Unsupervised Morphological Segmentation Based on Segment Predictability and Word Segments Alignment

Word segments are relevant cues for the au-tomatic acquisition of semantic relationships from morphologically related words. Indeed, morphemes are the smallest meaning-bearing units. We present an unsupervised method for the segmentation of words into sub-units de-vised for this objective. The system relies on segment predictability to discover a set of pre-fixes and suffixes and performs word segments alignment to detect morpheme boundaries.