The effects of collaborative models in second life on French learning

Abstract

French is the ninth most widely used language globally, but French-learning environments in Taiwan have been insufficient. Language acquisition is easier in a natural setting, and so such a setting should be available to language learners wherever possible. This study aimed to (1) create an authentic environment for learning French in Second Life (SL), and then (2) determine the effects of different collaborative models on learners’ French performance and their perceptions about learning French. Twenty-three college students participated in the study. They were asked to collaboratively execute tasks involving the creation of French-language movies in SL. Both the movies produced by the participants and the collaborative processes were collected and analyzed, with the results indicating that the students adopted different collaborative models that led to different learning outcomes. Three types of collaborative models were identified: context-inclusive collaboration, context-exclusive collaboration, and context-exclusive centralization. The results also indicated that the context-inclusive collaboration model had positive impacts on students’ speaking capability and other aspects of their learning performance, and at the same time facilitated the effects of collaboration on how the students performed.

Introduction

French is considered to be one of the most beautiful languages in the world, and UNESCO statistics indicate that the French-speaking population constitutes the ninth largest among all languages. The number of French speakers exceeds 0.2 billion (statistics from http://www.quid.fr/francophonie) worldwide, representing more than 2.5 % of the total population and with these speakers residing across five continents. In Taiwan more than 15,000 students learn French in high schools, universities, and language centers, with French being the second most popular foreign language chosen as an elective course by high-school students.

Even though French is a world language, Taiwan’s foreign-language education has been mainly focused on the teaching of English, and the environment provided for learning French is insufficient domestically (Kuo 2010). According to the theory of second-language acquisition, Krashen (1981) stated that language learning occurs best in a natural setting. Hence, it is important that an appropriate setting is provided for language learning, such as can be found in IVWs (Immersive Virtual Worlds) that offer virtual simulation environments in which learners can become immersed and left with an impression of actively experiencing the setting; this enhances learning performance, efficacy, and satisfaction (Bulu 2012; Witmer and Singer 1998).

Lave and Wenger (1991) used the term “situated learning” to describe how knowledge learning is constructed by the learning process interacting with the environment in which it occurs. This model is beneficial for enhancing the learning process and experience (Efe et al. 2011), while a realistic 3D setting is beneficial for situated learning, since learners can truly feel the sense of place, which enhances the learning efficacy (Bellotti et al. 2010; Slater et al. 2009). Many researches have focused on situated English learning whereas few have focused on situated French learning, which makes it worthwhile to investigate if situated French learning has the same effect as situated English learning. Second Life (SL) is a multiuser virtual environment that is the most widely used by educational institutions and educators (Warburton 2009) due to its high flexibility and expandability features, which allow educators to design situated learning environments that provide learners with realistic learning settings (Chow et al. 2012).

Reading French is easier to learn in the classroom than are listening comprehension and speaking skills (https://www.lttc.ntu.edu.tw/SFLPT/report/%E5%9C%8B%E5%85%A7%E6%B3%95%E8%AA%9E%E5%88%9D%E5%AD%B8%E9%9A%8E%E6%AE%B5%E6%95%99%E5%AD%B8%E7%8F%BE%E6%B3%81%E8%AA%BF%E6%9F%A5%E8%88%87%E5%AD%B8%E7%BF%92%E6%88%90%E6%9E%9C%E5%88%86%E6%9E%90.pdf). Foreign languages are conventionally taught in Taiwanese classrooms with a focus on listening, reading, and writing, with there being insufficient opportunities for foreign-language learners to enhance their speaking capabilities, or to use foreign language as a communication tool. This has resulted in the speaking skills of learners of a foreign language generally being worse than theirs skills in reading and writing (Yang et al. 2012, 2013). Previous studies have focused on English teaching, and teaching English is more common than teaching French in Taiwan. It is therefore worth investigating if learning French has the same problems and outcomes as teaching English. Vygotsky (1978) stressed that interpersonal interactions are necessary to strengthen the effects of learning, and opportunities for this can be facilitated by collaborative learning for foreign-language learning. Collaborative learning involves dividing students into pairs or small groups to complete certain learning tasks through interacting with others (Andreas et al. 2010; Barkley et al. 2004).

Based on second-language acquisition theory, Long and Crookes (1992) suggested that giving “learning tasks” to students helped them to learn and use the language naturally as they completed the assigned tasks. One such approach, called task-based learning (TBL), was proposed by Willis (1996) with the aim of enabling learners to complete certain meaningful tasks using the target language, and hence construct knowledge and develop language skills. This has led to some educators applying TBL in SL (Bellotti et al. 2010; Lan et al. 2013).

Considering the above-described factors, the present study chose SL as a platform for French teaching, and it was hypothesized that combining TBL and CL would strengthen the speaking capability of the learners. The perceptions of these students about their learning experience were also investigated. This study recorded a series of French-learning processes with the aim of determining the collaborative models that the students adopted when conducting TBL learning in SL, and their impacts on the learning of French.

