborated to create a summary of the reading; in the reconstruction condition, small groups collaborated on a summary reconstruction task. This was an activity in which the teacher first wrote a summary of the reading and then removed all the grammar words, punctuation, and capitalized letters, leaving only the content words. Students had to reconstruct the original summary. In the last phase of the lesson, students wrote an essay on the same topic as the reading. The in-class talk was analysed for differences and the complexity, accuracy, and fluency of the writing were calculated. Results showed the collaborative summary prompted discussion about the ideational content of the summary, whereas the reconstruction task promoted talk about which grammatical forms to use. Subsequent individual writing was impacted; accuracy improved after the reconstruction task with no loss of fluency. Fluency increased in both conditions when compared to pre-tests, while syntactic and lexical complexity of the writing was unaffected by condition.
Improving the lexico-grammatical accuracy of learners’ writing is among the goals of university EFL writing instruction. Unfortunately, learners often repeat the same mistakes, despite being able to identify and correct these same mistakes when identified by the instructor – indicating that learners have linguistic knowledge but have not deployed it. This poses the problem of creating a situation in which learners produce accurate writing – to the limit of their competence – without reducing the fluency or complexity of the produced language.
A solution may lie in teachers taking advantage of the situated nature of the language classroom. Language classrooms, language learning tasks, teachers, and learners are interconnected elements of both larger and smaller systems (levels) existing over time (Larsen-Freeman & Cameron, 2008). The elements and agents in the classroom intersect at a point in time and space (a classroom on a day and time) with a socially defined event (a lesson) that has an objective (teach/learn “something”) with a longer-term goal (e.g., learners may want to get a job). “The lesson’s event connects to previous events and to events yet to happen,” (Larsen-Freeman & Cameron, 2008, p. 201). This suggests that a performance of a task in class will be influenced by: the conditions of that task (Robinson, 2001, 2011; Skehan, 1998, 2009); that lesson’s pre-task activities (Willis & Willis, 2007); and all the learning histories of each individual doing the task (Larsen-Freeman & Cameron, 2008).
Rather than look at the writing task as an isolated event and examining its performance as the result of the task design (Kuiken & Vedder, 2007; Robinson, 2011; Skehan, 2009), this study imagines it as a stage of a “learning history process” that is influenced by earlier lesson activity (Larsen-Freeman & Cameron, 2008), with the immediately preceding task being the most salient history. Viewing writing tasks this way encourages instructors to arrange task sequences in such a way that prior activities in the sequence prompt learners into performing later tasks in the way that is most beneficial to their learning needs. For example, performing tasks that require a high degree of linguistic accuracy (e.g., dictation) might create a classroom context in which there is a residual emphasis on accurate production, leading to more accurate language production in an immediately following individual writing activity. Should this be borne out, it would be of help to teachers when creating sequences of tasks to address their learners’ needs.
Tasks and language learning
Since the emergence of task-based language teaching (TBLT) in the 1990s, much has been researched about the various ways tasks are designed and implemented in ESOL classrooms. TBLT holds that language use leads to language learning, yet the mechanism for how this happens has been under considerable, as yet unresolved debate. Early explanations focused on the cognitive processes of noticing a gap between what one needed to say and what one could say (Schmidt, 1990). The task would be designed in a way that learners would notice this gap, clarify it through negotiation of meaning (Long, 1996), or they would engage in language related episodes – talk about language (Swain & Lapkin, 1998) and thereby learn what they did not know. This noticing, theorized to happen in real time, relied on learners to first notice and then to take the initiative to articulate what was noticed. A delayed approach to noticing was proposed by Willis (1996). She suggested providing samples of native speakers’ talk when doing the same task for learners to examine after doing the task themselves. Through this analysis, learners would notice what they had not known, and this noticing could then lead to learning (Willis, 1996; Willis & Willis, 2007). Long’s approach relied on incidental acquisition of items at the level and need of the learner, whereas the Willis’ (Willis & Willis, 2007) approach allowed for more formal pre-planning of items by the teacher, depending on what the teacher felt the learners required. However, they still insist that the curriculum be sequenced by tasks rather than by linguistic items – i.e. a task-based syllabus (Samuda & Bygate, 2008, p. 58).
Another way of conceptualizing how task performance leads to language learning stems from a sociocultural view of education in which learning is viewed as a process mediated through interaction between learners and experts (usually the teacher) (Vygotsky, 1986). From this view, the environment of the task performance affords learning opportunities through students engaging in languaging (Swain, 2010) or otherwise assisting each other, for example by waiting, using gestures, finishing each other’s utterances, etc. (Ohta, 2001).
While acknowledging the merit to a sociocultural post hoc explanation of how tasks aid language acquisition, many researchers and course designers argue that since we do not know a priori what sociocultural conditions will be present at the time a task is performed, only the characteristics of the task itself should be considered when sequencing them into a curriculum (Robinson, 2001, 2011; Skehan, 1998). Therefore, a great deal of research has been carried out investigating various aspects of tasks in order to develop a theoretically valid approach to sequencing.
