Effects of Immediate Repetition in L 2 Speaking Tasks : A Focused Study

This paper reports on a focused investigation into the immediate effects of oral narrative task repetition by two adult EFL learners of intermediate and high proficiency. Two participants performed a narrative speaking task after watching a cartoon video clip and repeated their performance three times, followed by a retrospective report in an interview. The results showed that repetition of narrative tasks increased fluency and accuracy, while complexity was the least sensitive to the practice effect. At the same time, it was found that the learners had generally correct self-perception of their performances, which was the interaction of enhanced repeated performance, fatigue, and their proficiency levels.


Introduction
The value of "practice" in learning has long been emphasized by teachers and educational researchers as it was supposed to render a certain skill "perfect".This is the epistemological basis for Ivan Pavlov and his disciple Frederic Skinner's behaviorist psychology that dominated early language learning theories.However, behaviorism was criticized by Chomsky as ignoring the innate human capacity for language, and hence gradually lost its influence in the field.Little significance is attached to repetition in second language (L2) classrooms nowadays.An intriguing fact is that children acquire their mother tongue with little effort, and at a fast rate.In contrast, the large population of L2 learners suffers from the phenomenally low success rate in their ultimate attainment.The plethora of never-frustrated L1 speakers and the paucity of the blessed L2 learners leave one wondering about the apparent difference between L1 and L2 learning.It seems that the over-optimistic L2 = L1 hypothesis in early SLA research needs to be questioned.Needless to say, formal instruction is still the most common way for L2 learners to acquire the target language.Ever since Bygate's (1996Bygate's ( , 2001) ) studies on task repetition, there has been a revival of research into the area of task repetition and its pedagogical value.This study aims to investigate the effects of task repetition in two focused case studies with both quantitative and qualitative data in a hope of presenting a more in-depth perspective on task performance and learner perception.Levelt's (1989) speaking model describes speech production in three stages, namely the conceptualizer, the formulator and the articulator.The conceptualizer controls the macro-planning stage, which provides general knowledge and discourse knowledge as input for the formulator in the next stage.The formulator in turn brings together the vocabulary, grammar and syllabary to generate a phonological plan, which is utilized by the last stage, namely the articulator, for actual oral production.These three stages are parallel processes in L1 speaking where native speakers are able to plan the content, organize language and make utterances at the same time.For most of the L2 speakers, however, speech making has to proceed step by step as a serial process.There is a monitor situated in the conceptualizer to supervise the whole course of speaking for appropriacy of the content and accuracy of the language and pronunciation.Over the years, different adaptations were proposed to this model in order to account specifically for L2 and bilingual speaking (de Bot, 1992;de Bot & Schreuder, 1993;Payne & Whitney, 2002;Poulisse & Bongaerts, 1994).Nevertheless, all would agree that one's working memory is limited, so are one's attentional resources and processing capacity (Skehan, 1998;VanPatten, 1990).As Bygate (1996) pointed out, L2 learners were expected to be more fluent in terms of pausing and speed after repeating the same task.This is because, all things being equal, one would expect that doing the task a second time could involve less planning work.Also it is likely that speakers would have fewer false starts and self-corrections because the task has already been formulated previously.Similarly, few errors can be achieved because learners have succeeded in assembling language items once.Some researchers have shown that L2 learners benefit from task repetition.To name a few, Gass et al. (1999) tested the frequency effects through story re-tellling.They found that after three repeated (or similar) tasks over approximately two weeks, the overall performance, morphosyntax and lexical sophistication all showed significant improvement, which they argued were the result of familiarization with the exercise content and the shift of attention from planning to linguistic forms.Lynch and McLean (2001) also reported generally positive effects of repetition in their poster carousel tasks.A clearer picture of the overall highly beneficial influence of task repetition was presented in Bygate's (2001) ten-week well-structured study.