Rose (2012) defined context-based learning (CBL) as a pedagogical methodology that focuses on the belief that a specific context is the key to obtaining and processing knowledge of the learning environment and the real situation. Rose pointed out that this method has the advantage that the learning environment is conducive to internalizing knowledge because it is connected to real-life experiences of the learner. In addition, CBL improves the confidence and independence of learners through their active participation and social collaboration (Rose 2012). Since the present research aimed to determine how collaborated learning influences the attitudes of learners, and the instrument used in this study (SL) provides a real environment, it is considered to be context-based research. Thus, we analyzed three types of collaborative models based on the theories of CBL and collaboration. Based on whether the learning activities are relevant to the contexts in SL, they were divided into context-inclusion and context-exclusion: the former occurs when the students’ discussion and videos are linked with the SL contexts, while the latter occurs when there are no discussions in SL or no contexts relevant to those in SL.

The following research questions were addressed in this study based on three types of collaborative models—context-inclusive collaboration (CICO), context-exclusive collaboration (CECO), and context-exclusive centralization (CECE):

• (1) How these three types of collaborative models influence the learning performance of students.

• (2) How these three types of collaborative models influence how students perceive the SL-based TBL activity.

Literature review

Language teaching in SL

In recent years the amount of research on language teaching using SL has been increasing rapidly, mainly due to SL offering a learning environment that can simulate real settings for foreign-language and second-language learners, resulting in them becoming more actively involved in language learning activities (Lan et al. 2013). Learners receiving language education in SL can undertake various learning tasks to engage in meaningful interactions, enhance their language capabilities, and express themselves properly. It has been suggested that using SL in combination with assistive tools will increase communication opportunities among learners (Jauregi and Canto 2012; Wang et al. 2009). Jauregi and Canto (2012) also stated that SL offers an interesting and useful learning environment for foreign-language teaching, since this particular combination of learning environment and designed tasks not only enhances the learning motivation for foreign-language learners and their willingness to communicate, but it also reduces their anxiety about speaking foreign languages (Jauregi and Canto 2012). Language learning strategies are helpful for language learning (Liu and Chu 2010); these include situated language teaching, the total physical response method, cooperative language learning, and TBL. Bellotti et al. (2010) applied these methods to teaching in the SL environment with the aim of improving the learning experience of learners.

Task-based learning

TBL was developed in the 1980s, and was further promoted by Prabhu, an Indian educator of languages, who claimed that TBL emphasized the acquisition and application of language skills when undertaking learning tasks (Prabhu 1987). In 1999, Foster summarized various perspectives of linguists on TBL, and concluded that “…giving learners tasks to transact, rather than items to learn, provides an environment which best promotes the natural language learning process” (Foster 1999). TBL is defined as a teaching method in which teachers give learners one or more learning tasks that facilitate learning while the tasks are performed. The TBL method enables learners to use the target language, develop language skills, and stay motivated in the learning process (Bellotti et al. 2010), and a sense of achievement can also be attained when the task is completed.

Willis (1996) considered that the following three-stage framework should be followed when employing TBL:

• 1. The pretask stage: In this stage, teachers assign a task to learners and give detailed explanations. Students need to familiarize themselves with vocabularies, phrases, grammar, and sentence patterns concerning this task. Some linguists have also stressed the importance of teachers fully explaining the task in the pretask stage, and recommend giving directions using task samples (Dörnyei 2001; Lee 2000).

• 2. The task-cycle stage: In this stage, teachers divide the learners into multiple groups according to the course materials designed, and the learners undertake the task assigned to the group they are in. They are expected to apply the vocabularies, phrases, grammar, and sentence patterns learned in the pretask stage to complete the task, and finally deliver the results to the other groups involved in a certain format. In this stage teachers only play a supervisory role by passively assisting learners in undertaking tasks without active interference; however, they can offer appropriate guidance about the learners’ performance once they have delivered their results.

• 3. The language-focus stage: In this stage, teachers and learners should discuss language-related problems that they experienced while undertaking the task, such as grammar mistakes, errors in sentence structure, and mispronunciations. Teachers will correct these problems and ask learners to keep the applicable rules in mind so as to enhance their overall language proficiency. There are three teaching objectives in this stage: (1) providing learners with the opportunities to present their performance of the task completed, (2) encouraging learners to reflect on their performance, and (3) encouraging learners to take note of and correct the problems they encountered during the task (Ellis, 2006).

There have been numerous researches on TBL in SL (Deutschmann et al. 2009; Jauregi and Canto 2012; Lan et al. 2013), but few have focused on role-play-based video production. Jauregi and Canto (2012) also used TBL in SL and utilized a role-playing task. The preliminary results showed that SL can indeed contribute to meaningful language conversation, resulting in students being more willing to communicate in a foreign language. The results of that research indicated that SL was a suitable platform to use in the present study. However, Jauregi and Canto (2012) focused on exploring the added value of implementing virtual interaction through SL in language teaching curricula, and did not investigate the reasons behind the added value. Therefore, one aim of the present study was to identify the factors that contribute to meaningful language conversation and the desire to communicate in a foreign language.

