Taguchi, N., Li, Q., & Tang, X. Learning Chinese formulaic expressions in a scenario-based
interactive environment. Foreign Language Annals.
This is the accepted version of an article published in Foreign Language Annals by the American
Council on the Teaching of Foreign Languages (ACTFL),
https://www.actfl.org/publications/all/foreign-language-annals. This article may be used for non-
While proficiency in a second language (L2) requires knowledge of grammar and
vocabulary, such linguistic forms provide only part of the knowledge and skills that are required
to perform communicative functions (e.g., greeting): alone, they do not guarantee successful
communication. Knowledge of context (whom to greet and when), cultural norms, and social
conventions is equally important to successful communication. Thus, pragmatics – knowledge of
linguistic forms, their social functions, and context of use – represents a critical aspect of L2
learning. Thomas (1983) argued that pragmatics entails two interrelated components:
pragmalinguistics and sociopragmatics. Pragmalinguistics concerns linguistic forms and their
communicative functions, while sociopragmatics involves understanding the social norms and
conventions that guide language users’ way of behaving. Both components need to be addressed
in teaching so that learners understand which linguistic forms to use to achieve their
communicative goals, and how their language use impacts others in a particular context.
A number of studies have been conducted to date using a variety of instructional methods
and tasks designed to teach pragmatics (for a review, see Taguchi, 2015). These studies
demonstrated the importance of five primary components: (1) input, (2) metapragmatic
explanations, (3) production tasks, (4) consciousness-raising tasks, and (5) feedback. However,
both the tasks and the findings are limited due to issues of contextualization, interaction, and
learner agency. Because most studies provided pragmatic input (pragmalinguistic forms and
contextual factors) in written or audio-recorded dialogues and although a few made use of video,
learners had to imagine, rather than actually participate in, the situation; thus, participants’
language use was only weakly contextualized. In addition, in many studies, learners took the role
of an observer, watching scenes as a third person. In the studies in which leaners could take on a
more active role, opportunities for interaction have been limited to structured output exercises
and/ or role plays, in which learners assumed assigned roles and interact with others within the
confines of the imagined situation. Finally, many studies used explicit metapragmatic
explanations, rather than having learners make and reflect on their own pragmatic choices or
revise their choices voluntarily. Thus, the impact of learner agency has been overlooked. In sum,
existing practice has been researcher-driven, using pre-selected tasks that were presented in a set
order, giving learners little autonomy to decide on their course of action.
In contrast, the study reported here made use of a computer-based instructional platform
that was specifically designed to avoid these limitations. Using this platform, L2 Chinese
learners engaged in a series of scenario-based, semi-interactive exchanges with characters who
appeared in videos that were recorded in Shanghai. The platform incorporated several gaming
elements that enabled learners to take on a character and select different situations in which to
practice. Learning outcomes and interview data from 30 students provide insights into the
usability of this approach to teaching pragmatics, in this case, formulaic expressions.
Virtual Participatory Environment and Gaming in Pragmatics Teaching
Pragmatics refers to the linguistic resources that are needed to perform communicative
acts, as well as the knowledge of cultural norms that allows one to understand how a
communicative act will be perceived (Thomas, 1983). The teaching of pragmatics thus must help
learners to bring together their knowledge of the culture, the social context, and the actual
language that is needed to carry out a communicative function (for a review, see Taguchi 2015).
Among various forms of technology-enhanced teaching, a virtual environment is particularly
promising for pragmatics learning because it provides immediate and convenient access to
contextualized communicative practice that can be altered to meet individual learners’ needs (for
a review, see Taguchi & Sykes, 2013; Taguchi & Roever, 2017). In the virtual world, learners
can simulate a variety of participant roles across diverse social situations, and develop
pragmalinguistic and sociopragmatic knowledge while completing goal-oriented tasks.
Several studies have examined the efficacy of a virtual environment for L2 learning
(Berns et al., 2013; Liou, 2012; Milton et al., 2012; Peterson, 2005, 2011). The most notable
trend is the use of digital games. With the expansion of the internet, various online games have
attracted millions of users who interact with each other to complete in-game tasks. This trend has
inspired L2 researchers to draw on the power of digital games when designing instruction
(Cornillie, Thorne, & Desmet, 2012; Gee, 2007; Reinhardt & Sykes, 2014; Squire, 2011; Sykes
& Reinhardt, 2012). Sykes and Reinhardt (2012) call this trend game-based research, i.e., using
game design principles to create instructional tasks that target specific pedagogical goals and
elicit certain language behaviors that are aligned with those learning objectives.
In L2 pragmatics, only a few studies have attempted game-based research (Holden &
Sykes, 2013; Sykes, 2009, 2013). These researchers developed original games to teach pre-
selected pragmatic features in existing L2 classrooms. Sykes (2009, 2013), for example, explored
the efficacy of an immersive gaming space in teaching Spanish speech acts. In a digital space
emulating local Spanish communities, learners practiced requests and apologies while
completing goal-directed tasks with computer-generated avatars. Results revealed only a small
gain in learners’ request-making strategies after a few sessions of gameplay, but the gain was
larger for apologies. The lack of gain in request-making may indicate that a mere in-game
practice may not generate sufficiently robust learning when systematic focus-on-form activities
and feedback are not embedded in the practice.
Sykes’ subsequent work focused on feedback as part of the learning mechanism. Holden
and Sykes (2013) created an iPod-based mobile game to teach Spanish pragmatics. They
developed a plot in which learners had to solve a murder mystery by collecting clues from non-
player characters (NPCs). Some NPCs preferred a direct style of speaking, while others preferred
an indirect manner. Learners had to manipulate different styles to obtain clues. Linguistic
choices that matched with NPCs’ preferred styles led to more clues, whereas inappropriate
linguistic choices resulted in game-end experiences. However, usability testing through
interviews, observations, and gameplay data revealed that the feedback was not salient enough
for learners to notice and act on it. Based on the findings, the authors made the feedback more
exaggerated, but this adjustment did not result in more significant learning outcomes.
