International Forum of Educational Technology & Society

The Effects of Using the Kinect Motion-sensing Interactive System to Enhance English
Learning for Elementary Students
Author(s): Wen Fu Pan
Source: Journal of Educational Technology & Society , Vol. 20, No. 2 (April 2017), pp. 188-
200
Published by: International Forum of Educational Technology & Society
Stable URL: https://www.jstor.org/stable/10.2307/90002174

JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide
range of content in a trusted digital archive. We use information technology and tools to increase productivity and
facilitate new forms of scholarship. For more information about JSTOR, please contact [email protected].

Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at
https://about.jstor.org/terms

International Forum of Educational Technology & Society is collaborating with JSTOR to digitize,
preserve and extend access to Journal of Educational Technology & Society

This content downloaded from

5.90.4.172 on Thu, 03 Dec 2020 16:57:01 UTC
All use subject to https://about.jstor.org/terms
Pan, W. F. (2017). The Effects of Using the Kinect Motion-sensing Interactive System to Enhance English Learning for
Elementary Students. Educational Technology & Society, 20 (2), 188–200.

The Effects of Using the Kinect Motion-sensing Interactive System to

Enhance English Learning for Elementary Students
Wen Fu Pan
Department of Educational Administration and Management, National Dong-Hwa University, Taiwan //
[email protected]

ABSTRACT
The objective of this study was to test whether the Kinect motion-sensing interactive system (KMIS)
enhanced students’ English vocabulary learning, while also comparing the system’s effectiveness against a
traditional computer-mouse interface. Both interfaces utilized an interactive game with a questioning
strategy. One-hundred and twenty participants were chosen from an elementary school. The students were
divided into three groups: Kinect, computer-mouse, and control. The participants’ vocabularies were
evaluated three times during a pre-test, a post-test, and a 1-month post-test. The following results were
obtained: (1) there was a partially disordinal interaction relationship between the three groups and the three
tests. Post-hoc comparison showed that the three tests have an order relationship. (2) The within group
comparisons, both for the motion-sensing and computer-mouse groups which utilized an interactive game
with a questioning strategy, displayed a relatively significant long-term retention. (3) In the between group
comparison, the two interactive groups (computer-mouse and motion-sensing group) did not reach
significant difference in English vocabulary learning. This means the motion-sensing interface of the KMIS
was not a key-factor to affecting short-term or long-term learning retention. Therefore, our suggestion is
that teachers can adopt interactive games with a questioning strategy to enhance students’ long-term English
vocabulary retention.

Keywords
Computer assisted learning, Human-computer interaction, Kinect sensor, Motion-sensing systems,
Vocabulary learning

Introduction
Traditional one-way rote memorization method for learning English vocabulary is frequently found in schools
(Smith, Li, Drobisz, Park, Kim, & Smith, 2013). However, the effects of such methods have not been found to be
better than interaction (Ge, 2015). Vygotsky (1978) considered that in second language acquisition, learners need
to interact with the socio-cultural environment via artifacts. These artifacts are referred to as “interfaces”
between the subject and object from the viewpoint of human-computer interaction (Engeström, 2000).

Early human-computer interface (HCI) studies mostly adopted usability testing (Buur & Bødker, 2000). In the
1990s, some scholars began to cite activity theory, proposed by Leont’ev in the 1930s, as a theoretical
framework of HCI design (Kaptelinin, 1996; Kuutti, 1996; Nardi, 1996). Activity theory emphasizes how to
construct meaning from interaction between subject and object via artifacts (such as rules, books, etc.) (Leont’ev,
1974). Subsequently, activity theory also became one of the theoretical frameworks for language learning
(Oxford, 1990). In 2003, Bedny and Karwowski divided activities into the following five levels: activity, task,
action, operation, and function; they also incorporated two design types: subject-oriented and object-oriented,
based on their proposed Systemic-Structural Theory of Activity (Bedny & Harris, 2005).

Subject-oriented design focuses on a subject’s socio-cultural context and has often been adopted by studies of
second language learning (Chapelle, 2009). On the other hand, in order to assess the usability of an emerging
technology, researchers have often adopted object-oriented design (Munassar & Govardhan, 2011). This study
also adopts a type of object-oriented design called “object-mental action” from activity theory. Specifically, a
subject (the learner) interacts with an object (game-based animation) via the Kinect Motion-sensing Interactive
System.

Edgar Dale’s cone of experience theory indicates that two-way interactive learning helps learners to obtain up to
90% learning retention (Dale, 1969). Human-computer interaction also benefits learning retention (Papastergiou,
2009; Prensky, 2005); however, is this effect derived from the human-computer “interactive content,” or
“operating interface”? This question is worthy of further research. Therefore, in this study, we designed a game-
based learning activity as the interactive “content” and a motion-sensing operation as the interactive “interface”
for English vocabulary learning. The related literature is reviewed as follows.

ISSN 1436-4522 (online) and 1176-3647 (print). This article of the Journal of Educational Technology & Society is available under Creative Commons CC-BY-ND-NC
188
3.0 license (https://creativecommons.org/licenses/by-nc-nd/3.0/). For further queries, please contact Journal Editors at [email protected].

This content downloaded from

5.90.4.172 on Thu, 03 Dec 2020 16:57:01 UTC
All use subject to https://about.jstor.org/terms
Applying game-based learning with a questioning strategy as interactive content

Previous research has shown that game-based English learning has resulted in better retention than traditional
rote memorization (Flores, 2015). Hwang, Chiu, and Chen (2015) also indicated that game-based learning is able
to improve students' inquiry-based learning performance, especially in an interactive environment. Also,
enjoying the game was cited as an important reason why students were willing to finish interactive tasks (Star,
Chen, & Dede, 2015). The design of digital games is an important and often used method for enhancing learning
motivation. A learner’s motivation to participate is enhanced through gamed-based learning (Birk, Atkins,
Bowey, & Mandryk, 2016; Ronimus & Lyytinen, 2015). The goal of this study is to design an English
vocabulary learning activity that integrates digitized game-based interaction.