A few researches have also explored the relationships between collaborative tasks and oral performance. Lan et al. (2013) discovered that using TBL to learn Chinese in SL improved the students’ oral performance and interactions. Their results showed that using TBL in SL may help to improve speaking skills, but the impact of student collaboration was not explored. However, Deutschmann et al. (2009) designed activities for English oral participation in SL, and their results seem to indicate that using a meaning-focused task, which involves authenticity and collaborative elements, has a direct impact on learner participation and engagement. This is why the present study aimed to understand the impact of collaborative learning on oral performance utilizing a task-based approach in SL. Thus, the results of this study can make up for this deficiency of previous work.

Collaborative learning

From the perspective of teaching, collaborative learning encourages learners to engage in the learning process and construct knowledge using their own individual strengths, enabling them to gain a deep understanding and develop language skills (Jara et al. 2009). The techniques of collaborative learning can be broadly grouped into the following five types (Barkley et al. 2004), while these techniques can be independently employed or used as a combination based on expected goals:

• 1. Discussion: Learners engage in interpersonal interactions and interchanges through oral communications. Chou (2011) found that students needed to discuss with their partners while preparing for group oral presentations, and that the resulting collaboration improved their English speaking ability.

• 2. Reciprocal teaching: Olteanu et al. (2014) found that learners assisted each other in learning the course material and developing skills.

• 3. Problem-solving: Lan et al. (2013) applied cooperative problem-solving techniques to learners in a cognition, usage, and expansion model.

• 4. Graphic information organizing: Groups used visualizable tools and messages in the study of Cheng (2014).

• 5. Writing: Learning vital course materials and skills through writing. The studies by Guasch et al. (2010) and Olteanu et al. (2014) used writing in collaborative activities.

Collaborative learning in virtual worlds such as SL can reduce inhibition among learners and increase their social interactions (Meadows 2007). SL is currently one of the most widely used 3D virtual platforms by education practitioners globally. De Lucia et al. (2009) suggested that SL can provide a real-time, low-cost cooperative learning setting for learners. In addition to offering a virtual environment to its users, SL is also known for its high degrees of flexibility and scalability (Kaplan and Haenlein 2009). Each SL user can own his/her space independently in SL (known as an “island”), and practitioners in the education field can tailor education programs to fit their needs. Many studies have focused on collaborative learning using SL; for example, a group at Università degli Studi di Salerno (University of Salerno) in Italy used SL to create an environment that enables cooperative learning (De Lucia et al. 2009). A group at Aristotle University of Thessaloniki in Greece studied the effects of SL on cooperative learning, and found that SL can be an effective tool for socializing (Andreas et al. 2010). A group at Linnaeus University in Sweden built a learning environment for students to fully interact with each other through SL to explore whether communication using online avatars is beneficial to cooperative learning (Petrakou 2010). Peterson (2010a) confirmed that SL provides an environment for students to learn target language collaboratively. His study showed that scenarios and assignments performed within the SL environment motivated learner-centered and active participation. Because of the student-centered activity used in the present study, SL was considered an appropriate platform for the research.

Collaborative learning is greatly dependent on interactions among learners and their interactions with the environment (Andreas et al. 2010), and so using an environment such as SL as a platform for communication should facilitate collaborative learning. Therefore, the present study wanted to enhance the French proficiency of students using collaborative learning and use a different way to improve students’ oral abilities by collaborative learning French through the production of videos in SL.

Based on the results of Arvaja (2011) and Linell (1998), the notion of contextual resources can be used as a type of tool to observe collaborative activities and thus understand how students make use of them and how these contextual resources help the students to carry out collaborative activities. Thus, the present study observed how the students collaborated, in terms of whether they were context-inclusive or context-exclusive.

Methodology

Participants

The subjects of this study comprised 23 students (21 females and 2 males) attending a private university in northern Taiwan. These students took a French conversation course in the second semester of the 2012 academic year, and they were divided into six groups (five groups of four people each and one group of three people) by their teacher to conduct the learning program. Each group contained at least one high-achieving student and one low-achieving student, with the former able to guide the low-achieving ones so as to elevate the performance of the low achievers. All of the groups were expected to achieve similar average performance levels. The academic achievement was evaluated based on end-of-term grades in the last semester.

Research method

This study adopted a qualitative approach to analyze how using different collaborative models influenced the students’ learning performance when they were producing videos. Video production was selected as the learning task because it required the writing of a script and a rehearsal: the script writing gave the students the opportunity to practice using vocabulary and phrases, while the rehearsals performed before shooting the video required the students to repeatedly practice using the dialogue learnt in SL. Videos of SL usage and chat history (including that of local chat, instant messages, and Facebook chat) recorded the discussions and conditions as the task was performed. The interview record was used to obtain feedback from the students about this learning program.