While pragmatics was not the target area, several studies created a game-based, virtual
participatory environment where pragmatics learning could take place. Wik and Hjalmarsson
(2009) developed a dialogue system that involved a role-play game with a built-in conversational
agent. Simulating a flea market scene in Sweden, learners of Swedish practiced communicative
functions such as requesting a product and negotiating price. Learners received feedback from
the agent, who smiled at deal closings and looked angry after bartering for a long time. In
another study, Si (2015) created a virtual platform for teaching Chinese. She designed an
immersive environment where learners interacted with each other using the Microsoft Kinect
camera and Teamspeak 3 (voice chat service). Working within a “school café” environment,
learners collaborated with their peers to complete a variety of activities, such as pushing a table
or moving a box. These activities elicited a variety of speech acts (requests and suggestions).
When interviewed, learners stated that it was a positive learning experience; however, the study
did not assess how much learning of Chinese (e.g., vocabulary, speech act strategies) actually
resulted from this immersive practice.
As illustrated above, there is a small body of research on interactive, immersive virtual
environments that integrate playful, game-like features to promote learning. These studies
demonstrated that rich contextual cues and high levels of interaction within the virtual
environment support pragmatics learning. The game design itself also provides an autonomous
learning space where learners can decide on their own course of action as they progress through
in-game tasks (Schwienhorst, 2002). Finally, previous studies show that salient feedback
mechanisms also assist learning, especially when feedback is pre-programmed in a way that the
system automatically responds to learners’ pragmatic choices and provides learners with
immediate opportunities to modify their choices (Holden & Sykes, 2013).
Definitions of formulaic expressions vary, but the literature points to several common
characteristics (Bardovi-Harlig, 2012; Wray, 2002; Wood, 2006). Formulaic expressions are: (1)
(semi) fixed multi-word sequences; (2) stored in memory as a holistic unit; (3) phonologically
coherent (produced without pause); (5) syntactically irregular (e.g., in the expression “beat
around the bush”, “bush” must be singular); (6) community-wide in use; and (7) bound to
specific speech events. Based on these characteristics, formulaic expressions in this paper were
defined as fixed syntactic strings tied to specific situations and communicative functions.
Formulaic expressions are useful linguistic resources because they permeate everyday
communication and assist social participation. Existing findings generally support positive
effects of residence abroad on formulaic learning (e.g., Taguchi, Li, & Xiao, 2013). However,
the effects differ between comprehension and production. Roever (2005) found that ESL learners
achieved better comprehension of formulae than EFL learners. Taguchi (2011) found that L2
English learners who studied abroad outperformed their counterparts without study abroad
experience on comprehension of formulae. However, the study abroad advantage was not found
in production studies. Bardovi-Harlig (2009) compared comprehension and production among
ESL learners in a U.S. university and found that learners’ comprehension was better than their
production. In Taguchi (2013) L2 English learners with and without study abroad experiences
evidenced similar profiles in their production of formulaic expressions.
These findings suggest two implications for the teaching and learning of formulae. First,
the modality effect found in previous studies indicates that formulaic expressions are best
practiced when production and comprehension are combined because these skills call for
different linguistic and cognitive resources. Comprehension occurs without precise linguistic
parsing by relying on contextual cues, but production requires a fine-tuned grammatical analysis.
Hence, comprehension practice can assist the initial entry of formulaic expressions into memory,
but interpretive practice needs to be supplemented with production experiences so that learners
can fine-tune their use across different situations. This is particularly important for formulaic
learning because incorrect forms, for example in word order or word choice, can lead to non-
target-like expressions: Omitting the word “to” in the formulaic question “For here or to go?”
results in a non-native-like expression.
In addition, previous studies documented the positive influence of study abroad
experience, at least on comprehension, suggesting that formulaic expressions are best taught
through participation in communicative events in a local community. Because formulaic
expressions are community-wide in use and are tied to ordinary speech events, it makes sense
that formulaic development happens in a context where the target language is widely spoken.
However, since observing learners’ uptake and use of formulaic expressions is difficult if not
impossible in regular, on-going daily social events, technology-supported instruction, for
example in the form of a multi-modal situation-based platform, can provide learners access to
formulaic language use in real-life situations so they can both understand the function of highly
context-dependent, culture-specific formulae and use them in the context of on-going,
Drawing on previous research into these two critical domains -- pragmatics and
formulaic expressions-- this study investigated the advantages of teaching pragmatics, in this
case, formulaic expressions, in the context of an interactional space that allowed L2 Chinese
learners to practice understanding and producing formulae in realistic, video-recorded, semi-
immersive situations. Instruction incorporated several gaming elements, including clearly-stated
learning objectives contextualized practice, decision-making mechanisms, interactions with
video-based characters, and timely feedback and assessment criteria. The following research
questions guided the investigation:
1. To what extent do L2 learners of Chinese improve their knowledge of formulaic
expressions as a result of engaging in scenario-based interactive practice, and do they
maintain those gains?
2. To what extent do learners perceive scenario-interactive practice to be a useful,
immersive-like, and game-like experience?
Thirty students (13 males and 17 females; mean age of 20.5) who were enrolled in the
Chinese language program in a private university in the United States volunteered to participate
in the study, including five students from the elementary-level courses, 12 intermediate-level
students, and 13 at the advanced-level. Twenty-five were native speakers of English, four of
Korean, and one was a native speaker of Hindi. On average, they had 33 months of formal
Chinese study, ranging from seven to 120 months. Seven had prior study abroad experience
ranging from one to six months. All participants reported using a computer every day; however,
18 participants reported that they rarely played computer or online games, while 12 students
reported playing games from one to four days per week. From the pool of 30 students, 24 were
selected to participate in a follow-up interview, including all elementary-level students (five
total) and all intermediate-level students except for one who did not agree to be interviewed (11
total). Because the advanced-level group had the largest number of participants (13 total), in
order to balance the number of participants at each level, eight students were selected for the
interview. Maximum variation sampling was used to select these eight students so as to counter-
balance gender, age, and L1 backgrounds.
