Promoting interaction and lowering speaking anxiety in an EMI course: a group dynamic assessment approach

2.1 Group dynamic assessment and mediation

Dynamic Assessment (DA) is a classroom-based assessment tool that has recently been discussed and applied in the field of language education (see Poehner and Wang 2021 for details). In particular, there is a growth in DA research related to L2 speaking performance (Antón 2009; Hill and Sabet 2009; Kao 2020a, 2020b; Oskoz 2013). The theoretical roots of DA can be traced to Vygotsky’s theory of cognitive development and his notion of the zone of proximal development (ZPD) (Minick 1987; Vygotsky 1978). The idea of ZPD describes the gap between what a learner can do independently and what the learner can achieve with the guidance and support of a more knowledgeable person, often an instructor or a peer (Vygotsky 1978: 86). In this sense, understanding an individual student’s ZPD can help teachers know their underlying abilities that have not yet fully developed and are still in the process of being internalized. Figure 1 helps visualize the concept of ZPD – learners expand their current ZPD when being supported by a teacher or more capable peers. With proper instruction or collaboration, learners gain opportunities to achieve future potentials.

Figure 1:

A pictorial representation of the zone of proximal development.

The essence of Vygotsky’s theory of ZPD is mediation. Language, in particular, when it is provided and aligned with the interlocutors’ level of ZPD; it becomes the mediation that could guide cognitive development. Yet, not all the interactions could be considered as mediation; only the ones that maximize the impact on student’s emerging abilities are counted. According to how the mediation is provided during the assessment procedures, two approaches of DA are discussed: interventionist and interactionist. The interventionist approach of DA follows rigid assessment procedures and implements a pre-determined list of hints for generating a weighted score. The mediation is usually tightly scripted, prepared in advance, and offered through a series of hints or prompts that go from the most explicit to the most implicit (Antón 2009; Kozulin and Garb 2002; Lantolf and Poehner 2011). The interactionist approach of DA, instead, is usually administrated in the interactive format that examiner provides the mediation and the learner response to it accordingly. The mediation is thus adjustable based on learner’s need and does not limit to the types of assistance available to help learner’s development, such as oral, written responses or nonverbal behaviors.

Group-based and one-on-one DA procedures follow the same principle of providing learners mediation, but they differ in that Group Dynamic Assessment (G-DA) evaluates multiple individuals who may have similar but not identical needs and capabilities. Two frameworks of G-DA are discussed (Kao 2020b; Poehner 2009). They are concurrent and cumulative G-DA. The concurrent G-DA means that the teacher provides mediation to a particular learner within a group; the procedures of it function similar to the teacher-student questioning. The teacher’s primary focus is on diagnosing an individual’s ZPD and then promoting the whole group’s mutual understanding. The cumulative G-DA, on the other hand, is that the teacher conducts a series of one-on-one DA mediations with each group member until the group has mastered the target learning goal. All the group members are welcome to respond to the mediation. Corresponding to Petrovsky’s (1985) principles of group activities, especially the ‘group-as-cooperation’ and ‘group-as-collective’ model (see Poehner 2009 for details), both frameworks of G-DA offers opportunities to examine various types of interaction, specifically, mediation, in a regular language classroom. However, very few empirical research has investigated the potential of applying G-DA in a language classroom (Davin and Donato 2013; Poehner 2009; Poehner et al. 2018). Kao (2020b) suggested that both concurrent and cumulative G-DA approaches could promote learners’ understanding of Chinese rhetorical structures. Through receiving tailored mediation, learners could co-construct awareness and knowledge within groups, internalize these concepts at an intra-personal level, and ultimately apply them in authentic learning contexts. Consequently, this study addresses this research gap by applying the interactionist approach of DA to (1) examine the effectiveness of teacher’s mediation in group settings to assist university students’ English-speaking performance and (2) understand how the mediation provided through either concurrent or cumulative approach, functioning as scaffolding, could lower university students’ speaking anxiety in the EMI learning context.