Instruments

Learning platform

The study used the SL 3D virtual world as the main learning platform. A virtual French landscape (shown in Fig. 1) was designed based on the objectives of the French teacher and the course materials. The Linden Scripting Language was used to write programs, and learning aids were provided to assist in learning French and the production of a video. Students then logged into SL to find suitable scenes for their French conversations (Fig. 2), and recorded the screen and sounds during the practice. The students first logged into SL while running the Fraps software (http://www.fraps.com/) so that what was seen and heard in SL could be recorded. They then used PowerDirector (http://tw.cyberlink.com/products/powerdirector-ultra/) to edit the videos. Also, technical assistants recorded the entire activities of students in SL to facilitate further observations, with the chat history of each group being analyzed.

Fig. 1

A virtual French landscape in SL

Fig. 2

Screen capture of the SL learning program

Learning content

Based on the teaching material of a French conversation course offered by the teacher, we constructed a 3D virtual setting featuring French landscaping in SL, combined with TBL theory, to design a 6-week-long French-learning task. The French teachers first provided related knowledge instruction in a traditional classroom setting in three lessons: an inquiry about a travel agency, a time inquiry, and purchasing tickets at a train station.

TBL activity design

The procedure of the video-based experiment in this study was based on the framework of the three-stage TBL theory developed by Willis (1996). The course content was designed according to teaching material provided by the French teachers. Figure 3 shows a flow chart of the learning activity.

Fig. 3

Flow chart of the learning activity

In the pretask stage, the teacher spent 3 days instructing students in a traditional classroom about the vocabulary, phrases, grammar, and sentence patterns related to the task, the aim of the task, the methods of performing, and the expected results. The teacher then divided the students into six groups, each containing a mixture of high- and low-achieving students based on their end-of-term grades in the last semester. Next, technical assistants taught the students how to use SL and to shoot videos. The students were given 7 days to master the skills required to finish the task before the second stage. This stage took 10 days in total.

In the task-cycle stage, the teacher assigned tasks to students and asked them to search for information, discuss the video script, and make the video themselves within 15 days after the classwork had ended. Students had to use SL in order to find suitable scenes for their French conversations and then record their role-playing. The teacher was not involved in the student-centered discussions.

The teacher, a teaching assistant, and students spent 2 h watching and discussing the task videos together in the language-focus stage. The teacher and the teaching assistant made comments to the students about the videos and corrected vocabulary, phrasing, and grammar errors in the videos in order to improve their French proficiency.

The students were then interviewed about their experiences of and impressions about the teaching curriculum 1 week after the language-focus stage.

Interview

After the course had been completed, we interviewed the 23 subjects who participated in the study as well as the teacher to explore their learning experiences and their perceptions about the course. They are interviewed to obtain their feedback about SL and the overall French course, their satisfaction with their interactions with other group members, and the presentation manner of the video-production task. The whole interview was documented and audio-recorded. We first transcribed all of the interview data. Next, based on the following three features, we divided the students’ responses into positive and negative categories in order to determine if the students found it user-friendly to carry out tasks in SL: (1) the benefits of this program, in terms of whether it helped in carrying out group work, and whether it enhanced their grammar and oral communication abilities; (2) the affective perceptions, in terms of how the students felt about the program; and (3) the user experience, in terms of whether the students found it user-friendly to carry out tasks in SL. All of the data were coded and analyzed.

Rubrics

The rubrics used for evaluating the videos consisted of five dimensions: conformity to the work requirement, plot-setting fit, video content, sentence structure, and pronunciation. These rubrics were adopted since the video-production task required the students to write a script and practice oral communication. Additionally, since rubrics have been practiced for years and have been well-received, they serve as the standard for the evaluation of learners’ assignments, projects, and research papers, and they are generally recognized as being suitable for evaluations in the education field (Andrade, 2000). Some dimensions were adapted from the approaches of Andrade et al. (2009), Saddler and Andrade (2004), and East (2009), and the oral evaluation standards adopted in the present study were developed by The International English Language Testing System (http://www.chinaielts.org/downloads/UOBDs_SpeakingFinal.pdf) with some additional items, such as the sentence patterns, pronunciations, appropriateness, and contents outline in the video. We also sought advice from an experienced French teacher, and had the evaluation chart reviewed many times by an expert in the field of language education to ensure that all of the items listed were correctly and completely presented. The advice provided by these teachers resulted in the addition of the following three items: (1) appropriateness, to evaluate whether the setting selected by the students matched the plots, and thus examining how much they were immersed into SL; (2) consistency, to evaluate whether the students could produce uniform writings; and (3) creativity and video effects, to evaluate how much effort the students put into the work.

