A Virtual Environment
For the Promotion
Of Language Learning

Design Specification including
Methodological Aims
Research Possibilities
Language Learning Activities


The purpose of Babel-M is to investigate the potential of 3-D virtual-reality environments for the creation of communities which can be used for the learning of second and foreign languages. It is intended that there should be the highest level of theoretical weight behind this project, at the same time as using knowledge of the technology to achieve this in practice in the most effective way.

The phenomenon of computer mediated communication (CMC) being used to create and foster collaborative, or even competitive, multi-user environments has become well established since 1980 when the first Multi-user dungeon (MUD) was created. However, it is only now that technology has developed to the extent that real-time interactivity in 3-D is possible for limited financial outlay. The main question that needs to be investigated is whether this type of technology can be used successfully to facilitate language learning. A second consideration that needs to be considered is whether there may be some reason for concern over possible excessive use of CMC for socialization.

The following information outlines the specification for a working environment that could evolve into a fully-functional, multi-language virtual environment. The environment will incorporate the latest models of constructionist learning at the same time as using traditional methodologies. This is because it is essential to utilise a wide range of pedagogy to convince the often conservative teaching profession, and students.


Finding users to participate in innovative and technologically impressive projects initially appears quite simple. Logging on to the Adobe Atmosphere web-site shows this as it is a popular site for various reasons. This fact can be misleading though, a fundamental problem with online communities is capturing a solid base of users to work together on developing the environment in a sustainable manner. For example, many users will log on to such environments once and never return, preferring to use their own, established messenger networks. For this reason the users of the advanced features of the site will be invited from a variety of language schools and universities. This means that use can be controlled and also encouraged by the external influence of teachers using participation as an assessed component of language courses and integrating it into teacher led CALL classes. The software that is being considered for this project allows chat logging which is useful for assessment, research and, most importantly, for the true evaluation of students’ strengths and weaknesses.

After these primary users there would be several other interested parties; teachers, researchers (be they concerned with linguistics or the social aspects of computers), students of CALL, software developers and finally builders who would be selected from any of the other groups. The different users would all have the ability to have some influence over the evolution of the community. In line with the ideals of its development, the utmost effort would be made to ensure that pertinent decisions were subject to group negotiation.

To foster a sense of power, control and belonging in student users, certain worlds will be customizable by users at run-time. There is the possibility for users to select the décor and audio of rooms at run-time, which will take effect on their own computer, or if there is a pedagogical need on all the computers connected to that instance of the world. Conventional web discussion boards will also be used for students to discuss and plan the design of any expansion of the environment. There is scope for the addition of worlds to the environment as it is possible to access hyperlinks by typing commands into the chat box. This would require the user to have write access to the web host for adding hyperlink sub-worlds.

The web is essentially a modular medium and the use of open source code in the form of HTML with Javascript means that it is almost organic in the way it expands. This allows a large amount of flexibility in the way the community is constructed. The file system can be organized in such a way that core worlds will only be changed by group negotiation. However, the distribution of servers will allow teachers to have control over certain aspects of their environments very rapidly as teaching demands require. As Atmosphere loads textures and audio as separate files it only requires a simple upload to update teaching materials without having to change the fundamental structure of the world.

The initial design of the environment will include a number of worlds which will have specific pedagogical aims. These will be more structured and controlled than social and cultural areas, containing activities with various amounts of task prescription which will facilitate the use of various language features and functions. Teachers will have the ability to suggest and specify new types of activity that designers would implement. There would be the possibility for CALL students and academic researchers to prototype and evaluate experimental and perhaps controversial language learning tasks.

Builders would have some freedom to add features to the world but they would need to earn this right by demonstrating some exemplary behaviour over an extended period as the world will essentially be a meritocracy. The security and rights for builder access can be controlled through conventional means, files uploads can be through FTP, or perhaps CGI pages for the upload of single files like textures and audio clips. It has to be noted that Atmosphere can use the same types of security as normal HTML pages but this security is not as simple to implement at the moment. This means that it is possible for anyone to link their own web-page or Atmosphere world to our site, if they have the URL. There will be a requirement for experienced javascript, database and CGI scripters if the project is to develop to its full potential as a living, breathing entity.

A final point when considering users, which may well be difficult to implement, is the possibility of using tandem learning. This concept has involved the pairing of speakers who want to learn each other’s languages. They support each other in the construction of linguistic rules. It has to be added that this requires much more discipline and reflection than traditional pen-pal activities.
“Tandem partners need to go beyond traditional classroom methods, not only by planning, monitoring, and evaluating their own learning process, but by exploiting their partner’s native speaker competence and developing an insight into the language learning process.”

