The Balkans are a melting pot of grammars, where several features of different languages blend and resemble one another due to centuries-long extensive language contact. MindContact examines the cognitive aspects modulating how languages change due to areal contact with an interdisciplinary approach by connecting contact linguistics and psycholinguistics –two fields of linguistics that have barely interacted. Blending perspectives from psycholinguistics and contact linguistics, this project takes the concept of ‘areal contact’ phenomena in the Balkans under scrutiny, focusing on four understudied minority language varieties: Romani, Ladino, Istanbul Greek, and Rumelian Turkish. We will critically investigate on-going language change patterns including future/perfect forms, evidentiality, indefinite articles, adjective-noun order in those particular languages. MindContact aims to (1) advance knowledge regarding how areal convergence occurs and which cognitive factors contribute to it, (2) validate theories and lab-based protocols with experimental fieldwork data from minority languages, and (3) generate tools and resources for studying language change and loss under minority language conditions. The novel character of this proposal is the involvement of an ‘experimental fieldwork’ approach to language contact by using powerful psycholinguistic techniques such as eye-movement monitoring and electrophysiology. These time-sensitive tools provide a unique window into understanding the cognitive underpinnings of language contact outcomes. By synthesising several experimental data from its subprojects, MindContact seeks to arrive at a unified hypothesis of explaining how cognitive factors are involved in areal contact, setting the foundations of a new stream of research.

Most of our cognitive activities require processing and memorizing sequences of information. Whether sequences of motor gestures, sounds, letters, or words, we memorize them by associating the elements that make up these sequences. The field of implicit statistical learning aims at understanding the associative mechanisms that allow us to memorize these sequences of information. The HEBBIAN project aims to better understand these fundamental associative mechanisms by relying on recent theoretical models whose general framework is Hebbian learning. This project aims to experimentally study three main questions concerning 1) the role of the spacing between two repetitions of the same sequence in the memorization of this sequence ; 2) the dynamic of sequence encoding as a function of sequence size, number, and learning context ; 3) the problem of parts/whole relations between sequences of different sizes. This experimental work will be carried out in a comparative perspective with humans and non-human primates (Guinea baboons, Papio papio) using serial pointing tasks and classical psycholinguistic tasks, such as the naming or lexical decision tasks, in which a sequence is systematically repeated without the subjects being informed. This experimental work will be done in conjunction with the development and evaluation of two types of models based on the principles of Hebbian learning (psychological models and others that are more plausible on the neurobiological level). All of this work should allow for significant progress in our understanding and conception of these fundamental associative mechanisms.

Information compression is an essential aspect of memory and a fundamental building block of cognition. The goal of the present project is to study a crucial cognitive mechanism called chunking, which is a critical ingredient of short-term and long-term memory optimization. This project is twofold: the first part is related to implicit chunking, which shapes our perceptual systems by extracting the statistical regularities that are present in our environment, and the second part focuses on explicit chunking, which helps re-encoding information. It is generally accepted that individuals have a tendency to group information in order to make it easier to retain by recoding it in chunks. The process of chunking allows the simplification of a memory task by taking advantage of regularities in order to reduce the quantity of information to retain. Chunking using long-term memory is usually thought as an example of an unwanted artifact in memory experiments. Accordingly, most memory span tasks use controlled material to prohibit the processing of the to-be-stored items as chunks. The current project takes an opposite direction, by inviting participants in our experiments to chunk. We developed a series of tasks designed to assess immediate memory capacity, while measuring the compressibility of information likely to allow the formation of chunks. Half of the present project is based on the idea that explicit chunking abilities directly impact individual differences in storage capacity because chunking is an optimization process of storage capacity in immediate memory. Accordingly, we believe that this new paradigm can shake up how the storage and processing components in working memory are conceptualized, and we think it can also potentially offer a better account of memory development and intelligence. A second goal is to better identify the fundamental principles underpinning the creation of chunks in long-term memory. We think that data compression is a key idea to study the nonsupervised segmentation of complex material. Effectively, chunking is also considered as an implicit key learning mechanism and it has had considerable impact on the study of the structuration of our perceptual and linguistic systems. Since the seminal study by Saffran, Aslin and Newport (1996), several empirical reports have shown that humans (and nonhuman primates) have an innate ability to extract and code the regularities of their environment. Regularities correspond for instance to the repeated co-occurrence of two sources of information A and B. It has been proposed that the systematic co-occurrence of A and B leads to the implicit grouping of these pieces of information. Several theoretical explanations for these results have been proposed so far and chunking frequently appears as a key process in accounting for the empirical evidence. It has also been shown recently that the ability to extract regularities (by chunking) directly impacts individual differences in learning capacity. In these implicit learning situations where transitional probability learning occurs, chunks reflect the statistical regularities of the material to-be-learned and our hypothesis is that implicit chunking can also be studied as an optimization process based on data compression. The main goal of the present project is to gain a better understanding of these crucial perceptual and conceptual cognitive processes that operate as information compression devices, on the basis of real data collected in human adults, children and nonhuman primates. We also plan on showing that the explicit encoding of information in immediate memory can be detrimental to implicit learning, because attentional processes can be directed away. The central idea of this experimental and modeling project is that chunking is a fundamental cognitive process by which perceptual and conceptual information can be compressed to form rich permanent (or temporary) memory structures.

