Vous êtes sur la page 1sur 3

Eighth IEEE International Conference on Advanced Learning Technologies

Design and Management of Material Sharing System for Ubiquitous


Vocabulary Learning

Xin Jin, Masatoshi Ishikawa, Keiichi Kaneko, Haruko Miyakoda and Norihide Shinagawa
Graduate School of Engineering, Tokyo University of Agriculture and Technology, Japan
50007646135@st.tuat.ac.jp, {masato-i, k1kaneko, miyakoda, siena}@cc.tuat.ac.jp

Abstract have developed a system, SIGMA (Special-Interested-


Group Material Accumulator). In this paper, we will
In our project, we are developing a system that discuss the reliability of the evaluation scores given by
helps learners in vocabulary acquirement. In the sys- learners for selection of appropriate learning materials
tem, learners can create their own materials for mobile and the management policy of the system.
learning, and they can also exchange their materials.
This paper focuses on a sub system, SIGMA, to share 2. System and materials
and exchange materials among learners. In addition,
SIGMA enables learners to give evaluation scores and 2.1. Vocabulary learning system
comments to all learning materials stored in the system.
In this paper, we will discuss the management policy of In this subsection, we will explain the vocabulary
the system based on the reliability of the evaluation learning system, PHI (Personal Handy Instructor),
scores given by the learners to select appropriate based on iPods. In PHI, a five-second movie is used for
learning materials. From the result of our experiment, each word as a learning material. Figure 1 shows a flow
we concluded that the evaluation scores have some re- of vocabulary learning by PHI. First, the learner selects
liability but an instructor should guide the learners. materials from a material list managed by PHI. The
chosen materials are copied to a folder called ‘vocabu-
1. Introduction lary book.’ Then, the learner downloads the materials
into iPod by dragging and dropping the folder onto the
When acquiring vocabulary, it is said to be effective iTunes window. PHI has the learning history of the
to memorize words with related images and sounds as learner and it can display the number of acquired words
well as to spend time on repetitive memorization activi- to encourage the sustainable learning of the learner.
ties [4], [5]. We have developed a system called PHI
(Personal Handy Instructor) with which learners can
acquire vocabulary at anywhere and anytime by using
iPods, and discussed its effectiveness [1]. Furthermore,
we have developed a system, PSI (Personal Super-
Imposer) that helps creation of learning materials for
PHI, and we have succeeded in reducing the heavy load
of learning material creation imposed on instructors [2].
We have also examined the effectiveness of the learn- Figure 1. Personal handy instructor
ing materials created by PSI.
There are many previous and related works with re- 2.2. Personal superimposer
spect to mobile learning and/or vocabulary acquirement
based on information technology. From the survey, we PSI is a system that supports in the creation of
concluded that we should provide a system where 1) a learning materials by learners. By providing the system
learning material created by a learner can be uploaded with a five-second movie related to the word, spelling
and distributed to reduce the burden of material crea- and meaning, learning materials can automatically be
tion, 2) a learner can create a learning material easily to created. Since the learners may use the system to create
collect many learning materials, and 3) a learner can e- their own materials, the burden of instructors is greatly
valuate the materials created by others to encourage the reduced because it would not be necessary for them to
creation of learners. According to this conclusion, we solely prepare a large number of materials beforehand.

978-0-7695-3167-0/08 $25.00 © 2008 IEEE 338


DOI 10.1109/ICALT.2008.49
Furthermore, a high learning effect can be anticipated uation by learners into the system. A learning material
because the learners themselves are involved in cre- is evaluated by asking learners a question: “Do you rec-
ating the materials (Figure 2). ommend this learning material to other people?” Then
server they answer it by selecting one of the seven choices:
“Absolutely Yes”, “Yes”, “Relatively Yes”, “Neutral”,

http server
movie, word, “Relatively No”, “No”, and “Absolutely No”. The an-
PSI
meaning execute swers are changed into scores from +3 to −3, and their
user average is calculated for the evaluation score of the
(WWW browser)
create
learning
material. In the process of evaluation, learners are re-
use
materials
data quested to use nicknames to establish anonymity to ex-
learning
material pronunciation
clude some bias based on human relation. In addition,
files db learners are able to browse comments of learning mate-
rials. From this function, learners can find out advan-
Figure 2. Personal SuperImposer Online tages of materials that cannot be identified only by their
evaluation scores.
2.3. Vocabulary learning materials To exclude malicious learners, the SIGMA system
manages learners by accounts and surveys their behav-
Figure 3 shows snapshots of materials. From a pre- iors. For instance, SIGMA can detect a learner who
liminary experiment, we fixed the length of materials to uses multiple nicknames and gives affirmative or nega-
be five seconds so that learners can browse the materi- tive scores to some specific materials. SIGMA can also
als repeatedly without stress. The pronunciation is re- detect a learner who registers very poor or inadequate
peated twice in the material. The first and the second materials by checking the average score of each ac-
pronunciation start from one and three seconds after the count. Based a report by the system, an instructor can
beginning of the movie, respectively. The spelling is give appropriate guidance to the malicious learners.
displayed from the beginning, whereas the meaning two
seconds afterwards. By postponing display of the 3.2. Implementation
meaning, we expect that a learner concentrates on the
spelling at first, and connects it to the pronunciation SIGMA is a Web application that uses Apache, PHP,
and the movie for memorization. and MySQL. Figure 4 shows the main frame of the
SIGMA system. Without any login operation, a learner
can browse the evaluation scores and comments of
learning materials, and also download learning materi-
als from the list in Figure 4. Moreover, an authorized
learner can register and manage his/her own materials,
create word tests, and give evaluation scores and com-
ments to all materials after login operation. SIGMA has
Figure 3. Examples of Learning Materials a test function by which score histories of authorized
learners are stored.
3. Learning material accumulator A management tool is prepared for instructors in the
SIGMA system. The tool has functions for account
3.1. Design management, vocabulary management, authorized
learners’ behavior management, and score management.
We have designed the SIGMA system so that learn-
ers can register their own learning materials, and down-
load the materials created by other learners. In addition,
we added a function that allows learners to evaluate the
learning materials and to give comments to them be-
cause user-driven reputation systems are often useful
when there are many materials [3]. In the designing
process, we encountered two problems: selection of ap-
propriate materials and exclusion of malicious learners.
To help learners to select appropriate learning mate-
rials from the material list, we introduced material eval- Figure 4. Main frame of SIGMA system