The study adopted these rubrics to evaluate the students’ performance when producing videos, with the evaluation being carried out by two French teachers who were not involved in the teaching of any of the participants. Both of these teachers repeatedly watched the videos produced by the students and marked them based on the rubrics, with their assessments being completed within 1 month of the experiments ending. Each dimension was evaluated on a scale from 1 to 4 points, with a score of 4 indicating the highest performance. The scores for all items were summed to provide the final score for each group, with a higher score indicating that that group had performed better. A detailed description of the standard used in the video evaluations is provided in “Appendix”.

Results

Collaborative models adopted by students in the 3D situated TBL learning program

The learning task performed by each group was evaluated by French teachers who had not participated in the program using the rubrics provided in “Appendix”. To explore how different collaborative models influence the effects of learning, we analyzed the recordings of the SL interactions and the interviews to understand the way that students performed tasks after the learning program. We then divided them into various types that were subsequently compared.

Based on the video-context consistency, video presentation methods, and the discussion types, we separated the students into the three types of the collaborative models: CICO contained two groups with a total of eight people, CECO contained two groups with a total of seven people, and CECE contained two groups with a total of eight people. Figure 4 shows a diagram with the video context consistency defined as the vertical axis and the presentation methods and discussion types defined as the horizontal axis. The figure reflects that we did not find a context-inclusive centralization model in the results, which is probably because the students found it challenging to complete a task that matched the context within the given time when they worked alone. In addition, the difference between being context-inclusive and context-exclusive—according to the present study—is whether the contents in the videos and the discussions are highly relevant to the context. Although some students are considered to have performed context-inclusive collaboration, most of them are considered to have performed context-exclusive collaboration. This means that it was unlikely that the context-inclusive centralization model was used.

Fig. 4

Three types of collaborative models based on video-context consistency and collaboration

In the video-production process, CICO emphasized teamwork with all members gathered together to discuss and work on the script and video-creation process in SL; the end product was thus highly consistent with its context. Teamwork was also a vital feature for CECO; however, it was used only in script discussions and when members attended the video shooting. In CECO, video shooting started after the script had been finalized in the group discussions. The video was shot based on a previously developed script that had not subsequently been adjusted to the different environment of SL, and this resulted in the video having lower video-context consistency. Lastly, teamwork was absent from CECE since the process was directed by one member who wrote the script and who later convened other members for video shooting in SL. Everyone followed his/her directions for video shooting and production. Since the script had been written before the actual shooting began, the video produced using this model did not exhibit a high context consistency. The features of the three types of collaborative models are presented in Table 1.

Table 1 Features of the three types of collaborative models

Impacts of different collaborative models adopted by students to learn French in the 3D situated TBL learning process

Basing on the qualitative data obtained during interviews conducted by the French teacher, it was found that the students’ oral communication, listening, and writing skills had improved. A part of the interview transcript is provided below.

Interviewer V2:

Was the students’ French oral communication skill enhanced after the learning program?

French teacher:

Judging from my and the teaching assistant’s observations for a month after the activity, we found that the students’ performance had been enhanced. They tended to review the conversations by using Second Life. Also, they tended to express themselves more through various word choices and sentence patterns. They have been more willing to try using different patterns not taught in the class but learnt from the Internet.

Interviewer V2:

Was there any positive influence on students’ French listening, reading, and writing skills?

French teacher:

Obviously yes. Some low-achieving students have started to understand what is going on in the class. In terms of their writing skills, I found that they have been more willing to use various words and sentence patterns, although they make some errors. However, since the usage of Second Life focuses on enhancing oral skills, it may take a while to expand oral skills to writing skills.

To explore how different learning models impacted on learning French, the study evaluated the video-production performance of each group based on collaborative models as demonstrated in Table 2. Each score in the table is the mean value for the groups with the same collaborative model. The scores in the evaluations ranged from 1 to 4 points, with a score of 4 indicating the highest performance.

Table 2 Video evaluation scores

Table 2 indicates that the performances for CICO were evidently better in every aspect than those for CECO and CECE. Except for the “plot-setting fit,” which fell between CICO and CECE, other performances for CECO were similar to that for CECE, while the “sentence structure” item received the lowest scores, especially because grammar correctness was scored lower than the other two learning models and thus reduced the overall score for sentence structure. The subcategory of video “sound effects” received a score of only 1.3 points. The scores for CECE were lowest for plot-setting fit and video content; the “relevance of plot and setting” subcategory of the former item was especially lower than for the other two models, and thus reduced the overall score, while the “conformity to work requirement” subcategory of the latter item was clearly lower than for the other two models, and thus reduced the overall score. In addition, “complexity of sentence patterns” in the sentence-structure category scored only 1.9 points, while sound effects scored only 1.5 points.

