Angelapia Massaro

In questo articolo presentiamo il progetto RESYNC (Resilient Syntax in Contact), che ha preso l’avvio a fine 2023 con lo scopo di studiare la vitalità interna delle lingue di minoranza ladino fodom, friulano occidentale, saurano e timavese. Per vitalità interna intendiamo il mantenimento, ossia la trasmissione intergenerazionale, dei principali tratti linguistici che caratterizzano una determinata lingua di minoranza. Nel caso del progetto RESYNC ci siamo concentrati su una selezione delle loro principali proprietà sintattiche, guardando alla sintassi del soggetto, alle frasi subordinate e ai sintagmi del determinante (ossia ai sintagmi che includono un elemento nominale). Le analisi si basano su inchieste che coinvolgono sia parlanti adulti, sia bambini dell’età di 3-6 anni; nel caso del ladino fodom, che è al centro del presente contributo, la raccolta dati si è svolta in due fasi: la prima inchiesta si è svolta nel novembre 2024, la seconda nell’ottobre 2025. In questo articolo presentiamo alcuni risultati preliminari riguardanti la prima fase, che permettono già di intravedere alcuni dati importanti.

Large-language models (LLMs) have recently become the object of syntactic investigation, whether via standard probability scores in masked modeling or the interaction with conversationalAI. This study reports the results on some syntactic patterns of the Left Periphery of the clause in two Romance varieties: V2 and violations to V2 in Romansh and topic-subject agreement in Neapolitan. We study masking models with multilingual transformers (multilingual BERT and monolingual via adapters, Swiss-BERT for Romansh; multilingual BERT for Neapolitan) and through interactions with two ConversationalAI (ChatGPT, DeepSeek) by prompting an evaluation task in three languages (English, Italian for ChatGPT, Chinese for DeepSeek). Our results show asymmetries across models and structures: the theoretical predictions for Romansh are confirmed by the monolingual transformer and partially by conversationalAIs, while the Neapolitan topic-subject greement remains challenging.

This paper explores the syntactic and semantic properties of some mirative strategies in Italian, and their possibility to co-occur without producing any redundancy in the utterance. By proposing a specific setting and an event that triggers the speaker's unprepared mind, we analyse the nature of some (syntactic) markers used to convey mirativity, from Ethical and Conversational Datives, to GO and TAKE periphrases, to the expletive negation within a rhetorical question, first individually and then in some combinations. We build a featural geometry to explain why markers of different origins can yield a mirative interpretation. We assign a +ZONAL feature to the markers in question, which represents a semantic space tied to the speaker's expectations. Lastly, we investigate the issue of why mirative obliques might be featurally more complex than other markers. We conclude that the acceptable stackability of two or more mirative markers depends on the fact that mirativity is a pragmatic inference arising as a byproduct of the manipulation of the speaker's expectations.

Large-language models (LLMs) have recently become the object of syntactic investigation, whether via standard probability scores in masked modeling or the interaction with conversationalAI. This study reports the results on some syntactic patterns of the Left Periphery of the clause in two Romance varieties: V2 and violations to V2 in Romansh and topic-subject agreement in Neapolitan. We study masking models with multilingual transformers (multilingual BERT and monolingual via adapters, Swiss-BERT for Romansh; multilingual BERT for Neapolitan) and through interactions with two ConversationalAI (ChatGPT, DeepSeek) by prompting an evaluation task in three languages (English, Italian for ChatGPT, Chinese for DeepSeek). Our results show asymmetries across models and structures: the theoretical predictions for Romansh are confirmed by the monolingual transformer and partially by conversationalAIs, while the Neapolitan topic-subject agreement remains challenging.

We present a statistical analysis of the realization of definiteness in Tunisian Arabic (TA) texts written in Arabizi, a hybrid system reflecting some features of TA phonetics (assimilation), but also showing orthographic features, as the use of arithmographs. In §1, we give an overview of definiteness in TA from a semantic and syntactic point of view. In §2 we outline a typology of definite articles and show that TA normally marks definiteness with articles or similar devices, but also presents zero-markings or weak definites. In §3 we discuss TA and how definiteness is instantiated in TA. In §4, we present data from the Tunisian Arabizi Corpus (TAC), a multidisciplinary work with a hybrid approach based on dialectological questions, corpus linguistics standards, and deep learning techniques. In §5 we define the behavior of TA with respect to what we observed in §1, §2 and §3, describing our TAC-based analysis.

