Promoting young EFL learners’ oral production through storytelling: coursebook adaptation in the Vietnamese classroom

2.1 Benefits of stories and storytelling

Storytelling have been shown to contribute to the development of linguistic skills for young learners in an EFL environment, including vocabulary (Gomez 2010; Kirsch 2016; Soe 2016), grammar (Garcia 2017; Kalantari and Hashemian 2016) and pronunciation (Lucarevschi 2018). As an example, Kalantari and Hashemian (2016) performed an intervention study on the effectiveness of storytelling on Iranian young learners’ vocabulary and attitudes. They compared a regular class with lessons based on the Backpack textbook and an intervention class with storytelling lessons based on the textbook prepared by the teachers and the researcher. The teachers used storytelling with pre-, while- and post-activities. Before reading the stories, the participants received interesting and comprehensible input through the teacher’s talk, games, reading and listening activities which familiarized them with the new language and vocabulary. While reading the story, the teacher directed the participants’ attention to a PowerPoint presentation which included the visual representation of the story to facilitate comprehension. In the post-storytelling stage, the teacher played vocabulary games with the participants and asked them to role-play the story by memorizing the dialogue. To assess the students’ improvement, a vocabulary pre- and post-test was conducted. The results showed that the intervention group experienced a significant increase in their vocabulary knowledge, compared to the comparison group. In addition, they become more motivated in learning with storytelling. However, it is worth noting that the intervention group received more and richer input than the comparison class.

Other studies report that storytelling can develop specific sets of skills, such as reading comprehension, critical reading (Belet and Dala 2010), listening (Oduolowu and Oluwakemi 2014), speaking (Fikriah 2016) and writing (Alkaaf and Al-Bulushi 2017). Gupta (2009) reported that storytelling plays an important role in the development of language skills by promoting social interaction and mutual collaboration, as storytelling encourages learners to interact with the teachers and their classmates by listening and telling stories to each other, thus providing opportunities to receive support from their teachers and their classmates.

Kim (2013) used classroom observations and teacher interviews to compare Korean elementary students’ interactions in task-based lessons and storytelling-based lessons. The findings indicate that interaction patterns for task-based lessons and storytelling-based lessons were different. In task-based lessons, some pitfalls were found such as spending time explaining how to perform a task, mainly using low-level lessons, and frequent use of native language. The frequency of students’ interaction was significantly higher in the storytelling classrooms. The overall process of the storytelling lessons provided students with opportunities for dynamic interactions in the classroom, while in the task-based lessons, the teacher and students talked more in their native language. Kim (2013) also suggests that the storytelling approach was more efficient in developing vocabulary, background knowledge and skills due to the enjoyment created in story-based lessons.

In terms of oral production, the focus of this paper, Li and Seedhouse (2010) investigated the role of storytelling in the development of oral interaction in primary Taiwanese EFL classes. They compared learners’ interactions in standard classes with textbook support (including choral drills, task-based activities and playing games) and storytelling classes (including talking about characters, prediction, storytelling, story discussion and pupils retelling the story). The researchers found there was an increase in oral interactions, vocabulary and expressions of different language functions among students in storytelling lessons as the story-based approach created an entertaining environment, causing a higher level of motivation and engagement from students. Li and Seedhouse (2010) suggest storytelling as an effective tool for promoting social interactions and, through interaction, improving their language skills.

