D: This took place in, probably the 1970s when stores weren’t quite so sophis- ticated as they are today. I was in the clothing department of a very busy department store and decided to try a pair of trousers on, and the shop assis- tant referred me to the changing room, and I went in. They had the sort of louvre doors, the sort of saloon room door, saloon bar door, you know, on hinges, you know the sort, and I went in, took my trousers off, and I put my left leg in the trousers and I went to put my right leg in the trousers I’d picked up in the store, and some clown had sewn the price ticket half way-they’d actually stapled the price ticket half way up the trouser leg, and I went to put my foot through and of course couldn’t. I tripped backwards through these doors, went cata-cata-cata straight out onto the shop floor (⬍laugh⬎), with one leg in the trousers, one leg half way up the other trousers, jumping around, fell flat on my back, right in front of all the shoppers, on a Saturday morning!
M: ⬍Laugh⬎That’s pretty good. I don’t know if I can top that, but the only thing I can think of that’s even mentionable on tape is in Japan, in a good old onsen, which is a hot spring. It’s happened to many, many people, in the days before I could read any Japanese characters, least of all the ones that said ‘man’ and the ones that said ‘woman’.
D: ⬍Laugh⬎I think I’m ahead of you here!
M: Yes, I think you probably are! Just, well, stripping off naked, as you do, and going into the bath with all the beautiful 80-year-old women, that’s about it⬍laugh⬎. I won’t elaborate. So whose do you think was the most embar- rassing? I’d have to say yours was.
D: Yeah, I think probably, um … Given that, um … Well you can imagine the scene, can’t you? This man wearing a pair of white underpants dancing around on one leg …
M: Oh! White underpants? That’s it, that’s the winner!
Part D
Investigating Variables: Task Conditions and Task Types
There are many alternative ways of designing and implementing tasks.
But how do we select which ways are best for our particular learners? In order to make sensible choices, we need to know what is likely to hap- pen when tasks are set up and implemented under different conditions.
The teachers in the five chapters in Part D describe how they explored the differences in learners’ interactions when they changed the way that they designed or set up a task, or followed up a task.
The first three papers explore the use of tasks within the context of a broader task cycle. In Chapter 15, Craig Johnston recorded mature adult learners doing tasks and found that quite often, the language they used in the privacy of their pairs to achieve the task goals successfully was very simple and stilted, sometimes nothing but a few words or phrases put together. He was afraid that their language would fossilize, in other words remain forever at that low level. So he experimented by giving learners time after a decision-making task to plan how to tell the whole class what they had decided together during the task, and then he recorded their more public reports. To examine precisely what differ- ences occurred in the language at Task and Report phases, Craig used four different kinds of measure, assessing syntactical accuracy, lexical accuracy, syntactical complexity and lexical variety. He illustrates clearly what each measure involves (so that you could use the same measures) and comes up with some interesting results.
In Chapter 16, William Essig, convinced of the relevance of story- telling to our every day lives, explores the different effects that result from telling a story spontaneously (without time to plan or rehearse). But he also looks at what happens to learners’ language when they are asked to re-tell their story to someone else later on, having had time to plan how they will re-tell it using dictionaries and other resources. He also
compares private and public contexts. He made all his story recordings in one lesson, then, having come up with eight hypotheses, he tested each one by looking closely at the recorded data. Not all his hypotheses were borne out, which shows how useful it is to investigate rather than simply assume things are true. And his students really enjoyed telling and re-telling their stories.
While Johnston and Essig both explored the effects of post-taskactiv- ities, Antigone Djapoura, in Chapter 17, looked at the pre-taskstage of a task-based cycle. She investigated whether allowing her learners’ pre- task planning time affected their task-performance, and in what ways.
Like Essig, she sets out a number of hypotheses which she then goes on to test. Her research design is similar to that used by Foster and Skehan (1996) and she compares her results with the findings of that study. This is an excellent example of what is often called a replication study. The value of a replication study, as she herself says, is that ‘looking for pat- terns in results across these two studies gives us the opportunity to make more powerful generalizations’.
