Logic in Semantic Universals (Renewal)

Language is one of humanity’s most remarkable inventions, yet across the thousands of languages spoken/signed worldwide, we see striking similarities in how meaning is structured. These recurring patterns—semantic universals—suggest deep constraints on how languages evolve. But what explains these universals? Do they stem from the way humans communicate, or are they rooted in innate structures specific to our linguistic ability? This project investigates whether the communicative function of language plays a key role in shaping semantic universals. It will explore how fundamental principles governing the evolution of communicative systems—such as the need for clarity and learnability—constrain the meanings that languages can encode. Understanding these principles will not only shed a light on why languages take the forms they do but also provide insights into human cognition and the development of artificial communication systems. Despite the broad recognition of semantic universals, their origins remain debated. Some researchers argue that these patterns arise from innate cognitive biases, while others suggest that they are products of cultural evolution of communicative systems. A major challenge is distinguishing between these explanations and identifying whether semantic universals emerge due to the demands of communication. This project comes this challenge by taking an innovative, interdisciplinary approach, combining experimental, computational, and theoretical methods. The goal is to address the question: to what extent do communicative pressures shape semantic universals, and ultimately human linguistic cognition? To do this, we will focus on three key areas: Compositionality and Lexicalisation. We will investigate how languages balance between simple building blocks (words) and more complex expressions (phrases built from words). As an approach, we will use artificial language learning experiments and computational models to explore why certain meanings tend to be lexicalised (expressed as a single word) while others are left for compositional constructions. 2. Logicality and Grammar. We will examine why languages often avoid trivial or contradictory meanings in their grammatical structures. To do this, we will conduct iterated learning experiments and computational modelling to test whether these patterns arise from minimum assumptions about how pragmatic speakers communicate and learn the grammar from others. 3. Semantic Universals in AI. We will test whether large language models (LLMs)—trained on vast amounts of human language data—exhibit semantic biases similar to those found in human languages. We will investigate whether these models can help us understand the cognitive and communicative foundations of linguistic universals. This research has broad implications beyond language sciences. Understanding the forces that shape language can inform fields such as: Artificial Intelligence & NLP: Insights from this study could improve how AI systems process and generate natural language, leading to more effective human-computer interaction. Language Learning & Education: Findings could contribute to better language teaching strategies by revealing what structures are more easily learnable and why. Cognitive Science: The project will shed light on fundamental aspects of human cognition, such as how we categorize and communicate meaning. Preserving Linguistic Diversity: A deeper understanding of universal patterns can aid efforts to document and revitalize endangered languages. By integrating linguistic theory with experimental and computational methods, this project aims to uncover the deep principles that shape all human languages. Its findings will enhance our understanding of both the nature of language and the cognitive and social factors that drive its evolution.