Focus on form through task repetition in TBLT

Abstract

Because there has been little research on focus on form during the post-task phase in task-based language teaching, this experimental study investigates the effects of task repetition after having directed learners’ attention to form during the main task. The study comprises two interventions, where each consisted of a task with a focus on a particular language structure. Forty-eight ninth-grade students learning German as a foreign language were randomly assigned to two conditions: one group repeated a similar task (R); the other group did not (NR). The first intervention targeted the German dative case after a preposition; the second German comparatives. Pretests, immediate and delayed posttests measured metalinguistic knowledge, written and oral accuracy as well as oral fluency. Results showed that on written accuracy and metalinguistic knowledge, the R condition outperformed the NR condition on both structures. No significant differences between conditions were found on oral accuracy and fluency.

Keywords

Accuracy/fluency trade-off focus on form German as a foreign language similar task repetition task-based language teaching

I Introduction

Over the last 30 years, task-based language teaching (TBLT) has become a field of interest to researchers of second language acquisition (SLA), curriculum designers, language teachers, and teacher educators (Ellis, 2003; Shehadeh & Coombe, 2012; Skehan, 1998; 2014; Van den Branden, Bygate & Norris, 2009; Willis, 1996). Several researchers, however, have expressed their concerns whether TBLT is able to teach specific grammar forms to learners of second and foreign languages (Sato, 2010; Swan, 2005). Ellis (2009a) responded to this criticism by stating that, although TBLT may not have an explicit grammar syllabus, all supporters of TBLT see a role for grammar in terms of a focus on form during meaningful communication.

Several studies have made learners focus on form at all three stages of the TBLT cycle (Skehan, 1996; Willis, 1996). According to this cycle learners start the TBLT sequence by carrying out a pre-task, subsequently perform the main task and then optionally perform some post-task activities. Strategies such as guided planning (Foster & Skehan, 1999), different forms of input enhancement (see Doughty & Williams,1998) and modeling (Kim, 2013) have been used to make learners attend to form at the pre-task stage. A focus on form at the during-task stage is mostly achieved through the provision of corrective feedback (see Lyster, Saito & Sato, 2013). In the post-task researchers have experimented with the anticipation of a public performance (Foster & Skehan, 2013) and post-task transcription of the oral performance (Qian, 2014). In this article we will investigate another post-task strategy that may predispose learners’ attention to accuracy, namely the repetition of a task.

Task repetition has been shown to positively change learners’ task performance in terms of complexity, accuracy and fluency (Ahmadian & Tavakoli, 2011; Bygate, 1996, 2001; Gass, Mackey, Alvarez-Torres, & Fernandez-Garcia, 1999; Lynch & McLean, 2001). Since this article investigates the effects of task repetition on accuracy measures, we are especially interested in accuracy effects. Following Bygate (1996), we argue that giving learners the possibility to perform a task again may lead to gains in oral accuracy. That is to say, because learners are familiar with the content through the first performance, the second time they may pay more attention to its correct formulation.

In most studies task repetition is operationalized as repetition of the same task (Bygate, 2001). In a pilot study on same task repetition we observed, however, that learners found it ‘boring’ to repeat the task and were not motivated to perform it. Therefore, in the present study we opted for repetition of a similar task, which was another version of the initial task. Since several studies have demonstrated that the beneficial effects of task repetition do not carry over to a new task (Bygate, 2001; Gass et al., 1999), Ahmadian (2012) suggested that learners may need feedback on their initial performance (Sheppard, cited in Ellis, 2009b) for effects to carry over to new contexts. Therefore, in the present study learners were provided with feedback during task performance.

Both Swan (2005) and Sato (2010) question whether TBLT is successful in teaching learners predetermined grammar structures. For that reason, the present study did not use overall measures of accuracy (occurrence of errors per t-unit; Bygate 2001) but investigates the effects of similar task repetition on the accurate use of two different German grammar structures. In addition, we examined whether learners’ focus on accuracy would come at the cost of fluency (see trade-off effect; Skehan, 1998, 2014).

II Review of literature

1 Theoretical background

Why would task repetition be an effective strategy to make learners focus on form and promote the correct use of grammar structures? Based on theories on learners having limited attentional resources, Skehan (1996, 1998) argues that the first time learners carry out a task they are so preoccupied with completing the task that they focus almost completely on the content and not necessarily on linguistic accuracy. Levelt (1989) defines this process as the conceptualization stage in which learners perform different kinds of mental activities, such as selecting and ordering information and thinking about ‘what will I say?’ It is not until the formulation stage that learners try to find the right words, grammar rules, and ‘phonetic plan’ (p. 12) to bring their message across. According to Bygate (2001), it is exactly during task repetition that learners can redistribute their focus from conceptualization to formulation (words and grammar) and articulation (pronunciation) because they are already familiar with the content and language knowledge.

2 Empirical evidence

One of the earliest studies on task repetition was carried out by Bygate (1996). In his small-scale laboratory study, he showed one learner of English a Tom and Jerry cartoon and asked to re-tell the story. Three days later, the student was asked to perform the same task again. Analysis of the oral output showed changes in complexity and grammatical variety and small gains in accuracy. In Bygate’s (2001) second study he investigated the effect of learners repeating a task versus learners practicing a type of task. Forty-eight ESL students carried out a narrative and an interview, twice with 10 weeks in between. Results showed a strong effect for task repetition in increasing fluency and complexity but not on accuracy. This effect did not carry over to a new task. Bygate commented, however, that some results showed a trend in that direction and he concludes that the ‘notion of “discourse competence” – the capacity to process certain types of discourse more easily than others – does appear to have some empirically identifiable psychological reality …’ (p. 43).

