In this thesis, I propose an unsupervised computational model of language acquisition through visual grounding. I especially focus on a case where the language input is in a form of variable length sentences. The ...