English is one of the most widely spoken languages in the world these days. Most of the commercial websites are also being designed in English language. So, we need a language translation system to understand all websites and all information from the internet. This paper is one part of the English-Myanmar Machine Translation system. In our system, tagged English text is accepted as input. Firstly, these tags are tokenized. Each phrase is tokenized. The structure is basically a list of tokens. The linguistics clues are normalized. And then dictionary look up is performed. The guessing process is applied. Since the files are sorted according to alphabetically, required information can be searched quickly with a binary search. The dictionary-based translation technique produces one or more translation terms in the target language for each term in the source language. We propose a method called word sense disambiguation to solve the ambiguity of words. By using this technique and bilingual lexicon, the system may retrieve correct translated words.