Gass et al. (1999) investigated whether task repetition led to more sophisticated language use and whether this would transfer to a new context. One hundred and three students of Spanish as a foreign language were assigned to a control, same content, or different content group. They watched video fragments of Mr Bean and recorded their own version of what was happening, at the same time. Gass et al. observed that task repetition led to greater overall proficiency, greater accuracy in morphosyntax (trend towards more target-like production of the verb estar), and lexical sophistication. It seemed that ‘… freeing up attention to meaning allows learners to gain greater control over their linguistic knowledge’ (p. 573). These results, though, did not transfer to a new context.

A variation on task repetition, task recycling, was investigated by Lynch and McLean (2001). They asked 14 oncology students, learning English for specific purposes, to present their research six times in a poster presentation carousel. In contrast to other repetition studies with larger intervals, repetition in this study was immediate with an interval of 3 minutes. Findings revealed that participants’ speech was more accurate and fluent.

Sheppard’s (2006) study is the only study that demonstrated that effects of repetition can transfer to new contexts. In addition to task repetition, Japanese students were provided with feedback on their initial performance. After repetition students advanced in fluency, complexity, and clearly in accuracy. Transfer effects were observed for grammatical complexity.

Hawkes’ (2012) study is the only one that investigated the effects of task repetition after having directed learners to form in the main task. After having carried out the main task, 13–14-year-old Japanese students studying English as a foreign language focused on form by highlighting and practicing different target structures and vocabulary. During the final phase participants repeated the exact same task as the main task. Results showed an increase in the number of form and pronunciation-focused corrections. From these findings Hawkes concluded that students paid more attention to form during the repeated task performance.

The most recent study to date on task repetition has been carried out by Thai and Boers (2015) who investigated the effects of learners repeating a monologue under time pressure on accuracy, complexity and fluency. Thai and Boers used a 4/3/2 format of task repetition, in which augmented time pressure gradually decreased learners’ opportunities for online planning. Participants were twenty 10-grade Vietnamese EFL students who were asked to talk about their favorite movie. Results showed positive effects on fluency. Thai and Boers concluded that when teachers aim to achieve a more accurate or complex performance, the 4/3/2 sequence may not be right format and in addition learners should be provided with opportunities to modify their performance early in the task sequence.

III Present study

Drawing on theories on limited attentional resources and Levelt’s (1989) Model of Speech we designed an experimental study that investigates the effects of similar task repetition, preceded by form-focused feedback during the main task, on the acquisition of two German grammar structures. The design incorporated two subsequent interventions with each a different German grammar structure because we attempted to examine whether the effects of task repetition would depend on the target structure. In addition, this study enabled us to investigate Skehan’s (1998; 2014) trade-off hypothesis.

To this end, three research questions were formulated:

Does task repetition as a post-task activity promote the accurate use of two new grammar structures?

Does the effect depend on the grammar structure?

Does students’ focus on oral accuracy have a negative effect on oral fluency?

IV Method

1 Participants

The study involved 48 ninth-grade students learning German as a foreign language (A2 level of the Common European Framework of Reference for Languages; CEFR, Council of Europe, 2001). Participants (30 female; 18 male) were recruited from two classes of a Dutch secondary school. The majority of the participants (95%) were native speakers of Dutch . Apart from German, participants also learn English and French as foreign languages at school. Their mean age was 14.2 years. The students had been engaged in learning German for 18 months, for 2 hours per week, 40 weeks per year. In accordance with the protocol of the University’s Faculty of Humanities’ Ethics Committee, all parents were informed about the study and the possibility of non-participation. No parent objected.

2 Design

In this study we carried out an experimental pretest–posttest–delayed posttest research design for two subsequent interventions (see Figure 1). Both interventions examined the effects of task repetition, but each on a task with a different target structure. Each intervention was spread out over a period of three weeks. We performed a randomized experiment in two intact classes. Within each class students were randomly assigned to the Repetition (R) (n = 24) and No-Repetition (NR) (n = 24) condition. Thus, in each class were students of the R and the NR condition. The operationalization of the two conditions was equal for both interventions. All students received form-focused feedback on the particular target structure during main task preparations. Two weeks later, the R condition repeated a similar task, the NR condition did not and instead performed a filler task, which was not related to the target structure. The performance of both the repetition and filler task, including preparations, took students approximately 25 minutes. To avoid cooperation between students of both conditions, the Repetition group was placed in the right part of the classroom and the No- Repetition group in the left part. In addition, lesson materials were designed for independent work.

Figure 1.

Research design repetition task.

Because of earlier negative experiences in a pilot study with students immediately repeating the same task afterwards (students were bored and not engaged) we altered the timing and the content of the post-task. The repetition task did not immediately follow the performed main task but was performed 2 weeks later as a post-task of the subsequent main task (see Figure 1), and the content was changed slightly. As a result, the content of the repetition tasks needed to be adjusted to the topic of the subsequent main tasks. For example, instead of asking students to describe their dream bedroom at home (Main Task 1) they were required to describe their dream bedroom on a winter holiday (Repetition Task 1). Although some adjustments to the content of the repetition tasks were made, the linguistic demands and grammar structures were the same as in the main tasks. For that reason, the main and repetition tasks can be regarded as ‘ “task families”, where a group of tasks resemble one another and may well have similar language or cognitive demands’ (Skehan, 1996, p. 56).

The tasks, designed for this study, were embedded in an ongoing school program, where the first author is active as a teacher of German. As part of the school curriculum, students’ performances of the tasks, needed to be assessed with the school grade scale ranging from 1 to 10 (where 1 is low, 10 is high and 6 is sufficient).

