Language between Redundancy and Deficiency

In the wake of remarkable advancements in artificial intelligence, particularly as large language models approach human-like language comprehension and production, we are prompted to re-examine the role of language in human cognition. The question looms larger than ever: if language was the defining trait of humanity, as Aristotle and countless others have suggested, how do we reconcile this with machines that seemingly share this capacity? As AI begins to mirror human language use, we need to explore the essence of language, its integration into human cognition, and its uniquely human attributes. One possible answer is that the way in which state of the art AI handles language is in fact quite similar to human language processing: both are stochastic, connectionist systems and, if anything, the difference between them is the strength of connection with other cognitive domains such as vision or the motor system. The collaborative research center Language between Redundancy and Deficiency presents a different perspective, motivated by the wealth of observations and results of formal language studies: language at its core is a symbolic and rule-based system unique to human cognition and akin to logic and mathematics. In particular, syntax, the core of grammar, is a pure symbolic system. Thus, the similarity between AI-language, as currently implemented, and human language is only superficial. Yet, language is also intertwined with a probabilistic cognitive system, i.e. a system governed more by patterns of experience and likelihood than hardwired rules of grammar. This gives the unique grammar system its adaptability to diverse cognitive, social, cultural, geographic, and political landscapes. The embedding of language in the overall cognitive system can be explicated by flexible rules that accommodate linguistic operations in situations that deviate from their precise definitions, characterized by either a scarcity of information (deficiency) or an excess (redundancy). Rooted in Austria`s distinguished tradition of formal linguistics, the collaborative research center brings together experts from the Universities of Graz, Vienna, and Salzburg to work in close collaboration to elucidate the way in which the linguistic system is adapted to its cognitive environment by focusing on the core phenomena that govern the context dependent usage of language. The research spotlights pronouns and ellipsis as pivotal in context-dependent language use. Pronouns allow for the reuse of established meanings without explicit syntactic encoding, while ellipsis enables the recycling of linguistic forms like sounds or signswithout their actual articulation. In studying these, the SFB employs state of the art methods of empirical linguistics from experiments to corpus study and cutting-edge mathematical formalisms. Through meticulous theoretical development around these phenomena, the collaborative research centre aspires to forge a theoretical framework that harmonizes symbolic and probabilistic systems, potentially revolutionizing cognitive science.