Employing Machine Translation in Glocalization Tasks

Today, we and our customers are faced with a huge amount of continuous data streams in multiple languages and different forms and formats. Therefore, our business communications requirements and strategies demand for an effective employment of various language resources to economically and efficiently administer, master and monitor information processes and workflows across languages, cultures and time zones. We have thoroughly investigated into what language resources are mostly suited for our needs, and what are the important enablers in different translingual technical deployment scenarios that guarantee throughput, scalability, quality and successful operations and applications. Although Machine Translation (MT) is still a gadget because neither individual nor business users do share the usability and quality of MT as a real user experience, MT is an intrinsic part of our solution. With this paper we want to share and discuss our findings on MT with the community. 1. Global Communications Landscape In the last decade, private and business communications have changed dramatically with the Internet being the ultimate communications platform for everyone across time zones, languages and cultures. Translation and cultural adaptation play an ever increasing critical role in this global communications landscape and are no longer restricted to business and technical communication only. With the ubiquitous web access from differ© 2010 European Association for Machine Translation. ent even mobile devices such as smartphones, the need for competent and effective language services increases exponentially, and in particular these services shall be highly configurable and available from everywhere, on demand, and preferable as a pay-per-use service offering. Obviously, this new, highly proactive communications landscape with its associated demands for multiple language services cannot be handled, administered and controlled properly with traditional translation technology service set-ups because currently they are not flexible enough to account for the various communications requirements. The existing translation management systems are mainly designed to manage and monitor human-oriented tasks which in general are comprised of single, disruptive steps without direct interprocess relationship and interaction. Full automation and ambient adaptability are key to keep pace with the speed and variety of the multilingual transcultural demands, and the intrinsic characteristics of the processes from start to finish with persisted states. The term “glocalization” that is derived from the Japanese term “dochakuka” meaning “global localization” names the just described global communications landscape most appropriately. In Section 2, we outline a possible technical solution that guided our language technology investigation with a focus on translation automation in particular machine translation (MT), and show what is needed to serve the demands of a global communications landscape. How this solution might be set into operation is discussed in Section 3, which also points to some serious shortcomings that exist with state-of-the-art technical and technological MT offerings with a focus on sharable language resources. In Section 4, we conclude and list the necessary steps in terms of short-, mediumand longterm investments, and how the MT [EAMT May 2010 St Raphael, France]