3 Target structures

We selected two tasks, each with a different target structure. We opted for tasks with a simple target structure with similarities to the L1, the comparative, and for a task with a more complex target structure with no similarities to the L1, the dative structure.

a Dative case after a preposition of place

The target structure in Main Task 1 and Repetition Task 1 concerned the dative case of an article after a preposition of place (in, an, auf, hinter, neben, unter, über, vor, zwischen¹). We consider this structure complex because (1) learners have to undertake several analytical steps to arrive at the correct form (see Hulstijn & De Graaff, 1994); and (2) learners have no existing knowledge regarding the dative structure because it has no equivalents in Dutch. Transfer between L1 and L2 is therefore unlikely.

b Comparative

In contrast to the dative case, we consider German comparatives to be a more simple structure because (1) most comparative forms in German are composed by a simple rule: Attach -er to the adjective or adverb. In addition, some German irregular comparative forms get an umlaut mark (groβ–gröβer [big–bigger] – hoch–höher [high–higher]) or have suppletive forms (gut–besser [good–better] – viel–mehr [many–more]); and (2) the forming of German comparatives is realized similarly as comparatives in the L1 (Dutch).

4 Treatment tasks

a Main Task 1: Room of my dreams

This task is a description task for which each student was asked to design and describe the bedroom of his or her dreams with a virtual budget of 10,000 euros (adapted from Van de Guchte, Braaksma, Rijlaarsdam, & Bimmel, 2015). To increase the chance that the target structure, a preposition of place plus Dative case, would be used students were required to describe the location of the furniture and accessories (minimal number of 15 pieces) in the room to two other students. These other students were not able to see the room and were asked to draw the described room on paper. Afterwards, they compared their drawings and posed questions on any wrongly positioned furniture. After finishing the task, each student presented his or her dream bedroom individually to a teacher trainee, who recorded their performance on video camera. This oral presentation was assessed afterwards by the teacher with the help of a grade scale ranging from 1 to 10.

b Main Task 2: Compare two ski holidays

Main Task 2 is a comparison task for which students compared two ski holidays in Austria. The use of the target structure Comparatives was triggered by the additional instruction that students needed to convince their parents in an oral discussion with at least 7 arguments to go to either the quiet St Johann or the more vibrant St Anton. The information that was needed to compare the two holiday destinations was put in a written structured chart. The elements that students compared in the chart were, for example, the number of ski lifts, the price of a ski pass, the altitude of the area, size of the apartment, the temperature, quality of the apartment, beauty of the area, and kindness of the staff. After finishing this task, students presented their seven arguments individually to a research assistant, who recorded their performance on video camera. This oral presentation was assessed afterwards.

c Repetition Task 1: Holiday room of my dreams

All students in the R condition were told to design the bedroom of their dreams for a ski holiday at the destination of their choice and to make a phone call to a reservations center to find such a room. During this phone call (with a teacher or teacher assistant) students were required to describe the location of the furniture and accessories (minimal amount of 15 pieces) in the room.

d Main Task 3: Summer memories

Main Task 3 was designed to embed Repetition Task 2 as a post-task. This task did not take part in the measurements. For this task all students were asked to bring a holiday picture into the class and to talk about the picture and the holiday during a speed date session with their fellow students.

e Repetition Task 2: Compare two summer holidays

Students were asked to compare aloud two summer holidays in Italy and convince their friend in an oral discussion with at least 7 arguments to go to either one of them. The information that was needed to compare the two holiday destinations was put in a written structured chart. The elements that students compared in the chart were, in a few cases, adjusted to the topic ‘summer holiday’ but elicited the exact same comparative forms as those in Main Task 2. Students compared, for example, the number of swimming pools, the price of the flight, the altitude of the area, size of the apartment, the temperature, quality of the apartment, beauty of the area, and kindness of the staff.

5 Procedures

The two groups were taught by two different teachers. Because the lessons for the two groups were scheduled in parallel it was impossible to have both groups taught by the same teacher. The groups were randomly assigned to the teachers. Both teachers held a Master’s degree in teaching German language and literature and were trained in providing the feedback type prompts by means of a protocol and a training session.

All students’ attention was directed to form during main task preparations by means of corrective feedback. Prompts were selected as the type of feedback students received, because several studies have demonstrated their effectiveness in fostering the acquisition of grammar structures (Ammar, 2008; Ellis, 2007; Van de Guchte et al., 2015; Yang & Lyster, 2010). When students made an error in one of the target structures, they first received metalinguistic information on the target structure, followed by an elicitation part in which they were asked to answer again (see example in Figure 2). Both teachers wrote down the number of feedback moments per student and were observed by a research assistant who also noted the amount of feedback for each student. The mean sum of feedback moments for the R condition on the dative structure of Task 1 was 2.88 (SD = .68) and on the comparative structure of Task 2, 2.08 (SD = .41). For the NR condition the mean sum of feedback moments on the dative structure of Task 1 was 3.04 (SD = .95) and on the comparative structure of Task 2.2 (SD = 0). No significant differences in the number of feedback moments were found between conditions on either the Dative (F (1,46) = .49, p = . 49) or the Comparative structure (F (1,46) = 1, p = .32).

Figure 2.

Example feedback episode at during-task phase of Repetition Task 2 (comparative).

6 Dependent measures

Drawing on Macrory and Stone’s (2000) study, investigating the relationship between knowing and using the perfect tense in French, we attempted to examine both students’ implicit and explicit knowledge (Ellis, 2005) on the target structures, by means of the following three ways of examining acquisition.

Learners verbalizing the rules on (a) a dative case after a preposition of place and (b) the comparative in German in a metalinguistic knowledge test (MKT).

Their actual knowledge on using these rules in a fill-in-the gap exercise (FITG²).

Their ability to use both structures in two meaning-based oral tasks.

We did not measure the complexity of the performance because the testing tasks were structured and focused on the use of the two target structures. The task elicited only short sentences and there was not much variety in the way the tasks could be carried out.

