__________________________________________________________________________________________Recently, the Taiwan government has been enthusiastically promoting the study of the languages of her native inhabitants, including Taiwanese. Hence the focus of our research is to design and develop an efficient retrieval technique for phonetic transcriptions. This new technique will make possible widespread utilization of Taiwanese, e.g., in PDA (mobile phone applications). In this paper we propose a minimal perfect hashing function for the 3028 existing Taiwanese phonetic transcriptions. Compared to the hashing designs based on the Chinese remainder theorem for various data sets, the proposed design is shown to be superior in space utilization.