The project “Computational Processing of Croatian Morphology (MORF-HR)” aims to develop an integrated linguistic resource describing both derivational and inflectional morphology of Croatian. Two complementary databases will be created: the Derivational Lexicon, containing information on morphological structure and derivational relations between words, and the Inflectional Lexicon, providing complete paradigms for all inflective word classes. Both resources will be accessible through a unified online interface and available for download under open-science principles. The interconnected lexicons will offer structured insight into word composition, derivational families, formation processes, affixes used, and full inflectional inventories. By combining two core levels of morphological analysis, the project introduces an innovative approach to computational processing of Croatian. The resulting resources will support the development of language technologies, facilitate linguistic research and teaching, enhance Croatian-language instruction for native and non-native speakers, and promote linguistic and digital literacy.
Team
| Krešimir Šojat | Faculty of Humanities and Social Sciences, University of Zagreb | project leader | ksojat(et)m.ffzg.hr |
| Matea Filko | Faculty of Humanities and Social Sciences, University of Zagreb | researcher | msrebaci(et)m.ffzg.hr |
| Marko Tadić | Faculty of Humanities and Social Sciences, University of Zagreb | researcher | marko.tadic(et)ffzg.hr |
| Ida Raffaelli | Faculty of Humanities and Social Sciences, University of Zagreb | researcher | iraffaelli(et)m.ffzg.hr |
| Tatjana Pišković | Faculty of Humanities and Social Sciences, University of Zagreb | researcher | tpiskovi(et)m.ffzg.hr |
| Ranko Matosović | Faculty of Humanities and Social Sciences, University of Zagreb | researcher | rmatasov(et)m.ffzg.hr |
| Kristina Kocijan | Faculty of Humanities and Social Sciences, University of Zagreb | researcher | kvuckovic(et)m.ffzg.hr |
| Kristina Hrastov | Faculty of Humanities and Social Sciences, University of Zagreb | researcher | khrastov(et)m.ffzg.hr |
| Daša Farkaš | Faculty of Humanities and Social Sciences, University of Zagreb | researcher | dberovic(et)m.ffzg.hr |
| Magda Ševčíková | Charles University, Prague | researcher | sevcikova(et)ufal.mff.cuni.cz |
| Zdeněk Žabokrtský | Charles University, Prague | researcher | zabokrtsky(et)ufal.mff.cuni.cz |
MORF‑HR is a linguistic resource that contains data on derivational relations between words as well as the inflection of variable word classes in Croatian. This linguistic resource with integrated data on the morphology of the Croatian lexicon (MORF-HR) consists of a Derivational and an Inflectional Lexicon.
Each entry in the Derivational Lexicon includes information on the word class and basic grammatical features. It also provides data on the morphological structure of the word, with words segmented into lexical, derivational, and inflectional morphemes. This structure is presented on two levels – surface and deep. On the deep level, morphs are recorded that serve as representatives of their allomorphs, which are conditioned by morphophonological processes. A deep morph is one from which its surface realizations can be derived using the smallest possible set of established morphophonological rules. Thus, the deep level shows word structure in terms of morphs that serve as the base forms for deriving surface-level morphs, while the surface level reflects their actual phonological realization in a given word. For example, the noun potpalublje is represented as pot‑palubl‑j‑e on the surface level and pod‑palub‑j‑e on the deep level.
The primary goal of word segmentation into morphemes is to identify derivational affixes and lexical morphemes. In the development of the Derivational Lexicon, identifying lexical morphemes plays a key role in its organization. Word formation is divided into two basic processes based on the number of participating lexical morphemes: derivation, which involves a single lexical morpheme (e.g., pisati – ispisati), and compounding, which involves (usually) two or more lexical morphemes (e.g., noga + metati > nogomet). The Derivational Lexicon contains only data on words formed through derivation. Derivation is described within derivational families, i.e., groups of words sharing the same lexical morpheme.
The entry description in the Derivational Lexicon provides the user with data on the part of speech and its morphological structure with a two-level representation. Based on morphological segmentation, words are grouped into derivational families defined by a shared lexical morpheme. This approach allows for an explicit representation of the relationships between base forms or source words and derivatives, as well as a clear specification of the word-formation processes and affixes used. The computational component of the project also enables the visualization of derivational chains from the center of a word family toward derivatives of various degrees. In other words, it allows for a graphical representation of the derivational path from root to derivative for each lemma in the lexicon, as well as a display of entire word families. The planned scope of the Derivational Lexicon is 40,000 lemmas of the general standard language.
The main goal of the Inflectional Lexicon is to systematically display all forms of selected lemmas belonging to variable word classes. In addition to searching by individual words, classifying words into specific declension or conjugation classes enables users to search for all words that inflect according to these classes. The projected scope of the Inflectional Lexicon is 60,000 lemmas of the general standard language.
Compounds in the Croatian language are processed in separate lexicons: a lexicon of Croatian compounds and a lexicon of compounds consisting of affixoids of predominantly Greek or Latin origin (e.g., biolog), or combinations of such affixoids and bases of domestic origin (e.g., bioznanost). Entries for compounds include the source elements they are built from, as well as affixes that are often used simultaneously in their formation (e.g., čast-o-hlep(iti)-j-e). Morphological segmentation will also be conducted for the description of Croatian compounds. Entries from both compound lexicons that belong to variable word classes will be linked to the Inflectional Lexicon.
The overall goal of the project is to provide users, through a single web interface, with access to the morphological structure of a queried word, comprehensive data on its derivational relations with other words, and all of its inflected forms.