SPSS software (version 20) was used to analyze the collected data. The results obtained in one-way ANOVA for the video scoring are presented in Tables 3, 4, 5, 6, and 7. The total score differed significantly between the three types of collaborative models (F = 82.808, p < .05), being higher for CICO (mean = 2.84) than for CECO (mean = 2.39), and higher for CECO than for CECE (mean = 2.33). The score for plot-setting fit also differed significantly between the three types of collaborative models (F = 3.635, p < .05), being much higher for CICO (mean = 3.38) than for CECE (mean = 2.50). The pronunciation score differed markedly between the three types of collaborative models (F = 9.965, p < .05), being higher for CICO (mean = 2.91) than for CECE (mean = 2.44), and higher for CECE than for CECO (mean = 2.41). The score for video content differed significantly between the three types of collaborative models (F = 20.637, p < .05), being higher for CICO (mean = 3.08) than for CECO (mean = 2.64), and higher for CECO than for CECE (mean = 2.55). Finally, the score for sentence structure also differed significantly between the three types of collaborative models (F = 73.718, p < .05), being higher for CICO (mean = 2.48) than for CECE (mean = 2.07), and higher for CECE than for CECO (mean = 2.01).

Table 3 Descriptive statistics of the rubrics and pronunciation

Table 4 Descriptive statistics of the plot-setting fit, content, and sentence structure

Table 5 Test of homogeneity of variances

Table 6 ANOVA of the rubrics, pronunciation, plot-setting fit, content, and sentence structure

Table 7 Post-hoc Tukey honestly significant difference

Perceptions of different collaborative models adopted by students learning French in the 3D situated TBL learning process

How students’ collaborative learning differences affected their perceptions about learning French was determined; the present study numbers and the results of analyses of the student interviews are presented in Table 8. Perceptions were classified into those related to learning benefits, affective perceptions, and the user interface. Each of these subcategories accounted for a percentage of all of the perceptions, as quantified below.

Table 8 Numbering students’ interviews

The negative feedback received from the students primarily related to time pressure, anxiety, and feeling troubled (26.24 %); for example, “I had to discuss with classmates in SL within a limited time. The time was kind of short, but there were many things to do.” Operating the user interface was another problem (21.98 %); for example, “…It was kind of abrupt to use SL, and I was not accustomed to the operation. It took more time to learn.” The students also gave positive feedback about this curriculum, such as about it providing more opportunities to speak French with peers. They were able to repeatedly practice and correct each other, which was helpful for their speaking (26.95 %); for example, “We can correct each others’ intonation and pronunciation in conversation” and “Repeated practices can enhance speaking, and I looked up vocabulary in order to write scripts.” Moreover, the students found SL interesting, refreshing, and natural for them to use, enabling them to speak out (16.31 %); for example, “Unlike other games or social networking websites, SL is refreshing.” Additionally, SL gave the students a sense of authenticity and achievement (8.52 %); for example, “In the SL virtual situation, I felt as if I was truly in France, having more senses when practicing speaking.”

The students’ negative affective perceptions did not vary much between the three types of collaborative models, being slightly higher for CECO (11.35 %) than for CICO (7.80 %) and CECE (7.09 %). The students’ perceptions differed mainly in the positive feedback they provided. The students exhibited extremely diverse perceptions about positive benefits and about positive affective perceptions between the three types of collaborative models. The students’ perceptions about positive benefits were 14.89, .71, and 11.35 % for CICO, CECO, and CECE, respectively; the corresponding values for their positive affective perceptions were 2.13, 5.67, and 8.51 %. CECO was the only learning model for which the positive affective perceptions of its category (31.43 %) were lower than the negative affective perceptions of its category (68.57 %).

χ² tests revealed significant differences in the students’ perceptions between the three types of collaborative models (p = .001; Table 9). These differences stemmed from the students’ perceptions about positive benefits of CECO (.71 %), which was vastly lower than for CICO (14.89 %) and CECO (11.35 %).

Table 9 χ² tests

Discussion

The purpose of the present study was to identify the collaborative models used by students and the impacts of these different models on their learning of French in a 3D situated TBL learning setting. This section explores the features of the learning models used by the students—as summarized in the “Results” section—and their effects on the students’ learning performances and perceptions.

The context consistency, video production, and discussion types of CICO, CECO, and CECE learning models used by the students were analyzed. To analyze the three types of collaborative models in terms of the context consistency, CICO was the only one based on CBL. The main differences between the other two models were in the methods of video production and discussion types. CICO was the only model used in discussions in SL, adjusting the script and shooting in accordance with the SL setting, so it had a much higher contextual consistency than the other two models.

To analyze three types of collaborative models in terms of video production and discussion types, both CICO and CECO require teamwork to complete discussion and video production. The difference between CICO and CECO is that the former spontaneously cultivates positive interdependence, individual accountability, face-to-face interactions, interpersonal and small-group skills, and group processing, which are the five cooperative learning elements reported by Johnson and Johnson (Roger and Johnson 1994). The authentic environment provided by SL allowed the students to have more diverse and abundant interactions; for example, the interactivity of massively multiplayer online role-playing games (MMORPG) promotes situated learning, and contributes to cooperation and conversation, negotiation, and self-restoration (Peterson, 2010b). In our study we found that the tasks of role-playing and video production in SL also have these functions, giving students a sense of participation. By working together, students build stronger interpersonal and teamwork skills.