In an ideal situation there could be this type of teaching occurring between entire classes. However, there are problems with different countries having school holidays at different times, on top of time zone problems.


In order to have control over specific language learning tasks and prevent the entry of disruptive users there is a need for security. Security is a complex issue and Atmosphere cannot use exactly the same systems as HTML as it uses its own browser and it is quite difficult getting variables, such as validation codes, into the virtual environments. Fortunately there is a simple way of introducing a low level of security into the worlds using Javascript. In this mode of security, doors (or hyperlinks) to new worlds are dynamically loaded by typing a command into the chat box. Users need to know both the command and the name of the hyperlink to open the new world. If there were security breaches by students passing on the commands to friends it is possible to change the names of hyperlinks quite simply using Windows Explorer type software, and then pass on the new names to users. This level of security would probably be sufficient for the period of time needed to evaluate the true level required by the system in the longer term, as well as being particularly cost effective.

Design of the Environment

One of the current problems with Atmosphere is the fact that the Chat function does not have as many features as some others. All of the users in an environment communicate through the same chat box, this means that if there are twenty users (the current maximum) in one room it will be very difficult to follow the flow of conversation. There are also no built-in features to allow for private chats, or for the “ignoring” of particular users. Design and building of the various rooms/worlds will need to take this into account.

Babel-M is designed as a world portal (here portal defines the entrance to a range of worlds) which will allow access to a wide range of language specific worlds, which in turn will link to task or theme specific rooms. This first contact is designed on a massive scale, as a metaphor for the scope of the project, and the edifice is modeled on a classical representation of the tower of Babel. There are no security constraints on entry to this world and the limit of users is the current maximum of twenty.

Inside the tower there are thirteen floors which are assigned to different regions and are accessed by an elevator. In this world, and many of the others, the ability to fly and walk through solid objects will be disabled. This means that there can be more control over tasks and students will have to discover how to achieve things through communicating with other users. For example, the elevator in Babel-M is controlled by commands typed into the chat box, these commands can be modified to resemble natural language. Furthermore, it will only be possible to open hyperlinks to other worlds if you are on the correct language floor. The hyperlinks are also opened by typing commands into the chat box, here it is essential to know the floor needed, the command needed and the name of the desired world.

The ability to dynamically load hyperlinks into environments allows for flexibility in the expansion of the world. For example, to add links to worlds requires no change to the hard coding of the environment, simple worlds simply consisting of a hyperlink are added to the file system. These work as dynamic hyperlinks and there are also named entry points which can be added to worlds to act as bookmarks.

Underneath Babel-M in the hierarchy are the language worlds which can be conceptualized as meeting places and secondary portals. A similar system of dynamically loaded hyperlinks will be used to access task specific and themed rooms. For example, libraries, bars, offices, games rooms and puzzles will be included for each language. A detailed description of possible activities and rooms will be discussed later in this document.

As mentioned at the beginning of this section Atmosphere does not have specific functions for private rooms. However, it is possible to have simple private rooms with the maximum number of users hard-coded into them. Therefore, if you had two friends and wanted to talk in a private area you could use the command “go_to_world private3”, where 3 is the number of users private3 will allow at one time. The same world could be created for 2, 4, 5 users, and so on.

Surrounding Babel-M is a supporting framework of HTML and CGI powered pages. The main pages will be the entrance, a list of users, a discussion board and library pages. The entrance will give basic information of the theory and purpose of Babel-M and provide access to the world, which will be embedded in an HTML page. It is essential to embed the browser in HTML as this ensures that access to other HTML pages is as stable as possible. The discussion board can be used for users to give a “hard copy” of their opinions regarding the development and evolution of the community. The list of users will allow for the participants to have access to information about others in the language group while library pages will give access to associated textual information.

Language Activities

To ensure the use of the environment by as many teaching professionals as possible it is essential to give as wide a range as possible of language learning affordances. Currently these affordances can range from totally unstructured chat, chat in environments with a variety of conversational (or more specifically discursive) prompts, negotiation during task completion activities and individual interaction with unsynchronized worlds. Although some methodologies (for example those based on behaviourism) are out of fashion with academia this by no means suggests that they are no longer popular with teachers.