Target Formulaic Expressions
The study targeted 28 formulaic expressions that most naturally occur in specific
situations to accomplish particular communicative functions (Appendix A). For example, the
[A little cheaper] occurs when bargaining with a street vendor in China,
while the phrase
[How have you been?] occurs when opening a conversation. To
identify target expressions, the authors selected 11 formulaic expressions from Taguchi et al.’s
(2013) study, which examined L2 Chinese learners’ knowledge of formulaic expressions that
were useful in study abroad contexts. These 11 expressions were included in Chinese reference
books and recorded in the authors’ field notes, along with the situations in which they were used.
Situations were subsequently piloted with Chinese native speakers in China to see if they did
indeed elicit the intended formulaic expressions. The native speakers also rated the commonality
of each situation by indicating the extent to which the situation occurred regularly in their own
experience. In addition, 17 new expressions and situations were created for the present study.
Following Taguchi et al., the authors identified a set of possible expressions and the situations in
which they would be used and then piloted the situations with 32 native speakers of Chinese.
Following Bardovi-Harlig’s (2009) criteria, situations that were judged as occurring regularly by
50% of the native speakers were retained. Of the 28 selected expressions, 15 appeared in
participants’ textbooks, but were presented with English translations but without opportunities
for communicative practice.
Unlike existing practice opportunities in virtual learning environments and games, the
design opted not to create a 3D immersive space where participants interacted with built-in
avatars, primarily because most researchers do not have expertise in game design, programming,
computer graphics, or animation. Instead, after identifying the target formulaic expressions, a
series of dialogues were developed in which those expressions naturally occurred. Three native
Chinese speakers checked whether each dialogue flowed naturally and several dialogues were
revised based on their feedback, resulting in ten scenarios, formatted like the sample situation
and dialogue between
[a shop assistant] and
[a customer] below (English translation
Situation: You are shopping at Raffles City Shanghai. A shop assistant walks up to you
to offer help, but you don’t need her help. While browsing, you see a T-shirt you like.
You wonder how much it is and if it fits you.
？[Hello, how can I help you?]
。[I’m just looking.]
。[Alright, please let me know if you need any help.]
? [How much is this T-shirt?]
。[It’s 120 RMB.]
？[Can I try it on?]
。[Sure, the fitting room is over there].
In this dialogue, which purposively includes three formulaic expressions (lines 2, 4, and
6), participants took the role of the customer. The authors then video-recorded native Chinese
speakers acting out the dialogues and videos were edited (e.g., sound effects) using a ready-made
video-editing software like iMovie. During the editing process, the targeted formulaic lines, in
the above example, the customer’s, were eliminated. To play their role in the scene, participants
were asked to select, and later produce, an appropriate formulaic expression from among four
For each dialogue, one correct option and four error options were provided. The error
options involved three error types: pragmalinguistic, sociopragmatic, and sequential. In the
sample item below, Option A is the correct response.
Situation: At a fruit vendor, you want to buy some apples. So you go to the vendor. Ask
the price of the apple and bargain for a lower price.
？[Do you want any apples?]
[How much is the apple?]
? (pragmalinguistic error)
[How much is the apple?]
? (sociopragmatic error)
[How much value is the apple worth?]
? (sequential error)
[How many apples do you have?]
Pragmalinguistic errors contained syntactic/lexical errors or syntactic/lexical deviations
from the target formulaic expression (e.g., unnecessary characters are added or some characters
are missing, which make the expression non-formulaic). Option B above illustrates this error
[money] is missing when asking for price. Sociopragmatic errors involved
expressions that presented a mismatch with the given situation for one of the following reasons:
(1) the expression is too formal/elaborate or too informal/rude for the situation or (2) the
expression does not convey the illocutionary force required in the situation. In the above
example, Option C illustrates this error type because
[worth] is typically used for an valuable
object in a more formal situation. Sequential errors involved an expression that is grammatically
correct but not sequentially relevant to the preceding utterance. Option D illustrates this error
type because the sentence is about the number of apples, not about the price, which is irrelevant
in the preceding discourse. When participants made a wrong choice, a hint corresponding to each
error type popped up on the screen.
The ten video tasks involving the 28 targeted expressions were incorporated into a coherent
set of connected interactive scenes, mapped in Figure 1, that was created with Adobe Captivate
software and took into consideration Sykes & Reinhardt’s (2012) ten critical components.
Insert Figure 1 about here
Figure 1. Task situations
The overall plot of the ten scenarios involved a college student (Zack or Mary depending on
participant’s choice of main character) who was studying abroad in Shanghai. The learning
objective (to use formulaic expressions) and goal of the game (to successfully navigate the ten
real-life scenarios by interacting with local Chinese speakers) were clearly stated. Three
feedback mechanisms were also incorporated: leveling, points, and hints:
(1) Leveling: Ten tasks were categorized into three levels of difficulty based on the length.
Higher-level tasks yielded more points upon completion.
(2) Points: Participants received four points if they answered correctly on the first try, three
points on the second try, two points on the third, one point on the fourth.
(3) Hints: Participants received hints when making a wrong choice.
These features added playfulness, supported imaginative language use, and provided deep
contextualization to what would otherwise be purely rule-based instruction. Because learners
could complete the scenarios in their preferred order, the game also offered learners a certain
degree of autonomy.
For each task, participants first read a situational scenario. After reading the
scenario, they played a short video clip in which a native Chinese speaker enacted his or her role
in the dialogue. The speaker’s utterances were also displayed at the top of the screen in Chinese
with Pinyin support. Participants then responded to what the character said by choosing the
correct formulaic expression from a multiple-choice list displayed in Chinese, also with Pinyin
support. If players made a wrong choice, a hint was provided, for example “Hint: The expression
is not appropriate for this situation. Think about the situation (setting, characters’ roles, topic,
communicative function) and try again.” After choosing the correct option, they moved to the
next part of the dialogue.
Once they had completed all of the multiple-choice questions for a particular scene,
participants reviewed the targeted formulaic expressions by filling in the blanks in the dialogue.
