This paper describes a Hierarchical Composition Recurrent Network (HCRN) consisting of a 3-level hierarchy of compositional models: character, word and sentence. This model is designed to overcome two problems of representing a sentence on the basis of a constituent word sequence. The first is a data sparsity problem when estimating the embedding of rare words, and the other is no usage of inter-sentence dependency. In the HCRN, word representations are built from characters, thus resolving the data-sparsity problem, and inter-sentence dependency is embedded into sentence representation at the level of sentence composition. We propose a hierarchy-wise language learning scheme in order to alleviate the optimization difficulties when training deep hierarchical recurrent networks in an end-to-end fashion. The HCRN was quantitatively and qualitatively evaluated on a dialogue act classification task. In the end, the HCRN achieved the state-of-the-art performance with a test error rate of 22.7% for dialogue act classification on the SWBD-DAMSL database.