Recently, Kim, Jung, and Tracey-Ventura (2017) pointed out that less work has been done investigating task performance in an existing learning context. Some prominent studies that have looked at situated tasks include McDonough and Chaikitmongkol (2007) and Van den Braden (2006), who both report longitudinal studies. Others have investigated single tasks in classrooms, either as proposed by educational authorities (e.g., Carless, 2004), or as part of a researcher’s agenda (e.g., Kim & McDonough, 2011).
What none of these studies has investigated is the dynamic impact a task may have on an immediately following activity within the same lesson. Ellis’ (2017) position paper on task-based teaching, while critiquing the previous research, discusses the issue of implementation conditions – how students go about doing the task – affecting task performance and task sequencing. Implementation conditions include for example, how much time students have to plan what they will say prior to performing the task or whether or not they can practice the task prior to doing it (Ellis, 2017, p. 513). However, in his review, the implementation conditions refer only to the task under investigation, not to any previous classroom activity. He does not address the idea that a previous activity, even one focusing on a different skill, could serve as an implementation condition for a later task, and in turn, alter the performance of that later task. In short, there has been no research to date on the idea that when students do one task that requires a certain mindset and then do another task, there might be a carryover of the first mindset that influences performance of the second task. To date, tasks have only been investigated as standalone activities, and efforts at sequencing tasks have focused only on the characteristics of the individual tasks and not at their relative effects on each other.
It is hypothesized that performing a task that naturally promotes discussion of grammatical items will lead to activation of grammatical knowledge in the learners, which remains activated during a subsequent in-class activity. This “activation” could work either by making the sub-conscious conscious (cf. Schmidt, 1990) or by creating a classroom ecology (Van Lier, 2004) or trajectory (Larsen-Freeman & Cameron, 2008) in which grammatical accuracy is given greater standing than it would have if following typical tasks that prize meaning over form (Ellis, 2003).
If it could be shown that the focus (on meaning or accuracy) of a preceding task influenced a later task, it would be of use to teachers who wish to integrate tasks into an existing curriculum – those using a task-supported language teaching (TSLT) approach (Samuda & Bygate, 2008, p. 60), which has been reported to be the main way that tasks are used in Asian contexts (Butler, 2011).
Methods
This study investigated changes in individual essay writing performance created by changed requirements of its preceding task. Specifically, it asked if performance of a collaborative task requiring linguistic accuracy prompted talk about language (i.e., languaging, Swain, 2010), and if this were performed immediately before an individual writing assignment, would the writing be more accurate, without sacrificing fluency or complexity? The research questions were:
- Do students differentially attend to form or meaning when performing a collaborative summary vis-à-vis a collaborative reconstruction task?
- Is the complexity, accuracy, or fluency of an individual’s writing differentially influenced by performance of a summarizing or reconstruction task immediately prior to writing?
Setting and participants
The study took place in two essay writing classes for general students (non-English majors) at a university in Seoul, South Korea. Each class met once per week for two hours. The lessons followed a “linked skills approach” (Nation, 2009). Skill 1 was reading and analyzing authentic texts (essays, news editorials, and the like), and followed by (skill 2) discussing the contents in small groups, followed by skill 3, independent writing in class (see Table 1). For the present investigation, skill 2 (discussion) was changed to collaborative summarizing, in which speaking and writing were integrated. The nature of the summary – freely written or reconstructed – was varied. The in-class essays were collected and photocopied, and the originals returned to the students for finishing as homework.
Prior to data collection, I asked the students for permission to use their classroom recordings and essays. They were given an opt-in sheet to sign; any sheets unreturned or left blank were assumed to be opt-outs. Students received no bonus for opting in and no penalty for opting out. Opt-outs attended the lessons and did the tasks because the tasks were part of their curriculum. I did not use non-participants’ data, or the data of their group members, in this study. [1]
There were 18 participants. There were more students registered in the class, but some did not wish to participate, and the remainder did not attend all three lessons. This reduced the N to nine per class because the non-participants’ data, along with that of their group members, was removed. The author was the classroom teacher for both classes.
Data collection
The study took place between the 10th and 12th week of a 16-week semester (See Table 1). By the 11th week of the semester, students were familiar with the class procedure, and the researcher could create a study that served the existing goals and needs of the participants – i.e., if it had been established that accurate production was not a need for these students, this study would not have taken place. An in-class essay in the 10th week served as a baseline, and then the classes alternated treatments. In weeks 11 and 12, each class did the same reading activity, followed by either a collaborative summary or a collaborative reconstruction task. Finally, the students did a 30-minute, timed individual writing. All the essays received feedback on their contents, organization, and language but the week 10 and week 11 papers were returned at the end of week 12 class, and week 12’s paper returned in the next class (week 14, due to a national holiday) to avoid any influence my feedback might have had.