2.2 Speaking anxiety

DA is a kind of assessment and instruction, and any process of assessment and learning involves some anxiety. Anxiety in speaking then can be interpreted as a fear of expressing oneself orally (Hanifa 2018; Mak 2011). In a language learning context, anxiety includes “self-perceptions, beliefs, feelings, and behaviors related to classroom language learning arising from the processof language learning” (Horwitz et al. 1986: 28). Second or foreign language learning process is a unique process because learners are required to communicate using a language that they have not yet mastered. Previous research indicated that second language anxiety has a debilitating effect on the oral performance of EFL learners (Aida 1994; Woodrow 2006).

Horwitz et al. (1986) considered anxiety comprises three components: (1) communication apprehension; (2) test anxiety; and (3) fear of negative evaluation. The three anxiety components are highly related to the procedures of the G-DA intervention implemented in this study. First, communication apprehension refers to difficulties in oral communication with a partner or in groups. In the G-DA intervention, the teacher will provide one-on-one mediation in a group context or request the individual participant to summarize the information heard or read from the task; the participant might experience communication apprehension. Second, test anxiety is “the harsh emotions that have physiological and behavioral concomitants one experiences in evaluative situations” (Sapp 1999: 272). Such debilitative feelings may be doubled in oral production testing situations wherein the anxiety produced within the learners interferes with their ability to perform to their full potential during a test or evaluation situation. For instance, at the end of each G-DA session, participants are required to record their oral responses independently – a setting similar to a test situation – they might feel anxious and fail to answer well in the test. Third, fear of negative evaluation is triggered by the teacher or classmates as fluent speakers. Young (1991) argued that the fear of committing a verbal error is the main reason learners do not participate in classroom activities. Koch and Terrell (1991) also indicated that speaking in front of their peers is a source of anxiety in learning a foreign language. For example, participants are required to speak in front of the small group members and share the information they read or hear from the test materials – these actions might trigger participants’ anxiety, such as the fear of negative evaluation.

Since the G-DA intervention considers a different approach to assessing and instructing foreign language learners’ speaking performance, the extent of language learners’ emotional reactions, such as anxiety, becomes an important issue to explore.

3.1 Research question

The research goal was to investigate the effectiveness of the G-DA framework in supporting university students’ English-speaking performance regarding their awareness of language functions, organizations, and presentation techniques emphasized in the integrated speaking tasks. Figure 2 represents a comprehensive view of the methodological outline of this study. Through comparing with a non-DA (NDA) group, the results showed the effectiveness of the teacher-led G-DA approach in promoting university students’ academic English-speaking performance in an EMI course. Two research questions are listed in the following:

1. Does the application of the G-DA intervention have an impact on university students’ speaking performance?

2. Is there any statistically significant difference between teacher-led G-DA and NDA concerning their effects on university students’ speaking anxiety?

Figure 2:

Research outline.

3.2 Participants

Fifty first-year students at a university in southern Taiwan were selected from an English-Medium Instruction (EMI) course focused on Academic English. Within this academic institution, approximately nineteen percent of the courses employed EMI, encompassing a range of subjects, including Economics and Instructional Design. The fundamental characteristic of an EMI course at this university involves delivering content knowledge in English, with the potential incorporation of translanguaging pedagogy (Cenoz and Gorter 2021). Translanguaging strategies, such as utilizing multimodalities or students’ native language (L1), may be employed to enhance content comprehension and in-class discussion. Notably, there are no specified language proficiency requirements for enrollment in the EMI program at this university.

Aged between eighteen and twenty, 36 female and 14 male students participated. Their English language background is the B2 level of the Common European Framework Reference (CEFR) standard or the intermediate level of the General English Proficiency Test (GEPT) certificate. The reason for this proficiency criterion is that participants can process better English reading and listening abilities before responding in English. They could also better understand the teacher’s interactive support in English. The recruited participants were then randomly divided into two intact groups: (1) the G-DA experimental group (i.e., 28 participants): that receives the teacher’s mediating support in small group contexts, and (2) the non-dynamic assessment control group (i.e., 22 participants) that receives the teacher’s instruction aiming at practicing their English-speaking skills (i.e., without the teacher’s interactive support).

