A Segmentation Approach to Morpheme Analysis

We present a simple algorithm that is also psychologically plausible to perform un-supervised learning of morphemes. The algorithm is most suited to Indo-European languages with a concatenative morphol-ogy, and in particular English. We will describe the two approaches that work to-gether to detect morphemes: 1) finding words that appear as substrings of other words, and 2) detecting changes in transi-tional probabilities. This algorithm, while most suited to the task of segmenting words into morphemes, also suffices to an-alyze morphemes.