To check their work, they clicked on the “submit & compare” button and compared their
utterances with correct utterances that popped up on the screen. Players were asked to revise
their utterances when mistakes were made. After completing the interaction (choosing responses,
writing responses, comparing and editing responses), participants returned to the initial task map
showing all 10 situations. Their total score appeared on the screen and, if desired, they selected a
Participants’ knowledge of formulaic expressions was assessed with a production and
comprehension test prior to, immediately after, and two weeks after the scenario-based
interactive practice. To reduce the practice effect, two parallel test versions were created by
making minor changes in the scenarios and choice of non-target expressions. To avoid priming
participants with target expressions, the production test was administered before the
The production test was composed of seven background scenarios, presented in English,
and seven corresponding dialogues, all in Chinese. Four of the scenarios and target dialogues
elicited target formulaic expressions and three distractor dialogues elicited non-target
expressions. These background scenarios and dialogues were presented on a computer screen.
Each target dialogue contained four to eight blanks into which students typed formulaic
expressions in Chinese. The scenarios and dialogues differed from those that were used in the
The comprehension test had 28 multiple-choice items. Each item contained a situational
scenario, provided in English, which prompted participants to select the utterance that was most
appropriate in the given situation. Each scenario was followed by a multiple-choice question
with four answer options, one correct answer (target formulaic expression) and three error
options (in Chinese). The error options contained three types of errors: pragmalinguistic,
sociopragmatic, and sequential. The scenarios and option sentences differed from those used in
the interactive practice.
In order to gauge participants’ perceptions of the scenario-based interactive practice
experience, one-to-one interviews were conducted with 24 participants in English. Interview
questions addressed participants’ like and dislikes, as well as most and least favorite and/or
useful aspects of the experience. Questions also addressed the extent to which participants felt
like they were interacting with, or observing “others” interact with, the characters and the extent
to which they considered the experience to be a game or game-like experience (e.g., fun, re-
Prior to launching the main study, the entire pre-assessment, game experience, and post-
assessment process reported above was piloted with five students who were enrolled in Chinese
classes at the target institution, but who did not subsequently participate in the main study. Slight
modifications (e.g., making the font of Pinyin smaller, changing the location of several icons,
incorporating longer video clips) were made based on their performance and feedback. The main
study followed the same procedures. On Day 1, after signing the consent form, participants
completed the pretest online. No time limit was imposed, and all participants completed both the
production and comprehension pretests in about 50–70 minutes. Participants completed all 10
scenarios at their own pace, also on Day 1. On Day 2, they completed all 10 scenarios again and
then completed the posttest. Finally, on Day 3, which took place about two weeks later,
participants completed the delayed posttest. Participants were interviewed at the end of either
Day 2 or Day 3. Interviews ranged from 15 to 45 minutes in English, and were audio-recorded in
a quiet office on campus.
This study addressed two research questions: (1) participants’ gain in the knowledge of
formulaic expressions as a result of two sessions of scenario-based interactive practice; and (2)
participants’ perceptions of interactive, virtual game-based practice.
To address the first question, pre, post, and delayed posttests were scored and compared
using within-subject analysis. The comprehension test had 28 multiple-choice question items.
One point was assigned per correct answer. For the production test, two native Chinese speakers
with Chinese teaching experience rated participants’ utterances using a six-point scale ranging
from 1 (cannot evaluate) to 6 (excellent). Six points (full score) were assigned to an utterance
that represented a target-like expression. Five points were given when the utterance was slightly
different from the target expression due to minor grammatical and/or lexical errors that did not
obscure meaning. Four points were given when the utterance differed from the targeted
expression, but was still comprehensible and appropriate and if the grammatical and/or lexical
errors did not obscure meaning. Three points were given when the utterance contained notable
grammatical and/or lexical errors. Two points were awarded when the utterance was
incomprehensible or too limited to evaluate. One point was awarded when participants did not
respond. After the initial norming session, two raters assessed each utterance independently.
Utterances from the pre, post, and delayed posttest were randomized to avoid rater bias. The
inter-rater reliability was determined by conducting Cohen’s κ (κ = .83, p < .0005), which
showed the strong rater agreement (81.9%). When raters’ judgments differed by a single point,
an average score was assigned. Discrepancies of two points or more were discussed until
consensus was reached. Non-parametric tests were used because the posttest data did not confirm
To address the second research question, the researchers conducted a content analysis of
the written interview transcriptions. Participants’ responses to each interview question were read
together and analyzed for recurring patterns, which were summarized as notable trends.
Learning and Retention of Formulaic Knowledge After the Scenario-based Interactive Practice
Table 1 displays descriptive statistics of pre-, post-, and delayed post-test results (n=30).
Descriptive statistics of comprehension and production test scores
Note. Score range of the comprehension test was 0–28 (28 items total), while that of the production test
was 28–168 (28 items total, each scored on the range of 1–6).
On the comprehension test, participants made sharp progress after just two practice
gaming sessions, and they largely maintained this gain two weeks later at the delayed posttest.
This was confirmed by the Friedman test. There was a significant difference across the three test
sessions (χ2 = 51.6, p < .0005). The Wilcoxon test detected a difference between the pre- and
posttest and between the pre- and delayed posttest (Pre-Post; Z = -4.8, p < .0005; Pre-Delay: Z =
-4.8, p < 0.0005), but detected no difference between the post- and delayed posttest, meaning that
the learning was maintained.
Table 2 provides data on the distribution of error options among the multiple-choice
responses on the comprehension test (see the methods for the definition of each error type). At
all three testing points (pre-, post-, delayed posttest), participants who did not select the correct
response the first time most frequently selected the option containing a pragmalinguistic error,
followed by the option containing a sociopragmatic error. The percentage of both error types
decreased after the scenario-based interactional practice, but the degree of decrease was much
larger for sociopragmatic errors than for pragmalinguistic errors. Throughout the tests, response
options containing a sequential error were infrequently chosen.
Distractor error analysis in the comprehension test
Turning to the production data, participants showed a gain from the pre- to the posttest,
but their scores dropped at the delayed posttest. The Friedman test revealed a significant
difference across test sessions (χ2 = 49.6, p < .0005). The Wilcoxon test detected a significant
difference between the pre- and the posttest (Z = -4.8, p < .0005), and the pre- and the delayed
posttest (Z = -4.7, p < 0.0005). Unlike the comprehension results, there was a difference between
the post and the delayed posttest (Z = -3.7, p < 0.0005), indicating that the learning was not
maintained two weeks later.