Informed consent was obtained from all participants before their involvement in the study. Participants were provided with a detailed information sheet outlining the purpose, procedures, and potential risks of the research. They were assured of the voluntary nature of their participation and were informed that they could withdraw from the study at any point without consequence.

3.3 Study design

Different treatments were applied in the G-DA experimental group and non-dynamic assessment (NDA) control group. Although the two groups differed in their treatments, the intervention time and the materials used in both groups remained the same. For instance, both groups received ten sessions of intervention (a total of 20 h) that focused on enriching participants’ academic English-speaking performance. The teaching materials were the speaking tasks selected from the TOEFL iBT Official guide (2020, 2021) and other related resource books, such as the Keynote TED talk (Stephenson et al. 2017). A total of four integrated speaking tasks were implemented in each group. Table 1 shows the content and topics for speaking tasks.

The instructor overseeing both participant groups possesses an English language teaching certificate and has undergone formal training in DA. With a teaching tenure exceeding ten years, she has dedicated the past four years specifically to delivering the EMI course.

3.4 Pretest, posttest and delayed posttest

The pretest, posttest, and delayed posttest method was employed to understand the effect of G-DA in promoting university students’ speaking performance in an EMI course. Three TOEFL iBT mock tests were chosen due to their alignment with the study’s goal of assessing the test taker’s ability to use and comprehend English in an academic context. Specifically, emphasis was placed on two question types, campus announcement, and academic lecture, as they evaluate students’ integrated language abilities. The topic of the pretest discusses a campus, daily-life situation, and the posttest focuses on an academic subject area. At last, the delayed posttest focuses on current news or issue.

In these three tests, participants needed to produce oral responses by combining provided textual and aural information and to apply their language skills to real-life situations. These integrated speaking tests aim to simulate real-life academic situations where participants need to both comprehend and communicate complex information. The advantages of integrated speaking tasks are their high level of authenticity (Yu 2013) and the possibility of generating positive washback (Huang and Hung 2010). Participants in both groups were required first to read a reading passage (75–100 words) that presents recent news or an issue and then listen to an excerpt (60–80 s; 150–180 words) that comments on the case mentioned in the reading passage. After receiving reading and listening prompts, the participants were required to summarize the speaker’s opinion based on the provided information. Following the time allowance on the official TOEFL iBT, participants have 45 s for reading, roughly 90 s for listening, 30 s for preparation, and 60 s for oral responses.

Participants’ oral performance in these tests were video-recorded and scored using integrated speaking rubrics developed by the Educational Testing Service (ETS 2008). Four criteria are indicated in the scoring rubrics: (1) General description; (2) Delivery; (3) Language use; and (4) Topic development. Twenty-five points for each criterion, and the maximum grade is 100 points. Detailed descriptors for each category of the scoring criteria are presented in Appendix A. For the purpose of ensuring the reliability of the scoring, two raters were involved – one is the researcher, and the other is a university instructor specializing in language assessment with seven years of teaching experience. A Pearson product-moment correlation coefficient has been calculated to assess the evaluation scale of the two raters. A positive correlation (e.g., above 0.89) has been attained to ensure the reliability of the test result.

As for the timeline, the pretest and posttest were conducted at the second and thirteen weeks, respectively. Poehner stated (2007) that learners’ potential development in DA can be defined as individuals’ abilities to re-contextualize their learning and to apply it to new, more demanding problems. The posttest, considered as the ‘near transfer’ task, was conducted right after the DA intervention. The delayed posttest, considered the ‘far transfer’ task, was conducted one month after the posttest to examine the extent to which participants applied their learning (i.e., the 17th week).

3.5 Speaking anxiety questionnaire

The Foreign Language Classroom Anxiety Scale (FLCAS) designed by Horwitz et al. (1986) was adapted to understand the extent of participants’ speaking anxiety before and after joining the G-DA experimental or the control group. This scale is selected because of its popularity in the field of SLA and its high test-retest reliability (r = 0.83) over eight weeks (Horwitz et al. 1986). It has also been applied and validated in the Taiwanese context (Huang 2013), and acceptable validity and reliability indexes have been reported for the scale (Ho 2010; Sung 2014).

