Projects

I am excited to share with you some of my current research projects, aimed at pushing the boundaries of knowledge and making significant contributions to the field. Through this webpage, I hope to foster collaboration, exchange ideas, and establish connections that can lead to fruitful partnerships. This page showcases my current projects, many of which are funded via competitive grants. The projects marked with a coffee cup icon are on a temporary coffee break. If you are a UoA student and would like to work on any projects listed here, do get in touch with me.

Authorship analysis

Authorship analysis is a fascinating field that encompasses various aspects, such as authorship attribution, profiling, and verification. In my research team, we are actively engaged in three core projects within this domain, namely:

Machine Learning Approach for Detecting AI-Generated Text

The advent of artificial intelligence has raised new challenges in determining the authenticity and origin of textual content. Our project focuses on developing a cutting-edge machine learning approach that can effectively detect AI-generated text. By leveraging advanced algorithms and computational techniques, we aim to enhance the identification and classification of AI-generated content, enabling us to better understand its impact on various applications.

eXplainable AI (XAI) Approach for Detecting AI-Generated Text

In order to build trust and transparency in AI systems, explainability plays a crucial role. Our team is dedicated to exploring XAI techniques specifically tailored for detecting AI-generated text. By providing interpretable insights and explanations for the detection process, we aim to empower users and stakeholders to make informed decisions regarding the authenticity and reliability of textual content.

Corpus Tool for Authorship Analysis

This project focuses on developing a besoke corpus tool for forensic linguists that facilitates the comparison and contrast of similarities between or among datasets. This tool aims to streamline the analysis of linguistic patterns, aiding in the investigation of authorship and uncovering valuable insights from textual data. Our goal is to provide forensic linguists with a robust and user-friendly resource that enhances their ability to analyze and interpret complex language-related phenomena.

Intelligent CALL: Pronunciation tools

CAPT tool: StudyIntonation

The working prototype of StudyIntonation Computer-Assisted Pronunciation Training (CAPT) system is a tool for mobile devices, which offers a set of tasks based on a “listen and repeat” approach and gives the audio-visual feedback in real time.

Reading aloud tool: Pronunciation Scaffolder

Non-native speakers of English frequently have difficulty reading aloud. This is particularly the case in high-stress environments, such as a research presentation. The Pronunciation Scaffolder is a web app, which is designed to help learners of English read scripts aloud more appropriately. This open-access tool visualizes pronunciation features using size, symbols and colour. The tool was written in Elm, which compiles to JavaScript.

Intelligent CALL: Writing tools

Intelligent Computer-assisted Language Learning (iCALL) was coined to describe computer-assisted language learning that harnesses the processing power of natural language processing pipelines. We are creating a suite of iCALL tools that help learners of English understand and use English.

Trend description generator

The Trend Description Generator is a natural language generation (NLG) tool which creates textual descriptions to accompany charts and graphs. The description is generated from the same structured dataset used to create the visuals. This open-access online tool enables learners of English to experiment and see how altering datapoints impacts the language features used to describe trends.

Genre-specific error detector with multimodal feedback

A purpose-built online error detection tool was developed to provide genre-specific corpus-based feedback on errors occurring in draft research articles and graduation theses. Feedback is divided in five categories, namely accuracy, brevity, clarity, objectivity and formality.

Grammatical error checker for Japanese learners of English

This bespoke tool was created on request to assist peripatetic teachers of English who provide indidualized tuition for hikikomori, that is students who are socially withdrawn and are unable to attend state or private schools. The tool identifies typical grammatical errors that learners make when completing homework assignments. The tool provides explanations in Japanese.

Feature visualizer

The feature visualizer shows and explains commonly-used language features present in a corpus of fully-annotated short research articles. Users can hide or reveal the features and their associated explanations on demand. Features that are incorporated include sectopms, moves, tense, voice, coherence and cohesion.

Feature detector

The feature detector enables users to check the readability, word profile and information structure of user-submitted texts. Multiple readability statistics are shared. Word profiles based on various vocabulary lists are available. Information structure, namely the principles of information focus, information flow and end weight, are identified and categorised.

Dynamic Language Assessment (DLA)

DLA environment for pragmatic failure in email requests

This bespoke dynamic language assessment is designed to raise students' awareness of pragmatic failure and the actions that can be taken to avoid such failure.

DLA environment for pragmatic failure in spoken requests

This tool is currently under construction. Expect an update here soon.

Mathematic Education

Interactive arithmetic

This project aims to create an interactive learning resource for young learners to master the four core operations of addition, subtraction, multiplication and division. The initial focus is on providing tools for two-digit multiplication using various methods, and creating accompanying visualizations.