Hyperspacing the Verb

Different words can share a stem (for instance, Austria and Austrian) or an ending (Slovenian and Austrian); in such cases we speak of related words. Relatedness is not an all-or-nothing affair: some words are more closely related than others. Linguists have noticed that words which share a stress pattern (e.g. génerate and génerating) tend to be more closely related than words which do not (e.g. génerate and genertion). Since it is not always clear which part of the word is responsible for stress, or perhaps even because most languages do not mark stress in writing, a lot of research on relatedness refers to pieces of words, without paying special attention to stress. Moreover, we do know that there are important differences between nouns and verbs in this respect, but no complete analysis is available to date. In this project we investigate how stress and tone work as markers of relatedness in Slovenian and Serbo- Croatian verbs. These two languages are especially suitable for this research because they abundantly use stress and tone to mark different levels of relatedness, which results in words consisting of the same pieces but with different stress or tone. For instance, the Serbo-Croatian verb mati to have (stress on the first syllable) can derive two nouns using the same suffix: the closely related manje, which means having and has the stress on the same syllable as the verb and a not so closely related imnje, which means property and, unlike the verb, has stress on the second syllable. In order to get the most complete picture we focus on the best described versions of these languages: standard Slovenian (which uses stress) and standard Serbo-Croatian (which uses stress and tone). The two languages are quite similar in their word structures, but they are quite different in how they use stress, which makes them especially valuable for our purposes. The verbs and words derived from them are chosen because each verb in these languages has dozens of forms and many verbs also derive various nouns, adjectives and other verbs, this all makes it possible to observe how stress changes in interaction with different kinds of endings. The main empirical goal of this project is a complete description of all changes in stress and tone that verbs undergo in all different combinations. For this we will use the existing descriptions and corpus data, but we will also collect speakers intuitions using the project websites and the projects social media presence. Based on the exhaustive descriptions, a full-fledged model will be developed of the ways stress and tone mark relatedness between words.

The project was a bilateral research collaboration between scientists from the universities of Graz and Nova Gorica. Its aim was to advance the more intensive and essential use of quantitative data in linguistic theorizing, particularly in assessing the mutual interactions and dependencies of phenomena from different levels of grammar: phonology, morphology, syntax, and semantics. The focus of the investigation was the verb system in the Western South Slavic languages, specifically Bosnian, Croatian, Montenegrin, Serbian, and Slovenian-languages known for their rich verbal morphology and semantics, as well as notable mutual differences. Ultimately, the target was those properties of verbs typically considered arbitrary, such as theme vowel selection or the availability of particular deverbal derivations. We explored the extent to which these properties actually display regular patterns, with complexities that could only be revealed through insights derived from the rich databases produced in the project. The published papers indeed include some unexpected yet insightful findings, the kind that had not previously been reported. As for the output, four large and richly manually annotated databases have been produced. Two are already published on CLARIN's repository, and two are in the final stages of preparation. We are particularly proud of these outputs, as they represent some of the richest databases available for linguistic research, cutting across phonology, morphology, syntax, and semantics of thousands of verbs, and providing valuable resources for sophisticated quantitative interdisciplinary research. The databases may also be utilized for various purposes in language learning and teaching, as well as serving as balanced stimuli for linguistic experiments. Other results include the publication of more than 20 articles stemming from the project, including in journals such as Linguistics, Glossa, and the Journal of Slavic Linguistics. Additionally, a monograph has been completed, accepted for publication, and is in the process of being published by Language Science Press, a leading open-access publisher in linguistics. Three events were also organized by the project, either independently or as satellites to major conferences. The project has greatly facilitated the development of research collaboration between the research groups in Graz and Nova Gorica. One PhD student has successfully completed their studies as a result of the project. Furthermore, a follow-up project on multifunctionality in grammar has been prepared and granted funding, while another project, focusing on the application of computational network analysis to the databases produced, has been submitted and is currently under evaluation.