engleski

Higher-order Functional Representation of Croatian Inflectional Morphology

Many language processing tasks rely on some sort of morphological processing, such as word-form generation, lemmatization, or morphological tagging. These are inherently difficult tasks for morphologically complex Slavic languages. For these languages morphology representation becomes an issue as well. This paper describes higher-order functional morphology (HOFM), a convenient formalism for representing inflectional concatenative morphologies, and its application to morphologically complex Croatian language. In HOFM, inflectional paradigms are represented by inflectional rules and morphological transformations are represented by higher-order functions. HOFM representations resemble the highly expressive morphology descriptions often found in traditional grammar-books, allowing for compact and comprehensible morphology models. We describe a HOFM implementation in Haskell, a modern functional programming language, and demonstrate its use on the task of generating and lemmatizing Croatian words.

Natural language processing; morphology modeling; inflectional morphology; lemmatization; Croatian language

nije evidentirano