339
1
3.3. Management 0.9
0.8

Retention Rates
In putting the SIGMA system into practice, there are 0.7
two choices with regard to its learning environment: 1) 0.6
0.5
an environment such as SNS, in which blog and bulle- 0.4
tin board systems are provided so that learners can 0.3
form a community to learn voluntarily; 2) an environ- 0.2
ment such as a classroom, where instructors take the 0.1
0
lead and give appropriate pieces of advice to learners.
-3 -2 -1 0 1 2 3
The former has the merit of reducing the burden of
Evaluation Scores
instructors, but we are not yet quite sure of the effects Figure 5. Average Evaluation Scores and Retention
of autonomous learning. In the next section, we report Rates (2 Weeks After) of Learning Materials
on our preliminary findings.
the scores and the effectiveness was 0.41 and we con-
4. Evaluation cluded that the average evaluation scores have some re-
liability, but it is preferable to as-sign an instructor to a
To investigate the possibility of autonomous learn- learning environment based on the SIGMA system.
ing environment by learners, we measured the correla- Future works include an extension of categorization
tion between the evaluation scores by learners and the of learning materials by giving parts of speech, key
effects of learning materials taking 23 undergraduate words, and so on to them.
and graduate school students in our university as sub-
jects. The experiment is conducted as follows: 1) Sub- Acknowledgements
jects were divided into Group A (13 subjects) and
Group B (10 subjects); 2) Group A subjects browsed This study is partly supported by the Special Coor-
62 learning materials in the word list we prepared, and dination Funds for Promoting Science and Technology,
they gave evaluation scores to them; 3) Group B sub- Ministry of Education, Culture, Sports, Science and
jects selected unfamiliar words from the word list; 4) Technology (MEXT) Japan. It is also partly supported
We selected 13 words from the unfamiliar words that by MEXT Fund for Promoting Research on Symbiotic
the subjects in Group B selected so that the evaluation Information Technology.
scores of the corresponding materials are scattered
evenly; 5) Each subject in Group B learns the selected
10 words for 10 minutes by the materials; 6) Tests are
References
conducted for subjects in Group B just after and two [1] S. Amemiya, K. Hasegawa, K. Kaneko, H. Miyakoda, and
weeks after the learning activity. We selected the reten- W. Tsukahara, “Long-term Memory of Foreign-word
tion rates of materials to represent their effects. Learning by Short Movies for iPods,” Proceedings of the 7th
Figure 5 shows the result of the experiment. The IEEE International Conference on Advanced Learning Tech-
correlation coefficient between the average evaluation nologies, July 2007, pp. 561−563.
scores and the retention rates of the meaning of the
learning materials turned out to be 0.41 meaning that [2] M. Ishikawa, K. Hasegawa, S. Amemiya, K. Kaneko, H.
there is weak positive correlation between them. The Miyakoda, and W. Tsukahara, “Automatic Creation of Vo-
result seems to suggest that the average scores have cabulary Learning Materials from Short Movies,” Proceed-
ings of the World Conference on E-Learning in Corporate,
some reliability, but it is preferable to assign an instruc-
Government, Healthcare, & Higher Education, Oct. 2007, pp.
tor to a learning environment based on the SIGMA 6044−6051.
system to manage it as a classroom.
[3] P. Resnick, R. Zeckhauser, E. Friedman, and K.
5. Conclusion and future works Kuwabara, “Reputation Systems'”, Communications of the
ACM, Vol. 43, No. 12, 2000, pp .45−48.
SIGMA, the sub system in our learning system for
vocabulary acquirement, enables learners to give evalu- [4] N. Schmitt, and M. McCarthy (editors), Vocabulary: De-
scription, Acquisition and Pedagogy, Cambridge University
ation scores and comments to all materials. Hence, in Press, 2005.
this paper, we conducted an experiment to investigate
that the evaluation scores of materials correctly reflect [5] A. Wright, Pictures for Language Learning, Cambridge
their effects. As a result, the correlation coefficient of University Press, 2005.

340

Vous aimerez peut-être aussi