We must solve the various ambiguity problems in developing English-Korean machine translation system. The method of part-of-speech determination is studied for part-of-speech disambiguation. The part-of-speech determination chooses one part-of-speech among the candidate part-of-speeches of a word. It is expected to help to make translation faster by reducing the parsing complexity and to improve parsing accuracy. Existing part-of-speech tagger or determiner assigns only one part-of-speech to a given word. But, even an error can cause parsing failure. This paper studies a method of preventing parsing failure due to part-of-speech determination errors. We try to give multiple part-of-speeches to a word for which is difficult to determine its part-of-speeches. That is, this paper proposes an extension method of existing probabilistic model for part-of-speech determination. So we expect the determination module will be applicable to English-Korean machine translation system. In experiments, we show the practicality of the proposed extension methods.