However, Chwo and Chen’s (2015) found that storytelling was not very effective. This study had a large population of 40 classes including 1,036 primary school learners, from 10 schools, lasting 34 weeks. In the intervention group, the students, in addition to their normal English classes, had a daily 10‒15-min storytelling session led by the general class teacher who played a CD of an English story being read aloud, with no attendant preparation or associated activities. Occasionally, foreign English teachers told a story in person in a 40-min session with pre-, while- and post- activities. Students also read and listened to stories from magazines. The comparison group had normal reading and listening input and they could borrow magazines to read and listen to outside the classroom. Results showed many favourable attitudes, together with considerable gains between pre- and post-tests of reading and listening, but interestingly no overall significant difference between the score improvements in the storytelling classes and those in the comparison classes. The unsuccessful storytelling may be due to two reasons. The first is that it was not clear whether the students listened to the stories during the daily 10–15 min storytelling sessions conducted by the general teachers. They did not have any reason for listening to the stories and it was not clear how the students interacted with the story. The second reason was that they did not have any interactions after listening to the story to use what they had learned. The children had exposure to many stories, but they did not interact with the story or with friends to construct meaning and to use the language. McGee and Schickedanz (2007) suggest that just exposing children to stories is not enough to develop language and cognitive skills through storytelling. Storytelling should include activities planned around specific learning outcomes and the story serves as the reference point for teaching cognitive and linguistic skills.

Although the studies explore the effects of storytelling in different areas, a common feature across their findings is that storytelling positively influences motivation, language knowledge and language skills for young learners. Most studies report the positive effects on learners’ motivation and interest in English classrooms. The use of stories encourages learners to actively participate in the learning process by not only listening to stories but also discussing them and telling their own stories. The students enjoy the meaningfulness of the story and its creative activities. A significant finding from the empirical studies is that interaction plays an important role in storytelling lessons for language learning development.

2.2 The Vietnamese context

Despite the theoretical and research support for the use of storytelling that emerges from the literature, its use in classrooms is far from universal. As noted earlier, Vietnam has ambitious goals for English language learning and teaching but faces challenges in achieving them (Le and Do 2012). The challenges vary depending on context. Le and Barnard (2009), for instance, investigated the introduction of communicative language teaching in a rural public secondary school and found that even when the teacher attempted a small innovation, such as group work, learners did not change their behaviour; they worked individually and silently. On the other hand, however, Nguyen et al. (2018) reported successful transformation of textbook activities applying task-based learning principles in an urban private secondary school. Bui and Newton (2021) also found the potential for TBLT in the adaptations of PPP among a select group of primary teachers, but as they themselves noted these teachers were not ‘typical’. Among the constraints innovation in English language education face are the dominance of the textbook, the demands of assessment, and teachers’ own proficiency. A further issue, again not restricted to Vietnam, is the history of top-down directives in education and the lack of involvement of teachers in formulating policy and guidelines (Enever 2020). In the context of this article, a background study (Le 2020) indicated that teachers had to use the textbook, and that although the textbooks contained stories the quality of the texts was insufficiently interesting; that the stories were not employed to provide effective learning opportunities for young learners; and that student engagement was low. The challenge, therefore, was how to find a way to work both within and against the existing constraints so that learners could benefit from storytelling.

2.3 Research question

Studies on storytelling are popularly conducted in countries where the curriculum allows teachers to choose their own materials and teaching methods or where their textbooks were designed with stories and interactive storytelling. They had freedom to choose storybooks and create stories that were suitable for the students’ interests, proficiency levels, and age or were supported with storybooks and by the authorities. In countries such as Vietnam, on the other hand, it can be a challenge to find this curriculum space or opportunities for the exercise of teacher agency. Thus, there is a need for more empirical research on effective innovation in resource poor EFL environments where teaching is strictly regulated, and textbooks are considered to be the curriculum. If an innovation, such as storytelling, can be demonstrated to be successful and feasible, it creates the opportunity for positive change. The question that guided this study, then, was:

What was the effect of the storytelling innovation on young EFL learners’ spoken language performance?

3.1 Context and participants

This study was conducted in an urban primary school in southern Vietnam. This school had six Grade 5 classes and two of these classes, each with 46 students, joined the study. English was taught for four periods a week using the Tieng Anh textbook. The coursebook had ten units and two review units, with each unit consisting of three lessons, and each lesson of six sections. One coursebook lesson was taught over two classroom lessons, with the storytelling innovation being based on the three sections of one lesson (Appendix A) and occupying one classroom lesson. Over the 15 weeks of the study, the intervention class had the storytelling innovation for 22 lessons, while the comparison class covered the same material following the textbook.