Even though CECO also involves teamwork, it only achieves individual accountability and promotes face-to-face interactions. Like CICO, CECO also promotes face-to-face interactions, but only face-to-face discussions are possible, as well as online discussions using Facebook or Skype, while in CICO the students can also perform discussions in the SL setting. The authenticity of the environment provided by SL enables students to be fully engaged in the task, thereby promoting and making the best use of teamwork. As a result, CICO results in more tacit acknowledgment and solidarity among the students in a group. Due to CECO not being better than CICO in terms of group maturity and interpersonal and small-group skills, the collaborative learning performance of CECO is not as effective as that of CICO (Johnson and Johnson, Roger and Johnson 1994).

CECE primarily involves leading by a single member, which corresponds to the lowest amount of teamwork of the three types of collaborative models considered in this study. Similar to CECO, CECE lacks group maturity as well as interpersonal and small-group skills. Moreover, CECE makes it possible for certain students to benefit from the results without them making equitable contributions. These factors all reflect the disadvantage of collaborative learning that not everyone contributes equally. For example, one member could perform most of the tasks himself, from script-writing to acting-teaching, if other members are too busy to share the work, which also reflects that a more conscientious student can end up performing an unfair proportion of the duties (Çelik et al. 2013; Thornton 1999).

Impacts of different collaborative models adopted by students on their performance in learning French

In this study it was found that CBL is significantly correlated with the performance of learning French in terms of the following four aspects: plot-setting fit, video content, sentence structure, and pronunciation. The finding of this research that CBL is helpful for students’ pronunciation is consistent with Lan et al. (2013) suggesting that CBL can facilitate speaking ability, and Jauregi and Canto (2012) indicating that utilizing SL for contextualized-speaking research contributed to excellent achievement. The present research also indicated that CBL is beneficial to sentence-structure correctness, including the difficulty of the vocabularies used, the complexity of sentence patterns, and the grammar correctness. The students in this study used more difficult words and a wider variety of vocabulary and sentence patterns for CICO. Rankin et al. (2006) found that students also increased their English vocabulary in an MMORPG, and that they improved by interacting with the nonplayer characters. However, in the present study, the students learned from the interactions between their avatars in the game. It is possible that CICO involves additional communication and interaction in a virtual environment, so students use more difficult and complicated words and sentences when writing scripts. In spite of this, both results illustrate that CBL plays a critical role in language learning. Furthermore, CBL is also crucial to plot-setting fit and video content. Since CBL is associated with the setting in which it occurs, students perform better in logical transitions of scenes and consistency of dialogues, exhibit more creativity, and use more appropriate sound effects to fit the setting. In fact, CECO simply requires the students to talk face-to-face or to use instant messaging. In addition, the videos were shot in SL after the scripts had been finalized, and so these scripts did not match the SL setting. The students also tended to choose an easier storyline that did not take the setting into consideration. Therefore, the mean score for the sentence structure when using CECO was only 2.0, lower than the score achieved (2.5) when using CICO, and even lower than the score achieved (2.1) when using CECE, which had the same problem as CECO. Consequently, the students’ performances for CICO were much better than those for the other two models that did not include CBL.

The results have shown that the students performed better when they were embodied in the contexts. This finding is consistent with the results of four previous studies: Lan et al. (2013) found that CBL facilitates students’ speaking ability; Jauregi and Canto (2012) obtained excellent results when using SL with contextualized speaking; Rankin et al. (2006) found that students’ knowledge of English vocabulary and conversational abilities improved in a massively multiplayer online (MMO) learning context, which is similar to the results of the present study; and Peterson (2010b) found that learning in the MMO context also enhances the usage of collaborative dialog, negotiation, and self-repair.

The results of the present study indicate that the overall performances for two collaborative models, CICO and CECO, are superior to those for centralization (CECE), which is consistent with Chou (2011) and Zeng and Takatsuka (2009) reporting that the collaborative model is helpful for language learning. The difference between them is that in the present study the use of the collaborative model was only clearly correlated with plot-setting fit, and partially correlated with video content (and not very obviously in most items), which indicates that not all types of collaborative models are beneficial to speaking and sentence structure. Chou (2011) and Zahedi (2012) found that the collaborative model is helpful for speaking, where this was only the case for CICO in the present study, which clearly highlights that situated cooperative learning has this effect. Besides, Jamshidnejad (2011) pointed out that the use of specific communication strategies in second-language exchanges broadened the language horizons of the participants. The communication strategies used for CICO by the students in the present study were consistent with the settings in which they were applied, which possibly encouraged them to use more language knowledge and lexicons that were not included in their textbooks. The student interviews indicated that most of them considered that repeated practices and mutual correction in this situation contributed to the enhancement of their speaking ability, which correlates with their video-production scores. However, since CECO does not involve mutual reliance and interpersonal and small-group skills, students are not able to practice repeatedly, correct each other, and improve speaking. In their interviews they reported that they did not consider that this would have helped them in speaking and sentence structure.