A fundamental structure which will run throughout the array of worlds is the addition of message filtering. This procedure is essential for the functioning of semi-natural language commands in the world but can also be extended to control the use or misuse of words. It is possible to make lists of disallowed words which will have various effects on messages attached to them. For example, the excessive use of a word could cause a warning message to appear and the message to be blocked, the use of offensive language could cause the user to be sent to ‘hell’ and the use of common English words would result in warnings if they were used in LOTE situations.

To encourage communication in the world it is suggested that only the simplest and most essential of commands are given out to new users. Additional commands and tips will need to be discovered either by exploration and the understanding of language based clues, or by the fostering of relationships with established users.

Currently Atmosphere does not support the easy synchronization of worlds which means that some of the more creative tasks will remain in the planning stage. At some point in the future it would be interesting to have an environment in which the users had to govern in a sustainable manner considering social and natural concerns. However, this is currently beyond the scope of the project. Despite this there are many possible tasks that are achievable.

To facilitate and smooth the use by class-based teachers it is possible to transfer some established tasks into the electronic medium. The added dimension of virtual distance can be used effectively in information gap activities. A common task with low level learners is the description of two pictures with small differences between them. This requires discipline on the part of learners as they can, and often do, simply look at each other’s picture. If this activity was conducted in a virtual room with the pictures on opposite sides of a wall it would be impossible to cheat. A similar use of pictures is made in the Cambridge suite of general English exams for higher level learners. Pictures are provided as discursive prompts (I believe that “discursive” is a more appropriate descriptor for the type of communication expected here) and a wide range of linguistic functions are expected here. There could be topic-based galleries of images connected with cultural and international concerns that students could explore. Taking this immersion into prompts one stage further there is the possibility to create worlds that are discursive prompts in their own right by putting students into unusual situations. For example, students could be transported to the seabed or the peak of Everest to encourage visualization and personal development.

Another activity derived from class use is the common part game where an object has to be described without using some previously prescribed words. For example, if we were considering the word “teacher” we would be prohibited from using the words; student, school, class, blackboard or subject. The chat filtering feature would allow this prohibition to be enforced by not allowing those messages through. Similar to this would be a world containing large objects that were difficult to see in their entirety and students would have to hypothesise about them and only be allowed to progress if the object was guessed correctly.

A very powerful information gap activity can be constructed by using the fact that the worlds are not synchronized. It is possible to create quite large and complicated worlds and there can also be many different entrances to these worlds. A command in the entry world could generate a random portal that would send the user to one of these entrances. Once the participants were in the world they would be able to communicate in the chat box, but be in different areas. Their task could be to find each other and the exit portal before they were able to leave the world. The teacher could be waiting in the exit room to ensure that they all entered at the same time.

As Atmosphere has the built-in chat feature it is easy to forget that it can also be used as stand-alone software, without the collaborative features. It could be used for creating 3-D multimedia environments with computer-based interactivity. Simple javascript features can be implemented to give students individual exploratory tasks led by script based audio or text instructions. Being able to follow a sequence of audio instructions might result in some form of motivation to continue. Javascript could also be used to create interactive resources or 3-D movies of events to facilitate the learning of vocabulary.

Finally the chat in Atmosphere could be combined with the external discussion boards to coordinate decisions about the development of the community. Following the example given by the Adobe homeworlds it is possible to add billboards advertising future events, such as a debate or vote about the design of the world. It is also possible to use javascript to display when the event will take place relative to UCT (universal coordinated time), an important fact to consider when participants cross multiple time zones.

For contributors, the modular nature of Atmosphere is very useful. For items discussed in this section it is possible to specify rigid attributes that will allow for the simple swapping of features. The pixel size of picture activities and billboards can be made public so that work and teacher-led activities can be submitted without causing any difficulties for the world builders. Exhibitions could be held where the submissions could be chosen for inclusion in the world, or there could be certain areas where all submissions were allowed on display.

A particularly social aspect will be added in some rooms to foster a feeling of belonging to the community and having individual control. This will be the ability to change certain aspects within your own instance of a world. Decoration, pictures and audio can be customised and will act only on your computer. This facility can be used so users can express their individuality, without affecting others.


With a view to encouraging participation by possibly skeptical teachers there are several possible avenues for research in this environment. The two main ones that I would pursue are direct, quantitative research on language output and qualitative surveys of user satisfaction and their perception of their learning. These two avenues are the most important from my point of view because the results can be used by teachers to evaluate how worthwhile the project would be if they joined.