Sentences such as ‘Mary believes/says/hopes that John is an angel; Maria glaubt/sagt/hofft, dass Hans ein Engel ist; Marie croit/dit/espère que Jean est/soit un ange’ describe a mental state or act with propositional content. This propositional content is materialized by a clause which is traditionally referred to as ‘complement/argument/object clause’. The aim of the project is to develop a theory that essentially abandons the classical view that finite clauses are syntactically complements and semantically propositional arguments of their matrix predicates. A generalized theory will be developed that systematically relates apparent proposition-denoting ‘complement’ clauses with different types of subordinating elements (complementizers) to the other types of clausal embedding (relative clauses, adverbial clauses) and semantic objects (e.g. attitudinal objects, Moltmann 2003), taking into account various factors (morphology of complementizers, types of embedding predicates, possibility of correlation within the matrix). In fact, this unification of all cases of subordination/hypotaxis is desirable under the view that they all display recursion (a clause within a clause within a clause…) and in the frame of a parsimonious approach to human language. To this aim, the recent theory according to which clauses are adjuncts will be revised under the hypothesis that they modify an overt or covert item (the “anchor”) that is the actual complement to the predicate (which we dub the “tripartite hypothesis”). This innovative idea independently originated in the works of the participants in this project, who decided to join their forces to develop it further and provide an account that is fully satisfactory in terms of syntax/semantics interfacing. The project will concentrate on Germanic, Greek and Romance and test the central hypothesis against non-Indo-European languages like Basque and Hungarian. It furthermore takes seriously diachronic and dialectal variation as important information sources for the understanding of finite embedding as a central component of the grammar of human language.

Surveys on the quantitative representation of women in visual media are insufficient to grasp the issue of gender inequality, the visual modality being key to do so. In film studies, two relevant visual discursive regimes have been identified and recently revisited: the male gaze and the female gaze. Yet, how can we pinpoint such complex, subtle, but wide-spread visual discourse patterns, that may convey biased gender representation, and how to quantify the extent of their respective usage? With the advances in Artificial Intelligence (AI) and computational linguistics, we are now in place to conduct quantitative analysis to identify and extract recurrent visual and textual patterns from media content. Yet, such an analysis requires an iterative approach in concert with qualitative media studies, to recognize what is characteristic of a discourse style in visual media, and how the computational findings fit into the wider narrative and socio-historical context for which the content was produced. To enable such a bi-directional dialog, the AI models will have to provide sufficient explainability and allow expert-in-the-loop analysis and refinement. TRACTIVE’s objective is to characterize and quantify gender representation and women objectification in films and visual media, by designing an AI-driven multimodal (visual and textual) discourse analysis. TRACTIVE aims to establish a novel framework for the analysis of gender representation in visual media. We integrate AI, linguistics, and media studies in an iterative approach that both pinpoints the multimodal discourse patterns of gender in film, and quantitatively reveals their prevalence. We devise a new interpretative framework for media and gender studies incorporating modern AI capabilities. Our models, published through an online tool, will engage the general public through participative science to raise awareness towards gender-in-media issues from a multi-disciplinary perspective.