Filetype PDF | Posted on 21 Sep 2022 | 3 years ago
Authentication
digital resources
135x Filetype PDF File size 0.12 MB Source: conference.pixel-online.net
CALL in Second Language Acquisition: Introducing a German
Textbook Based Translation Game in a Geneva Middle School Class
Claudia Baur
Université de Genève (Switzerland)
claudia.baur@unige.ch
Abstract
This paper presents an evaluation of an exercise where CALL-SLT, a web-based spoken translation
game, was integrated into in a Geneva middle school class. It shows how the usage of a CALL
application as a complement to traditional school teaching can help pupils improve their productive
language skills and how it can help them overcome their inhibition to communicate in a foreign
language. In this paper the CALL-SLT platform and its adaption to a German textbook are described,
as well as the experiment and results of integrating a translation game application on a middle school
level.
1. Introduction
Computer Assisted Language Learning (CALL) has become increasingly interesting with the rapid
technological innovations of the last decades. Thanks to the inclusion of speech recognition
technology in CALL software, such programs are no longer restricted to receptive exercises, but can
also handle the training of productive skills.
A promising possibility for training oral skills via productive exercises was demonstrated by Wang and
Seneff with their idea of a translation game [1]. In this approach, speech and machine translation
technology are used to show the language learner a sentence (prompt) in his L1, which has to be
translated into the L2. While Wang and Seneff suggested using sentences in the learner’s natural L1,
the CALL-SLT project [2] went about the same idea with the goal of creating more abstract prompts,
which can help to prevent the language learner from simply translating on a word-to-word basis.
Since many language learners are first confronted with a second language in school, they usually start
learning with traditional study techniques, and the advantages of CALL have to take a back seat. The
strategy of this project was therefore to start integrating CALL-SLT where the first contact with the L2
is made, namely at school. The goal of this preliminary experiment was to see if CALL-SLT can help
French-speaking pupils improve their productive German skills and if their motivation to learn a difficult
second language can be boosted at the same time.
This paper first describes the architecture of CALL-SLT and its integration in school teaching and then
explains the experiment set-up in more detail. The last section presents the results of the experiment.
2. CALL-SLT
CALL-SLT is an open-source speech-based translation game that helps the student learn a foreign
language by improving their productive skills, such as pronunciation, vocabulary, grammar and
sentence structure or word order. The system is accessible with a personalized login via the internet,
though nearly all the processing (speech recognition and language processing) is carried out on the
server side. Currently there are various versions of CALL-SLT in use, which cover a multitude of
language combinations (L1: EN, FR, DE, JAP, IT, CHIN, AR; L2: FR, EN, JAP, DE, SWE, GRE)
[2][6][7].
In the game the learners are confronted with an abstract sense (prompt) in their L1, which they have
to utter in the respective L2. The system uses a grammar-based speech recogniser and an
interlingua-based machine translation system, which have both been developed with the Regulus
platform [3]. The system first transforms the spoken input into its written form, based on the grammar.
In a second step the machine translation system transforms the utterance into its interlingua
representation (meaning) and matches it against the interlingua representation of the initial prompt.
This second step determines if the learner’s answer corresponds to the meaning of the prompt. After
every attempt, a feedback in form of a coloured bar in green (successful recognition and meaning
transmission) or red (unsuccessful attempt) is generated. In case of an unsuccessful attempt, the
learner can make use of the help function, to access correct examples in both written and spoken form
(recorded by native speakers). If the attempts remain unsuccessful, the difficulty level (easy,
intermediate, hard) may also be adapted. Figure 1 shows the web-based interface of CALL-SLT with
the above-described functionalities.