In addition, a questioning strategy was implemented in this study to enhance the two-way interactive learning.
The questioning strategy is defined as actively presenting a question and waiting for the students’ answer.
Research has indicated that implementing a questioning strategy in English learning can also result in better
retention (Basturkmen, 2001; Boyd & Rubin, 2006; Shomoossi, 2004; Yang, 2010).

Applying motion-sensing operation as an interactive interface

Developments in emerging technology have transformed the types of interaction between humans and computers.
Past research (Chuang & Kuo, 2016; Hsiao & Chen, 2016; Sheehan & Katz, 2012) has shown that applying
various motion-sensing technology to learning environments benefits students. Microsoft released the source
code of Kinect (3D depth sensor) in 2012, and since then there has been a lot of development in motion-sensing
applications. Currently all short-distance motion (or gesture) sensors use either infrared light emitters and
sensors, ultrasonic sensors, or 3D depth sensors (Kinect) (Kumaragurubaran, 2011). The principle of the Kinect
motion-sensing technique is that it employs three lenses, as well as a diffuser lens to expand or diffuse projected
laser speckles. For the speckles that reach the human body, a separate camera coordinated with a light coding
technique is employed to collect the 3D depth of field information regarding the human body within a 5-m
tapered space (Pan, Chien, & Tu, 2012a). Pan et al. (2012a) compared the differences between infrared,
ultrasonic, and Kinect sensing techniques. Applying Kinect in learning offers the following benefits: (1) it does
not require a handheld controller; (2) it provides real-time feedback; (3) it is able to distinguish humans from
objects; (4) it provides teachers (or developers) with a way to customize interactive content.

A lot of research has been conducted and is ongoing in applying Kinect technology to various fields (Nissimov,
Goldberger, & Alchanatis, 2015; Yao, Wang, Cai, & Zhang, 2015). Kinect motion-sensors have been integrated
into interactive learning, and research in this area has become a growing trend (Chuang & Kuo, 2016). For
example, Sommool, Battulga, Shih, and Hwang (2013) applied Kinect motion-sensing technology to create and
evaluate interactive learning classrooms; Tutwiler, Lin, and Chang (2013) applied it to multiple intelligence
instruction; and Levinger, Zeina, Teshome, Skinner, Begg, and Abbott (2016) utilized Kinect in gait practice for
knee replacement rehabilitation. In recent years, Pan led a team focused on the development of Kinect
applications for educational situations (Pan, Tu, & Chien, 2014). Their research covered a range of applications
including campus safety (Pan, Chien, Liu, & Chan, 2012b), accessible learning (Pan et al., 2012a), and
interactive learning (Pan, Lin, & Wu, 2011). Pan (2013) indicated that the Kinect motion-sensing interface can
better enhance students’ learning motivation compared to a more traditional computer-mouse interface. Some
studies (Sommool et al., 2013; Vrellis, Moutsioulis, & Mikropoulos, 2014; Yuan, Hsieh, Chew, & Chen, 2015)
have also supported the idea that the novelty of the Kinect system can attract students’ attention and increase
learning motivation.

Concept framework of this study

Based on the above literature review, the concept framework of this study is shown in Figure 1.

A one-way interaction rote memorization method is frequently used in schools for learning of English
vocabulary (Smith et al., 2013). In order to improve learning methods, two-way interactive learning and human-
computer interfaces should be applied based on activity theory. The Kinect Motion-sensing Interactive System,
or KMIS, designed in this study includes two parts: (1) applying game-based learning with a questioning
strategy as interactive content; (2) applying the Kinect motion-sensing operation as an interactive interface. Past
research (Pan, 2013; Sommool et al., 2013; Vrellis et al., 2014; Yuan et al., 2015) regarding Kinect motion-
sensing applications in learning has usually adopted an empirical method, so an experimental design was used in

189

This content downloaded from

5.90.4.172 on Thu, 03 Dec 2020 16:57:01 UTC
All use subject to https://about.jstor.org/terms
this study. In addition, although some studies (Pan, 2013; Vrellis et al., 2014; Yuan et al., 2015) have also made
the comparisons between Kinect motion-sensing and computer-mouse interface, these studies did not focus on
the interactive relationship of the learning “content” and “interface,” or its effects on English learning. Therefore,
through this experimental design, the researcher wants to assess the within-group (interactive content) and
between-group (operating interface) interactive relationship, as well as compare their effects on short-term and
long-term English vocabulary retention. It was expected that the findings from this KMIS design would show
improvement over one-way rote memorization learning.

Activity

Task

Function

Figure 1. The concept framework of using the KMIS to improve English vocabulary learning

Objectives of the study

According to the above analysis, this study has the following three research purposes:
● To do statistical analyses on the interactive relationship between three groups (no-interaction, Kinect
motion-sensing and computer-mouse) and three tests (pre-test, post-test and delayed post-test).
● To analyze the performances on the three tests (pre-test, post-test and delayed post-test) in interactive
learning with an integrated questioning strategy game for English vocabulary learning.
● To compare the short-term and long-term retention effects on English vocabulary learning in the three
groups.