The final two chapters in Part D look at variables in task design, rather than the effects of different phases in the task-cycle.
Greg Birch in Chapter 18 is worried (like Craig Johnston) that task- based interactions might promote fluency rather than accuracy and thus lead to fossilization. He sees the need to try to attain a balanced focus and wonders whether different task types systematically offer different learning opportunities. He uses the work of Skehan (1998) on assessing task difficulty and examining task characteristics in order to predict whether learners doing particular types of task are more likely to focus their attention primarily on fluency, accuracy or complexity of lan- guage. He selected two one-way information gap tasks suitable for his 16-year-old students, and recorded three classes of 40 students doing both tasks. He predicted, on the basis of Skehan’s work, that both tasks (but for different reasons) would push learners to a reasonably accurate and fluent interaction, but would not push them to use complex con- structions. Was he right? Read it and see.
In Chapter 19, Glen Poupore compares learner interaction arising from two different types of task: problem-solving/prediction tasks, where all learners have the same information but need to reach some kind of joint decision, and jigsaw tasks, where each learner has some information that the others don’t have but need. However, in order to make this comparison, and to find out which type of task is more likely to lead to learners’ language development, he needs some way of meas- uring and defining what he feels is ‘quality interaction’. Assuming, as
Long (1988) and others have since done, that when learners negotiate meaning (eg ask for clarification) they are likely to learn more, Poupore explores his data to discover what other aspects his learners negotiate and uncovers how far these negotiations push learners to a higher quality of interaction.
Both Birch and Poupore discover that some kinds of task may be bet- ter than others for achieving different kinds of interaction and learning opportunities. It is now up to us as readers to continue the explorations, so that we can gain a larger and clearer picture of which types of task are likely to achieve what.
Investigating Variables 189
15
Fighting Fossilization: Language at the Task Versus Report Stages
Craig Johnston
191
Summary I wanted to test the claim that a public report stage in the task cycle can help learners monitor the quality of their language output. To do this, I used four different measures (syntactical and lexical accuracy, syntactical complexity and lexical variety) to investigate the differences between task- stage and report-stage language. I found some marked differences, suggesting that a report stage is indeed valuable.
Context and rationale
Students at the conversation school where I work in Japan place a high - priority on talk time and meaning-focused language use. Communicative tasks serve very well in this environment, however the quality of language employed by students in completing tasks is sometimes a concern. To take an extreme example, here is an excerpt from an exchange between two low-intermediate students who are trying to decide on a seating plan for well known guests at a dinner party:
(Note: Koizumi is Japan’s Prime Minister at the time of writing. Soseki and Tokugawa are well known historical figures in Japan. More details on the context of this excerpt will follow).
Kumiko Koizumi here?
Hiroko: Koizumi Kumiko: That’s good idea Hiroko: Soseki … Kumiko: Ah, Tokugawa
Hiroko: next to Tokugawa ruler same ruler I …
Kumiko: OK, so our final agreement Hiroko: Partner? (inaudible Japanese) Kumiko: No, uh
Hiroko: Final?
Despite the obvious problems, Kumiko and Hiroko understood each other and went on to complete the task successfully but using very min- imal language. This highlights an area of concern in task-based learn- ing, as noted by Seedhouse (1999). With success defined by task completion, often under a time constraint, students may focus dispro- portionately on fluency (ie expressing their meanings quickly, using gestures and even their L1 to do this) at the expense of accuracy and complexity (Skehan 1996a: 22). This in turn may lead to ‘fossilization’, ie the stagnation of L2 development.
