Incremental syntactic parsing and coreference resolution. As computers become smaller, keyboards and screens become increasingly impractical. We'd like to be able to talk to our computers, but they'd have to understand what we say. This project will develop a computational model that tracks which things are talked about and identifies 'who did what to whom' in text or speech.
Computational models of synergies in human language acquisition. How do children learn language? Do they first learn to recognise words and then associate words with meanings, or do they use the meanings to figure out what the words are, or do they do both at the same time, and if so, how? This project will investigate questions like these using advanced computational models of the way children learn from their environment.
Investigation of the component distributions of pause duration in spontaneous speech: Constraints for models of language production. We have discovered that the distribution of pause durations in spontaneous speech of individual speakers can be decomposed into at least two log-normal distributions. Our project will investigate this finding and provide a foundation for new research relevant to language production models. This will be achieved by determining the semantic, lexical, psycholinguistic ....Investigation of the component distributions of pause duration in spontaneous speech: Constraints for models of language production. We have discovered that the distribution of pause durations in spontaneous speech of individual speakers can be decomposed into at least two log-normal distributions. Our project will investigate this finding and provide a foundation for new research relevant to language production models. This will be achieved by determining the semantic, lexical, psycholinguistic, physiological, and acoustic concomitants of each component distribution and by investigating the impact of selected variables on the shape and location of each. The project has important implications for models of language production and applied problems involving automatic speech recognition, forensic speaker identification, and human communication disorders.Read moreRead less
Using eye movements to study how past experiences shape expectations. We intend to examine how the brain decides where to look next with our eyes, a decision made approximately three times every second. Understanding how the normal brain makes decisions will in turn help us to understand what happens when things go wrong in diseases like dementia and Parkinson's disease, with the hope of better - and earlier - diagnosis, and improved monitoring of treatment. In addition, our research will establ ....Using eye movements to study how past experiences shape expectations. We intend to examine how the brain decides where to look next with our eyes, a decision made approximately three times every second. Understanding how the normal brain makes decisions will in turn help us to understand what happens when things go wrong in diseases like dementia and Parkinson's disease, with the hope of better - and earlier - diagnosis, and improved monitoring of treatment. In addition, our research will establish an important research link with The University of Cambridge, and allow Australia to be competitive with laboratories in North America and Europe that are currently studying how the brain makes decisions about where to look.Read moreRead less
You came TO DIE?! Perceptual adaptation to regional accents as a new lens on the puzzle of spoken word recognition. Investigating Australian, New Zealand and UK listeners adaptation to each others accents will reveal how we achieve stable word recognition via flexible adjustment to pronunciation differences. Results will inform word recognition theory and illuminate why unfamiliar accents are difficult for language learners and automatic speech recognisers.
A computational and experimental investigation of reading aloud: Dyslexia, disyllables, and beyond. Australia is a world leader in computational cognitive science, particularly with respect to language processing. This project will help maintain and extend this position. Insights from the project will help us understand the processes that underlie both normal reading and reading disorders, particularly in areas that are comparatively neglected yet extremely important, such as how people read wor ....A computational and experimental investigation of reading aloud: Dyslexia, disyllables, and beyond. Australia is a world leader in computational cognitive science, particularly with respect to language processing. This project will help maintain and extend this position. Insights from the project will help us understand the processes that underlie both normal reading and reading disorders, particularly in areas that are comparatively neglected yet extremely important, such as how people read words of more than one syllable. Given that everyone in Australian needs to learn to read and that acquired and developmental disorders of reading are common, providing the theoretical base on which the processes involved in reading can be understood (and hence learnt and remediated most effectively) is of utmost importance.Read moreRead less
Learning to read: Representations and mechanisms underlying orthographic learning. In literate societies, the ability to read is one of the most important skills an individual ever learns. There are now a number of precisely specified computational models of reading that explain detailed aspects of skilled reading behaviour. However, these models do not incorporate changes in reading ability over time - in particular, they do not explain how the ability to read is acquired. This project aims ....Learning to read: Representations and mechanisms underlying orthographic learning. In literate societies, the ability to read is one of the most important skills an individual ever learns. There are now a number of precisely specified computational models of reading that explain detailed aspects of skilled reading behaviour. However, these models do not incorporate changes in reading ability over time - in particular, they do not explain how the ability to read is acquired. This project aims to understand the learning mechanisms that give rise to skilled reading. Developing detailed models of the process of learning to read will greatly facilitate the early identification and treatment of children with reading difficulties.Read moreRead less
Learning from others: Inductive reasoning based on human-generated data. Most of the data we see every day, from politics to gossip, comes from other people. Making inferences about such data is difficult because the people who provided it may have biases or limitations in their knowledge that we do not know about and must figure out. This project uses a series of experiments tied to normative computational models of social reasoning to explore how people solve this problem. This work has the po ....Learning from others: Inductive reasoning based on human-generated data. Most of the data we see every day, from politics to gossip, comes from other people. Making inferences about such data is difficult because the people who provided it may have biases or limitations in their knowledge that we do not know about and must figure out. This project uses a series of experiments tied to normative computational models of social reasoning to explore how people solve this problem. This work has the potential to make a major impact in understanding how information is understood and shared, especially when it is about topics that people lack firsthand knowledge about, like climate change. The computational models also have applications to the development of expert systems upon which our information economy relies.Read moreRead less
Beyond Segments: Towards a lexical model for tonal bilinguals. Most people in the world today speak more than one language. Thus, they need to decide, unconsciously, which language to use at any given time. This project aims to understand how healthy adult bilinguals resolve competition from their unintended language to communicate successfully in the intended language. In both bilingual language comprehension and production, the project will characterise the role of an under-explored linguistic ....Beyond Segments: Towards a lexical model for tonal bilinguals. Most people in the world today speak more than one language. Thus, they need to decide, unconsciously, which language to use at any given time. This project aims to understand how healthy adult bilinguals resolve competition from their unintended language to communicate successfully in the intended language. In both bilingual language comprehension and production, the project will characterise the role of an under-explored linguistic dimension, lexical tone, in cross-language processing. Expected outcomes include enhanced understanding of bilingual communication and theories of bilingual language use, and practical implications for optimal language learning for bilinguals and intervention for clinical populations who speaks two languages. Read moreRead less
Overcoming language barriers in healthcare: towards safe and effective communication when patients or clinicians use a second language. An international team of scholars will study language barriers in Australia's healthcare syste; arising from growing numbers of patients and clinicians not speaking English as their first language. The project's results will lay the foundation for future research and point to practical solutions.