Building flexible language-learning systems: Perl and HTML vs. XML and XSL

The Humanities Computing Development Team at the University of Oxford has
recently (1999-2000) worked on the specification and development of two
new systems to aid and support language-learning. Each project provides a
different style and level of functionality, and uses new technologies such
as XML and Java to provide functionality for both end-users and systems
administrators that is not available in other language-learning software
and is, we believe, highly transferable to other similar projects.

This abstract provides an overview of the technical development of each
system with the intention of providing this information to conference
participants in advance of the session.The conference paper will briefly
present the different development strategies taken in each case to best
balance the needs of the academic project partners, and the target
end-users, with the technologies that are currently available. We will
then focus the majority of the paper upon an evaluation of the success of
each project in developing a complex, interactive system where the content
can be chosen and administered by non-expert users.

Centre for Teaching Chinese as a Foreign Language CD-ROM and web site
(version II)

The Centre for Teaching Chinese as a Foreign Language at the University of
Oxford is leading a three-year project to develop IT-based learning
materials for Mandarin Chinese language with a strong cultural
element. Since 1998 the HCDT has worked with the CTCFL to develop a web
site and most recently a CD-ROM that provides twenty-two consecutive
lessons in basic grammar, vocabulary, character-drawing, and listening
comprehension. The way users interact with the system, and are led by it,
has been designed with close attention to pedagogic needs, incorporating a
variety of media. The software gives the user control over the choice of
exercises that they wish to explore and in which order, control over the
audio and video clips that can be played, and a higher level of learning
and self-testing through interactive Chinese character-drawing, and
exercises on specific language points.

The software uses a number of different technologies including Java,
Javascript and Perl. One of the key issues during the development of the
system was that the content for each lesson would be added incrementally
and by a number of different members of the project team with varying
amounts of IT experience, and that the layout and functionality of the
lessons should be completely standard. For this reason, we decided to
develop a set of administration interfaces linked to templates that would
allow the project team to add, edit and remove content to and from a
development server which was geographically remote from them. It also gave
us the opportunity to standardise the content that was provided for each
section of the site and to ensure that all the information needed was
supplied at each stage, something that is notoriously difficult when
editing and managing a large publication with a diverse team. These
administration interfaces proved to be extremely valuable to the
completion of content for this version of the web site.

We decided that we also wished to create a CD-ROM version of the web site
to allow remote users easier and cheaper access to the materials,
something that is particularly important because of the numerous video
clips and images that are used through the site. The requirement to
develop a CD-ROM version had not been taken into consideration during the
development of the first version of the web site and, in fact, conflicted
in many ways with the 'content independent' approach that we had taken to
the development of the lessons. To give a concrete example, the templates
and administration sections separated the content from the system and then
re-integrated the content on-demand for the user with Perl scripts, adding
some Javascript to give additional interactivity. When we planned the
production of the CD-ROM it was clear that the method used to build the
user interface would not work without a web server to run the Perl
scripts, and that some of the more complex functions that used Java would
not work properly 'off-line'. An achievable timescale was set for the
re-programming of the system to also include some design changes, and to
add some additional support materials to the CD-ROM version. The system
that has now been developed will now allow the systems administrator to
create a CD-ROM output of the site, and to have control over certain
parameters during this process. Ultimately, this additional administration
interface will allow the administrator to create a new, static version of
the web site that will be easy to use and maintain, and have less server
load. This re-development will be completed in December 2000 and the
CD-ROM distributed in early 2001.

Interactive Aural Comprehension Materials using XML

The Faculty of Modern and Medieval Languages at the University of Oxford
teaches a wide range of languages through traditional face-to-face
contact. Students are encouraged to make use of computer-based language
and literature programs for private study and personal research, but this
usage is rarely built in to the teaching program. In 1999 the Faculty
decided to initiate a faculty-wide project to develop a web-based system
to help undergraduates prepare for their final aural comprehension
examinations. The aim of the project was to build a system that would
allow the student to have 'virtual' experience of the examination
situation through a close replication of that experience, but to also give
the student access to 'practice' sessions where she / he could control the
flow of information to suit their own requirements. An essential
requirement of the system to be chosen was that it should be easily
updated and expanded whenever required, and that new sets of exercises
should be added for all languages over a period of time. Ideally, the
system should also be easily updated by a non-expert user through a series
of web forms or, as a minimum, through the use of a standard template to
capture the information.

In the first phase of the project (1999-2000) we decided to focus upon
audio files and not yet consider video files, as this would create more
complexity for capture of the video data. We did however decide to build
in functionality so that video could be easily added in the future if
required. One of the most challenging requirements for the design of the
user interface was that the audio files should be linked to specific
sections of the text of about two or three lines in length. To manually
break up each file into the requisite number of sections and then encode
the links to each section would be time-consuming and
error-prone. Instead, we decided to investigate the ability of
Synchronised Multimedia Integration Language (the XML standard for
encoding multimedia materials) to synchronise the text with specific parts
of the audio files (Groves, 2000). Our research led us to believe that
SMIL could deliver what was required and we proceeded to develop mock-ups
of the user interface using Javascript to control embedded RealMedia
controls, and HTML, with some Perl used to capture, store and retrieve
user responses to the questions. These interfaces were used to generate
some user feedback and to make improvements to the user interface, for
example to give the user added guidance to the process that they should
follow by using prompt screens between each stage of the exercise.

When the user interface had been agreed we could then develop the dynamic
system for storing the data in a non-proprietary format, and allowing the
systems administrators to easily add new sets of exercises without needed
to re-program the system. We decided to use XML to store the standard sets
of data, and the SMIL timings, and constructed (with advice from Lou
Burnard) a suitable Document Type Description to record the information to
be stored for each exercise-set. A separate part of the system was used to
temporarily store and pass on the user responses to the questions in each
exercise, and allow them to print or email their answers together with the
model answers; this is processed on the server using Sablotron and
Perl. XSL and Saxon are used to dynamically create the exercises and store
them on the server as new exercises were added, to build the contents
pages, and to integrate the content for each exercise with the set of
standard templates (including the SMIL controlled RealAudio files) and
output static HTML files to the web server. The end-user therefore has
access to a stored HTML version of the site rather than the server
generating the pages 'live' on-demand, something which saves on processing
time and server load, as well as providing a back-up.

The systems administration interface was more complex to implement and we
finally decided to require the administrator to use a standard XML
template to add new text content. Standard audio manipulation tools (Real
Products) were used to create and record the audio files. This section of
the site will be completed in November 2000.

Compare and contrast

The most substantial section of the paper will compare and contrast the
success of each system. As we have seen, the two projects began with a
quite different target audience and brief, yet both had many common
requirements for high levels of interactivity for the end-user, and the
simple addition (or revision) of the content of the system by non-expert
systems administrators who were most likely to be academic staff with
limited time and IT experience. To create a complex interactive system
that requires the use of cutting-edge technologies is challenging; to
build in user-friendly systems administration tools to allow the control
and addition of content is an additional challenge. In the light of user
feedback that will be captured and analysed during January - June 2001, we
will examine how successful these two endeavours have been in meeting
these objectives, and draw some conclusions about the future possibilities
for the development of highly complex, user-controlled learning tools.