Segmentation: a remark on the Syncretism Principle

详细信息    查看全文

• 作者：Sebastian Bank
• 关键词：Inflection ; Homophony ; Morpheme segmentation ; Learning algorithms
• 刊名：Morphology
• 出版年：2017
• 出版时间：February 2017
• 年：2017
• 卷：27
• 期：1
• 页码：1-20
• 全文大小：
• 刊物主题：Linguistics, general; Comparative Linguistics; Phonology; Sign Language;
• 出版者：Springer Netherlands
• ISSN：1871-5656
• 卷排序：27

文摘

Morphological analyses usually prefer ‘deriving’ form-identities as systematic syncretism over just stating them in terms of accidental homophony. While such anti-homophony is mostly assumed implicitly, Müller (2004) spells it out more explicitly as violable Syncretism Principle guiding both language acquisition and linguistic analysis (‘same form → same meaning’). However, as soon as the child or linguist decomposes word forms into smaller formatives (morpheme segmentation, subanalysis), it is unclear what instances of form-identity exactly are to be avoided (e.g. substring-identities?). This paper frames the logical space of possible Syncretism Principle interpretations, which relate to their functional motivation (ambiguity avoidance) demonstrating their concrete consequences for analysis with a paradigm learning algorithm offering segmentation and meaning assignment.