This Habilitation Thesis provides a comprehensive overview of my research on information structure and implicit meaning, aiming to demonstrate my qualifications for obtaining the venia legendi in Comparative Romance Linguistics and Italian Linguistics. The thesis is organized into two main sections. The first section explores the relationship between non-canonical syntactic structures and discourse, focusing on their information-structural properties and functions. Specifically, it examines cleft sentences, clitic dislocations, and wh-in situ questions in Italian and other Romance and Germanic languages. The studies presented in this section adopt a contrastive, corpus-based perspective, allowing for the examination of these constructions in their natural discourse settings and the assessment of cross-linguistic variation on their licensing and use. The main findings indicate that there is no intrinsic link between specific structures (such as cleft sentences) and particular information-structural configurations. Instead, ‘typical’ configurations arise based on the language and text genres. For example, analyses of cleft sentences in French, Italian, and English across comparable text genres reveal that Italian clefts preferentially incorporate anaphoric elements and express cohesive functions, whereas English and French more frequently employ structures with contrastive/corrective and information focus, respectively. Building on these observations, I have also shown that information structure may be subject to interference effects in particular cases of language contact (i.e., translation). Rare information-structural configurations in Italian (e.g., all-focus and information-focus clefts) increase in frequency in texts translated from French, where they are more common. In general, the individual papers in this section contribute to a multifaceted, quantitative-oriented research on non-canonical syntax and advocate for analyses integrating formal and functional factors, including textual and interactional dimensions alongside cross-linguistic variation and genre-specific constraints. Additionally, the analysis of the discourse-pragmatic functions of non-canonical structures often involves examining aspects of their usage conveyed inferentially, either through conversational implicatures or presuppositions (as in the case of cleft sentences). The second section transitions to the empirical study of presuppositions, aiming to bridge the gap between quantitative research on pragmatic categories and other linguistic domains. In particular, the discourse functions of presuppositions are analyzed using corpus data from English, French, and Italian political discourse. My research indicates that presuppositions conveying questionable content (often referred to as non-bona fide true presuppositions) can be categorized into four main discourse functions: criticism, stance-taking, self-praise, and praise of others. Corpus analyses that I have conducted with colleagues reveal systematic associations between these functions and specific presupposition triggers. For example, definite descriptions tend to correlate with criticism, while change-of-state verbs align with stance-taking. Additionally, systematic associations emerge between discourse functions and individual politicians. Moreover, a comparative analysis between French and Italian politicians’ speeches and those generated by Large Language Models (LLMs) indicates that LLMs currently lack the capacity to replicate the nuanced use of presuppositions observed in political discourse. Instead, LLMs exhibit a skewed preference for specific presupposition triggers and functions, favoring change-of-state verbs and the stance-taking function. These corpus-based investigations also address the challenges inherent in the empirical analysis of presuppositions by offering concrete suggestions for developing effective research protocols in the study of implicit meaning. The final study in the volume adopts an experimental perspective, focusing on the processing and memory recall of definite descriptions and change-of-state verbs. The results suggest cognitive differences between these two presupposition triggers that may stem from their intrinsic linguistic nature (nominal vs. verbal) as well as their information-structural ‘packaging’ (topical vs. focal). In conclusion, the research presented in the Habilitation Thesis has also paved the way for my ongoing investigations into the development of pragmatic competences in didactic settings and the role of LLMs in assisting pragmatic annotation of corpus data.
Davide Garassino (Wed,) studied this question.