Methods
Experimental design

The quasi-experimental design was adopted as shown in Table 1. The subjects of the experiment were divided
into three groups: motion-sensing, computer-mouse, and the control group (no interaction). The motion-sensing
group (X1) was tested with the KMIS and the computer-mouse group (X2) was tested with a traditional
190

This content downloaded from

5.90.4.172 on Thu, 03 Dec 2020 16:57:01 UTC
All use subject to https://about.jstor.org/terms
computer-mouse interface. Both the motion-sensing group (X1) and computer-mouse group (X2) had the same
interactive content (a football game with a vocabulary quiz). Each group contained forty 6 th grade students, who
each completed the English vocabulary test three times: a pre-test (O1), a post-test (O2), and a 1-month delayed
post-test (O3). Types of interaction (three groups) and the three tests were the two factors in data analyses for a
two-way mixed-design ANOVA. The two-way mixed-designed ANOVA was a better choice than analysis of
covariance (ANCOVA) in this case because it was not only able to analyze the interactive relationships of three
groups and three tests, but also able to compare the within-group and between-group differences.

Table 1. Two-way mixed-design structure (N = 120) with distinct types of interaction and tests
Experiment group N Pre-test Treatment Post-test Delayed post-test
Control group 40 O1 - O2 O3
Computer-mouse group 40 O1 X1 O2 O3
Motion-sensing group 40 O1 X2 O2 O3
Note. O represents a 25 multiple-choice question test; the motion-sensing group (X2) used the Kinect motion-
sensing interactive system (KMIS) as the interactive interface

Research subjects

Six classes were randomly selected as research subjects with cluster sampling from 6 th grade classes at a large-
scale elementary school in Hualien City, Taiwan. Then two classes were randomly assigned to each of the three
groups. Each group had 40 students averaging 12 years old, and the three groups had a total of 120 participants.
Each group’s participants had similar academic performance in school and the subjects of the three groups were
examined by the homogeneity test. Statistically, the Box’s test did not reach the level of significance (F = 1.522,
p = .108) and the F-test for the pre-test did not reach the level of significance for difference (F = .22, p = .80,
shown in table 6). Therefore, the basic background of subjects and environmental factors of the three groups
were considered homogenous, and the two-way mixed-designed ANOVA was adopted. The experiment and test
data were collected from October to December of 2013.

Development of the English vocabulary cognition test

The English Vocabulary Cognition Test (EVCT) for 6 th grade students was created for evaluating students’
learning performance. To establish the content of the test, forty words were randomly sampled from the “1200
English Vocabulary Words” endorsed by the Ministry of Education in Taiwan for elementary and junior high
school students. These words were drafted into 40 multiple-choice questions for a test for 6th grade students in
another elementary school in Hualien City, Taiwan. The 169 valid tests were ordered by scores, and the top and
bottom 27% of the tests were grouped into high-score and low-score for item analysis of the questions. Under the
significant threshold of p < = .013, twenty-five of the best (excellent) questions were selected for the formal test.
Table 2 shows the Item Number, Initial Item Number in the pilot test, Item Difficulty (P), Item Discrimination
(D), Critical Ration (CR), and Significance (p). The P-values appeared in between .272 and .554 (P-values =
(Percentage correct High-score-group + Percentage correct Low-score-group) / 2; the excellent questions’ P-values ideally
should be between .2 and .8), and the D-values fell between .239 and .652 (D-values = PH -PL; all questions’ D-
values ideally should be at least .2 or more). The mean P-values of the 25 questions was .407 (for excellent
questions the mean of P-value ideally should be close to .5), the mean D-values was .403 (for excellent questions
the mean of D-values ideally should be close to 1.0), and the mean CR-values was 4.455 (all questions reached
statistical significance, p < .05).

Table 2. Item analysis of the elementary school 6th grade English vocabulary test
Item Item no. in pilot Difficulty Discrimination Critical ration Significance
no. test P D CR p
1 2 .478 .391 4.038 .000**
2 3 .380 .457 5.054 .000**
3 5 .380 .587 7.199 .000**
4 6 .402 .543 6.316 .000**
5 7 .500 .565 6.500 .000**
6 8 .467 .413 4.314 .000**
7 10 .500 .652 8.162 .000**
8 12 .522 .478 5.173 .000**
9 13 .359 .283 2.925 .004**

191

This content downloaded from

5.90.4.172 on Thu, 03 Dec 2020 16:57:01 UTC
All use subject to https://about.jstor.org/terms
 10 15 .489 .587 6.880 . 000**
11 16 .380 .370 3.905 . 000**
12 17 .457 .348 3.535 .001**
13 18 .413 .261 2.606 .011**
14 19 .326 .304 3.256 .002**
15 20 .554 .457 4.905 .000**
16 21 .293 .283 3.097 .003**
17 24 .304 .435 5.085 .000**
18 25 .315 .239 2.527 .013*
19 26 .272 .239 2.647 .010**
20 30 .435 .435 4.629 .000**
21 32 .424 .370 3.825 .000**
22 35 .467 .283 2.801 .006**
23 38 .478 .522 5.809 .000**
24 39 .293 .283 3.097 .003**
25 40 .293 .283 3.097 .003**
Mean: .407 .403 4.455 .002**
Note: * p < .05; **p < .01.

After item analysis, the final draft of the EVCT contained 25 multiple-choice questions composed of 10 word
meaning questions, 9 word form questions, and 6 word usage questions. Each question provided 4 choices, with
only one correct answer. Each correct answer was scored as 4 points for a full score of 100. The EVCT was used
for the pre-test, post-test, and 1-month delayed post-test. It was also used for the content of the interactive
learning game for both the Kinect and computer-mouse interface.

Experimental process

The experiment was performed from October to December 2013. The treatments of the three groups for their
review activity phase are outlined in Table 3. The pre-test, post-test, and delayed post-test were arranged on
October 17, November 7, and December 12 of 2013. Each group received identical content both for their EVCT
and review activities. The control group merely viewed the test paper with the correct answers for their review
activity while the motion-sensing and computer-mouse groups reviewed using the interactive football game on
their respective interfaces. The only difference between the latter two groups was the interface, point and click vs.
a physical kicking motion.

