KATR: Software for morphological studies in computational linguistics

• Student: Lei Shen 9/1999, now at Bell South Cellular
• Purpose: Compute queries about language morphology
• Method: Morphology of a language (for example, its declensional scheme) is encoded in default-inheritance hierarchies in the DATR language. This project extends DATR to include set-based rules, which are of value in Bulgarian and Swahili. The software is a freshly-written Java program that parses and compiles the KATR program and then accepts queries on it.
• What the student learned
  1. Writing a compiler in Java. (This part took considerable cleverness.)
  2. Building a detailed specification from a vague starting requirement.
  3. Adding new features to an existing software package while considering tradeoffs.
  4. Technical writing skills in preparing the write up and packaging the resulting tool (in gzip form and in zip form).