A total of 22 items in a five-point Likert scale of (1) strongly disagree, (2) disagree, (3) undecided, (4) agree, and (5) strongly agree were applied (cf. Table 7). A few original FLCAS items were changed to fit with the current study. For instance, the original item #14,‘I would not be nervous speaking the foreign language with native speakers’, and item #32, ‘I would probably feel comfortable around native speakers of the foreign language’ had been deleted because, in this study, there were no English native speakers involved. These two items were replaced by the following two descriptive statements related to the experimental procedures: ‘I feel anxious when listening to a passage in this course’ and ‘I feel anxious when reading a passage with a limited time in this course.’ The adapted questionnaire items were reviewed by a university professor who has seven years of experience in teaching English speaking courses.

Participants took this speaking anxiety questionnaire before and after the intervention, at weeks 1 and 18, respectively.

4.1 English speaking performance

To answer the first research question and determine the extent of the G-DA intervention supporting university students’ English-speaking performance, participants’ pretest scores were first analyzed through the independent sample test. The result indicated that the pretest mean scores of the two groups had no significant difference (p = 0.470 > 0.05). Second, the between-subject effect showed a significant difference (p = 0.033 < 0.05) was found in different treatment groups (Table 3). It indicated that the G-DA intervention had an impact on university students’ English-speaking performance.

Table 3:

Between-subjects’ effects.

In order to understand if participants in the G-DA group made progress throughout the time, a repeated measures ANOVA was also conducted to compare participants’ speaking performance after joining the intervention. In addition to descriptive statistics (e.g., means, modes, medians, bell curves, etc.), the effect size was reported to identify any practical significance.

Although the within-subject effects indicated there were significant differences found throughout the pretest, posttest, and delayed posttest (Table 4), the results of the pairwise comparisons (Table 5) indicated that the statistically significant difference was only found in the G-DA group’s performance between the pretest and posttest (p = 0.047 < 0.05). This result was also revealed through descriptive statistics. As shown in Table 6, both groups started with similar pretest grades (Cohen’s d = 0.100), and no significant differences were found in the pretest scores. It indicated that participants were generally unfamiliar with academic English speaking, which might result from their lack of experience integrating information from the listening clips and reading texts. They thus received ten weeks of intervention focusing on integrated oral practice – the G-DA group was involved in teacher-led discussion while the NDA group completed it individually. By the end of the intervention, the posttest score of the G-DA group (Mpost = 80.5; SDpost = 10.16) performed slightly better than the NDA group (Mpost = 79; SDpost = 8.2). Besides, comparing the posttest’s mean score between the two groups also showed that though the G-DA group’s mean score only outperformed the NDA group by 1.5, the deviation in the G-DA group was more significant than the NDA group. It might indicate that the teacher’s mediation triggered individual differences because some participants might benefit more from the teacher’s prompts while others might not.

Table 4:

Within-subjects effects.

Table 5:

Pairwise comparisons.

This study also conducted a delayed posttest (Cohen’s d = 0.174) to investigate participants’ transfer of learning after one month of the treatments. Although no significant difference was found in the delayed posttest of the two groups (Table 5, p = 1.0 > 0.05), the G-DA group was able to maintain the performance (Mdel. = 80.7; SDdel = 13.58) while the score of the NDA group decreased (Mdel. = 78.68; SDdel = 8.96). It indicated that the effect of the teacher’s interactive mediation was able to support students in an EMI class to complete a more challenging speaking task after a more extended period without practice. The same phenomenon of sizable standard deviation was also observed from the delayed posttest – some participants in the G-DA group grew significantly from the teacher’s mediation.

In summary, the posttest scores revealed a significant difference between the G-DA and NDA groups, signifying superior English-speaking performance in the G-DA group. While no significant difference emerged in the delayed posttest, the mean scores of the G-DA group underscored the lasting benefits of teacher’s mediation. Participants in the G-DA group not only maintained their performance after one month but also successfully completed a more challenging oral task.

4.2 English speaking anxiety

The Cronbach coefficient alpha (reliability) for the adapted five-scale FLCAS was computed to address the second research question. The resulting value is 0.80, indicates satisfactorily reliable, particularly given the limited number of participants included in the current study. Table 7 presents the ranking of the mean of each survey item. The situations described in items with a mean above 2.5 points are considered as comparatively more anxiety-provoking than those below 2.5.