Table 3. Experimental process of English vocabulary learning for three groups

Groups Pre-test Review activities Post-test Delayed
20 min 40 min 20 min 20 min
A1. Control group The volunteers viewed and practiced the
☑ ☑ ☑
test paper with answers.
A2. Computer-mouse The volunteers operated the game with
group computer mice, and the bystanders
☑ participated in watching and answering the ☑ ☑
vocabulary questions.
A3. Motion-sensing The volunteers played the game with
group kicking motions, and the bystanders
☑ participated in watching and answering the ☑ ☑
vocabulary questions.

Interactive system-design for the motion-sensing group

The subjects were divided into the following three groups for the experiment: the control group, which did not
have an interactive review activity; the computer-mouse group (X1), which utilized a traditional point and click
computer interaction; and the motion-sensing group (X2), which utilized the Kinect motion-sensing interactive
system (KMIS). The structure of the KMIS is described below.

The KMIS is a system composed of both software and hardware. Figure 2 shows the Kinect Software
Development Kit (SDK) from Microsoft and the Kinect Flexible Action and Articulated Skeleton Toolkit

192

This content downloaded from

5.90.4.172 on Thu, 03 Dec 2020 16:57:01 UTC
All use subject to https://about.jstor.org/terms
(FAAST) offered by Suma, Krum, Lange, Koenig, Rizzo, and Bolas (2013) integrated with Windows 7, a
projector, and Kinect hardware. When the participant stands at a proper distance from the Kinect sensor (about
1.5m-3.5m), the Kinect would convert the kicking motion into a computer command through FAAST and
become a selection operation. The four footballs on the screen represent the four options for the multiple-choice
questions. The researcher defined four kicking motions to respond to the four options: (1) the left foot kicking
towards the left for the first ball (Option 1), (2) the left foot kicking forward for the second ball (Option 2), (3)
the right foot kicking forward for the third ball (Option 3), and (4) the right foot kicking towards the right for the
fourth ball (Option 4).

Figure 2. The KMIS interactive system design for motion-sensing group

Integrated design of game-based interactive learning and a questioning strategy

For the purposes of this research, the definition of “English vocabulary learning experience” is that learners read
the vocabulary question on the review paper or screen, and then passively or actively find the correct answer
from the review paper or game screen. The effects of their English vocabulary learning experience were
evaluated by the three EVCT tests. The definition of “the questioning strategy” is that the questions of the
English vocabulary game are actively shown on a projector screen, the learner’s answer is waited for, and finally
a response is given by the learner (shown in Figure 3). The questioning strategy was only utilized with the
computer-mouse group and motion-sensing group.

The computer-mouse group and the motion-sensing group respectively represented traditional and novel human-
computer interaction. Both utilized the football game and questioning strategy (Figure 3). In contrast, the control
group only reviewed the vocabulary test paper. The football game consisted of 25 multiple-choice questions
where the participants gained points by selecting the correct answers. Whether or not the correct answer was
selected, the screen would still display the correct answer as feedback. This feedback was an integral part of the
questioning strategy which separated the subjects in the interaction groups (computer-mouse group and motion-
sensing group) with the subjects in the control group, who only reviewed the paper tests with correct answers.

Figure 3. Football game and questioning strategy integrated design for mouse and motion-sensing group

193

This content downloaded from

5.90.4.172 on Thu, 03 Dec 2020 16:57:01 UTC
All use subject to https://about.jstor.org/terms
The Flash football game used in this study was free software downloaded from the internet, at
http://goo.gl/dFXLKw. The software was revised by K. H. Yen, a teacher at Li Xing Primary School (Yen, 2012).
It allows users (teachers) to input customized questions and implement the questioning strategy to design
learning activities with its football game.

Data analysis method

A two-way mixed-design ANOVA was utilized for testing the relationship between the three types of interaction
(control or no-interaction, computer-mouse, motion-sensing) and the performance on the three tests (pre-test,
post-test, delayed post-test). A homogeneity test for variance was performed before the analysis, and the data
analyses were presented with a descriptive statistics summary, a two-way mixed-design ANOVA, a test of simple
main effects, and the LSD Method.

Results
The homogeneity test for variance was performed before the two-way mixed-design ANOVA, and the result did
not reach the level of significance (Box’s M = 18.972, F = 1.522, p = .108). This shows that the variance of the
test scores was homogenous and that we could proceed with the successive statistical analyses. The results of the
types of interaction and of the three tests, including a descriptive statistics summary, two-way mixed-design
ANOVA, and a test of simple main effects, are shown in Table 4 to Table 7.

The findings regarding descriptive statistics of the types of interaction (three groups) and the three tests

Table 4 presents the population distribution, mean, and standard deviation of the types of interaction and the
three tests. The overall mean of the three tests was also calculated for the post-test (M = 66.40, SD = 22.62),
delayed post-test (M = 63.00, SD = 21.94), and pre-test (M = 55.33, SD = 22.06). Overall, the performance of all
three groups improved, where the control group received the highest average score in the pre-test (M = 57.20),
while the motion-sensing group received the highest average scores in the post-test (M = 71.30) and the delayed
post-test (M = 65.60). The findings seem to reveal some variation within-group (between the three tests) or
between-group. This result needs to be further tested with a mixed-design two-factor ANOVA.