In order to avoid the retrieval of explicit knowledge of the structures, students performed the oral task first followed by the written task (FITG) and the metalinguistic knowledge test MKT. The same measurement design was used for Task 1 and Task 2 (see Figure 1). One week prior to the start of each task students performed pretests on written accuracy and oral accuracy and fluency. One week after completing the task students took part in posttests-1 on MK, written accuracy and oral accuracy and fluency. Posttests-1 only measured the effects of instruction and task performance, which enabled us to examine whether there were any significant differences between conditions. Results of the linear mixed model analysis showed no significant differences at posttest 1 for the two structures (MKT: t(230) = 1.73, p = .085; FITG: t(322) = 1.087, p = 2.78; Oral Accuracy: t(322) = .136, p = .892; Oral Fluency: t(322) = –.562, p = .575). Two weeks after each main task performance students of the R condition repeated a similar task. The NR condition performed a filler task that was not related to the use of the target structures. One week later students participated in posttests-2, which measured the effects of task repetition. Two weeks later students performed posttests-3, which measured the delayed effects of task repetition.

a Metalinguistic knowledge tests (MKT)

In the MKT on the dative structure students were asked to complete the following rule:

When you describe a room, using the following prepositions: in, an, auf, hinter, neben, unter, über, vor, zwischen, a der word changes into …; a die word changes into …; a das word changes into …; a die plural changes into ….

Scores were expressed in percentage correct (total amount of correct dative cases divided by four). For the MKT on comparatives, students were invited to write down the standard rule to form the comparative in German and two additional irregular ‘rules’ to form the comparative. Scores were expressed in percentage correct (total amount of correct rules divided by three). Students were not given the MKT as a pretest because we did not want to predispose them towards particular target structures, before the subsequent meaning-focused task performance

b Written accuracy tests

As in our earlier study (Van de Guchte et al., 2015) in which we compared the effect of prompts and recasts on the German dative structure and the comparative, we used a written gap-fill test to measure the students’ actual use of both rules. Written tests were administered by the teacher during the lessons, and took about 15 minutes. There were four similar versions of the written test: For the pretest, each participant received version A, for posttest 1 version B, for posttest 2 version C, for posttest 3 version D; thus all conditions received the same test at each measurement occasion.

The written tests on the dative included 14 gap-fill sentences in which students were asked to write down the correct case of an article after a locally used two-way preposition (Appendix 1). In order to disentangle application of the dative rule from knowledge of noun gender, grammatical gender was noted in brackets after the noun (Schrank (der)) [cabinet (the)]. Reliability (Cronbach’s alpha) showed .70 for the pretest, .90 for posttest 1, .92 for posttest 2, and .93 for posttest 3.

The written test on the comparative included 13 gap-fill sentences in which students were asked to write down the correct comparative. Reliability (Cronbach’s alpha) was .66 for the pretest, .66 for posttest 1, .72 for posttest 2, and .57 for posttest 3.

c Oral accuracy tests

Oral accuracy in both tasks was assessed by means of an oral communication test, held with one of the three interlocutors: two native speakers of German and the first author. The tests consisted of communicative tasks similar to those used in the treatment sessions. For both tasks, the oral test was administered in four versions: For the pretest, each participant received version A, for posttest 1 version B, for posttest 2 version C, for posttest 3 version D. The interlocutors, all seated in different rooms, administered the tests with help of a step-by-step plan to ensure that all tests were performed similarly. Each student was tested individually and audio-recorded, out of class. The recordings were presented to a native speaker of German and the first author who rated them blind for condition for accuracy. The first author evaluated all the recordings; the second rater evaluated 70%. For both tasks inter-rater reliability was very high: 99% for the dative and 86% for the comparative.

Dative: For the oral picture-description test (Appendix 2), the interlocutor used a 3D picture of a room, including the German translations (plus gender) of the furniture and accessories in the picture. Students were asked to look at the picture and to describe the room. For the same reason as in the written tests, grammatical gender was noted in brackets after the noun. The interlocutors were asked to elicit a minimum of 15 target structures. Accuracy scores were expressed as percentages correct (correct use of dative case after a two-way preposition / total number of articles used after a two-way preposition). When a student used other prepositions than the target structures to describe the room, these were not counted. Furthermore, repairs were assessed and not the utterance before the repair.

Comparative: During the comparison test the interlocutor showed the student a chart with information on two ski holidays, which contained 12 elements: the number of ski lifts, length of slopes, the cost of a ski pass, snow reliability, altitude of the ski area, après ski, beauty of area, weather, temperature, surface, quality of the apartment, and kindness of the staff (Appendix 3). Students were asked to compare 12 elements of the two holidays concerned. Accuracy scores were expressed as percentages correct (correct use of comparatives / total number of comparatives used). Incorrect use of the comparative meant that either the comparative form itself was incorrect or that the comparative was not used properly in the context.

d Oral fluency tests

With respect to fluency, we ran a script on the sound files written by De Jong and Wempe (2009) in PRAAT (Boersma & Weenink, 2007), with which we were able to calculate students’ speech rate (total number of syllables divided by total duration of the task performance). However, as De Jong et al. (2012) point out speech rate takes ‘breakdown fluency and speed fluency […] together into one measure that encompasses aspects of pausing as well as speed of delivery’ (p. 2). Therefore we included two measures that did not confound pausing and speed of delivery: articulation rate, that is, mean duration of syllables (speaking time divided by total number of syllables), and the number of silent pauses (number of silent pauses divided by speaking time). Following De Jong et al. (2012), silences of 250 ms or longer were considered to be hesitations or pauses, and thus silences shorter than 250 ms, so-called micropauses (see, for example, Riggenbach, 1991), were discarded.