This work aims to analyze complex noun phrases of the qualitative binominal kind (an N of an N) in a Southern Italo–Romance language, namely Apulian. The principle of compositionality – even if it represents a good working hypothesis for syntactic analysis – faces several challenges with linguistic expressions such as idioms (with which binominal qualitatives share several properties), whose meaning is not yielded compositionally (Szabò, 2004). The data under analysis here are particularly interesting for two main reasons. The first pertains to the nature of the preposition, and whether it is an a priori specified element depending on the syntactic embedding; two main working hypotheses are found: Den Dikken (2006) posits two different derivations, one for binominal qualitative of and one for genitive of. On the other hand, Kayne (1994) posits a unitary derivation for both instances of the preposition. The second reason pertains to the problem of the overall interpretation of the phrase, i.e., an intersectional interpretation of the predicate and the subject of the predication, and the focus/topic subdivision in the phrase (Villalba, 2007): the interpretation depends on the marked order of the predicate with respect to the subject. Given that N1 is a predicate and usually an adjectivized noun or a nominalized adjective and that in southern Italo–Romance only few adjectives can precede a noun, cases in which the predicate precedes the noun can be considered as displaying a marked order. Even though the case of the preposition might seem a challenge for compositionality, it will be shown that a proper characterization of such an element confirms that compositionality can still account for instances such as those of binominal qualitatives.

We explore ChatGPT’s handling of left-peripheral phenomena in Italian and Italian varieties through prompt engineering to investigate 1) forms of syntactic bias in the model, 2) the model’s metalinguistic awareness in relation to reorderings of canonical clauses (e.g., Topics) and certain grammatical categories (object clitics). A further question concerns the content of the model’s sources of training data: how are minor languages included in the model’s training? The results of our investigation show that 1) the model seems to be biased against reorderings, labelling them as archaic even though it is not the case; 2) the model seems to have difficulties with coindexed elements such as clitics and their anaphoric status, labeling them as ‘not referring to any element in the phrase’, and 3) major languages still seem to be dominant, overshadowing the positive effects of including minor languages in the model’s training.

We investigate adverbial agreement in Sandəmarkesə (S. Marco in Lamis, Apulia) proposing phase-bound, local agreement relations, reducible to coordination, as in past and absolute participial constructions, suggesting a copulaless analysis where arguments are subjects in a small clause. With disjunct nominals with matching φ-features, the adverb agrees separately with each part in the set, otherwise resulting in ‘non-agreeing’ forms, which we test also with negative polarity items (niʃun-, ‘nobody’ and nentə, ‘nothing’). With fragment answers, the negation scopes over adverbs agreeing with the two proposed topics: matching of the φ-features of both nouns values the negative operator with the same features. In fronted adverbs, agreement occurs when the question contains overt coordinated arguments, elements continuing a chain, and if coordinated arguments have matching φ-features. In agreement in topical contexts with fronted adverbs, agreement occurs with the aboutness-shift topic closely preceding them, rather than with the embedded direct object.