Table 4. Statistics summary of types of interaction (A) and three tests (B)
B. Three tests A. Types of interaction M SD N
B1. Pre-test A1. Control 57.20 22.16 40
A2. Computer-mouse 54.70 22.65 40
A3. Motion-sensing 54.10 21.82 40
Total 55.33 22.06 120
B2. Post-test A1. Control 60.70 22.22 40
A2. Computer-mouse 67.20 22.52 40
A3. Motion-sensing 71.30 22.41 40
Total 66.40 22.62 120
B3. Delayed post-test A1. Control 59.70 21.80 40
A2. Computer-mouse 63.70 21.97 40
A3. Motion-sensing 65.60 22.17 40
Total 63.00 21.94 120

The findings of mixed-design two-factor ANOVA for the types of interaction and the three tests

The two-factor analysis of types of interaction and the three tests is shown in Table 5. There was interaction (F =
14.98, p < .01, η² = .204) between the two factors (A×B), and the three tests (B) also showed significant
variation (F = 114.62, p < .01, η² = .495). A partially disordinal interaction relationship is shown in Figure 4,
where the students’ three tests before the interaction were initially ranked control group, computer-mouse group,
and motion-sensing group, but the ranking was reversed in the post-test and delayed post-test. This change
presents a significant interaction worth further analyses in the simple main effect test.

194

This content downloaded from

5.90.4.172 on Thu, 03 Dec 2020 16:57:01 UTC
All use subject to https://about.jstor.org/terms
 Figure 4. Statistical interaction between types of interaction and three tests

Table 5. Two-way mixed-design ANOVA of types of interaction (A) and test (B)
Variation source SS df MS F p η²
Types of interaction A 1212.09 2 606.04 .43 .652 .007
Test performance B 7712.36 1.52 5066.41 114.62** .000 .495
Interaction A×B 2015.38 3.05 661.97 14.98** .000 .204
Error
Error between groups 164995.73 117 1410.22
Residual 7872.27 178.10 44.20
Total 183807.82 301.67 609.301
Note. **p < .01.

The findings of the simple main effect for the types of interaction and the three tests

The simple main effect is shown in Table 6. The three types of interaction (A) did not reach a significant level of
difference (FB1 = .22; FB2 = 2.28; FB3 = .75) in the detailed items (B1.pre-test, B2.post-test, and B3.delayed post-test) of the
three tests (B), revealing no between-group difference among the three types. In other words, the motion-sensing
group was not superior to the computer-mouse and control groups in students’ vocabulary learning. The three
tests (B), on the other hand, achieved the significant difference (FA1 = 3.87, p < .05, η² = .090; FA2 = 65.00, p
< .01, η² = .625; FA3 = 73.55, p < .01, η² = .653) in the detailed items (A1.control, A2.computer-mouse, and A3.motion-sensing)
of the types of interaction (A), presenting the necessity of post-hoc comparison of the three tests (B).

Table 6. Simple main effect analyses of types of interaction and three tests
Variation source SS df MS F p η²
Types of interaction(A)
B1. Pre-test 216.27 2 108.13 .22 .80 .004
B2. Post-test 2285.60 2 1142.80 2.28 .11 .038
B3. Delayed Post-test 725.60 2 362.80 .75 .47 .013
Test performance(B)
A1. No-interaction 260.00 1.77 146.87 3.87* .03 .090
A2. Mouse 3326.67 1.49 2225.94 65.00** .00 .625
A3. Body-sensing 6141.07 1.25 4912.05 73.55** .00 .653
Error 172868.00 295.10 585.80
Note. *p < .05; **p < .01.

The findings of post-hoc comparisons for the types of interactive and the three tests

The post-hoc comparison (LSD Method) findings (Table 7) showed that the within-group (three tests)
comparison has an order relationship (B 2 > B1, p <.05). It revealed that the post-test performance was superior to
the pre-test regardless of the group. The researcher considered that the practice effect may have affected short-

195

This content downloaded from

5.90.4.172 on Thu, 03 Dec 2020 16:57:01 UTC
All use subject to https://about.jstor.org/terms
term learning retention. To further analyze each type of interaction, the students in the control group (A1)
presented significantly higher post-test (MB2 = 60.70) than pre-test (MB1 = 57.20) performance (B2 > B1, p < .05).
This reveals that simply memorizing vocabulary still presented a short-term memory benefit. However, there was
no significant difference in the performance between delayed post-test and pre-test, implying that over a
prolonged period of time the students forgot what they had memorized and statistical significance could not be
achieved. The next group, the students in the computer-mouse group (A2) displayed an even more significant
performance on the post-test (MB2 = 67.20) and delayed post-test (M B3 = 63.70) when compared to the pre-test
(M B1 = 54.70) (B2 > B1 and B3 > B1, p < .05). This reveals that the traditional interaction of the computer-mouse
group still led to a vocabulary learning effect. Finally, the learning performance of the motion-sensing group (A3)
was analyzed. From Table 7, the LSD method result of the motion-sensing group was identical to that of the
mouse group (post-test and delayed post-test performance were higher than pre-test), revealing that the type of
interaction was not a key factor affecting learning performance since the computer-mouse group also improved
in the post-test and delayed post-test. However, interactive games were a key factor for long-term retention, as
shown by the delayed post-test (B3 > B1, p < .05), since both the computer-mouse group and motion-sensing
group played the same interactive game with a questioning strategy. In addition, that the post-test results are
better than the delayed post-test results in both the motion-sensing and computer-mouse group (B2 > B3, p < .05).
This reveals that some loss of retention always occurs over time.

Table 7. Simple main effect of three tests (B) with types of interaction (A)
Types of interaction Three tests N M SD LSD method
A1. Control B1. Pre-test 40 57.20 3.50
B2. Post-test 40 60.70 3.51 B2 > B1*
B3. Delayed post-test 40 59.70 3.45
A2. Computer-mouse B1. Pre-test 40 54.70 3.58 B2 > B1*
B2. Post-test 40 67.20 3.56 B3 > B1*
B3. Delayed post-test 40 63.70 3.47 B2 > B3*
A3. Motion-sensing B1. Pre-test 40 54.10 3.45 B2 > B1*
B2. Post-test 40 71.30 3.54 B3 > B1*
B3. Delayed post-test 40 65.60 3.51 B2 > B3*
Note. *p < .05.