7 Statistical analysis

To examine whether task repetition was effective in promoting the acquisition of the two new grammar structures we carried out a linear mixed model analysis (LLM) for each dependent measure (metalinguistic knowledge, written accuracy, oral accuracy and oral fluency). We investigated the effects of two independent variables (i.e. grammar structure and condition) on four measurement occasions: pretest, posttest 1, measuring the effect of the instruction which was the same for both groups, posttest 2, measuring the effect of the repetition intervention, and posttest 3 measuring the delayed effect of the repetition intervention. The fact that students were exposed to a Repetition and No-Repetition condition, two target structures and four measurement occasions made the resulting observations correlated (i.e. not independent). For that reason, we opted for a statistical approach that could manage correlated errors accurately and therefore used a general linear mixed model (LMM). The model included fixed and random factors for each dependent measure. The three fixed factors were time of measurement, grammar structure and condition. We analysed the model, including three 2-way interactions and one 3-way interaction. The random factor was participants. The alpha level was set at .05.

V Results

1 Metalinguistic knowledge tests

Table 1 presents the means and standard deviations resulting from the metalinguistic knowledge tests. On the MKTs we found a main effect for time F(2, 230) = 3.47, p = .033, for condition F(1, 46) = 12.66, p = .001, and for structure F(1, 230) = 11.43, p = .001. Pairwise comparisons showed that the Repetition condition outperformed the No-Repetition condition (MD = .206, 95 % confidence interval [CI] = .089, .323, p = .001). The main effect for condition was moderated by time F(2, 230) = 4.4, p = .013. At posttest 2 (1 week after task repetition) the scores of the Repetition condition increased on both structures with large effect sizes, compared to the No-Repetition condition (Dative, d = .94; Comparative, d = 1,22). At posttest 3 (3 weeks after task-repetition) the Repetition condition showed a decrease on both the Dative structure (d = .62) and the comparative structure (d = .91).

Table 1.

Condition means (proportions correct) and standard deviations for metalinguistic knowledge tests for three measurement occasions.

		Posttest 1		Posttest 2		Posttest 3
Structure	Condition	M	SD	M	SD	M	SD
Dative	R	.688	.278	.917	.190	.750	.338
	NR	.583	.319	.646	.361	.542	.366
Comparative	R	.778	.306	.959	.113	.958	.149
	NR	.710	.384	.625	.372	.708	.359

Note. R = Repetition group (n = 24); NR = No-Repetition group (n = 24).

2 Written accuracy tests

Table 2 demonstrates the means and standard deviations resulting from the written accuracy tests. Results of the fill-in-the-gap tests showed a main effect for time F(3, 322) = 207.42, p < .001 and for condition F(1,46.24) = 9.74, p = .003. Pairwise comparisons revealed that the Repetition condition outperformed the No-Repetition condition (MD = .126, 95 % confidence interval [CI] = .045,.207, p = .003). In addition, the main effect for condition was moderated by time F(3, 322) = 4.22, p = .006. At posttest 2, for the Repetition condition, we observed an increase for both structures with large effect sizes, compared to the No-Repetition condition (Dative, d = .92; Comparative, d = .77). At posttest 3, we noted a decrease for both structures (Dative, d = .88; Comparative, d = .14).

Table 2.

Group means (proportions correct) and standard deviations for written accuracy tests for four measurement occasions.

		Pre		Posttest 1		Posttest 2		Posttest 3
Structure	Condition	M	SD	M	SD	M	SD	M	SD
Dative	R	.101	.101	.744	.271	.872	.195	.824	.286
	NR	.066	.081	.625	.310	.610	.350	.554	.329
Comparative	R	.280	.111	.900	.112	.875	.101	.817	.136
	NR	.292	.155	.804	.208	.731	.245	.740	.195

Note. R = Repetition group (n = 24); NR = No-Repetition group (n = 24).

3 Oral accuracy tests

Table 3 presents the means and standard deviations resulting from the oral accuracy tests. On the oral accuracy tests we found a main effect for time F(3, 322) = 90.11, p = .001 and for structure F(1, 322) = 281.38, p = .001. No significant effect was found for condition F(1, 46) = 1.44, p = .237. Pairwise comparisons showed that the mean scores on the comparative structure were significantly higher than the scores on the dative structure (MD = .367, 95 % confidence interval [CI] = .324,.401, p < .001). Furthermore, the main effect for structure was moderated by time F(3, 322) = 5.24, p = .002. For the comparative we observed a decline at posttest 2 and an incline at posttest 3; for the dative we observed an increasing line from posttest 1 to posttest 3.

Table 3.

Condition means (proportions correct) and standard deviations for oral accuracy tests for four measurement occasions.

		Pre		Posttest 1		Posttest 2		Posttest 3
Structure	Condition	M	SD	M	SD	M	SD	M	SD
Dative	R	.021	.034	.400	.401	.595	.380	.628	.394
	NR	.019	.048	.435	.390	.472	.381	.464	.375
Comparative	R	.511	.193	.868	.130	.839	.150	.875	.140
	NR	.461	.150	.845	.163	.759	.182	.813	.171

Note. R = Repetition group (n = 24); NR = No-Repetition group (n = 24).

4 Oral fluency tests

In Tables 4 –6 the means and standard deviations resulting from the fluency tests are presented. On speech rate we found a main effect for time (F(3, 322) = 66.19, p < .001), but not for condition (F(1, 46) = .079, p = .780) or structure (F(1, 334) = .376, p = .540). On articulation rate we found a main effect for time (F(3, 322) = 4, p = .008), but not for condition (F(1, 46) = .025, p = .875) or structure (F(1, 336) = 1.49, p = .223). On silent pauses we found a main effect for time (F(3, 322) = 11.6, p < .001), and structure (F(1, 322) = 139.77, p < .001), but not for condition (F(1, 46) = .510, p = .479). It appeared that students produced more silent pauses in the preposition plus dative task than in the comparative task.

