Unsupervised Morphological Segmentation Using Neural Word Embeddings


4th International Conference on Statistical Language and Speech Processing (SLSP), Pilsen, Czech Republic, 11 - 12 October 2016, vol.9918, pp.43-53 identifier identifier

  • Publication Type: Conference Paper / Full Text
  • Volume: 9918
  • Doi Number: 10.1007/978-3-319-45925-7_4
  • City: Pilsen
  • Country: Czech Republic
  • Page Numbers: pp.43-53
  • Keywords: Morphology, Semantics, Neural representation of speech and language, Morphological segmentation, Unsupervised learning, Word embeddings
  • Middle East Technical University Affiliated: Yes


We present a fully unsupervised method for morphological segmentation. Unlike many morphological segmentation systems, our method is based on semantic features rather than orthographic features. In order to capture word meanings, word embeddings are obtained from a two-level neural network [11]. We compute the semantic similarity between words using the neural word embeddings, which forms our baseline segmentation model. We model morphotactics with a bigram language model based on maximum likelihood estimates by using the initial segmentations from the baseline. Results show that using semantic features helps to improve morphological segmentation especially in agglutinating languages like Turkish. Our method shows competitive performance compared to other unsupervised morphological segmentation systems.