To build an Indonesian Machine Translation (MT), it is not only needed a related syntactic analysis to the correct spelling of words but also needed related contextual analysis, consist type and function of word, morphology, and semantic. The dictionaries usage is needed to translates Indonesian basic words and to captures good word translations through the semantic and context of words in a sentence or document. This study purposes to extracts Indonesian and Tolaki words for building a good MT by comparing the development of Indonesian MT which focuses on deep cases of morphology and syntactic. We developed morphtool to captures the morphological elements of Indonesian and Tolaki words. For working in deep syntactic case, we build a rule to captures the function and type of word that can affect the word itself translation in the sentence. We combine supervised and unsupervised techniques to work on the text extraction in the words, sentences, and documents through the morphonemic rules of Indonesian-Tolaki syntaxis manner. Then, we use hybrid MT, combining Statistical MT (SMT) and Rule Based MT (RBMT), for sentence translation process. The hybrid MT evaluation process from the Indonesian-Tolaki to English translation performance test shows the accuracy result is 0.74. Meanwhile, the performance test of the English to Indonesian-Tolaki translation shows the accuracy result is 0.71. These results indicate that the proposed MT method can work better than the SMT and RBMT methods with an average accuracy of around 70%.