Table 1. Research Design
Week / Topic |
Class A | Class B |
10 (Pre-test) Education |
Timed essay (30 min) | Timed essay (30 min) |
11 News Media |
Reading (45 min) | Reading (45 min) |
Reconstruction task (15 min) Feedback (10 min) |
Summary task (15 min) Feedback (10 min) |
|
Timed essay (30 min) | Timed essay (30 min) | |
12 Presidential Election |
Reading (45 min) Summary task (15 min) Feedback (10 min) Timed essay (30 min) |
Reading (45 min) Reconstruction task (15 min) Feedback (10 min) Timed essay (30 min) |
14 | All three essays returned in class. (Including teacher’s feedback) Individual consultations outside of class hours |
Materials
There were different materials for each phase of each lesson. The first phase was a reading lesson using teacher-made vocabulary and reading comprehension activities supporting the reading of an online news editorial.
The second phase of the lesson was the treatment phase. Class A received a collaborative summary task, while Class B performed the summary reconstruction task. (See Figs 1, 2, and 3 below for examples.) The following week, classes A and B switched tasks.
A meaning-oriented task: collaborative summary [2] (S). To do the summary, students work in teams of three or four to make an approximately 100-word summary of that article. One student served as the group’s writer, while the others suggested what to write.
A language-oriented task: reconstruction (R). In reconstruction tasks (R), learners must return a corrupted sample of language to its original form (cf. Willis & Willis, 2007). In this study, the original was a teacher-written summary. In this case, the corruption was the removal of all grammar words, capital letters, and punctuation (leaving only nouns, main verbs, adjectives, and adverbs) in their correct form and order. The students then had to re-create the summary by adding in any necessary items without changing the form or order of the words on the page.
The third phase was individual essay writing. Students were given 30 minutes to write an in-class TOEFL-style independent essay. The TOEFL independent essay is an argumentative or discursive essay in which the writer is expected to produce approximately 300 words in 30 minutes (a written fluency of 100 words per minute). It is rated according to development, organization, and facility of language use. Language use is described as: syntactic variety, appropriate choices of words and expressions, and lexical and grammatical accuracy (TOEFL Writing, 2021).
The Week 10 pretest was: “What are the qualities of a good university?”, Week 11 was, “What are the qualities of a good newspaper?”, and week 12 was “What are the qualities of a good president?” These prompts were on the same topic as the readings but did not require the student to directly respond to the reading because an essay responding directly to the reading might result in students lifting structures and ideas directly from the reading, skewing the results.
Data
The data for this study were: student essays written during class. audio recordings of task talk, and interviews with volunteers after the semester had ended. All classroom talk was recorded, transcribed, and analyzed for language related episodes – essentially points in the talk at which students talk about the language (vocabulary choice, form, etc.) they are using (see Swain & Lapkin, 1998) and propositional content (ideas).
The essays were photocopied before grading and returning. Because the essays were handwritten during class, after the semester ended, they were retyped taking care to preserve the text in its original form. They were entered into Text Inspector (Text Inspector, 2016) that gave a word count and vocabulary diversity (VOCD) measures. The essays were divided into t-units and their related clauses, which were then analysed for errors.
There were four complexity, accuracy, and fluency (CAF) measures: errors per 100 words (% error) as a measure of accuracy, words per minute as a measure of fluency, clauses per t-unit as a measure of syntactic complexity, and VOCD as a measure of lexical diversity (see Ellis & Barkhuizen, 2005; Housen et al., 2012 for more details on CAF measures). These measures correspond to the TOEFL independent writing test’s rating criteria of lexical variety, syntactic variety, linguistic accuracy, and a sufficient writing rate (approximately 100 words per minute) (See the previous section).
Results
Does the in-task talk differ between summarizing and reconstruction tasks?
Because it has been shown that students perform tasks differently than task designers intend (Coughlan & Duff, 1994), the study first investigated the interactive tasks to see if the collaborative summary prompted an orientation to meaning, and the collaborative reconstruction task prompted an orientation to language as predicted. Specifically:
- Do students attend to form or meaning when performing a collaborative summary?
- Do students attend to form or meaning when performing a reconstruction task?
To answer these questions, audio recordings of task performance and end-of-semester interviews were used.
Characterizing task performance of a collaborative summary. The collaborative summary was almost completely oriented to ensuring the summary’s contents matched the article. Table 2 demonstrates the meaning-oriented collaborative summary talk.