A closer analysis of the production data showed variations in the degree of learning
among formulaic expressions. For example, the expression
[A littler cheaper], became
target-like after the practice as shown in the following sample data from one participant.
[I don’t have that much money. (It’s) impossible to be that expensive.]
[A littler cheaper]
[A littler cheaper]
The utterance produced at the pretest could be considered as a hint because it did not directly
convey the communicative function of “bargaining for a lower price”. After two sessions of
practice, the native-like expression emerged at the posttest and was retained at the delayed
In contrast, the expression
([I’m just looking] when refusing a shop
assistant’s offer of help did not remain native-like at the delayed posttest, as shown in the
following data from the same participant:
[Hello, I want to take a look. I don’t have any questions.]
[I’m just looking.]
[I will first take a look.]
Although the participant’s utterance became target-like at the posttest, he was not able to retrieve
the correct form at the delayed posttest. A portion of the expression was native-like (i.e.,
meaning “take a look”), but the utterance was not entirely correct because both
convey the same meaning and it is incorrect to use both parts together.
Interview data were analyzed for content based on the questions encompassing several
areas: aspects of the practice that participants liked/disliked or felt useful/not useful, whether the
participants felt immersed, or engaged, in the scenes, and the extent to which they considered the
experience to be a game. Several notable trends were revealed.
The majority of participants reported that contextualization was their favorite aspect of
the practice experience (15 responses). This approach integrated textual, visual and auditory cues,
creating a multi-modal exercise, which further allowed learners to practice comprehension and
production in an integrated manner. On the other hand, the production practice (“putting-it-
together”) was revealed to be the least favorite part of the learning experience: Eight learners
mentioned the high cognitive demand due to the need to recall the utterances and type the
responses in Chinese. A larger number of participants (12) requested more diverse situations
beyond the ten that were included. Five of the 24 interviewees also requested greater agency, for
example, the option to select a different story line or create their own character.
In terms of the feeling of engagement with the characters, nine participants did not
express strong feelings on this topic. Seven participants felt that the learning experience offered
an immersive experience in which they could interact with the characters directly. Comments
from these participants suggest that the angle from which each video was recorded (one main
character who made direct contact with the game participant by looking the video camera and
speaking directly to the camera) contributed to the immersiveness of the learning experience.
For example, one participant responded, “[I felt like I was] interacting with the characters
because I was the only person. The perspective was me.” The use of authentic scenes that
captured real-life people in everyday situations in Shanghai helped to generate a strong sense of
authenticity and cultural immersion.
However, eight participants reported feeling that they were observing the situation as a
third person or playing someone else’s character rather than their own. Interview data suggest
that the appearance of the multiple-choice questions in the middle of the interaction, which
halted the video after the character’s utterance or question until the participant had selected the
correct response from among those offered, limited the extent to which the experience could be
viewed as a real-life interaction in which participants were able to respond freely using their own
words and seamlessly continue their conversation. Participants reported that the appearance of
the multiple-choice questions was “disruptive” and “exercise-like”. In addition, several
participants reported that they wanted to produce their own responses orally, rather than select an
option, because the pre-determined options were not personalized or realistic; one student
pointed out that, “in real life there are several ways of saying things”.
In terms of the game-like feeling, 10 participants reported that they had a game-like
experience thanks to features that allowed them to select their own character; navigate through
the map to go to different locations; receive feedback via the point system; and interact with
video-based characters. However, 13 participants felt that the experience was clearly a learning
activity and was not “gamey;” it was instructional but not recreational due to the multiple-choice
questions and production practice (“putting-it-together”), which resembled classroom exercises.
Participants also mentioned that several other factors that made the game “un-playable”. For
example, the point system was not motivating because gaining points was not linked to any
extrinsic reward or part of a larger competitive framework. As some participants commented that
a good reward system is crucial to make a game re-playable, that is, “after accumulating points,
you can use points to buy something or move to a different level.” Similarly, some participants
reported that there was “not much stake” and no “penalty” was imposed for wrong answers.
They suggested that adding some time constraints or a ranking system would make the
experience more competitive. In addition, the small number of scenarios made the “game”
repetitive, a limitation that could be eliminated by adding more situations in which players could
“learn new things,” and offering a more coherent story line that allowed more personalized
decision-making options (i.e., going through specific scenarios to achieve individual goals).
In summary, interview data showed that, while participants enjoyed learning formulaic
expressions by means of a scenario-based, semi-interactive experience, their perceptions of the
immersive quality and game-like nature of the experience varied. While the videos recorded in
real-life contexts in Shanghai made the practice more fun and engaging and clearly distinguished
it from textbook-based learning, the response formats (multiple-choice and production practice)
detracted from the immersive-feel and the repetitive format, limited range of scenarios and
absence of a reward system made the practice more like a learning activity than a re-playable
This study reported on a scenario-based interactional platform whereby L2 Chinese
learners completed a dialogue with a character in a video featuring a scene in Shanghai.
Formulaic expressions (targeted pragmatic knowledge) served as linguistic resources that
participants needed to complete each dialogue and progress through the task. Accumulation of
formulaic knowledge was considered to be a by-product of the participants’ successful
completion of the goal-oriented, interactive scenarios. Various game-like features incorporated
into the task design—plot and setting, role playing, competition (levels and rewards), timely
feedback—were expected to make this form of task-based practice fun and re-playable, thus
promoting the participants’ engagement in the task. These design features were incorporated to
address gaps in the existing practice of pragmatics teaching (Taguchi, 2015), which are largely
limited in contextualization, interaction, and affective considerations (e.g., motivation, learner
Indeed, learning of formulae did occur as a by-product of the practice. Initially, the
participants’ receptive knowledge of target expressions was only moderate (about 60% on the
comprehension test), but it jumped to 90% and remained at the same level two weeks later. Post
hoc error analyses showed that the participants’ sociopragmatic errors decreased the most,
suggesting that the learning experience helped them recognize the relationship between the form
and the context of use. Similarly, the participants’ productive knowledge showed a strong gain,
shifting from 64% at pretest to almost 80% at posttest.
