Sustainability of Research and Development: A case of Successful Technology Transfer in Spoken Language Technology Mucemi Gakuru University of Nairobi P.O. Box 30197, 00100 Nairobi, KENYA mucemi@teknobyte.ltd +254-722-306712 Roger C F Tucker Outside Echo Ltd Beracah House, Gloucester Rd, Tutshill Chepstow NP16 7DH, U.K. roger@outsideecho.com +44-7963-106173 ABSTRACT We share our experience on how to establish sustainable research and development based on successful collaborative research between the Local Language Speech Technology Initiative of UK and Kenyan partner - Teknobyte Ltd. Starting in 2003 when we embarked on spoken language technology development, technology and expertise were successfully transferred to the Kenyan partners, culminating in the launch in April 2008 of the National Farmers Information Service (NAFIS), a voice service accessed over the phone which offers a wide range of information in Kiswahili or Kenyan English to supplement the existing agricultural extension services. General Terms Management, Human Factors, Languages Keywords Spoken Language Technology, IVR 1. INTRODUCTION Most ICT projects in the developing world have had a difficult history, particularly those with a specific developmental aim [1]. Implementation of these projects usually follows a model where a donor gives funding and expertise to develop and deploy a system. On completion of the system local personnel is trained on how to use it while relying on foreign expertise for support and maintenance. This model often fails to build in certain pertinent aspects that would ensure success in acceptability and therefore proper utilisation. For one, the system may not provide the appropriate solution as the developers do not capture all the important local issues owing to their limited experience in the local environment. Secondly the project may fail to create local ownership which is key to acceptability that would ensure full utilisation and hence sustainability. In order to succeed in developing and deploying ICT technologies based on Spoken Language Technology (SLT) a well thought out model for the developing world is even more crucial as the languages are clearly local and hence require a lot of local knowledge. Many challenges would need to be dealt with which are due mostly to the lack of speech data and linguistic resources, the skills shortage in the countries where the languages are spoken, and the general lack of funding for research in the poorer countries of the world. However we still chose Spoken Language Technologies because they offer a great potential in the developing world for making IT more accessible for those without text or computer literacy, enabling widespread access to information on health, markets, agriculture, weather, microfinance etc. Compared to the use of SLT in the developed world, the potential has yet to be realized in the developing world. There are two key factors that help explain the difficulty – the lack of success of ICT for Development projects generally, and the specific problem of SLT for local languages in the developing world. One way to address these challenges is to nurture local expertise and ownership thereby creating a sustainable base for continued development and resource mobilisation. We therefore set from the beginning to address the key issue of technology transfer and ownership and found out that by allowing the partners in the developing world to take ownership of the technology development process early on, they had the motivation and maturity to see it through to successful deployment.