Table 2. Transcript of a collaborative summary talk*†
43 | Frank | so let’s start our summary . this article, okay. this article. |
44 | Robin | this article . . . is about the inter-corruption of the dsc |
45 | Frank | i think that write than inter-corruption ? it should be like 근본적인 문제(fundamental problem) . about the existence of the dsc . |
46 | Peter | 근본적인 문제? (fundamental problem) |
47 | Frank | should we? we can ? |
48 | Peter | fundamental problem ? basic problem ? |
49 | Robin | let’s just put then . this article talks about their inter-corruption of dsc and its fundamental problem. |
50 | Frank | OKAY! 끝! (finished!) let’s start writing . |
51 | Peter | (writing) this article is about |
52 | Frank | interna… and |
53 | Peter | 근데 이런 거 안 쓰는 거 라고 배웠는데, this article같은, summary 할 때. 그냥 바로 시작하는 거라고. (But I learned that it is not necessary to start with “this article” when you write a summary) uh…but dsc is full of corruption. |
54 | Frank | let’s say dsc DEFENSE SECURITY COMMAND |
55 | Robin | sorry ? |
56 | Frank | we have to write the full name of dsc . |
57 | Peter | 가장 중요한 거 뭐지 ? (what is the most important?) the most of important problem is that… it is corrupted but the army cannot [touch] |
58 | Frank | [touch it] yeah . |
59 | Peter | touching army 고칠 수 없다 (Cannot reform it.) not able to reform. |
60 | Frank | then write the . fundamental of all related problems of dsc is that although something, that things that they do ? you know ? they are the . you have to put fancy words here |
61 | Robin | Okay. |
62 | Frank | so fundamental problem of .. |
63 | Robin | um. |
64 | Frank | the fundamental problem of … dsc related. Ahhh (sigh) |
65 | Robin | 이거 쓰고 싶었는데. (I would like to write this.) is it the problem of dsc corruption or the existence of it |
66 | Peter | first of all the corruption and cannot be purified … again and then their-it is corrupted and why people cannot touch it . and then it comes to that the president himself gives this agency more power |
67 | Frank | uh huh |
68 | Peter | and then ? |
69 | Frank | and then, it points out that if (extending) |
70 | Peter | and then it points that if the gaining of political power is extended . it’s- it’s searching. |
71 | Robin | searching ? |
72 | Robin | uh, surveillance |
73 | Robin | OH |
Note: *Korean language is written in Hangul and translated in parentheses immediately following it. †Jeffersonian transcription conventions are used (See Appendix).
Frank, Robin, and Peter (pseudonyms) have read the article and now start to work out the writer’s meaning. The original article called for the elimination of a spy agency (the Defence Security Command), but one person is not sure. Frank says they should write that the fundamental problem is the existence of DSC (turn 45). In turn 65, Peter asks to confirm if the problem with the DSC is corruption or its very existence – indicating he is unsure of the writer’s intention. Turns 66-70 continue the discussion on this point.
Korean (L1) is used to clarify understanding (turns 45, 46 and 59). The use of L1 is not highlighted here because students are “not using English” or some other negative reason, but rather to point out that the students are not discussing English language, but the contents of the original article. Their L2 English is not sufficient for ensuring they each understand what is meant, so L1 is used.
There is also metalinguistic talk about the form of the summary (turn 53) and what one writer explicitly wants included (turn 65). Peter, in turn 53, says that they should not use “this article” when making a summary – an explicit reference to a discourse “rule” – one not taught in the present course – but one that is based on content. The words “this article” are redundant in a summary so they add no meaning and can be dropped. Throughout their conversation, the students rarely discuss the language choice of the final summary (i.e., engage in languaging) – with a potential case in turn 48 where Peter says, “fundamental problem? – basic problem?” He appears to be asking about a word choice, but no one responds, and Robin just inserts “fundamental” in turn 49. The talk is about the content and their (mis)understanding of it.
Their final summary is shown in Fig 1 (below). It emphasizes both the corruption of the organization and the fact that it is an untouchable (turns 58, 59, 66 of Table 2) “sacred cow” (Fig 1, line 2). Turn 66, 69 and 70 in Table 2 discuss the topic of changes in the agency’s power, which are shown in lines 3 and 4 of the summary. This shows the relation between the content of the talk and the content of the summary. There are student-made corrections in the summary: “it has” in line 1, and the removal of “it is not the” in line 5 but these were not mentioned in their talk – the writer made these changes himself without prompting. Moreover, the latter is not a grammatical change, but a change in the meaning of the sentence.
Figure 1. Example completed collaborative summary task (Peter, Frank, Robin)
When asked “What did you think about making a summary in a group?” in interviews after the semester had ended, Peter responded:
In the summary, we had … it was harder because we had to remember – uh understand the contents of the article and um sometimes we argued because we didn’t know what the writer means … sometimes it was good because we can checked our understandings [with] each other and learn what the article’s about… (Peter, post-semester interview)
Peter mentions that the challenge of the collaborative summary lay in demonstrating comprehension of the content. The advantage of the summary as a comprehension check is highlighted in his answer, and he refers to his partners as people who can help him understand the original. The relative lack of talk about language implies that the students’ attention is elsewhere, a conclusion borne out by Peter’s interview comment.
Characterizing task performance of a reconstruction task. In contrast, there is almost a total opposite focus when learners perform the reconstruction task. The reconstruction task (see Figs 2 and 3) involved the teacher first making a summary, and then deleting all the function words from it. The students then collaboratively rewrote it, one copy per group, by filling in the missing grammar words.
Figure 2. An example teacher-written summary
Figure 3. An example reconstruction task prompt with grammar words removed
Table 3 shows one small group, Julie, Henry, and Bruce, talking as they do the reconstruction task given in Figure 2. This is the same article that the students were summarizing in Table 2, above.
