This research work falls under TRAILs project (supervised by Roberto Zamparelli) and aims to explore the uses of technology to answer fundamental questions in the study of human language. One aspect that has been the current focus of the project is whether human beings acquire their mother tongue by using an innate, language-specific learning predisposition (the so-called Universal Grammar hypothesis), or rather by reusing learning skills that are common to multiple cognitive domains.
To explore this theory, the ability of various types of neural networks to acquire complex metalinguistic skills, in particular, the ability to judge the syntactic well-formedness of rare constructions, a peripheral but important human ability that has rarely been investigated in the natural language processing community, is being tested in the project.
And since artificial neural networks are not built with any language-related pre-set feature and learn from data alone, they are the best model of a potential human learner without any language-specific biological bias. Thus their ability to match human metalinguistic intuitions would, therefore, be a remarkable step in investigating whether innate linguistic biases are a necessary component.
Turn-taking is one of the key aspects of conversational dynamics in daily conversations and is an integral part of human-human, and human-machine interaction systems. It is used for discourse organization of a conversation by means of explicit phrasing, intonation, and pausing, and it involves intricate timing. In daily conversation, each designed turn has an underlining purpose, such as signaling the speaker’s intention, supporting and extending the ongoing conversation among others and providing us a window to understanding some aspects of human behavior. The aim of this research work is to uncover the dynamics of turn-switching and the discourse of competitiveness in interaction overlaps uncover the dynamics of turn-switching and the discourse of competitiveness in interaction overlaps along with understanding the influence and inner-meaning of silences in a conversation and how the whole turn-taking dynamics leads to the success of the conversation and more importantly how can we design more interactive Human-Machine model using the research findings.