Impacts of different collaborative models adopted by students on their perceptions about the French-learning activity

The present study indicates that both CBL and collaborative learning positively affect students’ perceptions about the benefits of repeated practices and mutual corrections on speaking ability enhancement. In CBL, since the motivating context promotes beneficial interactions in the target language (Peterson, 2010b), the repeated practices and mutual corrections in CICO can greatly enhance students’ speaking abilities. Moreover, CICO applies a good collaborative model. Students who adopted CICO indicated that cooperation with other members was very helpful—it not only increased their opportunities to practice speaking, but also reduced the task burden, and therefore yielded the best learning outcomes.

However, teamwork can also negatively influence students’ learning due to factors such as individual differences, emotional factors, divergence, and deadlines (Chou, 2011). The student interviews performed in this study revealed a large divergence in the students’ perceptions of CECO. Moreover, they considered it difficult to find the collective time for everyone. Thus, individual differences, divergence, and deadlines can influence CECO, and further affect the achievements possible from collaborative learning. The suggestion of needing the teacher’s guidance in CECO accounted for the largest proportion of opinions, which explains that the students need a person to guide them and help them negotiate with each other.

For CECE, despite the students generally providing positive feedback about their perceptions of the benefits brought by repeated practices and mutual correction on speaking ability enhancement, this was not reflected in their learning performances. This is because the positive feedback is given by the leaders, who were the only beneficiaries of the activities. Students’ perceptions about benefits were negatively correlated with the number of suggestions; that is, students had fewer suggestions about activities that they considered to be more helpful.

Factors related to students’ affective perceptions of freshness, fun, anxiety, and pressure are CBL and students’ perceptions about the benefits, and all of these factors were negatively correlated. The negative correlation between perceptions about benefits and affective perceptions can exemplify why CBL exerts a negative effect on affective perceptions. Chou (2011) indicated that activities are instrumental in fostering students’ speaking ability, but students might not consider the activities to be interesting, which is consistent with the findings of the present research. Students in the CICO groups reported that they learned a lot from the activity, but the time available was too short, and this time pressure may negatively affect learners’ learning performance (Zeng and Takatsuka 2009). Even though the students using the three types of collaborative models all worked under time pressure, CBL requires extra language activities, which increased the amounts of time and energy required for students in the CICO groups to complete the tasks. This can be seen from the interviews: students provided more negative perceptions about ease of use, such as problems with operating the user interface. Spending too much time on user-interface problems reduces the amount of interperson communication (Deutschmann et al. 2009), and the resulting stress reduces learners’ motivation and participation. This is related to the affective filter hypothesis (Krashen 1987): as the affective filter increases, anxiety also increases, while learners’ attitudes and motivation decline. Lan et al. (2013) found that CBL has a positive impact on learners’ attitudes. The main difference between that study and the present study is that the learners may have had a more flexible time schedule in the former, while in the latter the learners felt a large time pressure. Zahedi (2012) found that group learning was correlated with motivation; learners’ learning performances originate from group discussions, collective creativity, and thinking (and not at the individual level), and their stress also decreased as a result. In the present research, group learning yielded good results, but the anxiety caused by time pressure also affected the learners’ motivation. Accordingly, future research could attempt to reduce anxiety by lowering the affective filter.

In addition, it was observed in this research that a model with a higher score for the video content resulted in a higher satisfaction level among the students. However, since this study included only a relatively small number of samples, and hence further experiments are required.

Conclusion and future works

The purpose of this study was to identify the collaborative models used by students and the impacts of using different collaborative models on students’ perceptions about learning French in a 3D situated TBL learning setting. After observing students’ learning characteristics, this study divided learning models into three types based on the extent of CBL and collaboration: CICO, CECO, and CECE. Among the three types of collaborative models, CBL positively affected the students’ speaking as well as every phase of learning performance, thereby enhancing the effect of collaborative cooperation. CBL and collaborative learning also positively impacted students’ perceptions about the benefits of repeated practices and mutual corrections on speaking ability enhancement. Nonetheless, students’ anxiety caused by time pressure and user-interface problems negatively influenced their affective perceptions, and further affected their motivation and attitudes. The reductions in students’ anxiety induced by lowering the affective filter should be investigated in future research.

This study was subject to some limitations. First, the smallness of the sample means that the results are not generalizable. Whether the results reported herein apply to other cohorts should be addressed by future studies involving larger numbers of participants. Second, the present study examined the influence on French language learning, so whether the results can be applied to other languages remains to be determined. Third, the participants of the present study were university students in Taiwan, and so future studies should determine whether similar results are obtained for learners of French as a second or foreign language in other countries.

Most research on collaborative language learning focuses on students’ writing, while this study used video production to facilitate the learning of French by students, in the hope of exploring new possibilities in language learning research. Since the amount of research on speaking is relatively small, this study may prompt an increased interest in this research field.

Appendix

See Table 10.

Table 10 Rubrics