Collaborative Research: Advancing Open-Access Resources for Languages: Investigating evidentiality and focus features

This project investigates evidential markers, the grammatical coding of information source and speaker’s perspectives. Evidential markers indicate whether a statement is based on direct observation, inference, hearsay, or other types of evidence. Examining how different languages express source of information can provide insights into how speakers evaluate information, revealing deeper cognitive processes. The project's goal is to create treebanks/universal dependencies corpora to develop an automated deep-learning natural language processing (NLP) model. Considering the distributional characteristics around evidential markers, the model will be able to extract information from texts semi-automatically. The project explores a key linguistic question: why evidential markers in some languages behave differently from those in other languages, as they seem to have more than one function and allow more flexibility in how they appear in sentences. This flexibility may affect how language users show evidence or emphasis in longer conversations. The work will not only expand understanding of how these languages work but also support language technology development. Because the project will use a deep learning architecture to develop this NLP model, there are several other potential applications beyond the proposed linguistics goals: the development of machine translation tools; text summarization; and other types of human-machine interactions (e.g. chatbots) and AI systems. This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.