Table 3. Reconstruction task talk†
10 | Julie | this uh this editorial argues (2.0) um argues (3.0) |
11 | Henry | tch the editorial argues about the dis-mantling [of] (Bruce: [of]) dsc which starts about -with talking about recent incident which was . uh h* |
12 | Bruce | The editorial argues dismantling of dsc |
13 | Julie | (cough) (4.0) This editorial argues . about ? dis:mantling the d- the dsc? is it? (6.0) just dsc ? dismantling dsc |
14 | Teacher. | (approaching) you can say the dsc or dsc, |
15 | Julie | uh okay . [dsc] |
16 | Teacher | it depends on [the context] |
17 | Bruce | (ignores T) [talking] |
18 | Julie | it starts (1.0) talking about (2.0) about a recent incident ? (cough) uh |
19 | Bruce | a recent incident |
20 | Julie | that was about (3.0) |
21 | Bruce | that was a cover-up ? OF ? for ? |
22 | Julie | (11.0) is a cover-up senior officials something like that ? isn’t isn’t it that incident ? ((confirming what the word “incident” refers to)) |
23 | Bruce | this is the incident ((rustling papers – he may be pointing?)) |
24 | Julie | this part ? (cough) |
25 | Bruce | yes |
26 | Julie | to here. then. (3.0) how-how-what are we are supposed to do to WAS ? |
27 | Bruce | incident.. which was |
28 | Julie | it starts talking about a recent incident |
29 | Bruce | which was a cover-up |
30 | Julie | which was a ? which was a (cough) which was a cover-up, cover-up (2.0) for? or of ? h* |
31 | Bruce | which one is it? |
32 | Henry | cover-up OF something cover-up FOR something. |
33 | Julie | I think it is FOR . for for senior officials [paying sex] |
34 | Bruce | (softly) [paying for] |
35 | Julie | THEN: put (2.0) cover-up for |
36 | Bruce | paying for ? |
37 | Julie | yeah paying for and then ? or |
38 | Bruce | AND then blaming subordinates period |
† This transcript uses Jeffersonian transcription symbols
From the table, we can see that the entire interaction is an extended act of languaging (Swain, 2010). Their discussion is a process of students nominating possible combinations of words as they work through the prompt. They either accept the nomination by repeating it or moving on, or they reject it by suggesting something else or explicitly asking. Outright acceptance happens in turn 10 when both Bruce and Henry say “of” at the same time. Explicit talk about articles, prepositions, and how to connect words happens throughout the extract. For example, Julie asks about the need of the definite article in turn 14; Bruce asks about the preposition “of” or “for” in turn 21, which Julie returns to in turn 30. Also, in turn 26, Julie asks about connecting the verb “was” to the rest of the sentence. Bruce solves this in the next turn by saying, “incident, which was,” which Julie accepts. There is no reference to the meaning of the original article throughout the entire interaction.
Their final reconstructed summary is given in Figure 4. We see that the writer had made notations between the words on the prompt itself in an attempt to solve the puzzle. The final reconstructed text directly reflects the transcript. For example, the second sentence in Figure 4 reads: “It starts talking about a recent incident which was a cover-up for senior officials paying for sex then blaming on subordinates” (emphasis added). The underlined items, “which was a,” “for,” and “for” were all languaged in the discussion prior to inclusion in turns 29, 33, and 34 respectively.
In his interview, when asked “What did you think about reconstructing the summary from which I [the instructor] had taken out the grammar words?” Peter said:
That was like a puzzle … um- like checking out our grammar. We had to try to solve this by thinking of how the sentence makes uh- makes sense with the grammars…we know the teacher made the summary so it’s a good one (but) that makes a pressure to make a sentence in a right way… I wish you had put the punctuations … because that would make it more easier (Peter, end of semester interview)
The talk from the task performance and students’ recollections of how they performed the task indicate that indeed, these students were focusing on the grammar of what they were writing as they performed the reconstruction task. This is in contrast to how students typically performed the collaborative summary in which they spent most of their time discussing the meaning of what they had read with little emphasis on word choice or grammar.
Figure 4. Example completed reconstruction task
Does the reconstruction task prompt more accurate subsequent writing?
The next research question asked if doing either the collaborative summary or the reconstruction task prior to writing an individual essay would affect the complexity (VOCD and Clauses per t-unit), accuracy (errors per 100 words), or fluency (words per minute) of that essay. Specifically:
- Is the complexity, accuracy, or fluency of an individual’s writing differentially influenced by performance of a summarizing or reconstruction task immediately prior to writing?
There were two classes of students, each did both tasks, in a different order. Table 4 shows the mean and SD for each of the main variables for each essay. Looking down the table gives a within-group picture of changes, while looking across the table shows how each class related to the other. The first point to note is that fluency (words per minute) differed significantly between class A (9.73 wpm) and B (6.44 wpm) at the outset (Week 10), indicating that despite having been drawn from the same student population, they were not equivalent, weakening direct across group comparisons: class B was writing more slowly. It is unknown to what level initial proficiency level may be an intervening factor here.