Both teachers had similar backgrounds. They attended the same three-year course at college. Since graduation, they had been teaching at primary schools and together in the same school for over four years. Both had achieved the B2 certificate of English proficiency and attended a 180-h workshop on TEYL and textbook training workshops. They were in the same professional learning group. Together with teaching normal classes, both had been working as teaching assistants for foreign teachers.

3.2 A storytelling innovation

Adaptation of existing materials was undertaken to create the storytelling innovation (Macalister 2016). The format and presentation of the textbook speaking lessons were redesigned into storytelling lessons to offer interactive opportunities for young learners to use language (see Appendix A for an example of a textbook story). A story was created by adding some short sentences to provide the setting and the characters. In the Appendix A example, for instance, in telling the story the teacher began by saying “Mai and Quan were at a gym”, thus establishing the setting. Instead of asking students to ‘Look, listen and repeat’ as instructed in the textbook, the teacher told the story, using the pictures without showing the words. Table 1 presents the structure of a storytelling innovation lesson.

Table 1:

Structure of a storytelling innovation lesson.

The lesson structure consists of three phases: pre-, during- and post activities. It is an integrated skills lesson with a focus on speaking and it has five main activities. In the first activity, students in groups or pairs join in a fun and engaging game. In the second activity, the teacher tells the story, pointing to the pictures (without showing the text), or using body language. Students are asked to work in pairs to tell what the story is about. In the third activity, the teacher retells the story (similar to Activity 1). Students retell the story in small groups. In the fourth activity, students retell the story in writing and compare their group’s story with the original. In the fifth activity, students are required to make the next events of the story (the ending of the story) and retell the story with their own endings.

3.3 Data generation

The storytelling innovation was used regularly in class for one term, while the comparison class followed the textbook as usual. Pre- and post-tests were conducted at the start and end of term. The tests involved learners retelling the story of a cartoon story they had just seen.

After piloting, two cartoon stories, Little Red Riding Hood (LRRH) and Tom and Jerry (T&J), were used in the story retelling. The students were randomly put in the subgroups. Thirty-one students in each class participated in the retellings and were divided into four subgroups depending on which stories they were exposed to in the pre- and post-tests (Table 2). However, six students in the comparison group and one student in the intervention group were removed from the analysis because they did not supply enough data. Two of the children in the comparison group watched the cartoon but refused to say anything in the pre-test and post-test; one of the comparison group children moved to another school towards the end of the second semester; and two of the students in the comparison group and one in the intervention group did not say anything after watching the cartoon in the pre-test but they did tell the story in the post-test. Therefore, the data of story retelling for analysis were from participants of 25 students in the comparison group and 30 in the intervention group.

3.4 Story retelling test procedure

Pre- and post-storytelling tests were conducted in the staff room and the corridor in front of the staff room where it was quiet. One of the second author’s colleagues managed the cartoon watching. Her assistance reduced their time out of the classroom. She explained what the students would do, did a short modelling demonstration using the cartoon Three Little Pigs and a series of pictures from it, and invited them to watch a cartoon. Two learners watched a cartoon together which helped increase their enjoyment. They had time to prepare their retold story individually. Each of them then took turns to retell their stories to the second author, who listened and audiorecorded them. The story retelling task was not timed, and the participant students were free to retell the story in as much detail as possible. They were also encouraged to use English as much as possible. No comments on the children’s story retellings were revealed to the children or their teachers and parents to minimize external threat.