Five themes were identified to elucidate the concurrent impacts on participants’ speaking anxiety following their participation in the experiments. These themes include (1) speech anxiety and fear of negative evaluation, (2) negative self-evaluation, (3) negative attitudes toward the English class, (4) anxiety regarding class requirements, and (5) fear of failing the class or facing personal failure consequences. These names were assigned to the respective factors by the researcher.

First, speech anxiety and fear of negative evaluation, contains six items with examples such as “It frightens me when I do not understand what the teacher is saying in English. (item 4)” and “I start to panic when I have to speak without preparation in this course” (item 10). The analysis first shows that all six items from either the G-DA or NDA group were ranked above the mean score of 2.5 in pre- and post-session responses. It indicates that university students felt anxious or feared negative evaluation when communicating in English during their EMI course. Although this is the case, participants in the G-DA group showed less anxiety than the ones in the NDA group after receiving the experiment (the mean dropped from 0.23 to 0.64 points in the post-session response). In items 1(I never feel quite sure of myself when I am speaking English in this course) and 21 (I can feel my heart pounding when I am going to be called on in this course), the NDA group even showed that they were more anxious than before the experiment took place.

The second theme related to participants’ negative self-evaluation, such as “I keep thinking that the other students are better at English (i.e., listening and reading) than I am.” G-DA group participants showed slightly higher scores (Mpre 3.65; Mpost 3.69) than the NDA group (Mpre 3.07; Mpost 2.57). The reason could be that the intervention in the G-DA group required participants to work collaboratively with other group members while the NDA group worked individually; the G-DA group participants might compare their English performance with others.

Regarding participants’ attitude toward the experiment/course they enrolled in, the third theme encompasses six items such as “I get upset when I don’t understand what the teacher is correcting (item 16)” and “I am afraid that my teacher is ready to correct every mistake I make (item 20)”. Analysis indicated that participants in the G-DA group exhibited less apprehension about engaging in the intervention. Particularly noteworthy are the responses to the aforementioned two items regarding teacher correction (or mediation in this study), where participants’ mean scores decreased. This indicated that participants did not feel uncomfortable about receiving teacher-provided hints and prompts aimed at enhancing the quality of their oral responses.

The fourth theme anxiety toward the requirements in this class, relates to the reading and listening tasks that participants in both groups needed to complete before their oral responses. The comparison between the pre-and post-session surveys indicated that both groups were less anxious when listening to a passage, yet, they showed more anxiety when asked to read a passage in a limited time. For instance, the mean score of the G-DA group dropped toward the more anxious end (Mpre 3.0; Mpost 2.62). This result indicates that university students might experience difficulties when asked to summarize the key points from the reading text and then integrate the ideas into speaking.

The fifth theme, fear of failing the class or consequences of personal failure, contains three items: “I worry about the consequences of failing this course/tutoring session (item 11)”. The G-DA and NDA groups generally showed less fear of personal failure after participating in the experiments. However, only on item 11, the participants in the G-DA group worried more about failing this English course than the NDA group did. Possible explanations for this phenomenon could include the perception among participants in the G-DA group that they received a higher frequency of hints and prompts from the teacher. However, they considered this scaffolding as “correction” that would harm their grades rather than viewing the support as “mediation” that would support them to reach the next level of development.

To summarize, from comparing the English-speaking anxiety questionnaires, results show that although the university students generally felt anxious or feared negative evaluation when they were required to communicate in their EMI course, the G-DA intervention helped participants feel less anxiety when prompted to respond orally. However, two procedures in the G-DA intervention might result in university students’ anxious behavior. First, since the G-DA intervention required participants to collaborate with other group members, participants were easier to compare their own English performance with others and thus might evaluate themselves more negatively. Second, the G-DA intervention provided participants ample opportunities to receive the teacher’s oral prompts and feedback. Nevertheless, without knowing these prompts were aimed to support them in learning, participants tended to see them as correction and thus became anxious about failing the course.