A self-referential childlike model to acquire phones, syllables and words from acoustic speech | IEEE Conference Publication | IEEE Xplore