This repository contains a computational tool for automatic morphosyntactic analysis of Light Warlpiri text data, developed as part of my Bachelor's thesis project at the Australian National University in 2020. The analyser applies rules-based word analysis and syntactic disambiguation to annotate each word in a corpus with its morphosyntactic tag (gloss).
The tool leverages the Computerised Language Analysis (CLAN) software for processing and was adapted specifically to the morphosyntactic properties of Light Warlpiri.
Manual morphosyntactic glossing is highly time-consuming and difficult to scale. This project demonstrates how computational methods can automate morphosyntactic analysis for a corpus of Light Warlpiri, including:
- Adapting rules-based analysis to the unique morphosyntactic patterns of Light Warlpiri.
- Applying syntactic disambiguation to improve tagging accuracy.
- Identifying recurring challenges and variables affecting performance.
For a detailed description of the method, adaptation process, and performance evaluation, see my thesis
- Automatic morphosyntactic tagging of Light Warlpiri corpus data.
- Rules-based word analysis using CLAN.
- Syntactic disambiguation to improve gloss accuracy.
- Configurable workflow for adaptation to different corpus sizes.
- CLAN requires specific input formats to run the analysis (e.g., files ending in .cha extension). For further details, refer to its manual, found at TalkBank