Literature Review
In the days when Krashen's input hypothesis prevailed, Swain (1985) counter-offered her well-known output hypothesis.Based on her longitudinal research of Canadian emersion education, Swain maintains that comprehensible input is necessary but not sufficient, and there must be enough comprehensible output to reach native-like accuracy and fluency, because learners can test their hypotheses of the target language and notice the gaps in their language knowledge, thus rendering self-modification and correction possible.
It will then be intriguing to see how repetition and output can be combined to benefit learners' language processing.There is no denying that this practice will bring about the issue of boredom and fatigue, as exemplified by Plough and Gass (1993), which reported that immediate task repetition did not show any positive effects on learners' performance because they were likely to lose interests.However, as zhou (2006) suggested, there must be an optimal point where the positive effects of task repetition reach their height while the negative impacts of boredom are still not too devastating.This study tries to combine the methodology of mainly Zhou (2006) and Bygate (1996) in a hope to examine the repetition effects of L2 speaking tasks.Retrospective reports from post-task interviews were also conducted to explore learner perception of task performance and fill the gap of qualitative research in this area.

Research Questions
In light of the existing literature as well as the gaps in relevant research, the following three research questions guided the present study: 1. Does immediate task repetition lead to improvement in L2 oral task performance?2. At which time of the task repetition can learners reach their optimal performance?3. What is the participants' perception towards task performance at different times of task repetition?

Participants
Two adult L2 English learners participated in this research.Both of the participants were postgraduate students at a university in Hong Kong where they had lived for nearly one year by the time of the study.Their mother tongue was Chinese, with one speaking Cantonese and the other, Mandarin.Participant One was an L2 speaker of high proficiency with a score of 8 in IELTS oral test, while Participant Two was an intermediate learner who obtained 6.5 in IELTS oral test.Both of them were familiar with the carton Tom and Jerry employed as task input in this study.

Input Material
Following Bygate (1996), a three-minute video clip from the cartoon Tom and Jerry was adopted as the input material.The advantages of such choice were: 1) the clip contained no dialogue, and thus simplified the task since incorporating dialogue into a narrative was likely to give rise to grammatical and functional complications; Linguistic priming effect in the conversation could be avoided.2) the cartoon was well-known to both participants; 3) the cartoon ruled out the interferences of listening comprehension problems, making input comprehension more controllable.

Study Design
The participants watched the video abstract once and went on to narrate the scenario four consecutive times.They were allowed to take a break between each speaking so as to alleviate fatigue, but they were instructed not to practice during the break.All the speaking was recoded with an MP3 recorder, which was transcribed and analyzed in terms of their accuracy, fluency and complexity (see section 4.5 Data Transcribing and Coding below for the operationalization of these measures).

Procedure
Firstly, the researcher collected the background information from the participants by having them fill out a survey form (see Appendix 1) during which the researcher reassured them that the information would be kept confidential and used in this research only.
Secondly, the researcher performed a demonstration of such a task by playing a short sample video (different from the real task) and narrating it twice, while telling them they would be doing it four times.During the demonstration, the MP3 recorder had been placed on the desk and switched on to get the participants accustomed to it.
Thirdly, the researcher played the target video, and asked the participant "can you tell me what happened in the video?".The participant then immediately began speaking upon hearing the question.The two participants did the task separately on the same day.
Fourthly, following the completion of the last rounds of speaking, the participants were interviewed by the researcher to answer a few semi-structured questions (see Appendix 2).Through this, the researcher was able to probe into some of the mental processes the participants went through during their four times of speaking.To reduce their boredom and avoid the risk of obtaining wild answers, the participants were allowed the freedom to answer the questions and be recorded or to fill the form in Appendix 2 by themselves.They could opt for either English or Chinese in answering the questions.In the actual interviews, both participants chose to speak mainly in Chinese (one in Cantonese and the other in Mandarin) with code switching to English at times.

Data Transcribing and Coding
The recorded speaking performance was transcribed (see Appendix 3 for the transcription) and coded for the following three dependent variables: Accuracy: the proportion of error-free clauses.An error-free clause is a clause without syntactic, morphological, or word order errors (Foster & Skehan, 1996); Fluency: the total syllables per minute after deletion of reformulations, replacements, false starts, repetitions and hesitations (Ortega, 1999); Complexity: the ratio of clauses to AS units (Foster, Tonkyn, & Wigglesworth, 2000).It was calculated by dividing the data into syntactic clauses and AS-units and expressed as the ratio of clauses to AS-units; An AS-unit is defined as a single speaker's utterance consisting of either an independent clause, or sub-clausal unit, together with any subordinate clause(s) associated with it (Foster, Tonkyn, & Wigglesworth, 2000).That is, the more clauses per AS-unit, the higher the complexity score (Foster & Skehan, in review).

