Research in the field of text analysis will always be related to words, either the selection of words to be used or the position of the words in a sentence. Furthermore, a hypothesis that each language difference can cause different meanings, makes some researchers interested in doing research classifying words based on emotion or affective words. Research focuses on affective states as a continuous numerical value to the dimensions of valence and arousal. Sentiment analysis that is usually done with positive and negative category approaches, nowadays, the dimensional approach can provide more analysis of grained sentiments. On the other hand, the affective words dataset with valence and arousal rating are still very rare, especially for the Indonesian language. Therefore, this research does an affective lexicon dataset called Indonesian Valence and Arousal Words (IVAW) containing 1024 words by Self-Assessment Manikin (SAM) surveys. Furthermore, for the next study, we will also crawls status in twitter based on selected words from IVAW to get Indonesian Valence and Arousal Text (IVAT). To predict VA rating for obtaining the advance of annotation quality, experiment will be compared by brain signal using EEG tool.