Challenges and Perspectives in Italian Clitic Tagging: A Case Study on the SerbItaCor3 Corpus
Saša G. Moderc · 2025
Italian clitics are functionally heterogeneous: some function as pronouns, others as various types of adverbial adjuncts, or as markers of passive and impersonal constructions.Clitics are also used pleonastically and are integrated into pro-complement verbs (e.g., farcela 'to manage').They occupy three positions relative to the verb and can form clusters of two or three clitics; in such cases, the entire cluster usually occupies one of the three aforementioned positions.In the SerbItaCor3_it corpus, the tagging of Italian texts was performed using TreeTagger (by Achim Stein).During analysis, inaccuracies were observed in the tagging of the clitics ci and si, as well as inconsistencies in the processing of homographs.The clitic si was tagged either as a reflexive pronoun (PRO:refl) or as a personal pronoun (PRO:pers), while its passive and impersonal uses were not marked.These tagging inaccuracies compromise the reliability of the corpus's statistical data and limit its usefulness for linguistic analysis and language teaching.However, thanks to the accompanying Serbian translations, in most cases it was possible to determine the exact function of the clitics.This enabled the proposal of improvements to the Italian tagger, contributing to more accurate tagging of clitics and homographs.The paper presents examples of incorrect tagging and translation-based solutions, on the basis of which suggestions for improved tagging of Italian texts can be formulated.