New words and abbreviated words are being born every day in CGM (consumer generated media) on the Web, such as Facebook and Twitter. Those words are not in the standard dictionaries and cause many difficulties in morphological analysis. This paper proposes a method to increase vocabularies from Twitter using Crowdsourcing. At the first stage, unknown words are chosen as candidates of new abbreviated words using a standard morphological analysis. At the second stage, Crowdsourcing System is used to determine if a word is an abbreviated word. Couwdsourcing System is used at the third stage to obtain the correct reading and the proper word.