One response to this challenge is to employ a post-task ‘report stage’
during which the learners report to other groups, or possibly the whole class, on the outcomes they achieved during the task (Willis, J. 1996a:
54–60). The report stage, by taking advantage of the natural desire to present a higher quality product when one knows the product will go on public display, seeks to reassert a focus on accuracy and complexity (ibid: 55; see also Essig, this volume). This shift in focus is supported by a planning stage which precedes the report, allowing students to organ- ize their thoughts, select appropriate language, and consult resources such as a dictionary. Intuitively it seems that the planning and report stages should help, but I wanted to find out if empirical data would sup- port this and, if so, to what extent and in what ways? Does the differ- ence in quality warrant the extra time spent? This paper describes an investigation that I carried out in search of answers, or beginnings of answers, to these questions.
Method
The students and the task
I carried out this study with a group of three students. A larger group would have yielded more data, however the small size of the office space, paired with the need for clear recordings, necessitated a small group. The partici- pants, Hiroko, Kumiko and Asuka, are all middle aged women, studying English primarily for pleasure and the opportunity to meet with foreigners.
The task instructions were as follows:
Imagine that the following people are coming to your house for a dinner party.
● Junichiro Koizumi ( Japan’s Prime Minister at the time of writing)
● Makiko Tanaka (controversial former Foreign Minister, recently fired by Prime Minister Koizumi)
● Ieyasu Tokugawa (iron-fisted shogun of Japan, early seventeenth century)
● Buddha
● Soseki Natsume (one of Japan’s most celebrated writers, late eighteenth and early nineteenth centuries)
● Sanma Akashiya (universally loved funnyman, appearing on numerous TV shows)
Your mission is to create a seating arrangement, around a circular table, which will allow them to enjoy interesting conversations with their neighbors, hope- fully avoiding arguments.
Step 1: By yourself, decide on an arrangement.
Step 2: Explain your arrangement and your reasoning to your partner. Then lis- ten to your partner’s explanation of her arrangement and write it down.
Step 3: With your partner, create a final arrangement that you both agree is best.
Procedures
Grouping of students and report preparation
With only three students, I had to partner with one of them. The alter- native, a single group of three, would have made the report redundant;
why report on what you did if your audience did it with you? After the task had been completed they were asked to prepare a report for home- work describing how they and their partner solved the problem. The fol- lowing week everyone was paired with a new partner, to whom they presented their report. At both stages the two pairs worked in separate rooms to enable clear recordings.
Measurement
I used the transcribed task discussions and reports to measure syntacti- cal (ie grammatical) accuracy and complexity as well as lexical (ie vocab- ulary) variety and accuracy. These concepts, and their relevance, will be discussed in more detail as they arise, below.
Findings
The results are presented below in three sections, the first dealing with accuracy, both syntactical and lexical, the second with syntactical com- plexity and the third with lexical variety.
Fighting Fossilization 193
Before discussing the results, I should point out that Asuka’s report was cut short by a question from Hiroko. This launched the two into a spontaneous discussion and effectively terminated Asuka’s report. For this reason, analysis of Asuka’s report was not carried out beyond Hiroko’s question, making it roughly half as long as those of Hiroko and Kumiko. Some interesting results were observed nonetheless.
Accuracy
Syntactical accuracy
Syntactical accuracy was measured by comparing the number of error-free clauses with the total number of clauses (Foster, 1996: 133). For example, Kumiko’s (report stage) statement that ‘He’s a famous novelist and he seem to be gentle’ scores 50 per cent on this accuracy test:
He’s a famous novelist and he seem to be gentle.
Error-free Error (seems)
Since only one of the two clauses is error free, the score is 50 per cent.
The levels of syntactical accuracy for all three students, at both task and report stages, are shown in Table 1 below.
Table 1 Syntactical accuracy
Student Percentage of error-free Percentage of error-free clauses at the task stage clauses at the report stage
Asuka (6/33) 18.2% (6/20) 30 %
Hiroko (2/39) 5.1% (5/34) 14.7%
Kumiko (15/47) 31.9% (6/32) 18.8%
(Figures in parentheses indicate the number of error-free clauses and the total number of clauses, respectively.)