Discussion
In agreement with past research, the interactive game with a questioning strategy used by the two groups
benefited students’ long-term retention

Overall, we found a partially disordinal interaction relationship between the three groups and three tests. The
post-hoc comparison found the three tests to have an order relationship (the post-tests of the three groups were
better than the pre-test (B2 > B1, p < .05), showing that the three groups all had short-term learning retention;
however, the control group displayed a more apparent lack of long-term retention, as the delayed post-test was
not significantly superior to the pre-test. In addition, both the computer-mouse group and motion-sensing group
using interactive games not only displayed significantly better performance on the post-test than on the pre-test
(B2 > B1, p < .05), but also displayed significantly better performance on the delayed post-test than on the pre-
test (B3 > B1, p < .05), showing that the game-based learning with a questioning strategy used for the two
interactive groups resulted in improved long-term learning retention of English vocabulary. Past studies (Dale,
1969; Papastergiou, 2009; Prensky, 2005) indicated that interactive learning should result in higher learning
retention. They also indicated that a questioning strategy can be applied to English learning for better retention
(Basturkmen, 2001; Boyd & Rubin, 2006; Cecil & Pfeifer, 2011; Shomoossi, 2004; Yang, 2010). Research has
also showed that game-based English learning promotes retention (Flores, 2015). Therefore, affirming past
research, our experimental design (the football game with a questioning strategy) also promoted long-term
learning retention. Since the forty English Vocabulary Words were “randomly” sampled for the pilot test
regardless of how familiar they were to the subjects, this could be a reason that the passing rate of the three tests
were lower than those past tests held in case classes.

The type of interactive “interfaces” was not a key-factor affecting learning retention

The results in this study also show that there was no significant difference between the two interactive types
(three group comparison). That is, the KMIS motion-sensing interface did not outperform the computer-mouse

196

This content downloaded from

5.90.4.172 on Thu, 03 Dec 2020 16:57:01 UTC
All use subject to https://about.jstor.org/terms
interface. Therefore it seems that the type of interactive “interfaces” were not the key-factors affecting learning
retention. Nevertheless, previous research (Pan, 2013; Sommool et al., 2013; Vrellis et al., 2014; Yuan et al.,
2015) indicated that the novelty of the Kinect motion-sensing interface can better increase students’ attention or
motivation to participate when compared to a traditional computer-mouse interface did. This reveals that the
novel KMIS interface perhaps could be applied to attract students’ attention or enhance motivation to participate
in English vocabulary learning when compared to one-way rote memorization of English vocabulary commonly
practiced in schools (Smith et al., 2013).

Since the development of applications of the Kinect in education is only just starting to expand, learners can
expect much more innovation in interactive technologies. As learners adapt to motion-sensing technology, a new
generation of interfaces could easily emerge for learners, similar to users’ adaptation to computer-mouse
operations in the 1980s (Karat, McDonald, & Anderson, 1986). The research after two decades (Forlines, Wigdor,
Shen, & Balakrishnan, 2007) has discovered that, even compared to direct touch, users were still used to using
mice to operate tasks on personal computers. Motion-sensing is still relatively novel compared to learners’
familiarity with mouse operation. For the participants in our study, the computer-mouse group had the benefit of
familiarity (familiarity results in lower cognitive load), but the operation was comparatively less novel (possibly
leaving a more shallow impression in the memory) (Pan et al., 2014). As a result, we should consider that both
the computer-mouse and the KMIS motion-sensing interface have their advantages in English vocabulary
learning. The novel KMIS “interface” is helpful to attract students’ attention and motivation; however, in this
study this interface was not particularly beneficial to enhancing students’ retention in English vocabulary
learning.

The Kinect motion-sensing operating commands need to be intuitive in the future

In this study, participants were unfamiliar with the KMIS interface. The meanings of certain motion-sensing
postures which correspond to different computer commands could differ in various research studies. This could
result in unfamiliarity with operations if the definitions of postures vary across different learning environments.
This also could be a hindrance in educational applications. Therefore, in further research, it would be beneficial
to transform the posture definitions to make them intuitive and user friendly and avoid too complex postures to
correspond to computer commands. On the other hand, further work could also consider the extent to which the
KMIS approach with game-based learning can be used to learn other parts of English such as English grammar,
or even other languages. This above issues are topics for future research.

Conclusion
After discussing the statistical analyses and their inferred meanings, the following conclusions were made: (1)
the analyses of two-way mixed-design ANOVA reveal partially disordinal interaction relationship between the
interactive types and the three tests (F = 14.98, p < .01, η² = .204). The post-hoc comparison (LSD method)
shows that the three tests had an order relationship (B 2 > B1, p < .05). This reveals that all three groups have
short-term learning effects. (2) The within-group comparison, both for the motion-sensing and computer-mouse
groups utilized an interactive game with a questioning strategy, displaying significant long-term (1-month)
retention (B3 > B1, p < .05). Although the control group displayed a vocabulary learning effect in the post-test
(B2 > B1, p < .05), there was a more apparent lack of long-term retention (B3 not higher than B1). (3) The
between-group comparison of the two interactive groups did not reach a significant difference in English
vocabulary learning, meaning that the motion-sensing interface of the KMIS was not a key-factor affecting
short-term or long-term learning retention. The key-factor was the interactive content applied by the two groups.

Based on the experimental findings, our suggestion is that teachers can adopt interactive games with a
questioning strategy to enhance students’ long-term English vocabulary retention. Teachers also can use the
novel KMIS interface for interactive operation in order to attract students’ attention in English vocabulary
learning. Learners are still relatively unfamiliar with using the Kinect interface for educational applications. This
could be a hindrance in educational applications using Kinect. It would be beneficial to transform the posture
definitions to make them more intuitive and user friendly and to avoid complex postures. In this study, the quasi-
experimental design and cluster sampling were adopted for convenience; however, it could result in sampling
error affecting experimental validity. Therefore, in future study, a true-experimental design should be adopted for
better control of the interference factors.