Table 4.

Condition means (1–5) and standard deviations for fluency measures for four measurement occasions: Data for speech rate.

		Pre		Posttest 1		Posttest 2		Posttest 3
Structure	Condition	M	SD	M	SD	M	SD	M	SD
Dative	R	1.71	.39	2.03	.49	2.03	.46	1.8	.52
	NR	1.76	.43	1.78	.27	2.09	.39	1.99	.44
Comparative	R	1.98	.53	2.39	.47	2.22	.50	2.51	.55
	NR	1.93	.32	2.31	.39	2.21	.33	2.53	.46

Note. R = Repetition group (n = 24); NR = No-Repetition group (n = 24).

Table 5.

Condition means (1–5) and standard deviations for fluency measures for four measurement occasions: Data for articulation rate.

		Pre		Posttest 1		Posttest 2		Posttest 3
Structure	Condition	M	SD	M	SD	M	SD	M	SD
Dative	R	3.69	.31	3.83	.33	3.92	.40	3.73	.43
	NR	3.75	.57	3.69	.33	4	.51	3.85	.39
Comparative	R	3.61	.38	3.86	.53	3.69	.39	3.74	.45
	NR	3.52	.32	3.78	.64	3.64	.41	3.74	.37

Note. R = Repetition group (n = 24); NR = No-Repetition group (n = 24).

Table 6.

Condition means (1–5) and standard deviations for fluency measures for four measurement occasions: Data for silent pauses.

		Pre		Posttest 1		Posttest 2		Posttest 3
Structure	Condition	M	SD	M	SD	M	SD	M	SD
Dative	R	1.17	.18	1.13	.34	1	.22	1.13	.30
	NR	1.09	.25	1.13	.17	1	.25	1.10	.29
Comparative	R	1.03	.35	.84	.25	.91	.28	.69	.24
	NR	.97	.20	.85	.22	.89	.28	.69	.19

Note. R = Repetition group (n = 24); NR = No-Repetition group (n = 24).

From pretest to posttest 3 students of the R condition, on the dative task, increased on speech rate by .18, on articulation rate by .04, and decreased on the number of silent pauses by .04. On the comparative task students of the R condition increased on speech rate by .53, on articulation rate by .13, and decreased on the number of silent pauses by .34. Students of the NR condition, on the dative task, increased on speech rate by .23, on articulation rate by .10, and on the number of silent pauses by .01. On the comparative task students of the NR condition increased on speech rate by .60, on articulation rate by .22, and decreased on the number of silent pauses by .28.

5 Oral accuracy versus fluency

The trade-off effect between accuracy and fluency measures was tested through correlations. For the Repetition condition on the dative task we found a significant moderate negative correlation on posttest 2 (r(24) = –.57, p = .004) between oral accuracy and speech rate. Further, no significant correlations were found between accuracy and other fluency measures.

VI Discussion

In this experimental study we aimed to investigate whether task repetition could be an effective focus on form strategy to promote the accurate use of German grammar structures. In addition, we investigated whether the effects of task repetition were different for the two tasks with the different grammar structures. Furthermore, the present study made it possible to examine Skehan’s (1998; 2014) trade-off theory.

As for the first research question, ‘Does task repetition as a post-task activity promote the accurate use of two new grammar structures?’ results showed positive effects on both metalinguistic knowledge on the rule (MK) and on written application of the rule (FITG). Since we have no online task-performance data that show the students really focused on the grammar features during task repetition, we can only assume that the better posttest performance of the R group can be attributed to students’ attention to these structures during task repetition. Because the students in the R group had more opportunity to practice using the target forms, this may probably have raised or re-awakened their metalinguistic awareness. Nevertheless, this can only be claimed for declarative knowledge but not for procedural knowledge of the target structures. That is to say, the oral tests, which attempted to tap the procedural knowledge of the target structures, revealed no significant results in favor of the R condition. In addition, results showed significant correlations between rule-knowledge and oral accuracy for the dative structure but not for the comparative structure. These findings might be explained by the fact that the dative is considered a highly reliable syntactic rule, the acquisition of which – Hulstijn and de Graaff (1994) argue – strongly depends on rule-based learning. However, it might have been the case that some frequent dative forms such as in der Mitte (‘in the middle’) have been acquired through exemplars. The formation of the comparative, on the other hand, is only partly based on a rule (add -er to the adjective or adverb), and it also has many irregular forms, which are acquired through exemplars.

Several explanations why students in the R condition did not outperform the NR condition on oral task performance can be given. From the point of view of skill acquisition (Anderson, 2000), we might argue that repeating a task once does not lead to the amount of communicative practice (DeKeyser, 1998) that one needs to turn explicit knowledge into implicit knowledge. In other words, students need more practice to fully automatize the target structures. This point has also been made by Gass et al. (1999), who comment that ‘learners may have knowledge of certain features, but they may not have acquired control over that knowledge’ (p. 554).

Not finding an effect in the oral measurement might be caused by how we operationalized the task repetition in terms of (1) a similar not an identical task and (2) the interval of 2 weeks between the main task and the repeated task. Possibly, learners did not experience the two talks as belonging to a ‘family’ (Skehan, 1996) and did not take advantage of the transferability of content. In addition, practicing twice in three weeks may not have been enough to proceduralize the Dative and Comparative structures. In line with Bygate (2001) and Sato (2010) we suggest that, for learners to be more successful on similar tasks, ‘more – or more massed – task exposure might be needed’ (Bygate, 2001, p. 43).

