ERC FUNDED PROJECTS

"Children learn any language that they grow up with, adapting to any of the ca. 7000 languages of the world, no matter how divergent or complex their structures are. What cognitive processes make this extreme flexibility possible? This is one of the most burning questions in cognitive science and the ACQDIV project aims at answering it by testing and refining the following leading hypothesis: Language acquisition is flexible and adaptive to any kind of language because it relies on a small set of universal cognitive processes that variably target different structures at different times during acquisition in every language. The project aims at establishing the precise set of processes and at determining the conditions of variation across maximally diverse languages. This project focuses on three processes: (i) distributional learning, (ii) generalization-based learning and (iii) interaction-based learning. To investigate these processes I will work with a sample of five clusters of languages including longitudinal data of two languages each. The clusters were determined by a clustering algorithm seeking the structurally most divergent languages in a typological database. The languages are: Cluster 1: Slavey and Cree, Cluster 2: Indonesian and Yucatec, Cluster 3: Inuktitut and Chintang, Cluster 4: Sesotho and Russian, Cluster 5: Japanese and Turkish. For all languages, corpora are available, except for Slavey where fieldwork is planned. The leading hypothesis will be tested against the acquisition of aspect and negation in each language of the sample and also against the two structures in each language that are most salient and challenging in them (e. g. complex morphology in Chintang). The acquisition processes also depend on statistical patterns in the input children receive. I will examine these patterns across the sample with respect to repetitiveness effects, applying data-mining methods and systematically comparing child-directed and child-surrounding speech."

1 998 438 €

Start date: 2014-09-01, End date: 2019-08-31

Aftermath seeks to understand the legacy of the East Asian War of 1592-1598. This conflict involved over 500,000 combatants from Japan, China, and Korea; up to 100,000 Korean civilians were abducted to Japan. The war caused momentous demographic upheaval and widespread destruction, but also had long-lasting cultural impact as a result of the removal to Japan of Korean technology and skilled labourers. The conflict and its aftermath bear striking parallels to events in East Asia during World War 2, and memories of the 16th century war remain deeply resonant in the region. However, the war and its immediate aftermath are also significant because they occurred at the juncture of periods often characterized as “medieval” and “early modern” in the East Asian case. What were the implications for the social, economic, and cultural contours of early modern East Asia? What can this conflict tell us about war “aftermath” across historical periods and about such periodization itself? There is little Western scholarship on the war and few studies in any language cross linguistic, disciplinary, and national boundaries to achieve a regional perspective that reflects the interconnected history of East Asia. Aftermath will radically alter our understanding of the region’s history by providing the first analysis of the state of East Asia as a result of the war. The focus will be on the period up to the middle of the 17th century, but not precluding ongoing effects. The team, with expertise covering Japan, Korea, and China, will investigate three themes: the movement of people and demographic change, the impact on the natural environment, and technological diffusion. The project will be the first large scale investigation to use Japanese, Korean, and Chinese sources to understand the war’s aftermath. It will broaden understandings of the early modern world, and push the boundaries of war legacy studies by exploring the meanings of “aftermath” in the early modern East Asian context.

1 444 980 €

Start date: 2018-11-01, End date: 2023-10-31

"When I asked my seven-year-old daughter ""Who is the boy in your class who was also new in school last year, like you?"", she instantly replied ""Daniel"", using the descriptive content in my utterance to identify an entity in the real world and refer to it. The ability to use language to refer to reality is crucial for humans, and yet it is very difficult to model. AMORE breaks new ground in Computational Linguistics, Linguistics, and Artificial Intelligence by developing a model of linguistic reference to entities implemented as a computational system that can learn its own representations from data. This interdisciplinary project builds on two complementary semantic traditions: 1) Formal semantics, a symbolic approach that can delimit and track linguistic referents, but does not adequately match them with the descriptive content of linguistic expressions; 2) Distributional semantics, which can handle descriptive content but does not associate it to individuated referents. AMORE synthesizes the two approaches into a unified, scalable model of reference that operates with individuated referents and links them to referential expressions characterized by rich descriptive content. The model is a distributed (neural network) version of a formal semantic framework that is furthermore able to integrate perceptual (visual) and linguistic information about entities. We test it extensively in referential tasks that require matching noun phrases (“the Medicine student”, “the white cat”) with entity representations extracted from text and images. AMORE advances our scientific understanding of language and its computational modeling, and contributes to the far-reaching debate between symbolic and distributed approaches to cognition with an integrative proposal. I am in a privileged position to carry out this integration, since I have contributed top research in both distributional and formal semantics. "

1 499 805 €

Start date: 2017-02-01, End date: 2022-01-31

Based on the APMPAL (ERC) project we continue to develop the frameworks of internal rationality (IR) and optimal signal extraction (OSE). Under IR investors/consumers behave rationally given their subjective beliefs about prices, these beliefs are compatible with data. Under OSE the government has partial information, it knows how policy influences observed variables and signal extraction. We develop further the foundations of IR and OSE with an emphasis on heterogeneous agents. We study sovereign bond crisis and heterogeneity of beliefs in asset pricing models under IR, using survey data on expectations. Under IR the assets’ stochastic discount factor depends on the agents’ decision function and beliefs; this modifies some key asset pricing results. We extend OSE to models with state variables, forward-looking constraints and heterogeneity. Under IR agents’ prior beliefs determine the effects of a policy reform. If the government does not observe prior beliefs it has partial information, thus OSE should be used to analyse policy reforms under IR. If IR heterogeneous workers forecast their productivity either from their own wage or their neighbours’ in a network, low current wages discourage search and human capital accumulation, leading to low productivity. This can explain low development of a country or social exclusion of a group. Worker subsidies redistribute wealth and can increase productivity if they “teach” agents to exit a low-wage state. We build DSGE models under IR for prediction and policy analysis. We develop time-series tools for predicting macro and asset market variables, using information available to the analyst, and we introduce non-linearities and survey expectations using insights from models under IR. We study how IR and OSE change the view on macro policy issues such as tax smoothing, debt management, Taylor rule, level of inflation, fiscal/monetary policy coordination, factor taxation or redistribution.

1 524 144 €

Start date: 2018-09-01, End date: 2023-08-31