3.5 Story retelling data analysis

Lexical richness and idea units were used to analyse the participants’ language. Lexical richness is about the quality of vocabulary in a language text, and it is popularly employed to measure the proficiency level of a student. Educators (Laufer and Nation 1995; Read 2000) have suggested different models to measure lexical richness. The oldest and the most frequently used measure of lexical richness is based on the type-token ratio, which is generally concerned with counting how many different tokens there are for each type in a text (Van Hout and Vermeer 2007). However, in the current study, the students’ English proficiency in their story retellings was at a low level. Most of the students were less proficient English speakers, so they spoke in English and Vietnamese in story retellings. Only a few proficient students spoke completely in English. In this study, tokens and types were counted but no ratio was taken.

For the analysis, the independent variables were the normal lessons for the comparison group and the storytelling lessons for the treatment group and dependent variables were the outcomes of the students’ pre-test and post-test story retellings. Tokens, English types and idea units in the students’ retold stories were analysed to examine their central tendencies and degrees of deviation to measure the language development within the participating groups and between them. Because the data were collected from mixed-ability classes, preliminary analyses of the data were performed to examine their normality, homogeneity of variance and outliers in order to choose precise statistical tests. The results of Shapiro-Wilks tests demonstrated violations of normal distribution and box-plots showed some outliers, while Levene tests showed the data had equal variances. Following advice from a statistical consultant, medians (not means) were determined to be better measures of the centres of the data and Wilcoxon signed rank and Mann-Whitney U tests were used to determine whether the differences of story retelling measurements within and between the comparison group and the intervention group were statistically significant. In addition, their size effects were also calculated on the formulas recommended by Clark Carter and Cohen (cited in Allen et al. 2014, pp. 257–259). All the tests were computed with SPSS 25, a statistical software programme.

In addition to counting tokens and types, idea units were also counted to measure the amount of ideas in a text. Idea units have been shown as a useful measure of oral speech, especially narratives (Brenes 2005; Gee 1991, 2018).

3.6 Data transcription

For each of the students’ story retellings, two versions were created. In the first version, everything was transcribed verbatim. The only changes made were when the students pronounced words with mistakes in final sounds. For example, if they pronounced “mouse” as [mau], it was transcribed as a full word [maus]. Grammatical mistakes were not corrected as in the example “the mother cook soup”. In the second version, all fillers like “umm” and “ủa” (oops), and unclear words were omitted, and some changes were made based on notes in counting tokens (see Appendix B). The two versions were saved as.txt documents, and the latter version was used for data analysis. The story retellings were analysed in relation to the number of tokens and English tokens, English types and idea units.

3.7 Tokens in students’ stories

A token is an individual occurrence of a linguistic unit in speech or in writing. In other words, tokens are the total number of words in a text, regardless of how often they are repeated. Tokens reflect the speakers’ language use. In this study, there were two types of tokens in the students’ stories: tokens in English and tokens in Vietnamese. For example, in “On the street cô bé quàng khăn đỏ gặp one wolf and then cô bé nói là đang đi đến grandmother home”, there are 22 tokens, consisting of nine English tokens and 13 Vietnamese tokens.

3.8 Counting tokens and types

Here is an example of a student’s retold story.

The mouse is go the mouse get up at night, and he open the door, and go outside, he see a bag in the river, he open the bag, and he see many dogs in the bag, one dog is lick the mouse, the dogs and a mouse go, the dogs and a mouse go to the house, and the cat see the dog, and he is very angry, the dogs and the cat was in the bed, the cat the mouse cooking soup, and he đổ soup vào miệng con mèo, and the dogs is lick the cat, there are many dogs go to the house, the dogs is lick the bowl of milk.

The text was submitted to Lextutor https://lextutor.ca/ for analysis. The analysis was for English tokens and types, with Vietnamese words appearing as off-list. However, a careful manual check was necessary as some Vietnamese words are recognized as English words. In this example, con was misrecognized as an English token and the figures needed to be adjusted, leaving in this example 110 English tokens and 37 English types.