Another perspective on why no significant results were found for the Repetition condition on the oral accuracy test may be gained by analysing why the No-Repetition condition performed so well on this test. Probably, the assessed public performance (motivated by school report requirements) at the end of the main-task phase accounts for the considerable scores of this condition. Skehan and Foster (1997) investigated the effects of the ‘threat’ of a public performance, as a post-task activity, and indeed found improvement in accuracy. They concluded that because learners knew that a private task performance in pairs would lead into a public performance, their attention was, to some extent, more focused on form and learners ‘were more attentive to error during actual task performance’ (Skehan et al., 2012, p. 174).

A positive answer to the second research question, ‘Does the effect depend on the grammar structure?’, requires an interaction effect between condition and structure. The results did not reveal any of these interactions, so, we therefore conclude that the effect of task repetition does not depend on the structure.

We now turn to the third research question, ‘Does students’ focus on oral accuracy have a negative effect on oral fluency?’ With this question we aimed to contribute to the literature on attentional resources in task-based language learning. For the Repetition condition on the dative task we found a moderate negative linear relationship between accuracy and fluency on posttest 2 (one week after task repetition), which indicates that students who performed the task more accurately showed less fluency. These findings lend support to Skehan’s (1998; 2014) trade-off theory of limited attentional resources. We presume that retrieval and application of the dative rule, during oral performance, confronted students with a heavy cognitive load, resulting in less attention to fluency. Since students already had exemplars of comparative structures in their long-term memory, retrieval was much easier, and did not interfere with a fluent performance.

VII Limitations and directions for future research

Despite the positive findings of the present study, some limitations need to be acknowledged. First, we only have posttest data and no online data to measure the effects of task repetition. Online processing data would have provided us with more insights as to whether students actually paid attention to the grammatical structures and how this could have led to the significant findings on the MK and FITG tests. Second, our choice to use ‘accuracy as a reflection to focus on form’ (Foster & Skehan, 1996, p. 280) might have been too conservative to get a full insight into the effects of task repetition. We agree with Foster and Skehan that, with taking complexity into account, one can also capture the learner’s risk-taking and experimenting, which connects with gains in interlanguage development. A challenging task for further research on task repetition is to investigate how the use of similar tasks in task repetition could promote accurate use of linguistic structures in oral task performance and which interval of repetition would be effective.

VIII Conclusions and classroom implications

Altogether, the results of the present study have led us to conclude that repetition of a similar task may be an effective strategy to promote the acquisition of grammatical structures. It appeared that students who received corrective feedback at the during-task stage and in addition repeated a similar task were better at verbalizing and applying the rules on a simple and more complex German grammar structure. We find this encouraging because these findings suggest that previous knowledge of the target structure is available to the learners to build on in subsequent performance. Possibly, more practice is needed to automatize these structures in oral performance. In addition, no interactions were found between the effect of condition and task, which indicates that the effect of task repetition does not depend on the structure. Furthermore, this study provided additional evidence for Skehan’s (1998, 2014) trade-off theory, which claims that people have limited attentional resources. The negative interaction we found within the Repetition condition on the dative structure provides additional evidence that accuracy may come at the expense of fluency. However, this was not the case for the comparative structure, which has many similarities with the L1, suggesting that under specific conditions accuracy and fluency are not in competition with each other.

A strength of the present study is that it has been carried out in real classrooms and that task repetition has been embedded in an ongoing program. For that reason, we argue that the findings of our study may have an important implication for language pedagogy in a foreign task-based language learning context. The Common European Framework of Reference for Languages asks for a more communicative approach in language teaching but, at the same time, does not provide sufficient guidance on how grammar education can be a part of meaning-focused language education. The results of the present study underline that similar task repetition can be an effective strategy to focus on form in task-based language teaching, when students have been provided with corrective feedback at the during-task stage. Designing repetition tasks that repeat the same grammar structures as the main task but in which content is slightly adjusted to the topic of the subsequent task in the curriculum might certainly be of interest for teachers or curriculum designers. This design may integrate task repetition in a natural way into the school’s task-based curriculum and may keep students interested and motivated.

Footnotes

Appendix 1

Appendix 2

Appendix 3 Acknowledgements

This study is based on the first author’s PhD research (Van de Guchte, 2015). We gratefully acknowledge the cooperation of the participating teachers and students. We thank the anonymous Language Teaching Research reviewers for their valuable input and feedback on earlier versions of this article.

Funding

This study was funded by the Research Institute of Child Development and Education of the University of Amsterdam, the Gooise Scholen Federatie, Bussum, the Netherlands, and the Roland Holst College, Hilversum, the Netherlands.

Notes

References

Ahmadian

M.J.

(2012). Task repetition in ELT. ELT Journal, 66, 380–382.

Ahmadian

M.J.

Tavakoli

(2011). The effects of simultaneous use of careful online planning and task repetition on accuracy, complexity, and fluency in EFL learners’ oral production. Language Teaching Research, 15, 23–49.

Ammar

(2008). Prompts and recasts: Differential effects on second language morphosyntax. Language Teaching Research, 12, 183–210.

Anderson

J.R.

(2000). Learning and memory: An integrated approach. New York: John Wiley.

Boersma

Weenink

(2007). Praat: Doing Phonetics by Computer v.4.5.24. [www.praat.org].

Bygate

(1996). Effects of task repetition: Appraising the developing language of learners. In Willis

Willis

(Eds.), Challenge and change in language teaching (pp. 136–146). Oxford: Macmillan Heinemann.

Bygate

(2001). Effects of task repetition on the structure and control of oral language. In Bygate

Skehan

Swain

(Eds.), Researching pedagogic tasks: Second language learning, teaching and testing (pp. 23–48). Harlow: Longman.

Council of Europe (2001). Common European framework of reference for languages. Cambridge: Cambridge University Press.

De Jong

N.H.

Wempe

(2009). Praat script to detect syllable nuclei and measure speech rate automatically. Behavior Research Methods, 41, 385–390.

