LISA: Please describe your work at IBM and the value it provides to IBM.

Saroj Vohra: I manage IBM's Translation Services Centers (TSCs) in EMEA (Europe, Middle East, Africa) and AG (Americas Group: USA, Canada and Latin America). The TSCs manage the end-to-end translation process for over two dozen language pairs in more than 30 countries.

The major benefit we provide to the corporation is reduction in cost and turn-around time of translations. We operate on a 'cost recovery' basis, and by embracing teamwork, simplicity and consistency, we have become an extremely productive, effective and efficient organization.

The TSC team possesses unique skills in the area of translation, project management and related topics. The professionals of the TSCs provide a variety of services 'gratis' to IBM teams, including consultation on a variety of translation-related topics such as Web localization, translation test management and advising development teams on process improvements. In addition, we also perform some non-translation tasks, including serving as the bridge organization between the product development and the product distribution centers, and managing relationships.

LISA: How large are your operations and what tools do you use to handle the volume you deal with?

Saroj Vohra: There are approximately 14 major centers that handle over 175 million words, including Translation Verification Testing (TVT). The operations are derived from a "hybrid" model that consists of both centralized and non-centralized functions.

Tools play a key role in our end-to-end translation processes and are critical to our productivity, efficiency, and day-to-day operations. With our code development partners we have developed a consistent process designed to eliminate errors before and after the translation process. We use a variety of tools such as Translation Manager and Translation Work-bench, as well as planning, communication, and packaging tools.

LISA: How do you see things evolving?

Saroj Vohra: In general, over the past several years, things have been progressing well. Slow but steady progress is being made on many fronts, and just-in-time translation, automation in audio translation through speech recognition, and animation (Flash files) are some very exciting trends that are rapidly evolving.

Two areas where the industry needs to be aggressive are the development of standards and quality improvement.

LISA: What new challenges are you likely to face at IBM?

Saroj Vohra: In today's dynamic market place, challenges are key to our progress and a good catalyst to innovation, re-engineering, and making the job interesting and refreshing. Some of our key challenges are:

• enhancing return on investment for tools/processes
• reducing cost of translation
• increasing productivity
• reducing cycle time
• new translation formats such as Web, Multimedia, Wireless, Audio visual translations
• 'testing improvements by verification' of quality across multiple platforms
• Changing market dynamics leading to an increased translation into complex non-Latin languages

LISA: How do you view the translation/localization industry as a whole?

Saroj Vohra: The world seems to be getting smaller and smaller, putting the translation/localization industry on a significant growth path. Another key element contributing to this growth appears to be the onslaught of eBusiness. It seems that the survival of businesses on the Internet may be dictated to some extent by how well they are translated/localized. There is compelling evidence to believe that English-only is not an acceptable position to take for products/services on the Web. Migrating English-only websites to multilingual sites is a costly proposition with inherent challenges, especially in the current economic environment.

On the other hand, the industry seems to be highly unregulated in terms of recognized and uniform standards for both tools and translators. Translation-related standards (or lack of them) and slow progress in their development warrants increased attention in the industry. Another area of concern is the growing trend to put information in files (e.g., Flash animations) for which no good translation standards are in place. Lack of experienced technical translators familiar with media other than "plain" documents may create some difficulties in the future.

LISA: What are the key challenges the GILT industry faces?

Saroj Vohra: The tools used in translation are not keeping up with the innovations and advances in software and information development. Since we always want the localized version of the product to be of equal quality and functionality as the English version, we are sometimes handicapped and prevented from delivering technical innovations because of localization tool limitations.

LISA: Where do we need (new) standards?

Saroj Vohra: I would make the following suggestions to start with:

• Availability of market intelligence is important in determining how much of a product should be translated to retain customer satisfaction, especially when resources are limited.
• It would be helpful to have some standards for technical product translation/localization in terms of what need/should be and should not be translated/localized.
• I would also like to see some work done on a consistent definition and measurement of translation quality.
• A standard set of criteria to qualify translators would be very helpful to new-comers in this field.
• Glossaries for consistent translation of terms/phrases.
• Translation memory for reuse across diverse tools.
• Install and packaging of translated program-integrated information. This sort of standard would help with porting products from platform to platform and is especially important with Open Source code.

LISA: How could relations between vendors and buyers best evolve?

Saroj Vohra: On-going dialog, communication, partnership and training are needed, such as a monthly or quarterly newsletter specific to vendors, training on the use of tools and technology at regular intervals, an regular meetings with vendors to solicit their input on topics of mutual interest. Joint work sessions to chalk our long-range plans would also help.

LISA: How can localization buyers reduce costs given rising volumes?

Saroj Vohra: There are a number of ways:

• Focus on the information's audience. Clearly the end user information needs to have the highest priority when it comes to translation. This will reduce cost by translating just the right information.
• Reuse information at authoring time and previous translations with both exact and fuzzy matching methods prior to translation.
• Make use of human-assisted Machine Translation.
• Make systems and applications intuitive to use, such that the help behind any function is minimized.
• Link translation cycle processes directly into the development environments. The developing/translating websites/applications process should dynamically handle updates and have the information translated/available concurrently in all languages.
• A consortium of translation providers could adopt a set of requirements to define some kind of open source for automated translation that would reduce "manual" translation and create a new role for translators – that of an editor

LISA: What is the role of translation technology?