Data Analysis
Given such a small sample size and the explorative nature of this study, the results are delineated mainly in descriptive statistics.Nevertheless, correlations of dependent variables and Mann-Whitney U tests (since the data were not normally distributed) between two participants were employed.A qualitative approach was adopted to analyze the the post-task interview for the participant perception of their own performance..0259.4 .167

Performance of Participant One
Table 1 presents the results of the highly proficient Participant One's four rounds of speaking.The highlighted figures showed the highest score achieved among the four tasks.For accuracy, there was not much variation between the four times (S.D. = 0.025) because there were two syntactically wrong clauses in the first and the third tasks, together with only one in the fourth task.However, it may be too early to draw a conclusion that task repetition does not help in accuracy.The proficiency level of this participant should be taken into consideration.He managed to obtain such a high accuracy rate in the first round of speaking, leaving little room for improvement in subsequent tasks.That said, Participant One still produced error-free performance at the second time, in which the complexity score also reached its maximum.The post-task interview clearly showed that the participant had precise self-perception.He ticked the second time in the survey form as his best performance, arguing that in the second round he was able to still remember all the video content and at the same time the practice effect from the previous speaking remained highly facilitative.For fluency, the immediate repetition effect seemed to be more clear-cut.The participant took a two-minute break between task two and three, leaving two optimal performances in the second and the fourth time, but generally, a rising trend in fluency along the four rounds of speaking was evident.For complexity, there seemed to be no clear pattern showing the practice effect.Nevertheless, what appeared regular across the three dependent variables was the sudden downturn in the third speaking, which ran counter to the Zhou's (2006) claim that the third speaking is the optimal point for oral task repetition.This happening suggests that, for some L2 speakers at the advanced level at least, the performance would be optimized in the second try.Participant Two should be more representative of typical L2 learners who have some control over oral English but still speak with effort.Both of his fluency and accuracy arrived at the optimal point at time three, which was in line with Zhou (2006).There was a sharp increase in accuracy at time two after the previous two almost identical performances, with a strange lowest accuracy rate at time four.The interview showed that the participant had realized this: "I think my third time was the best.Though I had wanted to do even better in the fourth time, I went for wool and came home shorn-I made a lot of mistakes instead.Maybe (it was because) I was too tired."(Note 1) However, fluency at time four was still the second best, indicating that repetition did promote fluency, a trend similar to that of Participant One.For complexity, once again there existed no obvious task repetition effects, with almost no change over the four rounds of video narration (S.D. = 0.02).