The aim of this dissertation is an analysis of agreement in relation to genitival constructions. It proposes that the Apulian non-prepositional enitives of San Marco in Lamis can be described as regulated by a definiteness agreement mechanism manifesting itself in the necessity of articled heads (excluding vocatives) and genitival nouns, coupled with an adjacency requirement which limits the realization of post-nominal modifiers of the head in a post-genitival position, where they might only refer to the genitive noun. This work thus proposes that such agreement for definiteness is the same holding in Romanian non-al genitives which result in the linker construction when agreement is disrupted. In chapter 1 we thus introduce Kartvelian genitives by Suffixaufnahme which notoriously represent a genitive-head noun morphological agreement phenomenon. Plank (1995) also shows that in a series of genitives, Suffixaufnahme shows up only on the last one, demonstrating that when it comes to agreement in genitives, local dynamics of sorts are often at stake (as it happens with the Costruct State). In 2 we move to linkers; linkers have been connected to agreement in Suffixaufnahme genitives at least since Plank (1995) and later works such as Larson and Yamakido (2006), Manzini et al (2016), and Manzini (2018), according to which linkers can be assimilated to agreement. In fact, in synthetic systems such as Romanian and Aromanian, Albanian and Arbëresh, and Kurdish varieties, linkers agree for φ with either the head or the genitive noun. Giurgea and Dobrovie-Sorin (2013: 126) further show that in Romanian non-linker constructions possessives agree for case with the head noun. In linker constructions, agreement for case is not present: it’s the linker itself which agrees with the head noun, this time for φ. Chapter 3 deals with genitival modification in Hebrew and a number of Arabic varieties. It proposes that the pseudoprepositions found in Arabic varieties are nouns in the Construct State. This was previously proposed for Palestinian Arabic in Mohammad (1999), which also shows that such elements agree for φ with the modified noun. 3.2 takes into account the Semitic preposition l-, dealing with the question of whether this lement can be characterized as possessive and locative as it happens for Romance a. 4 analyzes Apulian non-prepositional genitives and proposes as anticipated that the necessity of articled nouns in the construction is to be linked to an agreement relation taking place via D. Lastly, 5 subsumes the conclusions of this work.

We investigate the morphosyntax of qualitative binominal constructions (QBCs) in a Southern Italo-Romance language from the Apulian town of San Marco in Lamis. QBCs are complex noun phrases like ‘a jewelN1 of a villageN2’, appearing here prepositionally (with the preposition də, ‘of’, allowing definites, indefinites, and demonstratives) and non-prepositionally (only allowing definites with definite articles and not proper names). We propose that in the latter, a categorial match in the determiner layer, which we call ‘match D’, relates N1 and N2. N1 is embedded as a noun, allowing for: 1) the recursive DP strategy of non-prepositional genitives, and 2) the extension of this mechanism to QBCs. This leads to the impossibility of syntactic extraction, which we connect to the concept of phase. With non-denominal N1s, N1’s article is treated as a head-agreeing adjectival linker, forming a constituent with the modifier but agreeing with the head. A phrase is interpreted as a QBC if N1 and N2 share the same number features and if the features of N1 do not allow for it to be interpreted as the possessum of N2 . We also discuss external agreement with the construction, presenting data supporting the relevance of the [+HUMAN] feature for agreement relations. (forthcoming, Italian Journal of Linguistics)

In this paper, which discusses data from Gargano Apulian Italo-Romance, I propose that prepositional and non-prepositional genitives are fundamentally two different types of phrases, and that the interpretation of a non-prepositional noun as the possessor is not due to a silent preposition or head-modifier inversion, but rather to an agreement mechanism taking place between the modifier and its head. We propose that, just as a genitive can agree with its head for gender and number features so it can for definiteness, and that agreement for definiteness yields a genitival interpretation of the non-prepositional noun. I.e., definiteness can externalize the syntactic relation between head and modifier. We also propose that in this Apulian variety, non-prepositional genitives are syntactic phases (Chomsky 2001), and that the same holds for non-prepositional ‘qualitative’ predicative phrases in the same language. This would explain the impossibility of accessing the phrase through syntactic operations such as extraction. DOI 10.1111/1467-968X.12229

This work aims at an initial description of prepositionless genitives in the Romance variety of San Marco in Lamis, spoken in the Southern Italian region of Apulia. The construction will be compared with other Romance, Semitic, Albanian, and Iranian varieties whereby the expression of possession is connected to the presence of D elements, or to morphology stemming from them. The paper deals, in particular, with the behaviour of the construction with elements such as definite and indefinite articles, demonstratives, proper names, and with how pre-nominal adjectival modification of genitives and post-nominal adjectival modification of heads can only occur in the prepositional kind of the construction. This is also the case with demonstratives preceding heads and genitives in the form raised nominals. It will be seen that genitives are only interpreted as such when they are non-raised, i.e. when they are articled. The pre-genitival article is thus understood to be a pivotal element in the interpretation of the second DP as genitival.