Figure 1: CALL-SLT interface
3. Integrating CALL-SLT in school teaching
A survey on language competences conducted by the European Commission [4] stated that
“information and communication technologies (ICT), offer more opportunities than ever before for
learners and teachers to be in direct contact with the TL”, and yet the results of the survey showed
that teachers reported “little use of web content for their classes, such as software or websites
specifically designed for learning languages”. And even though students use the computer often “for a
range of purposes, such as homework, games, entertainment and contact with others”, they “use the
computer less often for TL homework”. One of the reasons for this lack of usage might be that existing
ICT tools on the web are not customised for individual textbooks. This brought us to develop a CALL-
SLT version, which is specially adapted to the learning matter treated in school language classes. In
our case we customised the CALL-SLT translation game to the textbook geni@l [5] that is used in
Geneva middle schools for German classes.
Since humans are creatures of habit, the best way to introduce ICT to pupils is to do so in the very
beginning of their language learning activities. We therefore decided to conduct the experiment with
middle school pupils (12 year olds), when they first start learning German as a foreign language in a
structured way. The goal of this early integration is for them to accept ICT as an integral component of
second language acquisition.
4. Experimental setup
Two middle school classes with an average of 20 pupils were involved in the experiment. The
experimental group was directed to use CALL-SLT at home over a three month period, in parallel with
their regular German classes at school. We asked the pupils to use the application at home for two
reasons: first, this setup reduced distraction and noise pollution, which can easily occur at the school’s
computer lab, and second, the hypothesis was that the participants would feel more at ease if they
could use the tool in an environment where they felt comfortable and were not monitored by anyone
but the computer.
At the beginning of the experiment the pupils filled out a questionnaire requesting information on their
linguistic background. They were instructed in the computer lab, where they got to try out the
application and ask relevant questions. During the whole experiment they also had the possibility of
communicating with the experiment conductor via e-mail or a Facebook page. After three months they
filled out a second questionnaire concerning their experience with CALL-SLT.
The material used during the experiment covered two lessons of the regular curriculum, namely loisirs
(leisure time) and fête (party). The lessons with each about 80-90 exercises were based on the
German textbook geni@l A1 and the curriculum conventions of the participating Geneva middle
school. Three sub-lessons were created for each lesson, which differed in the complexity of their tasks
as shown in Table 1.
During the whole experiment, the pupils had access to both written and spoken help examples, which
assisted them in finding the correct vocabulary, sentence structures and pronunciation.
Main lessons Sub-lessons Examples of tasks Examples of correct responses
Loisirs 1 Dis ce que tu fais: BADMINTON - Ich spiele Federball.
- Ich spiele Badminton.
Loisirs Loisirs 2 Dis ce que tu aimes faire: SKI - Ich fahre gern Ski.
- Ich treffe meine Freundinnen im
Loisirs 3 Dis où tu rencontres tes copines : Park.
PARC - Im Park treffe ich meine
Freundinnen.
Fête 1 Dis où vous fêtez : JARDIN - Wir feiern im Garten.
Fête 2 Dis quand est ton anniversaire: 3 - Ich habe am dritten Juli Geburtstag.
JUILLET - Am dritten Juli habe ich Geburtstag.
Fête - Die Party fängt um halb acht an.
Fête 3 Dis quand commence la fête: - Die Party beginnt um halb acht.
19H30 - Die Party beginnt um sieben Uhr
dreissig
Table 1: CALL-SLT experiment lessons
5. Results
For the evaluation of the experiment, different components were considered. One evaluation is based
on the analysis of the session log files, which allow us to draw conclusions concerning the
improvement of the pupils’ spoken productive skills. Another evaluation concentrates on the user
feedback to the second questionnaire.
In order to identify whether the pupils improved their spoken productive skills, the percentage of
successfully recognised attempts was analysed in the beginning and at the end of the experiment. For
these calculations only the five main users who used the application regularly during the experiment
(on average each 50-100 prompts) were taken into consideration. Table 2 indicates that the
recognition of two users improved significantly with a difference of +28.35% and +27.44%
respectively. Another user showed a slight improvement (+6.02%) and two of the participants reflect a
slight decline in successful recognition results (-2.19%, resp. -2.79%). The small number of
participants obviously makes the results anecdotal, but they none the less seem encouraging.