3.9 Idea units in students’ stories

Idea units are the basic elements that are used by speakers to successfully transmit a message. In speaking, they can be identified as spurts of language that are typically separated by a long pause; therefore, pausing is an important factor that serves to identify idea units when transcribing an oral text (Chafe 1985; Gee 2018). An idea unit is associated with a clause and “is spoken with a single coherent intonation contour, ending in what is perceived as a clause-final intonation” (Chafe 1985, p. 106). However, in story retellings, the participating students rarely had clear rising and falling intonation in their English speech as they were affected by the way they spoke their mother tongue and were just English language beginners. Therefore, the keys to identify idea units were brief pauses, and commas were used to show their pauses in the transcriptions; after the retellings were transcribed, clauses were the main tool for counting idea units.

Grammatical mistakes were ignored when counting idea units. For example, a student said: “The dog is lick Tom” instead of saying: “The dog licked Tom”. “The dog is lick Tom” was counted as one idea unit.

There were four types of idea units in the students’ stories.

1. Idea units in Vietnamese such as “Ngày xửa ngày xưa ở làng kia có một cô bé quàng khăn đỏ” (Once upon a time, in a village lived Little Red Riding Hood).

2. Idea units in English such as “The wolf walk slowly and slowly”.

3. Idea units in Vietnamese and English such as “He see one cái bọc is trôi trên water” (He saw a bag floating on the water).

4. Idea units in Vietnamese with English characters’ names such as “Jerry nấu cháo cho Tom ăn” (Jerry cooked soup for Tom).

Further information about identifying and counting idea units is provided in Appendix C.

4.1 Tokens

Tokens were examined to determine whether the choice of texts affected the students’ language production and whether or not the length of their stories improved.

As seen from Figure 1, the median of total tokens in LRRH retellings was higher than T&J retellings. The longer character names of LRRH than in T&J possibly made LRRH longer; however, when character names were removed, the mean of LRRH is still higher than T&J. The median difference between T&J and LRRH with names is about 30 and without names is about 27.

Figure 1:

Medians of total tokens in each story.

Figure 2 shows that both groups produced longer stories in the post-test than they did in the pre-test in both LRRH and T&J. The comparison group produced longer stories than the intervention group in both pre-and post-test. Both groups produced longer LRRH than T&J stories in both pre- and post-test. When the characters’ names are removed, the medians of stories were a little lower than the medians with names, but the tendencies of language production did not appear to be changed. In sum, the finding from the examination of the central tendencies of total tokens of two stories and of tokens in each story in the pre-test and in the post-test lead to the conclusion that the choice of texts (two different cartoons, one with and one without speaking characters) had little effect on the students’ language production.

Figure 2:

Medians of total tokens x story x group.

Figure 3 consolidates the findings and reveals that both groups produced more tokens in the post-test than in the pre-test. From the pre-test to post-test, the median scores increased relatively modestly in the comparison group from 293 to 316 but more dramatically, from 187.5 to 287.5, in the intervention group.

Figure 3:

Medians of tokens in each story in pre-test and post-test.

The comparison group produced longer stories on both occasions but it is important to recall that up to this point the discussion has included both Vietnamese and English tokens. The following sections will present the analysis of English tokens, types and idea units to examine whether the students improved their oral English production.

4.1.1 English tokens

As can be seen in Figure 4, the intervention group produced a few more English tokens than the comparison group in the pre-test. The results of the Mann-Whitney U test showed that the intervention group’s production of English tokens was not statistically significantly higher than the comparison group (Mean Rank = 24.64 for the comparison group and Mean Rank = 30.80 for the intervention group, U = 291, z = 1.420, p = 0.155).

Figure 4:

Medians of English tokens.

In Figure 4, we can see that both groups produced more English tokens in the post-test than in the pre-test. From the pre-test to post-test the median scores increased from 44 to 53 in the comparison group and from 51 to 154 in the intervention group.