As has been discussed many times it is virtually impossible to determine the true effectiveness of any language program, due to a virtually infinite number of aspects that can contribute to the process, the products and attitudes towards it. However, as some of the tasks possible in Babel-M are derived from class activities it is possible to conduct some form of comparative investigation to see if the media has an appreciable effect on output. The comparison could be between the oral output given in a traditional class situation, chat output given in a standard chat room using hard-copy prompts and then chat output given within the virtual environment. Ideally this would be carried out on as large a population as possible. Presumably there would be little difference between the two forms of CMC, but this research might provide interesting results.

There can also be comparitive research carried out on logs taken before and after the use of a disallowed word filter. Although refusing the use of certain words can be considered a particularly behaviourist form of encouraging students to notice their own output, this feature would probably produce measurable results. The comparison could also be made indirectly by comparing the use of language in the free production, social rooms before and after the use of a disallowed word filter in task-based worlds.

If this feature was found to be effective it could be extended to include lists of words that were selected from logs of the worlds. Logs could be analysed statistically to check for excessive use, or common mistakes and the javascript controlling this altered accordingly.

Qualitative surveys of student attitudes and perceptions could be administered and used as program evaluation tools or even market research. Results from this could be used to improve the functionality and levels of use of the world. They could also be used in conjunction with the quantitative measures to encourage new language groups to join.

Other aspects which might prove enlightening would be a study of the use of gesture by avatars and the design of individual avatars. Gestures could be categorized and then a study of collocations between words and gestures could be made. Again this could be compared to worlds in which avatars were unable to use gestures. A vital question is whether the gestures inhibit or promote the use of traditional language within the virtual environment. The design of avatars may be seen to represent various features of socialization and the development of identity. It is entirely possible for users to change sex, to become androgynous, or even become a different species and a study of the extent of these changes could give valuable insight into the relationship between reality and virtual reality. The possibility that the design of avatars may influence the linguistic features of communication would be an interesting sphere of research. Drama has some history in language teaching and virtual reality can facilitate the integration of this into classes seamlessly. To foster the exploration of different identities there could be archives of avatars created which would support different roles; police, children, politicians and so on. Communication under these different roles could be compared to the same activity without avatars designed for the task.


As mentioned in the introduction there are some aspects of CMC and computer use in education that need to be taken into account. Firstly there is the
well -documented fact that it is possible for people to become addicted to chat and MUDs. A common cry amongst “normal” people is that these people should “get a life”. However, it has been suggested that young people use these facilities excessively due to the unlimited possibilities for social contact they present.

Considering this fact, and the justifiable concern that some educational professionals may have about institutionalizing this type of activity, it is also necessary to give students information about this facet of CMC. Students and teachers should be made aware of the negative aspects before they enter the environment. There should be support and counseling facilities provided for all the participants to make use of. In addition, some form of on-going monitoring of the use of the environment should be made from within classes.

With virtual environments there is the possibility that form may overshadow the function. The design of Babel-M is necessarily massive, to provide a metaphor for its extent and range. It is essential that the underlying theory is made clear as a feature which underlies the superficial appearance. There might be the necessity of having HTML pages open up when entering task-based worlds which would give a breakdown of the task and methodology behind it. The fact is that any simulation activity will be considered a game by some users, so the fundamental aim of such simulation needs to be made clear. As the technology develops the environment will become much more of a simulation and take on many aspects of computer gaming. There will need to be controls put on this to ensure that the use of language to succeed in tasks is a continuous thread.


Atmosphere is one of two VRML authoring environments currently available. It is in competition with an established product called Active Worlds which has some history of academic and commercial use. Atmosphere is currently free and the all chat is routed through an Adobe server, this is one possible reason for the lack of world synchronization. Active Worlds supplies a free browser but the builder and server, which is essential for the running of environments, are commercial products. At first sight there is little to choose between Atmosphere and Active Worlds. The only readily visible difference is the fact that chat in AW appears directly above the avatar which is “speaking”, this is a definite advantage when worlds have multiple users. However, on more careful investigation we can see that AW supports world building and synchronization through its programming language and dedicated server. Unfortunately this language is more complicated than Atmosphere’s combination of javascript with proprietary extensions. As javascript is one of the simplest professional programming languages, and is the web-standard in HTML and Flash pages, this could be considered as a major advantage.

A further advantage of these types of software are the readily available components and scripts which have been developed by the companies concerned and the web-community at large. This allows for the rapid creation of rooms and worlds by amateurs and professionals alike. There is no need for a high level of knowledge of the software before worlds can be made. It is obvious that this would empower all levels of participants in the project.

This site created by
Charlie Williams M.A.
Claire Weetman M.Ed.

23rd June 2003