Table 4. Means and SD for accuracy, fluency, and complexity measures
Class A n=9 |
Class B n=9 |
||||||||
Week | % error | WPM | Clauses/ T-unit | VOCD† | % error | WPM | Clauses/ T-unit | VOCD | |
10 pre-test |
mean | 7.26 | 9.73* | 2.46 | 81.23 | 9.88 | 6.44* | 2.33 | 81.14 |
SD | 2.69 | 3.50 | 0.45 | 16.90 | 5.14 | 2.86 | 0.38 | 17.51 | |
11 A: recon B: summ |
mean | 4.43 | 10.02 | 2.59 | 84.14 | 13.63 | 7.61 | 2.43 | 80.62 |
SD | 1.62 | 3.44 | 0.40 | 17.16 | 5.99 | 2.18 | 0.40 | 11.94 | |
12 A: summ B: recon |
mean | 13.57 | 11.44 | 2.27 | 86.74 | 9.48 | 8.90 | 2.72 | 84.60 |
SD | 6.00 | 3.00 | 0.43 | 24.70 | 5.26 | 2.55 | 0.62 | 23.54 |
* statistically significant difference, pcrit= 0.05; † = VOCD measures are an average of 10 calculations
Due to the small number of participants registered in the classes and willing to participate, non-parametric Friedman’s tests were used. The results are in Table 5. They show that only accuracy and fluency measures had statistically significant differences among the three essays. Kendall’s W indicates that the accuracy changes were large, and those in fluency were moderate. Complexity and lexical diversity measures did not change significantly, indicating that gains in accuracy or fluency did not influence the students’ range of syntax and vocabulary. This was somewhat expected over such a short time frame.
Table 5. Within-subject changes over three lessons (Friedman’s test)
Class A | Class B | ||||||||
Construct | Variable | Chi-square | df | p | Effect size | Chi-square | df | p | Effect size |
Accuracy | % error | 14.222 | 2 | 0.001* | 0.790† | 13.556 | 2 | 0.001* | 0.753† |
Fluency | WPM | 6.889 | 2 | 0.032* | 0.383‡ | 6.000 | 2 | 0.050 | 0.333‡ |
Syntactic complexity | Clauses / T-unit | 4.667 | 2 | 0.097 | 0.259 | 4.22 | 2 | 0.121 | 0.235 |
Lexical Diversity | VOCD | 0.667 | 2 | 0.717 | 0.037 | 0.00 | 2 | 1.000 | 0.000 |
* = statistically significant (pcrit =0.05), † = large effect size (Kendall’s W), ‡= moderate effect size
Table 6 gives the results of Post-hoc Wilcoxon signed-rank tests for accuracy and fluency measures, arranged by the conditions being compared: (P) individual essay pretest, (R) collaborative reconstruction before individual essay (R), and (S) collaborative summarizing before individual essay.
Table 6. Post-hoc test results for within-subject changes in Accuracy and Fluency (Wilcoxon signed rank tests)
Class A | Class B | ||||||||||
Var-iable | Con-dition | Difference | W | z | p | Effect size (r) |
Difference | W | z | p | Effect Size (r) |
% error | R vs P | -2.62 | 1 | -2.55 | 0.011* | 0.60† | 2.13 | 17 | -0.65 | 0.515 | 0.15 |
S vs P | 4.43 | 44 | 2.55 | 0.011* | 0.60† | 3.88 | 45 | -2.67 | 0.008* | 0.63† | |
S vs R | 7.05 | 0 | -2.67 | 0.008* | 0.60† | -1.75 | 0 | -2.67 | 0.008* | 0.63† | |
WPM | R vs P | 0.29 | 26 | -0.41 | 0.678 | 0.10 | 2.47 | 39 | 1.96 | 0.051 | 0.46‡ |
S vs P | 1.71 | 41 | 2.19 | 0.028* | 0.52† | 0.93 | 37 | 1.72 | 0.086 | 0.40‡ | |
S vs R | 1.42 | 5 | -2.07 | 0.038* | 0.49‡ | 1.53 | 11 | 1.36 | 0.173 | 0.32‡ |
* = statistically significant (pcrit =0.05); effect size, r = |z|/√N; † = large effect size; ‡= moderate effect size; P = pre-test; R= reconstruction task; S = summarizing task
Figure 5 shows the accuracy (% error) of class A (solid line) and B (dashed line). A reduction in error indicates improved accuracy. In week 11, class A did the reconstruction task prior to writing their essay and their errors decreased significantly (Table 4). Then in week 12, class A’s errors increased after doing the summarizing task – statistically significantly higher than both the pre-test (week 10) and after the reconstruction task (week 11).
Figure 5. Accuracy of student writing (% error)
Class B performed differently. Their pre-test essay and reconstruction condition essays (weeks 10 and 12) were not different in accuracy. However, their summarizing condition essay (week 11) was significantly less accurate than both the pre-test and reconstruction condition essays (Table 4). All the significant differences in accuracy had large effect sizes.
Figure 6 shows the fluency (words per minute) for both classes. Both classes improved in fluency over the three weeks of the study.