Unlike the results of token analysis, the Wilcoxon signed rank test for the comparison group indicated that improvement of English tokens was statistically significant in the post-test, T = 229.50, z = 2.272, p = 0.023, with 18 participants ranked as more improved, six participants ranked as negatively improved and one participant ranked as unchanged; this effect can be considered rather ‘large’, r = 0.45. Similarly, the same test for the intervention group indicated that the growth of their English tokens was statistically significant, T = 445.00, z = 4.371, p < 0.000, and the size effect can be considered ‘large’, r = 0.8, 28 participants ranked as more improved and two participants produced less in the post-test. The improvement of the treatment group was much higher than the comparison group as the Mann-Whitney U test indicated that the difference of English tokens between them (Mean Rank = 20.56 for the comparison group and Mean Rank = 34.20 for the intervention group) was statistically significant, U = 189, z = 3.144, p = 0.002, which can be described as a ‘medium’ size effect, r = 0.42.

4.1.2 English types

The intervention group produced more English types than the comparison group in the pre-test. The results of the Mann-Whitney U test showed that the intervention group’s production of English tokens was statistically significantly higher than the comparison group (Mean Rank = 23.22 for the comparison group and Mean Rank = 31.98 for the intervention group, U = 255.5, z = 2.021, p = 0.043).

As can be seen from the data in Figure 5, both groups produced more English types in the post-test than in the pre-test. From the pre-test to post-test the median scores increased from 11 to 21 in the comparison group and from 20 to 43 in the intervention group.

Figure 5:

Medians of English types.

Similar to the findings of the tests for English tokens, the Wilcoxon signed rank test for the comparison group and the intervention group’s English types indicated that their improvements were both statistically significant. For the comparison group, the results were T = 254, z = 2.977, p = 0.003, with 18 participants again ranked as more improved, six as negatively improved and one as unchanged; this effect can be considered rather ‘large’, r = 0.45. For the intervention group, the results were T = 381, z = 4.054, p < 0.000, with 28 participants again ranked as more improved and two as not; it is considered a ‘large’ effect, r = 0.8. The difference of English types between the comparison group (Mean Rank = 23.18) and the intervention group (Mean Rank = 32.02) was statistically significant as the Mann-Whitney U test computed, U = 254.50, z = 32.039, p = 0.041. This effect can be described as ‘medium’, r = 0.42.

Like the result of the analysis of English tokens, the English types in the intervention group’s retellings were much more than in the comparison group’s as the deviation was significantly different in the post-test. A proper name, no matter how many times it is used, is counted as one type. Therefore, the students’ growth in language use was not strongly affected by naming the characters. Obviously, the number of tokens and types in the students’ stories improved, but how did they use them to create meaningful units in their stories?

4.2 Idea units

It is apparent in Figure 6 that the comparison group did not increase their idea units in the post-test compared to the pre-test while the intervention group did; specifically, idea units in English increased from 3 to 10.5 and idea units in English and Vietnamese from 3.5 to 4 while idea units in only Vietnamese decreased from 4.5 to 3. Based on these changes, nonparametric tests were run for idea units in English and idea units in English and Vietnamese only.

Figure 6:

Medians of idea units.

In the pre-test, the intervention group produced more idea units in English and idea units in English and Vietnamese than the comparison group. The results of the Mann-Whitney U test showed that the differences were statistically significant (for idea units in English: Mean Rank = 24.28 for the comparison group and Mean Rank = 30.10 for the intervention group, U = 2.562, z = 2.021, p = 0.043; for idea units in English and Vietnamese: Mean Rank = 27.64 for the comparison group and Mean Rank = 28.30 for the intervention group, U = 366, z = 0.154, p = 0.877).

In the post-test, the results of Wilcoxon signed-rank tests for idea units in English and idea units in English and Vietnamese for both groups were similar. For the comparison group, the improvements of these two kinds of idea units were not statistically significant. The Wilcoxon test results for idea units in English were T = 97, z = 0.974, p = 0.33, with 10 participants ranked as more improved, seven participants ranked as negatively improved and eight participants ranked as unchanged. Those for idea units in English and Vietnamese were T = 116, z = 1.333, p = 0.183, 10 participants ranked as more improved, eight participants ranked as negatively improved and seven participants ranked as unchanged.