Surprisingly, Kumiko’s accuracy actually dropped. In a later interview, Kumiko indicated that she had been too busy to put much time into her report. This may account for the drop. It may also be the case that, as the most proficient English speaker in this small, friendly class, Kumiko felt very little to be at stake in presenting her report, and so may not have approached its preparation with the same care as Hiroko and Asuka. Both Asuka and Hiroko produced decidedly more accurate language in their reports, Asuka posting a 65 per cent gain and Hiroko almost tripling her
percentage of error-free clauses. Given the low level of accuracy in their task language, these are important results, suggesting that the planning/
report stage may indeed help boost accuracy and play a part in combating fossilization.
Lexical accuracy(or Lexical selection)
Accuracy, as defined above, does not help us evaluate lexical choices and so I will use the term lexical selectioninstead; ‘improved lexical selection’
meaning ‘a more native speaker-like choice of words’. Using lexical selection as a measure of accuracy, we find that the above results actu- ally understate the gains. Consider the following comparison of Hiroko’s task and report language:
Task Report
Soseki is a novel writer so Koizumi’s right hand side seat is
Koizumi Prime Soseki.
Minister is … very like reading Koizumi likes reading a book so books so next to the Soseki. they get along well with each other.
I think most readers will agree that the report version represents more native-like use of the language and yet, by the above test of syntactical accuracy, it actually scores lower than the task version. (Both have three clauses but only the task version has a grammatically error-free one:
Soseki is a novel writer. In context, they get along well with each other Fighting Fossilization 195
Table 2 Changes in lexical selection
Speaker Language used during Language used during
task stage report stage
Kumiko Koizumi and Tanaka … keep Koizumi and
must be sit at separate Tanaka and Tokugawa table. (Grammatically and apart from each other.
semantically flawed; separate tables were not an option.)
Kumiko Sanma-san is very good, He (Sanma) has a sense uh, has a nice character. of humor and friendly.
Kumiko Soseki is a famous He (Soseki) is a
nov, novel … famous novelist.
Hiroko Quiet, quiet person is Tokugawa, I separated to every second Soseki, Buddha. And talkative is seat a talkative person and Koizumi, Tanaka, Sanma. a calm person sit.
requires the insertion of ‘would’ to be considered error-free.) What then makes the report version the preferred one? I would suggest that it is the improved lexical selection represented by the phrase, ‘A’ gets along well with ‘B’. Note that the ‘correctness’ of this phrase cannot be explained by grammatical rules; it is a ‘chunk’, a single unit of vocabulary (Bolinger 1975). Table 2 presents other examples of differences in lexi- cal selection. In each case, the gains achieved spring not from gram- matical accuracy but from superior choices of words.
The improved lexical selection seen in the reports is another impor- tant result, further supporting the inclusion of planning and report stages in the task cycle.
Syntactical complexity
Complexity ‘concerns the elaboration or ambition of the language which is produced (Skehan, 1996a: 22)’. A review of the transcriptions showed that report language was markedly more ambitious. For example, task stage language was typified by simple utterances such as those in this four-turn excerpt:
Hiroko Now is Koizumi and Tanaka is this … Kumiko Don’t, don’t get along with each other.
Hiroko Each other, so … Tokugawa Kumiko Tanaka-san here
Even when taking longer turns at the task stage, the students’ language remained very simple as in the following example. (Putting aside the many syntactical errors, notice that Hiroko has relied heavily on the simple conjunctions andandso.)
My, my, ah, I thought Buddha is only sitting and very quiet, so Sanma between and Tanaka Makiko sitting Buddha, and Soseki is novel writer so Koizumi prime minister is … very like reading books so next to the Soseki.
The one example of more complex language use occurs at the beginning when ‘I thought Buddha is only sitting…’ sets up the dominant clause, I thought, and the dependent clause, Buddha is only sitting… This in fact was the only example of complex clause relations in Hiroko’s entire task performance. Compare this with the opening lines of Hiroko’s report:
Our final arrangement, the party. I separate, I separated to every sec- ond seat a talkative person and a calm person sit. Koizumi Prime