Formulaic expressions are tied to recurrent communicative situations and functions (e.g.,
Bardovi-Harlig, 2012; Wray, 2002). Hence, contextual parameters (settings, speakers’ roles,
topics, and communicative goals) need to be mapped onto forms in order to facilitate learning
and subsequent effective language use. Given their community-wide use, formulaic expressions
can be learned effectively when they are presented as occurring in a local community. Video
clips featuring local neighbourhoods, combined with simulated interactions with community
members, probably made the form-function-context mappings more salient for participants and
supported the retention of the expressions in participants’ long-term memory. The transition
from comprehension-based to production-based practice was designed to help participants re-
produce the expressions and to revise their work when errors were made. This learning sequence
involved multiple modalities (comprehension and production), input cues (visual, auditory, and
textual), and contextual representations (e.g., ordering at a restaurant, asking directions at a
major sightseeing spot), all of which probably contributed to strengthening participants’ learning
of formulae, and supported access and retrieval two weeks after the learning experience was
However, the findings showed that knowledge when measured in the production test
was not equally robust. The participants achieved almost 80% on average at the immediate
posttest, but the rate dropped to 74% at the delayed posttest. Unlike the comprehension test
results, the difference between the two tests was statistically significant. These findings
corroborate previous findings that formulaic knowledge does not develop in parallel in the
interpretive (comprehension) and presentational (productive) modes (e.g., Bardovi-Harlig, 2009;
Bardovi-Harlig & Bastos, 2011; Taguchi, 2011, 2013). Furthermore, while some existing
findings indicate that learners’ comprehension of formulae is supported by immersive
experiences abroad, this has not been shown to be the case regarding production: Taguchi’s
(2011, 2013) studies found no study abroad advantage in the production of formulae, although
such an advantage was found for comprehension. L2 learners also demonstrated stronger
formulaic knowledge in a comprehension task than in a production task (Bardovi-Harlig, 2009).
Unlike previous studies conducted in a naturalistic (uninstructed) setting, this study
implemented focused instruction with the clear objective of improving formulaic knowledge in
both the interpretive (receptive) and presentational (productive) modes. Still, the modality effect
on instructional outcomes found in the study suggests that, just as residence abroad is no panacea
to the productive knowledge of formulae, instructional effects may not manifest strongly in
production, either. This is the case even when the instruction is designed to maximize the
understanding of formulaic expressions as attempted in this study (e.g., contextualized input
promoting the saliency of form-function-context associations, incorporation of the community-
The rather weak instructional effect on production found in this study is understandable
considering the linguistic and cognitive demands posed by production. Although learners can
comprehend overall meaning without a bottom-up analysis of forms, production requires precise
linguistic processing. Furthermore, linguistic preciseness is particularly critical for formulaic
expressions because incorrect word order or word choice can obscure meaning. For example, the
difference between the questions “Do you have the time?” and “Do you have time?” is the matter
of only one word (“the” in the former), but their meanings and communicative functions are
totally different. As Pawley and Syder (1983) claimed, formulaic competence requires “native-
like selection”— the ability to select the exact strings of preferred forms, and native-like
selection is revealed more clearly in production than in comprehension because production data
show learners’ exact choice of lexis and grammar. Word choice errors and grammatical mistakes
found in the participants’ production data in this study (see examples in the results) indicate that
participants had difficulty producing the exact linguistic strings that are preferred among local
members of the Chinese-speaking community.
Perceptions of gaming features
In addition to learning gains, interview data from this study demonstrated that
technology that is readily accessible (videos, instructional software) can be used creatively to
generate focused and realistic, semi-immersive contexts for pragmatic language use. However, as
noted above, a number of participants nominated the production practice as their least favorite,
commenting that this portion of the experience was “difficult”, “boring”, and “time consuming”.
First, unlike the comprehension practice that was embedded within a culturally-rich, video-based
interaction, the production practice was presented as a stand-alone fill-in-the-blank exercise. It is
possible that this format failed to engage participants or motivate them to pursue precise forms.
In addition, unlike the comprehension portion of the experience, focused feedback was not
offered on the forms that participants produced. After filling in blanks, they simply had to hit the
“submit-and-compare” button to pull up the model responses and then copy the responses
verbatim. Since rewards (i.e., points) were not offered upon completion, it is possible that the
lack of game-generating features in the production task (no rewards, no feedback, no interaction)
made the production task less engaging, leading to the frail knowledge base that was revealed on
the delayed post-assessment.
Furthermore, although the study incorporated Sykes and Reinhardt’s (2012) components
for designing games for educational purposes (e.g., feedback, rewards, plot and settings), the
data suggest that the mere existence of those components does not automatically lead to a game-
like experience. The key seems to be about how to implement those components to generate
game-like feelings and behaviors (e.g., re-playablity, engagement, goal-orientation, motivation).
For example, while a reward system (i.e., points) was incorporated in the design, the system was
not motivating for some participants because of its low-stakes nature and the fact that points
served no instrumental purposes (i.e., using points to accomplish something within or beyond the
game). Similarly, although the characters, plot and setting served to connect the ten interactive
situations, several participants reported that the plot was not coherent -- success or failure in any
of the ten stand-alone situations did not have down-stream implications or consequences as the
learning experience progressed. Allowing participants to create their own character and produce
their own responses rather than selecting from the options provided would also promote
participants’ originality and autonomy. The feedback system could also be improved: Feedback
was provided in a fixed, linear format, appearing in a box every time a correct answer was
selected or a mistake was made. One participant mentioned that she did not even notice the
differences among feedback types (e.g., pragmalinguistic and sociopragmatic feedback) because
they all looked the same to her. These comments suggest that how feedback is given (e.g.,
explicitness, manner, timing, format) is important in generating a game-like feeling. These
operational details also affected the participants’ perceptions of immersiveness. Although scenes
were filmed from the participant’s angle and incorporated turn-taking mechanisms (taking turns
with the characters in the video by exchanging responses), some participants commented that it
was hard to feel immersed because ultimately the participant’s interaction with the character was
limited by the multiple-choice format and feedback mechanisms, which disrupted the
conversational flow. Thus, the way in which game design components are actualized, rather than
the components themselves, needs to be addressed. Playfulness and re-playability make a game
attractive (Sykes & Reinhardt, 2012). Lacking these features, participants will not choose to
“play” over and over again and thus accumulate the massive amounts of practice that lead to
robust learning. In conclusion, synchronizing content learning with learners’ affective and
behavioral needs has the potential to transform a mundane language practice experience into an
engaging, self-guided learning experience that leads to more robust and sustained learning.
