Unsupervised induction of inflectional families. (May 2022)

• Record Type: Journal Article
• Authors: Kirschenbaum, Amit
• Is Part Of: Computer speech & language. Volume 73(2022)
• Publication Date: 2022-05
• Subjects: Morphology -- Unsupervised learning -- Inflectional families -- Clique Percolation Method -- Multiple Sequence Alignment; Speech processing systems -- Periodicals; Automatic speech recognition -- Periodicals; Computers -- Periodicals; Linguistics -- Periodicals; Speech-Language Pathology -- Periodicals; Traitement automatique de la parole -- Périodiques; Reconnaissance automatique de la parole -- Périodiques; Automatic speech recognition; Speech processing systems; Electronic journals; Periodicals; 006.454
• DOI: 10.1016/j.csl.2021.101324
• Languages: English
• ISSNs: 0885-2308

Abstract

Abstract: This paper proposes a model for inducing inflectional families from text, without any specific language knowledge. The method is based on a combination of word embeddings and network theory, inspired by Bybee's network model for morphology. Experiments were conducted on English whose morphology is simple, and German, which has relatively rich morphology. The proposed method outperforms FastText, a state-of-the-art algorithm employing subword embeddings to model morphology, and achieves F 1 scores higher by more than 14% and 10%, for English and German, respectively.