Developing and testing measures of readability

Overview

Topic status: We're looking for students to study this topic.

Project Summary

This project aims at offering measures of sentence readability appropriate to specific linguistic backgrounds. For example, a sentence in French such as “les étudiants de l’Université sont contents” (the students of the University are contented) is reasonably easy to guess for an English native with a little knowledge of French, but the sentence “l’instituteur a fait un beau cours” (the teacher has done a good lecture) is much harder even though in average its constituent words are shorter. Indeed, classical readability measures rely on factors such as frequency of words in the language and length of words, whereas many more technical terms, that may be both infrequent and long, are shared by many languages, making them effectively easy to understand given the correct language background. Conversely, some words that sound and look very similar may convey radically different meanings, possibly leading to confusion.

The proposed measure will be tested against readability judgements collected in the first part of the project, based on a web-based survey to be developed as part of the project.

Technically, most of the coding will involve reading external resource files and using APIs or external tools.

Expected outcomes, applications and/or benefits

This project will deliver a methodology as well as an effective measure of readability of a sentence in a second language based on users level and background language. It will be part of a tool that delivers bi-lingual subtitles.
Bi-lingual subtitles are expected to generate a never-ending source of ubiquitous training for people learning second languages, as they can be used when watching every movie in the leaned language at home.

Required student skills/experience

Reasonable programming skills, either in Java or a scripting language (Python, Perl, etc. ). Some knowledge of natural language processing (INx344) and of a second language would be useful but not required.

Study level
Vacation research experience scholarship
Supervisors
QUT
Organisational unit

Science and Engineering Faculty

Research area

Computer Science

Keywords
readability
Contact
Contact the supervisor for more information