This study offers important pedagogical insights into the impact of video and gaming
experiences on learning outcomes. Since pragmatics involves using language appropriately to
carry out communicative functions across a range of social contexts, it can only be learned in
deeply-contextualized environments. Embedding language within culturally-rich interactions, for
example, using video to depict the many possible ways in which a situation could be negotiated,
is thus essential to the teaching and learning of pragmatics. Although videos nearly always are
provided by textbook publishers and are commonly used, students are usually positioned as
observers rather than participants. In the observer role, students are not expected, or even invited,
to analyze the socio-cultural context, make active linguistic choices, or engage in turn-taking.
Editing to remove one speaker’s lines in a video-recorded conversational exchange, as was done
in this study, or simply stopping the video at the end of a speaker’s turn, can set the stage for
class discussion about what could, or should, be said and how the subsequent turn at talk would
be perceived by the other interlocutor(s) in that particular cultural and social context. As
demonstrated in this study, exploring different ways of using video can promote situated,
interactive practice and make language learning more participatory, meaningful and effective. In
addition to the use of video, the study also demonstrates the efficacy of a multi-modal approach.
Using multi-modal input (texts, audio, graphics) and output (reading, writing, and listening) not
only reinforced learning, but also offered a more accessible and focused approach than short- or
long-term, or even repeated study abroad experiences. Furthermore, a multi-modal, video-based
approach is also more accommodating of students with different learning styles.
The study also has implications for gaming and game-like learning. The strong learning
outcomes that were observed among the participants in this study, along with their positive
perceptions of game-based learning, indicate that incorporating various gaming features (e.g.,
rich contextualization cues, interactions with video-based characters, autonomy in choosing a
character and the order in which the ten scenes would be enacted, and the point system as a
means of receiving immediate feedback) can motivate learners to play-- and hopefully re-play --
real-life interactions across multiple, varied, and socially-nuanced iterations. What is more, out-
of-class game-based opportunities to learn and practice essential language content and analyze
how and with whom particular expressions should be used also free up face-to-face instructional
time for other communicative purposes. The findings also point out that designers of virtual
learning environments must actively consider the extent to which affective factors (e.g.,
motivation, drive, and playfulness) are incorporated into the game, while simultaneously
ensuring that learners acquire the language skills that they need to meet learning targets and
Implications for future research
In future studies, researchers and practitioners should also consider ways to address
learners’ use of formulaic expressions in the interpersonal mode. Although it is challenging for
speech recognition software to identify variations in a spoken utterance, this is increasingly
feasible once all of the situations and their corresponding formulaic expressions have been stored
in the database. Additionally, in this study learners participated in the same 10 scenarios over
two practice sessions. While it is possible that the repetitive practice reinforced participants’
understanding of formulaic expressions, direct repetition also reduced learners’ use of these
expressions to a limited number of situations and rendered their interactions less game-like.
Since, in a real game, players always encounter new situations and challenges as they progress
through in-game tasks, future research should include more scenarios and more diverse
Finally, this study presented formulaic expressions as a series of syntactic strings that
need to be memorized through recognition and production-based practice. Although this
instructional approach is fitting for the construct of formulae (chunks and fixed expressions),
future research can explore a different approach that can strengthen learners’ memory of
formulaic expressions. There are several exemplary approaches (e.g., Boers, Eyckmans, Kappel,
Stengers, & Demecheleer, 2006; Boers & Lindstromberg, 2009). In Boers, et al. (2006), the
authors exposed L2 English learners to authentic language use (audio, video, and textual) while
directing their attention to formulaic sequences in the texts. The tasks that strategically promoted
noticing of formulaic sequences (i.e., gap-fill exercises focusing on formulae; exercises of
highlighting frequent word combinations) were found to be effective in promoting learners’
knowledge of formulaic expressions. Boers and Lindstromberg (2009), on the other hand,
presented a variety of activities, for example formulae elaboration activities that made use of
images to present the expressions and formulae consolidation activities that used rhymes
associated with the expressions. Future research can incorporate these instructional approaches
into the game-based interactive environment that was used in this study.
This study created an interactive participatory platform incorporating a virtual reality-
feel and some elements of game playing to advance the current practice of pragmatics teaching.
The platform was designed in a way that L2 learners of Chinese interacted with characters in a
simulation-based, multi- modal space while using target pragmatic forms (i.e., formulaic
expressions). The study revealed three major findings: (1) L2 Chinese learners showed strong
gains in their knowledge of formulaic expressions after two game-based learning sessions; (2)
learners retained their gains two weeks later, although their retention was more prominent in the
interpretive (comprehension) mode than in the interpersonal (production) mode; and (3) learners
had generally positive feelings about the scenario-based, semi-interactive experience, but they
expressed mixed feelings about the immersive quality and game-like nature of the experience
due to: (a) the exercise-like response format (i.e., multiple-choice questions, fill-in-the-blanks);
(b) limited options in the course of action they were allowed to take; (c) limited variations in the
situational scenarios; and (d) the absence of a truly motivating reward system. These findings
should inform future game development and research. Because empirical data are considerably
limited regarding the cause-and-effect relationship between learners’ participation in virtual
gaming and increased pragmatic knowledge, more research is clearly needed. Critically, such
research should employ a systematic experimental design to directly assess learning outcomes
resulting from game play. In addition, while virtual learning games can seek to incorporate the
features of recreational games, they also need to incorporate clear learning objectives in order to
ensure that the types of learning that are expected from game-based interaction actually occurs.
Bardovi-Harlig, K. (2009). Conventional expressions as a pragmalinguistic resource:
Recognition and production of conventional expressions in L2 pragmatics. Language
Learning, 59, 755–795.