10.

De Jong

N.H.

Steinel

M.P

Florijn

A.F.

Schoonen

Hulstijn

J.H.

(2012). Facets of speaking proficiency. Studies in Second Language Acquisition, 34, 5–34.

11.

DeKeyser

(1993). The effect of error correction on L2 grammar knowledge and oral proficiency. Modern Language Journal, 77, 501–514.

12.

Doughty

Williams

(1998). Pedagogical choices in focus on form. In Doughty

Williams

(Eds.), Focus on form in classroom second language acquisition (pp. 197–261). New York: Cambridge University Press.

13.

Ellis

(2003). Task-based language learning and teaching. Oxford: Oxford University Press.

14.

Ellis

(2005). Measuring implicit and explicit knowledge of a second language: A psychometric study. Studies in Second Language Acquisition, 27, 141–172 .

15.

Ellis

(2007). The differential effects of oral corrective feedback on two grammatical structures. In Mackey

(Ed.), Conversational interaction in second language acquisition: A collection of empirical studies (pp. 339–360). Oxford: Oxford University Press.

16.

Ellis

(2009a). Task-based language teaching: Sorting out the misunderstandings. International Journal of Applied Linguistics, 19, 221–246.

17.

Ellis

(2009b). The differential effects of three types of task planning on the fluency, complexity and accuracy in L2 oral production. Applied Linguistics, 30, 474–509.

18.

Foster

Skehan

(1996). The influence of planning and task type on second language performance. Studies in Second Language Acquisition, 18, 299–323.

19.

Foster

Skehan

(1999). The influence of source of planning and focus of planning on task-based performance. Language Teaching Research, 3, 215–47.

20.

Foster

Skehan

(2013). Anticipating a post-task activity: The effects on accuracy, complexity and fluency of L2 language performance. Canadian Modern Language Review 69, 249–273.

21.

Gass

Mackey

Alvarez-Torres

M.J.

Fernandez-Garcia

(1999). The effects of task repetition on linguistic output. Language Learning, 49, 549−581.

22.

Hawkes

M.L.

(2012). Using task repetition to direct learner attention and focus on form. ELT Journal, 66, 327–336.

23.

Hulstijn

J.H.

de Graaff

(1994). Under what conditions does explicit knowledge of a second language facilitate the acquisition of implicit knowledge? A research proposal. AILA Review, 11, 97–113.

24.

Kim

(2013). Effects of pretask modelling on attention to form and question development. TESOL Quarterly, 47, 8–35.

25.

Levelt

W.J.M.

(1989). Speaking: From intention to articulation. Cambridge, MA: MIT Press.

26.

Lynch

Maclean

(2001). A case of exercising: effects of immediate task repetition on learners’ performance. In Bygate

Skehan

Swain

(Eds.), Researching pedagogic tasks: Second language learning, teaching, and testing (pp. 141–162). Harlow: Longman.

27.

Lyster

Saito

Sato

(2013). Oral corrective feedback in second language classrooms. Language Teaching, 46, 1–40.

28.

Macrory

Stone

(2000). Pupil progress in the acquisition of the perfect tense in French: The relationship between knowledge and use. Language Teaching Research, 4, 54–82.

29.

Qian

(2014). The effects of post-task transcribing on learners’ oral performance. In: Skehan

(ed.), Processing perspectives on task performance (pp. 129–154). London: John Benjamins.

30.

Riggenbach

(1991). Toward an understanding of fluency: A microanalysis of nonnative speaker conversations. Discourse Processes, 14, 423–441.

31.

Sato

(2010). Reconsidering the effectiveness and suitability of PPP and TBLT in the Japanese EFL classroom. JALT Journal, 32, 189–200.

32.

Shehadeh

Coombe

C.A.

(2012). Task-based language teaching in foreign language contexts: Research and implementation. Philadelphia: John Benjamins.

33.

Sheppard

(2006). The effects of instruction directed at the gaps second language learners noticed in their oral production. Unpublished PhD thesis, University of Auckland, New Zealand. In Ellis

(2009b). The differential effects of three types of task planning on the fluency, complexity and accuracy in L2 oral production. Applied Linguistics, 30, 474–509.

34.

Skehan

(1996). A framework for the implementation of task-based instruction. Applied Linguistics, 17, 38–62.

35.

Skehan

(1998). A cognitive approach to language learning. Oxford: Oxford University Press.

36.

Skehan

(ed.) (2014). Processing perspectives on task performance. London: John Benjamins.

37.

Skehan

Foster

(1997). Task type and task processing conditions as influences on foreign language performance. Language Teaching Research, 1, 185–211.

38.

Swan

(2005). Legislation by hypothesis: The case of task-based instruction. Applied Linguistics 26, 376–401.

39.

Thai

Boers

(2015). Repeating a monologue under increasing time pressure: Effects on fluency, accuracy and complexity. TESOL Quarterly. Advance online publication. doi: 10.1002/tesq.232

40.

Van den Branden

Bygate

Norris

(Eds.) (2009). Introduction. In: Task-based language teaching: A reader (pp. 1–13). Amsterdam: John Benjamins.

41.

Van de Guchte

(2015). Focus on form in task-based language teaching. Unpublished doctoral dissertation, University of Amsterdam, The Netherlands.

42.

Van de Guchte

Braaksma

Rijlaarsdam

Bimmel

(2015). Learning new grammatical structures in task-based language learning: The effects of recasts and prompts. The Modern Language Journal, 99, 246–262.

43.

Willis

(1996). A framework for task-based learning. Harlow: Longman.

44.

Yang

Lyster

(2010). Effects of form-focused practice and feedback on Chinese EFL students’ acquisition of regular and irregular past tense forms. Studies in Second Language Acquisition, 32, 235–263.