197

This content downloaded from

5.90.4.172 on Thu, 03 Dec 2020 16:57:01 UTC
All use subject to https://about.jstor.org/terms
Acknowledgements
The deepest gratitude goes to the Ministry of Science and Technology of Taiwan for its funding (MOST 102-
2511-S-259-014), and the Review Panel for its utmost advice.

References
Basturkmen, H. (2001). Descriptions of spoken language for higher-level learners: The Example of questioning. ELT Journal,
55(1), 4-13.
Bedny, G. Z., & Harris, S. R. (2005). The Systemic-structural theory of activity: Applications to the study of human work.
Mind, Culture & Activity, 12(2), 128-147.
Bedny, G. Z., & Karwowski, W. (2003). A Systemic-structural activity approach to the design of human-computer interaction
tasks. International Journal of Human-Computer Interaction, 16, 235-260.
Birk, M. V., Atkins, C., Bowey, J. T., & Mandryk, R. L. (2016). Fostering intrinsic motivation through avatar identification in
digital games. In J. Kaye & A. Druin (Chairs), Proceedings of the 2016 CHI Conference on Human Factors in Computing
Systems (pp. 2982-2995). New York, NY: ACM.
Boyd, M., & Rubin, D. (2006). How contingent questioning promotes extended student talk: A Function of display questions.
Journal of Literacy Research, 38(2), 141-169.
Buur, J., & Bødker, S. (2000). From usability lab to “design collaboratorium”: Reframing usability practice. In D. Boyarski &
W. A. Kellogg (Eds.), Proceeding of the 3rd Conference on Designing Interactive Systems: Processes, Practices, Methods,
and Techniques (DIS’00) (pp. 297-307). New York, NY: ACM.
Cecil, N. L., & Pfeifer, J. (2011). The Art of inquiry: Questioning strategies for K-6 classrooms. Manitoba, Canada: Portage
& Main Press.
Chapelle, C. A. (2009). The Relationship between second language acquisition theory and computer‐assisted language
learning. The Modern Language Journal, 93(s1), 741-753.
Chuang, T. Y., & Kuo, M. S. (2016). A Motion-sensing game-based therapy to foster the learning of children with sensory
integration dysfunction. Educational Technology & Society, 19(1), 4-16.
Dale, E. (1969). Audiovisual methods in teaching. New York, NY: Dryden Press.
Engeström, Y. (2000). Activity theory as a framework for analyzing and redesigning work. Ergonomics, 43(7), 960-974.
Flores, J. F. F. (2015). Using gamification to enhance second language learning. Digital Education Review, 27, 32-54.
Forlines, C., Wigdor, D., Shen, C., & Balakrishnan, R. (2007). Direct-touch vs. mouse input for tabletop displays. In M. B.
Rosson & D. J. Gilmore (Eds.), Proceedings of the 2007 Conference on Human Factors in Computing Systems (pp. 647-656).
New York, NY: ACM.
Ge, Z. G. (2015). Enhancing vocabulary retention by embedding L2 target words in L1 stories: An Experiment with Chinese
adult e-Learners. Educational Technology & Society, 18 (3), 254-265.
Hsiao, H. S., & Chen, J. C. (2016). Using a gesture interactive game-based learning approach to improve preschool children’s
learning performance and motor skills. Computers & Education, 95, 151-162. doi:10.1016/j.compedu.2016.01.005
Hwang, G. J., Chiu, L. Y., & Chen, C. H. (2015). A Contextual game-based learning approach to improving students’ inquiry-
based learning performance in social studies courses. Computers & Education, 81, 13-25.
Kaptelinin, V. (1996). Activity theory: Implications for human-computer interaction. In B. A. Nardi (Ed.), Context and
Consciousness: Activity Theory and Human-Computer Interaction (pp. 103-116). Cambridge, MA: MIT Press.
Karat, J., McDonald, J. E., & Anderson, M. (1986). A Comparison of menu selection techniques: Touch panel, mouse and
keyboard. International Journal of Man-Machine Studies, 25(1), 73-88.
Kumaragurubaran, V. (2011). Sensing, actuating and processing in the built environment: A beginner’s guide to physical
computing tools. Retrieved from
http://quicksilver.be.washington.edu/courses/arch498cre/2.Readings/1.Manuals/Beginners%20Guide%20to%20Physical%20
Computing.pdf
Kuutti, K. (1996). Activity theory as a potential framework for human-computer interaction research. In B. Nardi (Ed.),
Context and consciousness: Activity theory and human-computer interaction (pp.17-44). Cambridge, MA: MIT Press.
Leont’ev, A. N. (1974). The Problem of activity in psychology. Soviet Psychology, 13(2), 4-33.