Bardovi-Harlig, K. (2012). Formulas, routines, and conventional expressions in
pragmatics research. Annual Review of Applied Linguistics, 32, 206-227.
Berns, A., Gonzalez-Pardo, A., & Camacho, D. (2013). Game-like language learning in 3-D
virtual environments. Computer & Education, 60, 210–220.
Boers, F., Eyckmans, J., Kappel, J., Stengers, H., & Demecheleer, M. (2006). Formulaic
sequences and perceived oral proficiency: Putting a lexical approach to the test.
Language Teaching Research, 10, 245–261.
Boers, F., & Lindstromberg, S. (2009). Optimizing a lexical approach to instructed second
language acquisition. New York: Palgrave MacMillan.
Cornillie, F., Thorne, S., & Desmet, P. (2012). Special issue: Digital games for language
learning: Challenges and opportunities. ReCALL, 24, 243-256.
Gee, J. P. (2007). Good video games and good learning. New York: Peter Lang.
Holden, C. & Sykes, J. M. (2013). Complex L2 pragmatic feedback via place-based mobile
games. In N. Taguchi & J. M. Sykes (Eds.), Technology in Interlanguage pragmatics
research and teaching (pp. 155–184). John Benjamins, Amsterdam/Philadelphia.
Liou, H-C. (2012). The role of Second Life in a college computer-assisted language learning
(CALL) course in Taiwan, ROC. Computer Assisted Language Learning, 25, 365–382,
Milton, J., Jonsen, S., Hirst, S., & Lindenburn, S. (2012). Foreign language vocabulary
development through activities in an online 3D environment. The Language Learning
Journal, 40, 99–112.
Pawley, A. & Syder, F. (1983). Two puzzles for linguistic theory: Native-like selection and
native-like fluency. In J. Richards & R. Schmidt (eds.), Language and communication
(pp. 191-226). London: Longman.
Peterson, M. (2005). Learning interaction in an avatar-based virtual environment: A preliminary
study. PacCALL Journal, 1, 29–40.
Peterson, M. (2011). Towards a research agenda for the use of three-dimensional virtual worlds
in language learning. CALICO Journal, 29, 67–80.
Reinhardt, J., & Sykes, J. (2014). Special issue on game-informed L2 teaching and learning.
Language Learning & Technology, 18. 2-8.
Roever, C. (2005). Testing ESL pragmatics. Frankfurt: Peter Lang.
Schwienhorst, K. (2002). Why virtual, why environments? Implementing virtual reality concepts
in computer-assisted language learning. Simulations & Gaming, 33, 196–209.
Si, M. (2015). A virtual space for children to meet and practice Chinese. International Journal of
Artificial Intelligence in Education, 25, 271–290.
Squire, K. (2009). Mobile media learning: multiplicities of place. On the Horizon, 17, 70-80.
Sykes, J. M. (2009). Learner request in Spanish: Examining the potential of multiuser virtual
environments for L2 pragmatics acquisition. In L. Lomika & G. Lord (eds.), The second
generation: Online collaboration and social networking in CALL (pp. 199–234).
CALICO Monograph. San Marcos, TX: CALICO.
Sykes, J. M. (2013). Multiuser virtual environments: Apologies in Spanish. In N. Taguchi & J. M.
Sykes (Eds.), Technology in interlanguage pragmatics research and teaching (pp.71–
100). Amsterdam/Philadelphia, John Benjamins.
Sykes, J. M., & Reinhardt, J. (2012). Language at play: Digital games in second and foreign
language teaching and learning. New York: Pearson-Prentice Hall.
Taguchi, N. (2011). The effect of L2 proficiency and study-abroad experience in pragmatic
comprehension. Language Learning, 61, 904-939.
Taguchi, N. (2015). Instructed pragmatics at a glance: Where instructional studies were, are, and
should be going. Language Teaching, 48, 1–50.
Taguchi, N. (2013). Production of routines in L2 English: Effect of proficiency and study-abroad
experience. System, 41, 109-121.
Taguchi, N., Li, S., & Xiao, F. (2013). Production of formulaic expressions in L2 Chinese: A
developmental investigation in a study abroad context. Chinese as a Second Language
Research Journal, 2, 23–58.
Taguchi, N., & Roever, C. (2017). Second language pragmatics. New York: Oxford University
Taguchi, N., & Sykes, J. M. (2013). Technology in interlanguage pragmatics research and
teaching. Philadelphia: John Benjamins.
Thomas, J. (1983). Cross-cultural pragmatic failure. Applied Linguistics, 4, 91–111.
Wik, P., & Hjalmarsson, A. (2009). Embodied conversational agents in computer assisted
language learning. Speech communication, 51, 1024-1037.
Wood, D. (2006). Uses and functions of formulaic sequences in second language speech: An
exploration of the foundations of fluency. Canadian Modern Language Journal, 63, 13-
Wray, A. (2002). Formulaic language and the lexicon. Cambridge: Cambridge University
Target formulaic expressions
Does (this bus) go to …?
Ask if this bus goes to a specific
Which bus goes to …?
Ask which bus goes to a specific
How much is …?
Ask for price
A little cheaper!
Bargain for a lower price
I’m sorry, I have to leave early.
Apologize for leaving early
I’m sorry. I’m late.
Apologize for being late
I’m sorry (for my absence).
Apologize for being absent
OK, I’m leaving.
Say goodbye (variant forms)
See you tomorrow!
Thank you for inviting me to dinner!
Express gratitude for an invitation
Thank you so much!
Express gratitude in formal
Express gratitude in informal
How have you been?
Greet with small talk
Excuse me, is this seat taken?
Ask for the availability of a seat
May I sit here?
Make a request (variant forms)
May I try it on?
May I …?
Excuse me, how to get to…?
Ask for directions
How long does it take?
Ask for distance
Refuse help from a shop assistant
What would you like to eat?
Ask for food preference
… is good here (a restaurant)?
Recommend a dish in a restaurant
Wait/Waitress, (we) want …!
Ask for warping up.
Is that … speaking (Caller requesting
who answered the phone)?
Confirm the identity of the
receiver on the phone
(Excuse me,) when will you be free?
Ask for someone’s availability