Human language displays both categorical/discrete aspects (phoneme perception, subject-verb agreement, distinction between questions and assertions etc.) and gradual/nuanced ones (perception of “deviance” in constructions like weak vs strong islands, Cinque 1990; processing measures such as reading times etc.). On the one hand, the foundations of generative grammar were largely built on the idea of a categorical linguistic competence, which can give rise to gradual judgments or preferences only through performance (individual fatigue, processing loads, memory limits etc.). On the other, an increasing use of quantitative methods in linguistics, both in the form of data (huge corpora, large-scale data collection of grammaticality judgments) and statistical techniques for their analysis (e.g. machine-learning-based predictions, Brown et al. 2020, including metalinguistic judgments, Sarti 2020) produced a reassessment of the importance of gradedness in linguistic theories (Lau et al. 2017), in the end making it difficult to explain more categorical distinctions: grammaticality contrasts expressed in structural terms may be hard to maintain if features and constituents are probabilistic entities. In this project we focus on children (hearing and deaf) and Primary Progressive Aphasic patients to demonstrate how categorial discrimination is crucial for a precise linguistic competence assessment and how this perspective can be reconciled with the observed gradience in various domains. Our starting assumption is that the identification of inherently discrete aspects of linguistic competence and performance requires us to factor out three types of emergent gradedness: 1. Noise due to extralinguistic factors (here we propose to improve the quality of the data collected using also implicit elicitation methods). 2. Gradedness emergent from the interaction of two or more independent categorical factors, especially at the interface of the structure building component with phonology (Zubizarreta 1998), and with semantics/pragmatics (Bianchi & Chesi 2014). 3. In atypical populations, the non-categorical manifestation of an inherently discrete factor may be due to the temporary inability to compute all the elements of a linguistic structure, resulting in an impoverished representation (Grillo 2008). These three points will be experimentally addressed by assessing the subjects\' responses to minimal contrasts concerning syntactic phenomena (agreement for person/number/gender and dislocation of constituents), where responses are expected to be categorical, and interface phenomena (question interpretation correlating with prosodic properties), where responses might vary across and within participants. In addition, we will reanalyse already available data, to develop techniques of noise reduction and find new signatures of implicit categorial discriminations. Finally, we will assess which theoretical linguistic model better fits the data thus obtained.

This project aims at contributing to the safeguard of minority languages (MLs) by investigating the level of language endangerment from a new perspective: we provide an objective measure of the transmission of their morphosyntactic structure. While the vitality of MLs is often assessed through sociolinguistic analyses, which target the frequency and contexts of use of minority languages, we will adopt a criterion that calculates the internal vitality, namely the degree of resilience of various morphosyntactic constructions. This is particularly important today, because the current demographic and social dynamics are characterised by increased mobility, mixed marriages and the spread of standard languages that affect the traditional transmission process, both quantitatively and qualitatively: children receive less and less input in the ML, and this input is restricted to a decreasing number of domains. As a first step, in this project we focus on three MLs, and three morphosyntactic features. The MLs are Ladin, Friulian and the German varieties spoken in Friuli (Northeast Italy). The morphosyntactic features are a) the syntax of the subject; b) subordination; c) nominal agreement. The crucial idea behind our proposal is that it is fundamental to understand not only if a regional language is alive, but also how alive it is. The qualitative aspect is fundamental in a minority setting.