Saroj Vohra: Translation technology is playing an important role in our processes. Its importance will continue to grow. The goal is to drive consistency in automation (tools) and related processes to minimize the manual elements of translation as it is done today.

Translation technology can help users to become more productive through automation and computer-aided support. For example, Translation Manager saves users a lot of time through the reuse of previous translations, which also improves consistency and quality.

End-to-end process support and tools integration should provide users a single interface for workflow integration.

Translation technology should provide e-business capabilities to integrate the supply chain with internal processes/resources on one end with customers/suppliers on the other end.

LISA: How do you see the role of technology in integrating humans with machines in the end-to-end translation process?

Saroj Vohra: This will be a difficult, though not impossible, change for everyone involved. We will have to be patient for this major jump to happen. Integrating humans with machines is a human factors concept. Technology should be easy and intuitive to use. People are not likely to use any technology that is difficult to use. Technology constructs should be guided by users rather than have users adapt to it. Technology must ensure that the processes it supports are fast, reliable, consistent, and unambiguous. Help information, if needed, should be presented according to the user's needs and situation and should utilize multi-media elements where appropriate to help users learn quickly.

Machine translation is equally important. Developments and enhancements in the arena of integration of voice recognition with Machine Translation will be of immense benefit.

LISA: How can developments in translation memory, terminology management, standardizing source English and machine translation provide solutions to the translation bottleneck, either together or separately?

Saroj Vohra: Each of these elements makes a partial but solid contribution to the overall problem of managing/reducing the rising cost of translation while promoting quality and consistency in results. None can completely solve the translation bottleneck on its own. Though all the four pieces of technology work hand in hand to optimize the results, we should not forget other elements such as terminology extraction, content management, and process-flow management. Each tool has its advantages but an integrated approach would definitely yield the best results.

Standardizing English terms is the first step in overcoming the bottlenecks. The potential strategic importance of standardized source English is enormous, but perhaps not yet fully recognized. It is also a key prerequisite in optimizing Machine Translation input. Similarly some terminology management programs can provide deliverables to enhance the output of Machine Translation and translation memory systems.

LISA: What is Global Content Management?

Saroj Vohra: GCM is the automated management of dynamic web site content across languages and the keeping of this content in synch at any time based on pre-defined, customable business rules. Primarily, it is software code that addresses the globalization process of content in an organization. GCM systems are also referred to as GMS (Globalization Management Systems) or eGSA (e-business Globalization Support Applications). The latter includes tools and is broader in scope than GMS. Fundamentally, they are all the same and most GMS solutions are composed of:

• a content detection and extraction engine for content to be localized,
• a set of connectors/adapters to connect various content repositories
• a set of workflows to manage the translation processes
• translation tools to optimize the translation process by speeding up translation time and minimizing costs
• business rules to automate the extraction based on business requirements
• reporting tools
• costing tools

LISA: How do you see the relationship between GCM and what we used to call the document management process? Are they the same? Are there new challenges?

Saroj Vohra: Document management provides the capability to integrate authoring, reviewing, editing, publishing and sometimes translation into one process in a shared environment in order to promote consistency and reuse. GCM is extending this to a global scale by integrating translation and localization in the same process. There appears to be a significant overlap between the two. Content detection and translation tools seem to be the key additional elements of GCM.

LISA: What is the best way for content management to integrate with the localization process?

Saroj Vohra: The answer could be the integration of the GMS features into ECMS products and provision of an open architecture capable of supporting multiple ECMS vendors. The best approach would be to bring together all the people involved in the web globalization process at one place in order to access projects, processes, tools, reporting features, etc. in an open-architecture environment,

LISA: Key challenges in this integration include:

Saroj Vohra:

• Product-centric approaches often do not take existing capabilities and requirements into consideration.
• Overlap and Integration. Customers don't want a brand new system working in parallel just for globalization purpose. They are looking to add globalization processes and capabilities to their existing systems/processes, leveraging their existing infrastructure in a secure environment. Globalization should not be an add-on, but rather should be thought from the ground up and built into the existing enterprise content management system.
• Integration of a set of heterogeneous tools coming from different vendors in the overall business process. There are many linguistic tools available on the market today, but they often do not work/integrate together.

LISA: How can LISA play a more effective role in the localization industry? How can LISA as a 'standards organization' best contribute to the development of the localization / translation industry?

Saroj Vohra: LISA could make the following contributions to the GILT industry as a standards body:

• LISA could create standards to promote text creation and tools standards targeted at software and electronic publishing requirements and based on standards in areas like XML and Unicode.
• LISA could also lead a consortium of translation providers to adopt a set of requirements for some sort of open source automated translation that would reduce "manual" translation and create a new role for translators.
• Promotion of open architecture to facilitate interoperability amongst automation devices developed by different vendors.
• Create a task force on consistent definition and measurement of Quality and related metrics.
• Conduct a survey of users to find out what is their expectation from LISA vis a vis its activities. It would also be helpful to understand vendor requirements and then bridge the gap between the two.

In addition increased participation of academia will enhance the value of LISA and attract a larger audience to LISA.