General Task Performances
When all eight speaking tasks are taken together, some clearer patterns emerge.First of all, the Mann-Whitney U tests showed that Participant One outperformed Participant Two in terms of accuracy (z = 2.31, p = .021)and fluency (z = 2.31, p = .021),but not complexity (z = 1.16, p = .25).Though the IELTS results served as a good indicator of the proficiency discrepancy between the two speakers, complexity did not seem sensitive to influence from either task repetition or proficiency levels.Zhou's (2006) belief that a repeated task can enhance complexity because speakers can be prepared to venture into more complex sentence structures is not supported in this study, not even in the first participant with high proficiency.One possible cause is that this study adopts only a syntactic complexity measure while leaving out the lexical complexity.Oral language is usually fragmentary and tends to be simple utterances, which could be more so in L2 speech.All this imposes difficulties on the measurement of its syntactic complexity.Secondly, statistics from Table 3 show that there is a very high correlation between fluency and accuracy (r=.86, p< .01),while the complexity is correlated with neither fluency (p > .05)nor accuracy (p > .05).This result seems to suggest that there is a trade-off effect (Foster & Skehan, 1996) in L2 learners' oral production processes.That said, fluency strongly correlates with accuracy (r = .86,p < .01),indicating that the two areas of performance can improve at the same time.Given limited processing capacity of L2 learners in speaking, the focus of attention in some performance areas would occur to the detriment of others (complexity in this case).In general, the dependent variables appear to compete for attentional resources for real time L2 speech making.
The trade-off effect is more evident in Table 4 where the mean scores of the four rounds of speaking by both speakers were calculated.To reach their optimal point, complexity needs twice practices, accuracy three times and fluency four times.A closer examination would reveal a clearer sign of competition that when accuracy reaches it highest point, complexity is at its lowest.This seems to support the trade-off hypothesis by Foster and Skehan (1996), in particular their claim that the major competition happens between accuracy and complexity.Thirdly, concerning the repetition effect per se, generally an ascending trend in these performances is observable.
Practice does appear to exert significant and positive effects on accuracy and fluency.However, complexity seems to be the least sensitive to task repetition in that an unequivocal pattern across the four rounds of speaking did not seem to be in place.The results reported in this study are almost identical with Zhou (2006) in terms of accuracy and fluency, but not complexity in that complexity also managed to get to its peak at time three in her research while it was quite the opposite in the present study.Swain's (1985) output hypothesis is largely confirmed in this study.Output can help learners realize the problems in their language, and the gaps between their interlanguage and the target language, which may enable them to improve in their language use and expression.Learners could be presented with the opportunity to respond to their linguistic deficiency through practice.That is why accuracy was on the increase over the first three times in the present research.To notice the problems in one's language requires one's conscious effort and learners need to pay enough attention to rectify mistakes or avoid habitual or idiosyncratic errors.In this sense, it is not that the more practice the more accurate because the boredom and the consequent impatience may be detrimental to one's attention concentration.Both speakers reported to be tired and bored after the third time and their accuracy went down significantly in the fourth time.
That being said, fluency touched the highest point at time four, suggesting that task repetition may help speakers automatize their performance.Fluency differs from accuracy in that essentially fluency is the result of proceduralization.It can be compared to the practice of tongue twisters-the more one repeats the faster s/he will become.Unlike accuracy, the attainment of fluency depends on automatization without much requirement for conscious effort.Therefore, fluency will be less influenced by attentional resource, which might be the reason why it is on an invariable up-going trend along the four speaking tasks.
The last intriguing point concerns the issue of proficiency.Though the mean scores appear to support a trade-off effect between performance areas, Participant One with high proficiency managed to strike a balance in these areas to a great extent.What seemed most interesting in his case was that he spoke the most accurately with the most complex sentence structure at the second time (see Table 1), seemingly "violating" the trade-off hypothesis.This result could be attributed to his high proficiency which was reflected in his native-like performance with highly autonomous skills where less attetentional resources are required for speech conceptualization and formulation (Levelt, 1989).Nevertheless, his most fluent speech still had to be made at the fourth time, but not the second time, which indicates that the competition between performance areas is not totally defied.

Conclusion
This study reports on an attempt to investigate the effects of immediate task repetition in L2 speech.It drew on methods from previous studies and measured the performance of narrative repetition of a short video clip.Based on the findings, the following conclusions are made in response to the three research questions: 1) task repetition appears to significantly improve fluency while accuracy also has quite some benefit to gain.Repeating a speaking task, on the other hand, had little influence on complexity in these two cases; 2) the third time in general is the optimal time for the general performance where accuracy is the best, fluency is the second best and complexity does not change much; 3) the participants had correct self-perception towards their own performances.The interview showed that they were aware of the best turn in their speaking, which they believed was a result from practice in previous speaking.Fatigue in the subsequent rounds of speaking was reported to bring about negative impact on task performance.This result offers preliminary evidence that the psychological or even physiological state of learners can have considerable influence on their performance.
A number of limitations existed in this Firstly, though this study was designed on purpose as an in-depth exploration into individual cases in task repetition performance, the small number of the participants prevents any grander claims or generalizations from being made.Secondly, the complexity measure employed in this study was only concerned with speakers' syntactic structures.Lexical aspects were largely ignored.Future studies should advance such a work further by examining task influence on the depth and the width of their vocabulary use.Then it has to be admitted that this study is exploratory in nature with results and discussions being more suggestive rather than conclusive.However, intriguing findings such as the observation that higher proficiency could help ease trade-off effects are worthy of future research at a larger scale and in a wider context.
I: Thank you.