198

This content downloaded from

5.90.4.172 on Thu, 03 Dec 2020 16:57:01 UTC
All use subject to https://about.jstor.org/terms
Levinger, P., Zeina, D., Teshome, A. K., Skinner, E., Begg, R., & Abbott, J. H. (2016). A Real time biofeedback using Kinect
and Wii to improve gait for post-total knee replacement rehabilitation: A Case study report. Disability and Rehabilitation:
Assistive Technology, 11(3), 251-262.
Munassar, N. M. A., & Govardhan, A. (2011). Comparison between traditional approach and object-oriented approach in
software engineering development. International Journal of Advanced Computer Science and Applications, 2(6), 70-76.
Nardi, B. A. (1996). Activity theory and human-computer interaction. In B. Nardi (Ed.), Context and consciousness: Activity
theory and human-computer interaction (pp. 7-16). Cambridge, MA: MIT Press.
Nissimov, S., Goldberger, J., & Alchanatis, V. (2015). Obstacle detection in a greenhouse environment using the Kinect
sensor. Computers and Electronics in Agriculture, 113, 104-115.
Oxford, R. (1990). Language learning strategies: What every teacher should know. Boston, MA: Heinle & Heinle.
Pan, W. F. (2013, July). The Different interactive interface differences on pupils’ Nano learning motivation. In C. S. Lin (Ed.),
GCCIL 2013 Global Chinese Inquiry Learning and Innovative Applications (pp. 14-17). Tainan, Taiwan: National University
of Tainan.
Pan, W. F., Chien, M. Y., & Tu, S. C. (2012a). The Feasibility assessment of using a Kinect-tablet integrated system to
improve electric wheelchair reversing safety. Industrial Technology and Management, 49, 121-125.
Pan, W. F., Chien, M. Y., Liu, C. C., & Chan, K. H. (2012b). Feasibility analysis of improving the accessibility and security
of Nano-Labs via Kinect. International Proceedings of Computer Science and Information Technology, 24, 138-141.
Pan, W. F., Lin, H. F., & Wu, M. Y. (2011, July). Using Kinect to create active-learning situations for nanotechnology
labs/classrooms. In C.-C. Chang (Ed.), 2011 3rd International Conference on Education Technology and Computer (pp. 54-
56). Singapore: IACSIT press.
Pan, W. F., Tu, S. C., & Chien, M. Y. (2014). Feasibility analysis of improving on-campus learning paths via a depth sensor.
Interactive Learning Environments, 22(4), 514-528. doi:10.1080/10494820.2012.682585
Papastergiou, M. (2009). Digital game-based learning in high school computer science education: Impact on educational
effectiveness and student motivation. Computers & Education, 52(1), 1-12.
Prensky, M. (2005). Computer games and learning: Digital game-based learning. In Handbook of computer game studies (pp.
97-122). Cambridge, MA: MIT Press.
Ronimus, M., & Lyytinen, H. (2015). Is school a better environment than home for digital game-based learning? The Case of
GraphoGame. Human Technology, 11 (2), 123-147. doi:10.17011/ht/urn.201511113637
Sheehan, D., & Katz, L. (2012). The Impact of a six week exergaming curriculum on balance with grade three school
children using the Wii FIT (TM). International Journal of Computer Science in Sport, 11(3), 5-22.
Shomoossi, N. (2004). The Effect of teachers’ questioning behavior on EFL classroom interaction: A Classroom research
study. The Reading Matrix, 4(2), 96-104.
Smith, G. G., Li, M., Drobisz, J., Park, H. R., Kim, D., & Smith, S. D. (2013). Play games or study? Computer games in
eBooks to learn English vocabulary. Computers & Education, 69, 274-286.
Sommool, W., Battulga, B., Shih, T. K., & Hwang, W. Y. (2013). Using Kinect for holodeck classroom: A Framework for
presentation and assessment. In J. F. Wang & R. Lau (Eds.), Advances in Web-Based Learning-ICWL 2013 (pp. 40-49). Berlin,
Heidelberg: Springer-Verlag.
Star, J. R., Chen, J., & Dede, C. (2015). Applying motivation theory to the design of game-based learning environments. In J.
Torbeyns, E. Lehtinen, & J. Elen (Eds.), Describing and Studying Domain-Specific Serious Games (pp. 83-91).
doi:10.1007/978-3-319-20276-1_6
Suma, E., Krum, D., Lange, B., Koenig, S., Rizzo, A., & Bolas, M. (2013). Adapting user interfaces for gestural interaction
with the flexible action and articulated skeleton toolkit. Computers & Graphics, 37(3), 193-201.
Tutwiler, S., Lin, M. C., & Chang, C. Y. (2013). The Use of a gesture‐ based system for teaching multiple intelligences: A
Pilot study. British Journal of Educational Technology, 44(5), 133-138.
Vrellis, I., Moutsioulis, A., & Mikropoulos, T. A. (2014, July). Primary school students’ attitude towards gesture based
interaction: A Comparison between Microsoft Kinect and mouse. In Proceedings of Advanced Learning Technologies
(ICALT), 2014 IEEE 14th International Conference (pp. 678-682). doi:10.1109/ICALT.2014.199
Vygotsky, L. S. (1978). Interaction between learning and development. In M. Cole, V. John-Steiner, S. Scribner, & E.
Souberman (Eds.), Mind in society: The development of higher psychological processes (pp. 79-91). Cambridge, MA:
Harvard University Press.
Yang, C. F. (2010). A Discourse analysis of a native and a non-native English teacher’s questioning in elementary EFL
classrooms: A Case study. Educational Linguistics Forum, 1(1), 93-125.
199

This content downloaded from

5.90.4.172 on Thu, 03 Dec 2020 16:57:01 UTC
All use subject to https://about.jstor.org/terms
Yao, N., Wang, K., Cai, Y., & Zhang, X. (2015). Application of somatosensory sensor Kinect in man-machine interaction
framework of gesture recognition. Sensor Letters, 13(12), 1050-1054.
Yen, K. H. (2012). 雄之語文高手(足球版) [Hsiung who is a language expert (Flash: Football game)]. Retrieved from
http://goo.gl/dFXLKw
Yuan, R. Q., Hsieh, S. W., Chew, S. W., & Chen, N. S. (2015). The Effects of gesture-based technology on memory training
in adaptive learning environment. In Proceedings of 2015 International Conference of Educational Innovation through
Technology (EITT) (pp. 190-193). doi:10.1109/EITT.2015.47

200

This content downloaded from

5.90.4.172 on Thu, 03 Dec 2020 16:57:01 UTC
All use subject to https://about.jstor.org/terms