Unlike results of the non-parametric t-test for the comparison group, the improvement of the intervention group’s idea units in English and idea units in English and Vietnamese were statistically significant. For idea units-in-English, the Wilcoxon test results were T = 323, z = 3.749, p = 0.000, with 22 participants ranked as more improved, four participants who produced less and four ranked as unchanged in the post-test; the effect can be considered ‘large’, r = 0.68. For idea units in English and Vietnamese, the results were T = 189.50, z = 2.047, p = 0.41, with 16 participants ranked as more improved, six participants who produced less and eight ranked as unchanged in the post-test. The effect can be considered ‘large’, r = 0.68.

Again, Mann-Whitney U tests indicated that the differences of idea units in English and idea units in English and Vietnamese between the comparison group (Mean Rank = 21.30 and 25.98 respectively) and the intervention group (Mean Rank = 33.58 and 29.68 respectively) were statistically significant; (for idea units in English U = 207, z = 2.848, p = 0.004, r = 0.38); for idea units in English and Vietnamese U = 189, z = 3.144, p = 0.002, r = 0.42. The size effects were ‘medium’.

In sum, the major findings show that at the beginning of the study, there was no significant difference in the English-speaking competence in terms of tokens, idea units in English and idea units in English and Vietnamese between the comparison and the intervention groups. At the end of the study, both groups improved their speaking competence, but the changes were quite different. The comparison group improved their oral English production but not as much as the treatment group. The students in the storytelling innovation class improved their oral English communication in dependent variables of English tokens, English types, idea units in English and idea units in English and Vietnamese; and their improvements were statistically significant not only inside each group but also compared to the comparison group (all p < 0.05 in Wilcoxon signed-rank tests and Mann-Whitney U tests).

5.1 Pedagogical considerations

A concern raised by the intervention group teacher, however, was that her students might suffer as a result of the intervention as the storytelling innovation was mainly to improve the students’ speaking while the normal tests focused on the four skills. This is a legitimate concern. Grade 5 students in Vietnam have two end-of-term tests per year. The test scores play a part in determining the students’ primary schooling achievement. One was done at the end of the first term, one week before they started the second term and one was done at the end of the second term. The storytelling intervention was conducted in term two; therefore, the end-of-term-one test was used as a form of pre-test and the end-of-term-two test was used as a form of post-test in this study.

The structure and the content of the tests were designed by the participating schools’ teachers of English on the basis of DOET’s test guides and test matrix. The tests appeared child-friendly with lovely pictures and clear instructions with examples. They were aimed to measure the four language skills with each macro-skill having just 10 questions. In speaking, for example, each student answered 10 questions that were previously learned in the textbook; the focus was on vocabulary and the structures learned in each lesson in the textbook. In the interviews with the participant teachers, they reported that the students wrote the answers to the questions and learned them by heart. This is not the place to raise questions of validity and reliability about the tests. Rather, the nature of the tests reinforces the idea that the comparison group students might be more advantaged because the form of the test matched with the normal learning that focused on language features, whereas the intervention group students might be disadvantaged because the process of their learning, which involved meaning-focused activities like listening to a story, constructing a story and creating a story, were not reflected in normal tests. However, when test scores were analysed, the results showed both classes got higher scores in the post-test than they did in the pre-test. In terms of means scores,^[1] the comparison group increased slightly about 0.2 from 7.12 in the pre-test to 7.30 in the post-test. The students in the intervention group improved their test scores a little more than the comparison group, over 0.3 from 7.98 in the pre-test to 8.32 in the post-test. In terms of median, the test scores increased 0.25 for both of the groups. Basically, the intervention had no discernible negative effect, and so allayed teacher concerns (Figure 7).

Figure 7:

Means and medians of test scores.