Figure 6. Fluency of writing (words per minute)
Class A and B improved in fluency, with class B’s most fluent writing coinciding with reconstruction (week 12) and class A’s with summarizing (also, week 12). Class A’s fluency (solid line) remained stable between weeks 10 and 11, and then improved statistically significantly in week 12 when writing after the collaborative summary (Table 4). Class B’s fluency (dashed line) was best after the reconstruction condition (week 12). Despite moderate effect sizes, the reconstruction task did not promote statistically significantly changes in writing fluency relative to either the pre-test or summarizing condition essays (Table 4). Looking at Figure 4, there is a steady increase in fluency (all moderate effect sizes per Table 4) from week 10 – 12 for class B.
Discussion and Implications
This study took place in two existing classrooms and the materials and tasks used were developed with the needs of these students in mind. This inherently limited the sample size and time available to conduct the study. The nature of the study allows it to catch the dynamics of this situation but costs it statistical power and generalizability. Despite its limitations, it points a possible way forward for future studies, larger in size and using repeated measures, that could investigate the relations between preceding and subsequent classroom tasks.
Allowing for the somewhat preliminary nature of this study, when we look at the findings a few key ones emerge. Combining the information from Table 6 and Figures 5 and 6 allows us to conclude that for class A, essay accuracy was best after the reconstruction task in which learners had been oriented to form, while fluency was best after the summarizing task in which they had oriented to meaning. The lack of change in fluency between the pre-test and the reconstruction condition indicates that the improvement in accuracy shown in Table 4 and Figure 5 did not bring about a loss of fluency. However, in week 12 (summary condition, oriented to meaning) the increase in errors combined with the increase in WPM implies that increasing fluency decreased accuracy. This suggests that the reconstruction task did not simply orient students to form, it activated their entire knowledge of language and made it accessible to them as they wrote, enabling them to write equally fast, with fewer mistakes. The summarizing task, on the other hand, oriented the students to the content of the article. Knowledge of form was not activated by the collaborative summary to the same extent as the reconstruction task, so in the follow-up writing task, linguistic forms were produced quickly, but with more mistakes.
This conclusion appears to not apply to class B whose accuracy was equivalent after the reconstruction task and the pre-test and was significantly reduced after the summarizing task (orienting to meaning). However, there was an increase in fluency between the pre-test and the reconstruction condition, where we might expect a decrease in fluency due to the orientation to form. Although Table 4’s comparison between R and P for class B shows the increase in fluency was not significant (p=0.051), there was a moderate effect size (r=0.46) indicating a gain that came with no loss of accuracy.
Class B’s starting point was different – it was less fluent and accurate than class A, yet it has drawn closer both in fluency (Figure 4) and accuracy (Figure 3). The latter revealing similar performances after summarizing tasks (meaning-oriented conditions). Their difference in the language-orienting condition (A- week 11, B-week 12) appears to be greater than in week 10. This suggests that the language-oriented task had a greater influence on the higher-level learners than the weaker ones. This seems reasonable; students can only activate prior knowledge. If there is less knowledge, any activation effect will be lower. Therefore, weaker students may get less benefit from a general orientation to form. This needs to be investigated further.
While much previous research on complexity, accuracy, and fluency of language focuses on single tasks, imagining the classroom as an integrated system (Larsen-Freeman & Cameron, 2008; Van Lier, 2004) requires the teacher-researcher to investigate the ongoing influence of one task on later activities. This also allows the teacher to use previous activities to set up an approach for later ones, without explicating this to the learner. Ellis (2003) and the Willis’ (1996, 2007) argue that telling the learners to attend to or to use a specific form decreases fluency and may produce unnatural expressions. Teachers can avoid this by not telling students to pay attention to grammar, but simply require them to do so by giving a reconstruction task. On a broader level, teachers and researchers should not only consider tasks as events but, once performed, as part of a learner’s history that influences performance on later activities.
Notes
[1] The study received institutional ethics approval: #RM18762. [back]
[2] Some readers may dispute that summary writing, or the reconstruction tasks are indeed language learning tasks. A full discussion of the definition of task is beyond the scope of this article, but we feel they fit the concept of ‘pedagogical’ task proposed by Ellis (2003) and the broader definition of task in Samuda and Bygate (2008, p. 69). [back]
About the author
Shaun Justin Manning is an associate professor in the Department of English Linguistics and Language Technology at Hankuk University of Foreign Studies. His research focuses on classroom dynamics, instructed language learning, task-based language teaching, and EFL writing.
To cite this article
Manning, J. S. (2021). Reconstructing better writing: Using summary reconstruction tasks to promote languaging while conveying meaning. Teaching English as a Second Language Electronic Journal (TESL-EJ), 25(3). https://tesl-ej.org/pdf/ej99/a4.pdf
References
Butler, Y. G. (2011). The implementation of communicative and task-based language teaching in the Asia-Pacific region. Annual Review of Applied Linguistics, 31, 36–57. https://doi.org/10.1017/S0267190511000122
Carless, D. (2004). Issues in Teachers’ Reinterpretation of a Task-Based Innovation in Primary Schools. TESOL Quarterly, 38(4), 639–662.