User Successful recognition in the Successful recognition in the end Difference
beginning
1 30.61% 28.42% -2.19%
2 18.37% 24.39% 6.02%
3 35.29% 63.64% 28.35%
4 3.33% 30.77% 27.44%
5 57.14% 54.35% -2.79%
Table 2: Successfully recognised attempts
The second evaluation is based on the user feedback given by the participants at the end of the
experiment. Table 3 gives an overview of the answers, divided into feedback by the five main users
and overall feedback from all participants. We can see that the main users’ feedback turned out to be
much more positive compared to the participants who did not use CALL-SLT regularly. This might
indicate that pupils tend to be sceptical about the use of ICT tools as a supplement to traditional
language learning activities until they actively work with them. The results show that the majority liked
using CALL-SLT and that they would also recommend it to their friends. 60% of the main users also
stated that they feel like they improved their German pronunciation skills.
Main users All users
+ + + - - - + + + - - -
I liked using CALL-SLT 40% 20% 20% 20% 20% 30% 40% 10%
CALL-SLT was a good supplement to regular German 40% 20% 20% 20% 20% 30% 40% 10%
classes
I easily understood what I was supposed to say 40% 60% 0% 0% 30% 50% 20% 0%
The help function was helpful 20% 40% 20% 20% 10% 40% 40% 10%
I improved my pronunciation skills 0% 60% 20% 20% 0% 40% 40% 20%
I feel more comfortable speaking German now 20% 40% 20% 20% 10% 20% 50% 20%
I would recommend CALL-SLT to my friends 20% 40% 20% 20% 20% 30% 20% 30%
Table 3: Feedback results
This evaluation seems to suggest that the pupils need to be motivated in order for them to use the
application regularly. Since there was no reward or bonus system involved, many pupils (75%) did not
use the program on a regular basis and hence their data could not be included in the two previous
evaluations. The CALL-SLT application has previously only been tested with adult learners, and has
not attempted to address the fact that 12-year-olds need more motivation and maybe also pressure for
an exercise of this kind, where they have to work independently at home.
6. Summary and conclusions
In the experiment described here, a set of twenty pupils was instructed to regularly use the CALL-SLT
system loaded with two lessons, each containing about 90 exercises (prompts). On average the pupils
went through 28.73 exercises per session. The overall percentage of successfully recognised
sentences totals 76.89%, meaning that only 23.11% of all tempted tasks could not be accomplished.
The average number of usages of 75% of the users was only between one and three times during the
experiment duration of three months. 25% of all subjects used the CALL-SLT system on a more
regular basis, meaning that they trained their productive skills between five and twelve times during
the experiment. Since the first session took place in a plenary session, some pupils might have
experienced problems with recognition due to the elevated noise level in the lab, which might have led
them to be frustrated and less motivated to continue using CALL-SLT. The 25% who used the system
regularly, however, gave a mostly positive feedback in regards to CALL-SLT and 40% of the main
users improved their productive oral skills by almost 30%.
These results suggest that the program can have a positive effect on language learning. As the
experiment has shown, there is some evidence that CALL-SLT can help pupils’ improve their
pronunciation skills and that it may be a helpful supplement to traditional classroom instruction.
However, the results of this preliminary experiment have also shown that only a fourth of the
experimental group worked well and trained their skills regularly. This leads us to the assumption that
teenage pupils need more motivation compared to adult learners (who are more determined and
motivated, as has been shown in previous experiments [6] [7]).
In order to get more usable data, future experiments might want to consider integrating the translation
game in the syllabus, so that it could be used at the school’s computer lab during class time. This
scenario would avoid requiring pupils to use it during their leisure time. However, the disadvantage of
no reviews yet
Please Login to review.
