This newsletter is a special issue on trying to bridge the gap between donors' resources and their effective use in targeting the poor, through the use of community development funds. Community development funds function like banks, but can work more flexibly and at different levels. Several case studies are presented from countries in South Asia, Southeast Asia, and southern Africa, and a number of tips and advantages in setting up a community development fund are outlined. Some of these are that they: " Set new standards of transparency and accountability; " Make multiple, small-scale investments in many community-initiated urban development projects; " Support tangible outputs of value to the urban poor, in different sectors and areas; " Help establish and strengthen long-term partnerships between community organisations, municipal authorities and the private sector, while stimulating new working practices; " Provide poor communities and their organisations with opportunities to learn by doing.
Rather than challenging the universal validity of PRA, this discussion paper focuses on the practical task of "doing PRAs" in a new and alien context. The authors advocate the acknowledgement and acceptance of local cultural trends, power relations and structures of authority when undertaking participatory research. This will allow to work with, rather than around these factors. Hence, the proposed "Vietnamisation" of PRA so as to allow local voices to shape the values and techniques of PRA itself. But, just how Vietnamised can PRA become until it comes into conflict with international liberal PRA values? This broad discussion originated in a workshop organised in Hanoi on Community Research Methods in February and March 1996. The issues covered by the paper include: introducing PRA and PRA values in Vietnamese communities, gender, local leadership and dominance; and international donors and PRA. Methodological issues covered are: sampling, recording of research information, interviews, focus group discussions and mapping.