Coughlan, P., & Duff, P. A. (1994). Same task, different activities: Analysis of an SLA task from an activity theory perspective. In J. P. Lantolf & G. Appel (Eds.), Vygotskian approaches to second language learning (pp. 173–194). Ablex.
Ellis, R. (2003). Task-Based Language Learning and Teaching. Oxford University Press.
Ellis, R. (2017). Position paper: Moving task-based language teaching forward. Language Teaching, 50(4), 507–526. https://doi.org/10.1017/S0261444817000179
Ellis, R., & Barkhuizen, G. P. (2005). Analysing Learner Language. Oxford University Press.
Housen, A., Kuiken, F., & Vedder, I. (Eds.). (2012). Dimensions of L2 performance and proficiency: Complexity, accuracy and fluency in SLA. John Benjamins Pub. Co.
Jefferson Transcription System—A guide to the symbols. (2021). University Transcription Services. https://www.universitytranscriptions.co.uk/jefferson-transcription-system-a-guide-to-the-symbols/
Kim, Y., Jung, Y., & Tracy-Ventura, N. (2017). Implementation of a Localized Task-Based Course in an EFL Context: A Study of Students’ Evolving Perceptions. TESOL Quarterly, 51(3), 632–660. https://doi.org/10.1002/tesq.381
Kim, Y., & McDonough, K. (2011). Using pretask modelling to encourage collaborative learning opportunities. Language Teaching Research, 15(2), 183–199. https://doi.org/10.1177/1362168810388711
Kuiken, F., & Vedder, I. (2007). Task complexity and measures of linguistic performance in L2 writing. IRAL – International Review of Applied Linguistics in Language Teaching, 45(3), 261–284. https://doi.org/10.1515/iral.2007.012
Larsen-Freeman, D., & Cameron, L. (2008). Complex systems and applied linguistics. Oxford University Press.
Long, M. H. (1996). The role of the linguistic environment in second language acquisition. In W. C. Ritchie & T. K. Bhatia (Eds.), Handbook of research on Language Acquisition: Second language acquisition, Vol. 2, pp. 413–468). Academic Press.
McDonough, K., & Chaikitmongkol, W. (2007). Teachers’ and learners’ reactions to a task-based EFL course in Thailand. TESOL Quarterly, 41(1), 107–132.
Nation, I. S. P. (2009). Teaching ESL/EFL Reading and Writing. Routledge.
Ohta, A. S. (2001). Second Language Acquisition Processes in the Classroom Learning Japanese. Lawrence Erlbaum Associates, Publishers.
Robinson, P. (2001). Task complexity, task difficulty and task production: Exploring interactions in a componential framework. Applied Linguistics, 22(1), 22–57.
Robinson, P. (2011). Second language task complexity, the cognition hypothesis, language learning and performance. In P. Robinson (Ed.), Second Language Task Complexity: Researching the Cognition Hypothesis of language learning and performance (Vol. 2, pp. 3–37). John Benjamins.
Samuda, V., & Bygate, M. (2008). Tasks in Second Language Learning. Palgrave Macmillan.
Schmidt, R. (1990). The role of consciousness in second language learning. Applied Linguistics, 11(2), 129–158.
Skehan, P. (1998). A Cognitive Approach to Language Learning. Oxford University Press.
Skehan, P. (2009). Modelling second language performance: integrating complexity, accuracy, fluency, and lexis. Applied Linguistics, 30(4), 510–532. https://doi.org/10.1093/applin/amp047
Swain, M. (2010). “Talking-it-through”: Languaging as a source of learning. In H. Byrnes (Ed.), Sociocognitive Perspectives on Language Use and Language Meaning (pp. 112–130). Oxford University Press.
Swain, M., & Lapkin, S. (1998). Interaction and second language learning: Two adolescent French immersion students working together.The Modern Language Journal, 82(3), 320–337. https://doi.org/10.1111/j.1540-4781.1998.tb01209.x
Text Inspector: Online lexis analysis tool at textinspector.com. (2016). http://textinspector.com
TOEFL iBT Writing Section (For Test Takers). (2021). ETS TOEFL. https://www.ets.org/toefl/test-takers/ibt/about/content/writing/
Van den Branden, K. (Ed.). (2006). Task-based Language Education: From theory to practice. Cambridge University Press.
Van Lier, L. (2004). The Ecology and Semiotics of Language Learning: A sociocultural perspective. Kluwer Academic.
Vygotsky, L. S. (1986). Thought and Language (A. Kozulin, Ed.; Revised). MIT Press.
Willis, J. (1996). A flexible framework for task-based teaching. In Jane. Willis & D. Willis (Eds.), Challenge and Change in Language Teaching. Heinemann English Language Teaching.
Willis, J., & Willis, D. (2007). Doing Task-Based Teaching. Oxford University Press.
Appendix: Jeffersonian transcription symbols
Figure 7. Jefferson Transcription symbols (Jefferson Transcriptions, 2021)
Copyright rests with authors. Please cite TESL-EJ appropriately. Editor’s Note: The HTML version contains no page numbers. Please use the PDF version of this article for citations. |