GLOSSASOFT is now in the case study phase where the initial guidelines and methods for the internationalisation and localisation of new and existing software are being evaluated. Before looking at the case studies in more detail a flavour of the material in the initial guidelines and methods is given in the next three sections.

Method for Localisation of Existing Software

The method consists of two basic phases: a feasibility study and an implementation phase consisting of a possible internationalisation step, followed by localisation and testing. The feasibility study consists of an analysis of the level of internationalisation provided by the existing software and its supporting platform and the business-related and language and culture specific market needs for localisation in the target markets.

The key questions answered by the study are:

• Given the target markets, should an internationalisation step be undertaken?
• What are the costs and risks involved in the internationalisation and localisation process?
• What current tools are available to support the internationalisation of existing software and what ideal tool set is required?
• What is the current positioning of products in the target markets and how does it affect the approaches which might be taken
• Given the current state of the target market, what sales and gross profits should be achieved for localised and non-localised products?

The decision making criteria for internationalisation and localisation from a business point of view are identified. The elements obtained from the feasibility study are resolved by identifying the trade-offs and the factors involved in making the decision to localise or not, and to what particular depth.

The implementation phase consists of an optional internationalisation step in which software reverse engineering, restructuring and re-engineering techniques can be used to identify and factor out locale dependencies and reconstruct the core of the software. The objective of this step is to create a international base version of the software which can be localised for a number of different markets with less effort.

The localisation step involves inserting in the locale dependencies required for particular markets. Trade-offs involved in the selection of different localisation strategies are identified and localisation methods and guidelines for tools are provided for the following categories:

• User interaction - menus, messages, dialogues and on-line help
• Data encoding
• Hardware devices
• Documentation
• Linguistic aspects
• Regression and usability testing

Linguistics

The objectives of this deliverable are to:

• give a short baseline on how to bridge the gap between software internationalisation and localisation and language technology by identifying language-dependent software problems and studying linguistic methods, techniques and resources
• to be aware of current state-of-the art language technology tools and provide guidelines for selecting and using the various types of tools
• investigate other approaches in language technology that could be used to make software internationalisation and localisation process more effective.
• It identifies the following problem areas: the inconsistent use of terms, the lack of uniformity in writing and translation style, errors, and version management of linguistic aspects.

It shows how language technology can help overcome these problems and at the same time satisfy the commercial requirements. It informs companies about language technology tools that may assist the process of internationalisation and localisation in the following categories:

• writing and proofreading tools
• translation tools and workbenches, and
• machine translation systems.

For each of these categories the deliverable specifies the corresponding types of tools and provides evaluation criteria and guidelines to help selectand use the appropriate tools

Software Structure

The software architectural foundations for software localisation and internationalisation are examined. Approaching software architectures with the intention of identifying where linguistic and culturally dependent parts of the system are placed, or should be placed. The requirements for internationalisation for basic operating systems and window system architectures to support internationalisation and localisation are identified.

Guidelines are presented on how to develop applications, or re-engineer existing applications, to exploit the facilities of the supporting software platform. Developments in operating systems and window management systems which would further support internationalisation are identified.

In summary, to internationalise well we need:

• to apply good software design practice
• to understand the locale requirements
• to know what facilities the support environment provides
• to be aware of the strengths and weaknesses of these facilities
• to understand possible trade-offs such as portability and performance

The Case Studies

There are four case studies:

1. The localisation of Hewlett Packard's VUE (Visual User Environment), a Graphical User Interface for Unix workstations for the Greek market.

   A localisation feasibility study will be done to evaluate the internationalisation level of VUE and identify the localisation needs of the Greek market. A localisation implementation will be done on the basis of the feasibility study. Messages, menus, icons, and help files will be extracted into resource files and transformed into appropriate formats for translation.

   An English-Greek terminological base terminological base of VUE terms will be created according to the linguistics guidelines. An English-Greek translation memory or database of some of the messages contained in VUE will also be created. The language technology tools used during the case study will be evaluated according to the evaluation criteria specified in the linguistics guidelines. Evaluation reports for each tool will be given to companies who supplied evaluation copies.

   Different translation scenarios will be examined using controlled languages and machine translation, terminological bases, translation workbenches and human translation. Comparison and evaluation of the different scenarios will then follow. The main evaluation criteria will be time, cost and translation quality.

2. The localisation of Claris' Help systems for the French market. This case study is an excellent base for the evaluation of existing translation tools. An evaluation will be done with particular emphasis on feasibility and practicality of use in real localisation situations. The guidelines and methods for developing and maintaining a terminological base will be evaluated. A particular issue is how the terminological base is "plugged in" to the overall localisation process. The information gained in the study will be generalised for the company and recommendations for an overall approach will be given.

   Sample texts will be used to study the potential use of controlled languages. How closely does the help match the requirements of a controlled language? How does text translated using a controlled language and an MT system compare with a human translation? What level of post-editing is required? What is the attitude of technical writers to the use of controlled languages?

   The importance of testing and quality assurance, and the level of resources involved in the process suggest that this is an area which would be ideal for some sort of automatic processing. The requirements for automatic methods for this application area will be outlined. This will be backed up with experiments to verify the recommended approaches.

3. The internationalisation and localisation for the Finnish and other markets of RiskTool and OsiCon. This case study focuses on two pieces of software: RiskTool and OsiCon.

   RiskTool is a tool for handling risks in software projects. OsiCon has been developed by VTT to ease development of message-based applications in heterogeneous network environments.

   Because of the availability of the source code of both of the systems and the nature of research and development work, the main emphasis in this case study is on internationalisation rather than on localisation.

   An interesting aspect of the RiskTool case study is that the product consists of the tool itself and the knowledge included in it. The Risk knowledge is often context dependent,: company-specific, and especially sensitive to cultural features. This makes the translation and transfer from one locale to another an important issue.

   A method for message generation will be developed including:

   • selection and refinement of appropriate formalisms for representation of knowledge and syntactical structures taking into account the need for interlinguality and practicality from the software developer's point of view,
   • solutions for morphological generation,
   • selection of tools used in the experimentation, and
   • selection of the languages considered in the experiments (e.g. English, Finnish, Greek).

4. A study on internationalised user interface messages.

   Effective, high-quality User Interfaces consist of texts and graphics that are simple, ergonomic and easy to localise. The textual part of the interface, called "messages" for the purpose of the study, will be the focus of the study. Normally, developers write and supply messages. Writing high-quality messages demands an understanding of the subject matter and skill with language. Developers have the first but not always the last.

   Writing for an international audience places additional demands on message providers. When they write for an international audience, they must keep in mind the needs of two kinds of users : the readers of that linguistic version (called the reference version), and the localisers who produce the other linguistic versions.

   The objectives of this case study are:

   • To produce guidelines, rules and techniques to help developers write English messages that are ergonomic, appropriate for an international audience and that can be easily, quickly and economically localised.
   • To specify an electronic tool to help developers write English messages which conform to the guidelines.

The Future

The final phase of the project will focus on integrating the case study findings into the final methods and guidelines and a dissemination programme to make the project results available to a wide audience. Internationalisation and localisation strategies for smaller software development companies is another area where the project aims to offer special guidance.

The Glossasoft consortium wishes to thank LISA and its members for their help, information and feedback. LISA members are eligible to join the Glossasoft Interest Group and there is no joining fee. On joining, IG members will receive copies of all the project deliverables to date. Any feedback from companies on the deliverables is much appreciated.