nd narrating of Participant One
Research: second time.
Participant One: Ok.Ah (1) I am not gonna repeat what I say and it is kind of stupid (laugh).So I am gonna start all over.So er it is a beautiful day.The sky is blue.So Tom is having a good time on the beach.Ah He is taking a rest in the shade of a huge parasol.Am and he is pulling (1) some sun cream over himself to (0.5) get away from the sun light.And ah he wears a pair of sunglasses to pre protect his eyes.He also pulls over a huge blanket over himself so he doesn't see anything or anybody.And all of sudden, Jerry the mouse with a huge fishing rod comes all over him.He basically rides across him.So Tom is very angry at what Jerry has done to him and he chases after him.So they go all the way to the pier, and (1) Tom ( 2) being (1) I mean he thinks he smart.He he, he will outsmart Jerry, but unfortunately, he is not.As Tom waits quietly somewhere by the pier in order to catch Jerry while Jerry is trying to fish something from the sea with a piece of cheese.Participant One: hmm, ok ah.I am feeling so refreshed with milk and everything (laugh).Ok the third time hopefully my memory still serves me.Erm It is a beautiful day with a blue sky and Tom the cat (0.5) is taking a rest in the shade of a huge parasol.Ah he is putting some sun cream on himself and he pulls a huge blanket over himself.And all of a sudden, Jerry the cat walks over him with a long and big fishing rod.So obviously, Tom becomes very angry at what Jerry has done to him.So he chases Jerry all the way to the pier where Jerry the mouse is trying to fish something from the sea with a piece of cheese.And Tom see his chance coming, he quietly waits somewhere at the pee pier, and em wants to catch Jerry.Of course Jerry is always the one who outsmart Tom so ah (2) along the story Tom unfortunately falls all the way to the bottom of the sea and at the bottom of the sea, he wakes up to a whole new world with different creatures.Em the first creature he sees is (1.5) ah a kind of colored fish.So he is likes I think he likes the fish.So he mimics the action and movements of the fish, which is very funny.And the second time he sees a huge turtle and he does the same thing as (0.5) the previous time.So all of a sudden Jerry appears, and this time Tom is not trying to mimic him, but trying to catch him, and something that obviously he manages (1) to do.But unfortunately I think Jerry (2) will find a way to get away.(…not trancribable) what's happening towards the very end of all the Tom and Jerry stories?So that will be the end of it?
4 th narrating of Participant One Participant One: I get so sick and tired of it (laugh).Ok em right, ah.So Tom the cat (1) is having wonderful time on the beach.He is putting some sun oil on himself.He is also wearing a pair of sunglasses.And he pulls (2) a huge blanket over himself.And all of a sudden, Jerry the mouse walks all over him with a huge fishing rod.So Tom (0.5) is angry with what Jerry has done to him.So he chases Jerry all the way to the pier.And Jerry, I mean strangely, he is trying to fish something from the bottom of the sea with a piece of cheese.I mean he is using the cheese as the bait.I don't know what he is going to do with that but, Tom quietsly, quietly waits at the pier.He is trying to catch Jerry.So ah, accidentally, he falls to the bottom of the sea and when he wakes up he sees a whole new world of different creatures.So first of all he sees a kind of colored fish which he mimicsly in a very funny way.And the second time he sees the turtle, and he does the same thing as the previous time.And the third time, well, it is not (1) a welcome visitor obviously because, well, it is a welcome visitor because, you know, Jerry is always on the mind of Tom.So this time Jerry appears and Tom catches him as quickly as he can.But I think Jerry can (1) always find a way to run away, because (0.5) he is the one what outsmarts Tom, otherwise we won't have the (1) Tom and Jerry story any more in the future.Ehm.

Table 1 .
Performance of Participant One (high proficiency)

Table 2 .
Performance of Participant Two (intermediate proficiency)

Table 4 .
Mean scores of both participants Tom accidentally ends up (1) being (3.5) fooled by (1) Jerry.So he falls all the way to the bottom of the sea.And (1) right then and there, he wakes up and he sees this world of sea creatures (1) going past him.So first of all, he is mimicking a colored fish.And the second time I saw him mimicking the turtle, which was funny to my friend but I don't think so.I said I don't because I didn't think so and I still don't (laugh).So ok, all of a suddenly like (1) magic, ok Jerry appears at a piece of rock.And Tom is so overjoyed at what he sees.And he simply seized Jerry and cobs him in his hands and we don't know what happens next.